The conservation and use of crop genetic resources
for food security

Colin K. Khoury

Thesis committee
Promotor
Prof. Dr P.C. Struik
Professor of Crop Physiology
Wageningen University
Co-promotor
Dr A. Jarvis
Research Area Director, Decision and Policy Analysis,
International Center for Tropical Agriculture (CIAT)
Flagship Leader for Climate Smart Agriculture,
CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS)
Other members
Prof. Dr M.E. Schranz, Wageningen University
Prof. Dr A. Graner, Leibnitz Institute of Plant Genetics and Crop Plant Research (IPK), OT
Gatersleben, Germany
Dr L. Visser, CGN, Wageningen University and Research Centre
Prof. Dr P.H. Feindt, Wageningen University
This research was conducted under the auspices of the C.T. de Wit Graduate School of
Production Ecology and Resource Conservation.

The conservation and use of crop genetic resources
for food security

Colin K. Khoury

Thesis
submitted in fulfillment of the requirements for the degree of doctor
at Wageningen University
by the authority of the Rector Magnificus
Prof. Dr A.P.J. Mol.
in the presence of the
Thesis Committee appointed by the Academic Board
to be defended in public
on Monday 31 August 2015
at 11 a.m. in the Aula.

Colin K. Khoury
The conservation and use of crop genetic resources for food security
305 pages.
PhD thesis, Wageningen University, Wageningen, NL (2015)
With references, with summary in English
ISBN: 978-94-6257-442-7

Abstract
Khoury, C.K. (2015). The conservation and use of crop genetic resources for food security.
PhD thesis, Wageningen University, Wageningen, The Netherlands. 305 pp.
Crop genetic diversity is a critical resource to address the nutrition and agronomic challenges facing
global food security. Increases in use of this diversity are expected but are dependent upon conservation,
availability, and access. Considerable erosion of crop genetic diversity has occurred in situ, i.e., in
farmers’ fields and natural habitats, and the variation conserved in ex situ genebanks is also vulnerable
due to insufficient resources. The window of opportunity to resolve these deficiencies and thus
accomplish a comprehensive global system for crop genetic diversity conservation and availability for
use will not remain open indefinitely.
Among the factors hindering the conservation of crop genetic resources is a lack of essential information
regarding this diversity. Questions include: (a) what is the status of diversity in our food systems, and
where are the greatest vulnerabilities?, (b) where can genetic diversity be found that can be useful in
increasing productivity and mitigating these vulnerabilities?, (c) is this genetic diversity available in the
present and in the long term?, and (d) what steps are needed to improve the ability for researchers to
access genetic resources critical for present and future crop improvement?
This thesis aims to contribute to the knowledge required to answer these questions through an
exploration of the need for, potential of, challenges and constraints regarding, and necessary steps to
enhance the conservation and use of crop genetic diversity. The research starts with an investigation of
the state of diversity in global food supplies, finding that national food supplies around the world have
become increasingly similar over the past 50 years, gaining in calories, protein, and fat, as animalderived foods and high-calorie plant foods have risen in importance. The proportion of diets consisting
of major cereals, sugar crops and oil crops has increased, while regionally and locally important cereals,
root crops, and oil crops have generally become further marginalized.
The thesis then delves into the potential for utilization of a particular set of genetic resources of
increasing interest globally – crop wild relatives. These wild cousins of cultivated species have been
used for many decades for crop improvement. Their utilization is likely only to increase due to
improvements in information on species and their diversity, advancements in breeding tools, and the
growing need for exotic genetic diversity in order to address compounding agronomic challenges. As
wild plants they are subject to a myriad of human caused threats to natural ecosystems. A focus on wild
genetic resources is thus timely both for biodiversity conservation and food security objectives.

Research on crop wild relatives first concentrated on the identification of potentially important wild
genetic resources at the national level in the United States. The resulting National Inventory listed close
to 5000 taxa. A prioritization of species based on value to food security emphasized close to 300 native
taxa that are most closely related to important food crops.
Once potentially valuable genetic resources are identified, subsequent information is needed regarding
where they occur, what diversity they may possess, and how well conserved and therefore available to
crop breeders they are. A ‘gap analysis’ methodology was proposed to answer these questions,
capitalizing on developments in digital occurrence and eco-geographic data as well as species
distribution modeling, intentionally utilizing freely available software and data, and incorporating a
novel expert evaluation tool.
The thesis then advances the gap analysis methodology, taking advantage of improvements in species
targeting, occurrence data, modeling, and expert feedback methods, and further utilizing eco-geographic
information to identify traits of value to crop improvement objectives. In case studies on the wild
relatives of bean (Phaseolus L.), sweetpotato [Ipomoea batatas (L.) Lam., I. series Batatas], and
pigeonpea [Cajanus cajan (L.) Millsp.], related species were found to be highly under-represented in ex
situ conservation systems and thus inadequately available to breeders and researchers. Species differed
among themselves and in comparison to the associated crop in their adaptations to temperature,
precipitation, and edaphic characteristics, and many species also showed considerable intraspecific
variation. Taxa and specific geographic locations were prioritized for further collecting in order to
improve the completeness of germplasm collections for these important crops.
While conservation of crop genetic diversity is fundamental to the availability of this diversity for
breeding, it is national and international policies that determine the real capacity for researchers to
acquire these resources. The thesis culminates in an exploration of the degree to which international
collaboration is required in order to achieve access to genetic resources where they are needed. Countries
were found to be highly dependent on crops whose genetic diversity largely sources from outside their
borders. This reliance is evident even in countries located in regions of high indigenous crop diversity
and has increased significantly over the past half century, bolstering evidence for the need for effective
national and international policies to promote genetic resource conservation and exchange.
Keywords: Agricultural development, Climate change adaptation, Crop diversity, Crop genetic
resources, Crop improvement, Crop wild relatives, Ex situ conservation, Food security, Gap analysis,
Germplasm conservation, Germplasm exploration, Interdependence, National inventory, Plant genetic
resources.

Preface
This thesis is dedicated to Luigi Guarino, Jane Toll, Andy Jarvis, and Hannes Dempewolf, for
their guidance, inspiration, and remarkable support.
My profound thanks to Cary Fowler and Gary P. Nabhan, for catalyzing this path long ago,
embracing my efforts along the way, and for continuing to inspire me with their visions.
My gratitude to my family for their patience and support.
My thanks to the Crop Wild Relatives team (Harold A. Achicanoy, Vivian Bernau, Nora P.
Castañeda-Álvarez, Julian Ramírez-Villegas, Chrystian C. Sosa, Steven Sotelo, Hector Tobon,
Alex Castañeda, Shirley Calderón, Ingrid Vanegas, and Ovidio Rivera) for your hard work, and
my particular appreciation to Nora, for being my academic and professional sister over these
years.
My gratitude to my advisors Paul C. Struik, Andy Jarvis, John H. Wiersema, and Stephanie
Greene, for your guidance and support.
Global efforts in plant genetic resources are successful only through diverse collaborations. I
thank the taxonomists, data managers, conservationists, curators, ecologists, plant breeders, and
other genetic resource professionals with whom I have had the honor to work and/or benefit
from their efforts in gathering and understanding pertinent data. Particular thanks to Harold A.
Achicanoy, Vivian Bernau, Anne D. Bjorkman, Nora P. Castañeda-Álvarez, Marleen Cobben,
Daniel G. Debouck, Stef de Haan, Hannes Dempewolf, Ruth J. Eastwood, Johannes M. M.
Engels, Lauren A. Eserman, Cary Fowler, Ximena Flores-Palacios, Stephanie Greene, Luigi
Guarino, Bettina Heider, Robert L. Jarret, Andy Jarvis, Henry Juarez, Michael Minion Kantar,
Mulualem T. Kassa, Sandy Knapp, Nigel Maxted, Richard E. Miller, Jonas V. Müller, Carlos
Navarro-Racines, Sally L. Norton, Julian Ramírez-Villegas, Loren H. Rieseberg, Genoveva
Rossel, Tiina Särkinen, Robert W. Scotland, Chrystian C. Sosa, Steven Sotelo, David Spooner,
Paul C. Struik, Mindy Syfert, Álvaro Toledo, Hari D. Upadhyaya, L. Jos G. van der Maesen,
John H. Wiersema, John R.I. Wood, and G. Craig Yencho for the intensive collaborations.

Thanks also to the staff of the Statistics Division of FAO for providing literature and clarifying
historical modifications to commodity and country listings, and Johannes M. M. Engels, Mark
Vellend, Sarah Elmendorf, Jay Bost, Gea Galuzzi, Gary P. Nabhan, Tim Crews, Alder
Keleman, Neil Palmer, and Carlos Navarro-Racines for inputs (Chapter 2); the USDA National
Plant Germplasm System (NPGS) curators, members of the NPGS Crop Germplasm
Committees, and Agricultural Research Service crop experts for their inputs on the Inventory,
and Gary Kinard, Edward Garvey, Karen Williams, and Larry Stritch for helpful comments on
the manuscript (Chapter 3); Nora P. Castañeda-Álvarez for preliminary inputs, and Silvia
Castano, Ovidio Rivera and Carlos Nagles for georeferencing (Chapter 4); Shirley Calderón,
Alex Castañeda, Ovidio Rivera, and Ingrid Vanegas for gathering and organizing occurrence
data (Chapters 5 and 6); Flora Mer for data gathering, and Steven Prager for comments (Chapter
6); and Gary Kinard, Álvaro Toledo, and Jan Borring for comments (Chapter 7).
Thanks to Charis Khoury for the cover illustrations, one of which was inspired by a photo taken
by Kasia Stepien.

Contents

Chapter 1. General introduction

1

Chapter 2. Increasing homogeneity in global food supplies and the
implications for food security

17

Chapter 3. An inventory of crop wild relatives of the United States

51

Chapter 4. A gap analysis methodology for collecting crop genepools:
a case study with Phaseolus beans

87

Chapter 5. Distributions, ex situ conservation priorities, and
genetic resource potential of crop wild relatives of
sweetpotato [Ipomoea batatas (L.) Lam., I. series Batatas]

129

Chapter 6. Crop wild relatives of pigeonpea [Cajanus cajan (L.) Millsp.]:
distributions, ex situ conservation status, and
potential genetic resources for abiotic stress tolerance

187

Chapter 7. Interdependence among countries on crop genetic diversity

237

Chapter 8. General discussion

275

Summary

291

Publications of author

297

PE&RC Training and Education Statement

301

Curriculum vitae

303

Funding

305

CHAPTER 1
General introduction
BACKGROUND
Human beings have been remarkably successful over the course of their history in expanding
both in total population as well as distributional range across the planet. Perceived limits to
population growth (e.g., Malthus 1803, Vogt 1948, Ehrlich 1975) have been overcome
historically through innovation in the use of natural resources and their derivatives, e.g., via the
transition from gathering and hunting to agricultural societies, improvements in hygiene and
waste management in increasingly densely populated areas, antibiotics and other advancements
in medicine, and the agricultural green revolution (Diamond 2004, Gepts 2006).
Global food production has thus far kept pace with demand (Gepts 2006), and both the number
of stunted as well as underweight persons worldwide is declining (Kearney 2010, International
Food Policy Research Institute 2014, WHO 2015), but a number of persistent and novel
challenges bring renewed concern to the question of our ability to feed ourselves into the future.
First, from the perspective of human nutrition, although adequate food energy (i.e., calories) is
available as a global sum, disparities in access and challenges in distribution continue to create
hunger (Hawkesworth et al. 2010, Kearney 2010, Remans et al. 2011, WHO 2015). Second,
food system and demographic change, and the historical emphasis on macronutrients (i.e.,
carbohydrates, protein, and fat) in agricultural research and in development policies have had
mixed effects on providing for essential micronutrients. Although improvements have occurred
in some geographic areas, deficits in vitamin A, iodine, folate, iron, and/or zinc, among others,
persist in diets in more than half of all countries, affecting more than two billion persons
globally (Remans et al. 2011, CDC 2015, WHO 2015). Finally, economic development,
urbanization, and many facets of globalization have increased consumer purchasing power and
the availability particularly of energy dense foods (Popkin 2006, Pingali 2007, Hawkesworth
et al. 2010, Kearney 2010). In concert with lifestyle change, this ‘nutrition transition’ toward
diets high in animal products, sugars, and fat has globalized the epidemic of overweight and
obesity and its associated non-communicable illnesses, including heart disease, Type 2
diabetes, and cancer (Popkin 2006, Pingali 2007, Kearney 2010, Tilman & Clark 2014). While

Chapter 1
each of these branches of malnutrition individually occur more severely in particular areas
worldwide, the so-called ‘triple-burden’ of malnutrition (i.e., the coexistence of food insecurity,
undernutrition, and overweight) can now appear within the same regions, communities, and
even households (Gomez et al. 2013).
Agricultural production challenges are equally concerning. Food demand may as much as
double in the coming few decades, due both to the rising global human population and to
changing dietary expectations (Hoisington et al. 1999, Tilman et al. 2011, Kastner et al. 2012).
Yet projected future yield trends are no longer expected to keep pace with this increase in
demand (Ray et al. 2013). Real limitations in the further expansion of arable lands and the use
of water, phosphorus, and other natural resources are also being reached (Cordell et al. 2009,
Foley et al. 2011), and pollution levels due to the excessive use of nitrogen fertilizers and other
inputs are considered to be well over those which may be considered safe (Rockstrom et al.
2009, Bodirsky et al. 2014). Arable lands and other natural resources dedicated to food
production are increasingly subject to competition from biofuel and other non-food crops (Ray
et al. 2012). There is also an increasing urgency to balance the use of lands and inputs for
agriculture with the maintenance of the integrity of wild habitat, biodiversity, and other
ecosystem services (Foley et al. 2005, Matson & Vitousek 2006, Phalan et al. 2011).
Compounding these challenges, global climatic change is generating greater uncertainty in
reliable agricultural production in an increasing number of regions due both to greater extremes
as well as to increased variation in temperatures and precipitation, and is additionally altering
pest and disease occurrences (Lobell et al. 2008, Burke et al. 2009, Jarvis et al. 2009, Wheeler
& von Braun 2013).
PROBLEM ANALYSIS
Realizing long-term global food and nutrition security will require achieving a level of balance
in demand from an informed and empowered populace, and sustainably producing sufficient
accessible food and nutrition in order to nourish this population (Ruel 2003, Alston et al. 2009,
Tilman & Clark 2015, Graham et al. 2007, Foley et al. 2011, Remans et al. 2011, West et al.
2014). While the pathway to this end is not completely evident, balancing food demand is likely
to require a wide range of actions including sustainable economic development and security,
education, reductions in inequality in human societies, and availability and access to family
planning services and technologies. In regard to food supply, it is expected to require
2

General introduction
maximizing productivity within defined arable lands with the use of renewable inputs, i.e.,
sustainable intensification, as well as reducing food waste (Godfray et al. 2010, Foley et al.
2011, Tilman et al. 2011, Struik & Kuyper 2014).
Producing more food on the same amount of land may be achieved through two main pathways
– making agronomic practices more effective to bring yields closer to potential, and increasing
potential yield through crop genetic improvement (Tilman et al. 2011). Innovations in
agronomy and in crop breeding have provided solutions to historical production constraints, but
have also contributed to the present challenges. Through the green revolution, traditional
practices of maintaining relatively low but reliable yields via the use of spatial crop genetic
diversity (i.e., the employ of diverse crop landrace populations, and numerous crops grown in
the same areas) have been replaced with technologies employing higher yielding, genetically
uniform crop varieties typically grown over large areas and supported by considerable use of
agricultural inputs (Gepts 2006, van de Wouw et al. 2009, 2010). Due to this spatial uniformity,
such varieties are typically substituted every few years in order to avoid crop failure resulting
from genetic vulnerability to pests and/or diseases (i.e., temporal diversity has mitigated the
loss of spatial diversity). Given the current and projected food demand, the global food system
is dependent upon the industry of modern crop varieties and agronomic practices to maintain
high yields (Hoisington et al. 1999).
The question therefore is how production and associated food supply practices in the present
and future can increase food availability while better conserving natural resources, maintaining
critical ecosystem services, and being robust against climatic change and other shocks. The use
of genetic diversity in crop improvement, with the aim of improving the ecological efficiency
of crop production through traits such as drought and heat tolerance, input use efficiency, and
extended storage life, is considered to be a key answer to this question (Xiao et al. 1996,
Hoisington et al. 1999, Zhu et al. 2000, Gepts 2006, Guarino & Lobell 2011, McCouch et al.
2013).
Crop genetic diversity is generated through genetic mutation and recombination, and further
transformed through natural and artificial selection, and is therefore the product of the evolution
over time of crops and closely related wild plants in their agricultural and natural habitats. As
the source of agronomic traits employed for adaptation to biotic and abiotic stresses and for
3

Chapter 1
yield increases through breeding, and the palette from which food systems may be further
diversified, this genetic diversity is a biological cornerstone of food security.
Expectations of increases in utilization of crop genetic diversity are made under the assumption
that adequate variation will be available for exploration. Ironically, considerable erosion of crop
genetic diversity has occurred through the adoption of improved crop varieties or substitute
crop species and subsequent neglect of traditional varieties and crops, economic development
and associated shifts in consumer demand, land use change and habitat destruction, and
urbanization and the displacement of cultures associated with particular crops and varieties,
among other factors (Harlan 1975, Hoisington et al. 1999, Gepts 2006, Wilkes 2007, van de
Wouw et al. 2009, 2010). In some crops, only a fraction of the diversity once present is thought
to still be found today in farmers’ fields, e.g., in wheat varieties in some areas of the Fertile
Crescent (Harlan 1971, Gepts 2006).
Speaking on the importance of crop genetic resources and expressing concerns regarding their
state of conservation four decades ago, Jack Harlan, a preeminent American agronomist and
professor of plant genetics, stated,
“These resources stand between us and catastrophic starvation on a scale we cannot
imagine. In a very real sense, the future of the human race rides on these materials.
The line between abundance and disaster is becoming thinner and thinner and the
public is unaware and unconcerned. Must we wait for disaster to be real before we
are heard? Will people listen only after it is too late?” (Harlan 1975).
Due to the disappearance of crop genetic diversity in farmer’s fields and wild habitats, the
world’s genebanks originally established to make plant genetic resources readily available to
breeders for crop improvement, have become essential repositories for crop diversity
conservation (Hoisington et al. 1999, Gepts 2006, FAO 2010, Thormann et al. 2015). A
substantial portion of the world’s remaining heritage of food crop genetic resources is likely
now conserved exclusively in genebanks, and important steps have been taken to begin to
ensure reliable long-term funding for the conservation, documentation, and distribution of these
resources (Global Crop Diversity Trust 2013), and to facilitate access to this diversity
worldwide (FAO 2002).
4

General introduction

Unfortunately, though, many genebank collections holding unique genetic diversity fail to
comply with international standards due to degrading infrastructure, inconsistent power
supplies, far from ideal regeneration cycles and protocols, poor safety duplication,
unsophisticated documentation systems, and/or inadequate staff and resources to fulfill their
conservation and distribution mandates (FAO 2010, Khoury et al. 2010). Two points regarding
the current state of the world’s heritage of crop genetic diversity are thus clear - much remains
to be done to secure the diversity threatened in situ and in under-funded genebanks, and the
window of opportunity to accomplish a comprehensive global system for genetic resources
conservation will not remain open indefinitely (Wilkes 2007, FAO 2010).
RESEARCH QUESTIONS AND OBJECTIVES
Among the factors hindering the actualization of a global system for crop genetic resources
conservation and availability is a lack of essential information regarding this diversity. A
number of primary questions have not been fully answered, including:
(a) what is the status of diversity in our food systems, and where are the greatest vulnerabilities?,
(b) where can genetic diversity be found that can be useful in increasing productivity and
mitigating these vulnerabilities?,
(c) is this genetic diversity available in the present and in the long term?, and
(d) if not available, what steps are needed to improve the ability for researchers to access genetic
resources critical for present and future crop improvement?
This thesis aims to contribute to the knowledge required to answer these questions through an
exploration of the need for, potential of, challenges and constraints regarding, and necessary
steps to enhance the conservation and use of crop genetic diversity.
RESEARCH APPROACH
The research starts with an investigation of the current state of diversity in global food supplies.
This also represents an exploration into a longstanding two-part assumption in the field of
genetic resources – that humanity relies on relatively few crops for its survival, and that this list
of important crops is growing smaller over time.
The thesis then delves into the potential for use of a particular set of crop genetic resources of
5

Chapter 1
increasing interest globally – crop wild relatives. These weedy and wild cousins of cultivated
crops have been used for many decades in order to introduce traits of value to crops through
breeding, particularly for pest and disease resistance (Iltis 1988, Xiao et al. 1996, Gur & Zamir
2004, Hajjar & Hodgkin 2007, McCouch et al. 2007, Khoury et al. 2010, Maxted et al. 2012).
Their use in crop improvement is likely only to increase for a variety of reasons. First,
information is improving regarding the identity, potential, and conservation status of these
plants, and digital data platforms are making this information more readily available (Harlan
and de Wet 1971, Andersson & de Vicente 2010, The Plant List 2010, GBIF 2012, Wiersema
et al. 2012, GRIN 2013, Vincent et al. 2013). Second, advancements in geographic information
systems, both through the use of increasingly high resolution ecological (e.g., climatic, soil,
ecosystem classification, and human impact) data and evolving models and methods, are
enabling a more comprehensive conceptualization of the geographic distribution of crop
diversity and its conservation needs (Hijmans & Spooner 2001, Hijmans et al. 2005, Jarvis et
al. 2005, Elith et al. 2006; Phillips et al. 2006, Phillips & Dudik 2008, Costa et al. 2010,
Ramírez-Villegas et al. 2010, Parra-Quijano et al. 2011, Hengl et al. 2014, Tapia et al. 2014).
In addition, rapidly progressing classical and genomic tools, technologies, and methods are
facilitating their use in crop breeding (Tanksley & McCouch 1997, Zamir 2001, Ford-Lloyd et
al. 2011, Volk & Richards 2011, McCouch et al. 2012). Finally, there is a growing interest in
the use of exotic genetic diversity in order to confront increasingly difficult pest and disease,
abiotic stress, and other agronomic challenges (Hoisington et al. 1999, Gepts 2006, Ortiz et al.
2008, Jarvis et al. 2009, Tester and Langridge 2010, Guarino & Lobell 2011, McCouch et al.
2013, Henry 2014).
Crop wild relatives are genetic resources at the nexus of a number of critical global challenges.
They have the potential to contribute significantly to crop improvement, helping to address food
security and development goals, while improving the adaptation of crops to climate change
(Hoisington et al. 1999, Guarino & Lobell 2011, Dempewolf et al. 2014). As wild plants, they
are subject to a myriad of human caused threats to natural ecosystems, including habitat
modification, urbanization, mining, logging, changing fire regimes, pollution, invasive species,
overharvesting, and climate change (Stein et al. 2000, Wilkes 2007, Jarvis et al. 2008, Loarie
et al. 2009, Brummitt & Bachman 2010, Khoury et al. 2010, Ureta et al. 2011). A focus on wild
genetic resources is thus timely both for conservation and crop improvement objectives.

6

General introduction
The research presented in this thesis was performed in contribution to a number of ongoing
crop wild relative initiatives active at different scales globally. "Adapting Agriculture to
Climate Change: Collecting, Protecting and Preparing Crop Wild Relatives", is a ten year global
project aimed at identifying important crop wild relatives worldwide, assessing their
representation in genebanks, filling gaps in these ex situ collections through targeted collecting,
and pre-breeding collected samples in order to provide breeders with materials of interest for
adapting their crop varieties to climate change (Dempewolf et al. 2014). On the national level,
the United States is bringing together information needed to conceptualize the crop wild relative
diversity present in the country, assess its conservation needs, and organize strategies for longterm management of this diversity by government, private, and other stakeholders (Plant
Germplasm Operations Committee 2010). Within the global CGIAR Research Programs,
including Roots, Tubers and Banana, and Grain Legumes, assessments are being made of the
state of conservation of crop wild relatives of mandate crops, with the aim of making the
diversity present in these species available to breeders worldwide (CGIAR 2015).
While conservation of crop genetic resources is fundamental to the availability of this diversity
for present and future breeding, it is not the only major constraint to utilization. National and
international policies on crop genetic resources determine the real capacity for researchers to
acquire diversity of potential interest (Fowler & Hodgkin 2004, Esquinas-Alcázar 2005,
Bjørnstad et al. 2013). The thesis culminates in an exploration of the implications of the global
geographic distribution of crop genetic diversity for food security, in particular the level to
which international collaboration is required in order to achieve access to genetic resources
where they are needed. This research was performed in contribution to current negotiations for
the potential expansion of scope and membership within the International Treaty on Plant
Genetic Resources for Food and Agriculture (Plant Treaty) (FAO 2002).
THESIS OUTLINE
This thesis is organized into eight chapters (Figure 1). Following this general introduction,
Chapter 2 contextualizes the need for the use of diverse genetic resources in crop breeding in
order to mitigate vulnerability in global food supplies. This chapter represents an exploration
of central assumptions in the field of plant genetic resources – that humanity relies on relatively
few crops for its survival, and that this list of important crops is growing smaller over time inspired by a seminal article in the field that was published 25 years ago using data from 35
7

Chapter 1
years before the present (Prescott-Allen & Prescott-Allen 1990). Research questions include:
what is the state of crop species diversity in global food supplies?, what is the trend over time
in change in this diversity?, what are the most important crops globally?, and, what options
might be available to diversify food supplies in the future?
From this global context, Chapter 3 concentrates on the identification of potentially important
wild genetic resources at the national level. Focusing on the United States, a large and ecogeographically diverse country with relatively advanced conservation policy and active national
genetic resources conservation efforts, the chapter presents an inventory of crop wild relatives
and other wild species of potential use in agricultural research. The chapter also develops a
method for prioritizing these wild species based upon their potential to contribute to food
security. Research questions include: how can crop wild relatives and other useful wild plants
be documented at the national level?, and, how can they be prioritized for conservation and
research actions?
Once potentially valuable crop genetic resources are identified, subsequent information is
needed regarding where they occur, what diversity they may possess, and how well conserved
and therefore available to crop breeders they are. Chapter 4 offers a methodology to answer
these questions at the crop genepool level, i.e., for the wild relatives associated with any
particular crop. This chapter builds upon developments in the generation of and access to digital
occurrence and eco-geographic data as well as improvements in modeling wild plant species
distributions, intentionally utilizing freely available software and data. The method also
includes a novel expert assessment methodology, using researchers knowledgeable in the
distributions and conservation concerns of crop wild relatives to evaluate the results. Research
questions include: how can modeled geographic distributions of crop wild relatives be used to
assess the comprehensiveness of the diversity represented in existing genebank collections?,
how can priorities for further collecting in order to fill gaps in the conservation of species in
genebanks be outlined at the species and population level?, and, how valid are these results
compared to expert opinion, and thus what are the potential limitations of the method?
Chapters 5 and 6 utilize the basic methodology outlined in Chapter 5, but also advance the
method by capitalizing on knowledge learned over the four years of doctoral research
completed since the publication of the original gap analysis case study. Along with
8

General introduction

Figure 1. Thesis structure

9

Chapter 1
improvements in species targeting, occurrence data, modeling, and expert feedback methods,
the chapters take an additional step by drawing upon eco-geographic information to indicate
the potential for species and specific populations to possess traits of value to crop improvement
objectives, particularly for abiotic stress tolerance. The chapters focus on crops differing
substantially in the state of existing information regarding associated wild relatives, as well as
historical use of the resources in crop improvement. Research questions include: what
constitutes a potentially useful wild relative of sweetpotato and pigeonpea?, where are these
species encountered?, what is the state of conservation and availability of these species to
researchers?, what are the highest taxonomic and ecogeographic priorities for further
collecting?, and, what traits do species and populations possess that may be valuable to specific
crop improvement objectives?
Chapter 7 focuses on access to crop genetic resources. Given the geography of crop genetic
diversity identified in previous chapters and additional literature stretching back to the works
of N. I. Vavilov (1926), the chapter investigates the disjunct between where crop genetic
resources originated and are particularly diverse, and where crops themselves are currently
important in diets and production systems. The overall goal is to assess the degree to which
countries are interdependent in regard to crop genetic resources, and therefore need to engage
in international exchange in order to support the crop breeding that underlies their national food
supplies and production systems. As with Chapter 2, this research was inspired by an important
historical article (Flores-Palacios 1998), which contributed to the initial arguments for the
creation of the Plant Treaty (FAO 2002). Publication of Chapter 7 is intended to increase
membership and contribute to the rationale for expansion of the crops covered under the Plant
Treaty. Research questions include: how does the distribution of crop genetic diversity
associated with the world’s major crops indicate interdependence among countries in regard to
crop breeding efforts?, how dependent are countries on crop diversity from outside their borders
in order to support their food supplies and production systems?, how is this dependence
changing over time?, and what are the implications of this dependence in regard to current
political efforts focused on access to crop genetic resources?
Chapter 8 provides a general discussion of the research and its main results. The chapter offers
a summary of the current status of conservation of the crop genetic resources targeted in the
thesis, identifying limitations in the study and suggesting future research in order to make
10

General introduction
further progress on key questions in the field. The chapter also discusses the impact of the
research to date, and highlights ongoing activities that are building upon the efforts documented
here.

11

Chapter 1
REFERENCES
Alston, J.M., Beddow, J.M., Pardey, P.G. (2009). Agricultural research, productivity, and food prices in
the long run. Science 325 (5945), 1209-1210.
Andersson, M.S., de Vicente, M.C. (2010). Gene Flow Between Crops and their Wild Relatives
(Baltimore: The Johns Hopkins University Press).
Bjørnstad, Å., Tekle, S., Göransson, M. (2013). ‘Facilitated access’ to plant genetic resources: does it
work? Gen. Res. Crop Evol. 60, 1959–1965.
Bodirsky, B.L., Popp, A., Lotze-Campen, H., Dietrich, J.P., Rolinski, S., Weindl, I., et al. (2014).
Reactive nitrogen requirements to feed the world in 2050 and potential to mitigate nitrogen pollution.
Nat. Commun. 5, 3858.
Brummitt, N., Bachman, S. (2010). Plants Under Pressure: A Global Assessment. The First Report of
the IUCN Sampled Red List Index for Plants (London: Natural History Museum, UK).
Burke, M.B., Lobell, D.B., Guarino, L. (2009). Shifts in African crop climates by 2050, and the
implications for crop improvement and genetic resources conservation. Global Environ. Chang. 19
(3), 317-325.
CDC (Centers for Disease Control and Prevention). (2015). Micronutrient Facts. http://www.cdc.gov/
immpact/micronutrients/.
CGIAR. 2015. CGIAR Research Programs. http://www.cgiar.org/our-research/cgiar-researchprograms/.
Cordell, D., Drangert, J-O., White, S. (2009). The story of phosphorus: global food security and food
for thought. Global Environ. Chang. 19 (2), 292–305.
Costa, G.C., Nogueira, C., Machado, R.B., Colli, G.R. (2010). Sampling bias and the use of ecological
niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodivers.
Conserv. 19, 883-899.
Dempewolf, H., Eastwood, R.J., Guarino, L., Khoury, C.K., Müller, J.V., Toll, J. (2014). Adapting
agriculture to climate change: A global initiative to collect, conserve, and use crop wild relatives.
Agroecol. Sust. Food Syst. 38 (4), 369-377.
Diamond, J. (2004). The wealth of nations. Nature 429, 616-617.
Ehrlich, P.R. (1975). The Population Bomb (Mattituck, New York: Ameron).
Elith, J., Graham, C.H., Anderson, R.P., Dudik, M., Ferrier, S., Guisan, A., et al. (2006). Novel methods
improve prediction of species’ distributions from occurrence data. Ecography 29, 129-151.
Esquinas-Alcázar, J. (2005). Protecting crop genetic diversity for food security: political, ethical and
technical challenges. Nat. Rev. Genet. 6 (12), 946–953.
FAO (Food and Agriculture Organization of the United Nations). (2010). The Second Report on the
State of the World’s Plant Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2002). International Treaty on Plant
Genetic Resources for Food and Agriculture (Rome: FAO).
Flores-Palacios, X. (1998). Contribution to the Estimation of Countries’ Interdependence in the Area of
Plant Genetic Resources. Commission on Genetic Resources for Food and Agriculture, Background
Study Paper No. 7, Rev. 1 (Rome: FAO).
Foley, J.A., Defries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., et al. (2005). Global
consequences of land use. Science 309 (5734), 570-574.
Foley, J.A., Ramankutty, N., Brauman, K.A., Cassidy, E.S., Gerber, J.S., Johnston, M., et al. (2011).
Solutions for a cultivated planet. Nature 478, 337–342.
Ford-Lloyd, B., Schmidt, M., Armstrong, S.J., Barazani, O., Engels, J., Hadas, R., et al. (2011). Crop
wild relatives - undervalued, underutilized and under threat? BioScience 61, 559-565.
12

General introduction
Fowler, C., Hodgkin, T. (2004). Plant genetic resources for food and agriculture: assessing global
availability. Annu. Rev. Env. Resour. 29 (1), 143-179.
GBIF (Global Biodiversity Information Facility). (2012). GBIF Data Portal (Copenhagen: GBIF).
http://data.gbif.org/ welcome.htm.
Gepts, P. (2006). Plant genetic resources conservation and utilization: the accomplishments and future
of a societal insurance policy. Crop Sci. 46 (5), 2278-2292.
Global Crop Diversity Trust. (2013). Fundraising Strategy 2014-2018 (Bonn: Global Crop Diversity
Trust).
Godfray, H.C.J., Beddington, J.R., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., et al. (2010). Food
security: The challenge of feeding 9 billion people. Science 327, 812–818.
Gomez, M.I., Barrett, C.B., Raney, T., Pinstrup-Andersen, P., Meerman, J., Croppenstedt, A. et al.
(2013). Post-Green Revolution Food Systems and the Triple Burden of Malnutrition. ESA Working
Paper No. 13-02 (Rome: Agricultural Development Economics Division, FAO).
Graham, R.D., Welsh, R.M., Saunders, D.A., Ortiz-Monasterio, I., Bouis, H.E., Bonierbale, M., et al.
(2007), Nutritious subsistence food systems. Adv. Agron. 92, 1-74.
GRIN (Germplasm Resources Information Network). (2013). USDA, ARS, National Genetic Resources
Program. http://www.ars-grin.gov/cgi-bin/npgs/html/wep.pl.
Guarino, L., Lobell, D.B. (2011). A walk on the wild side. Nat. Clim. Change 1 (8), 374-375.
Gur, A., Zamir, D. (2004). Unused natural variation can lift yield barriers in plant breeding. PLoS Biol.
2(10), e245.
Hajjar, R., Hodgkin, T. (2007). The use of wild relatives for crop improvement: a survey of
developments over the past 20 years. Euphytica 156, 1-13.
Harlan, J.R. (1971). Agricultural origins: centres and noncentres, Science 174, 468-474.
Harlan, J.R. (1975). Crops and Man (Madison: American Society of Agronomy).
Harlan, J.R., de Wet, J.M.J. (1971). Toward a rational classification of cultivated plants. Taxon 20 (4),
509-517.
Hawkesworth, S., Dangour, A.D., Johnston, D., Lock, K., Poole, N., Rushton, J., et al. (2010). Feeding
the world healthily: the challenge of measuring the effects of agriculture on health. Philos. T. R. Soc.
B. 365 (1554), 3083–3097.
Hengl, T., de Jesus, J.M., MacMillan, R.A., Batjes, N.H., Heuvelink, G.B.M., Ribeiro, E., et al. (2014).
SoilGrids1km - global soil information based on automated mapping. PLoS One 9 (8), e105992.
Henry, R.J. (2014). Genomics strategies for germplasm characterization and the development of climate
resilient crops. Front. Plant Sci. 5, 68.
Hijmans, R.J., Cameron, S., Parra, J., Jones, P.G., Jarvis, A. (2005). WorldClim, Version 1.3 (Berkeley:
University of California).
Hijmans, R.J., Spooner, D. (2001). Geographic distribution of wild potato species. Am. J. Bot. 88, 21012112.
Hoisington, D., Khairallah, M., Reeves, T., Ribout, J.-M., Skovmand, B., Taba, S., et al. (1999). Plant
genetic resources: what can they contribute toward increased crop productivity? Proc. Natl. Acad.
Sci. 96, 5937-5943.
Iltis, H.H. (1988). “Serendipity in the exploration of biodiversity. what good are weedy tomatoes?,” in:
Biodiversity, ed. E.O. Wilson (Washington D.C.: National Academy Press), 98-105.
International Food Policy Research Institute. 2014. Global Nutrition Report 2014: Actions and
Accountability to Accelerate the World’s Progress on Nutrition (Washington, DC: International Food
Policy Research Institute).
Jarvis, A., Lane, A., Hijmans, R.J. (2008). The effect of climate change on crop wild relatives. Agr.
Ecosyst. Environ. 126 (1-2), 13–23.
13

Chapter 1
Jarvis, A., Ramirez-Villegas, J., Hanson, J., Leibing, C. (2009). “Crop and forage genetic resources:
international interdependence in the face of climate change,” in: The Impact of Climate Change on
Countries’ Interdependence on Genetic Resources for Food and Agriculture. Background Study
Paper No.48, eds. S. Fujisaka, D. Williams, M. Halewood (Rome: Commission on Genetic Resources
for Food and Agriculture, FAO).
Jarvis, A., Williams, K., Williams, D., Guarino, L., Caballero, P.J., Mottram, G. (2005). Use of GIS for
optimizing a collecting mission for a rare wild pepper (Capsicum flexuosum Sendtn.) in Paraguay.
Genet. Resour. Crop Ev. 52 (6), 671-682.
Kastner, T., Rivas, M.J.I., Koch, W., Nonhebel, S. (2012). Global changes in diets and the consequences
for land requirements for food. Proc. Natl. Acad. Sci. 109 (18), 6868-6872.
Kearney, J. (2010). Food consumption trends and drivers. Philos. T. R. Soc. B. 365 (1554), 2793-2807.
Khoury, C., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic resources:
a review of global crop and regional conservation strategies. Genet. Resour. Crop Ev. 57 (4), 625639.
Loarie, S.R., Duffy, P.B., Hamilton, H., Asner, G.P., Field, C.B., Ackerly, D.D. (2009). The velocity of
climate change. Nature 462 (7276), 1052-1055.
Lobell, D.B., Burke, M.B., Tebaldi, C., Mastrandrea, M.D., Falcon, W.P., Naylor, R.L. (2008).
Prioritizing climate change adaptation needs for food security in 2030. Science 319 (5863), 607-610.
Malthus, T.R. (1803). An Essay on the Principle of Population; or, a view of its past and present effects
on human happiness; with an enquiry into our prospects respecting the future removal or mitigation
of the evils which it occasions (London: John Murray).
Matson, P.A., Vitousek, P.M. (2006). Agricultural intensification: Will land spared from farming be
land spared for nature? Conserv. Biol. 20 (3), 709–710.
Maxted, N., Kell, S.P., Ford-Lloyd, B.V., Dulloo, E., Toledo, Á. (2012). Toward the systematic
conservation of global crop wild relative diversity. Crop Sci. 52, 774-785.
McCouch, S., Baute, G.J., Bradeen, J., Bramel, P., Bretting, P.K., Buckler, E., et al. (2013). Agriculture:
feeding the future. Nature 499 (7456), 23-24.
McCouch, S.R., McNally, K.L., Wang, W., Sackville Hamilton, R. (2012). Genomics of gene banks: A
case study in rice. Am. J. Bot. 99 (2), 407-423.
McCouch, S.R., Sweeney, M., Li, J., Jiang, H., Thomson, M., Septinginsih, E., et al. (2007). Through
the genetic bottleneck: O. rufipogon as a source of trait-enhancing alleles for O. sativa. Euphytica
154, 317-339.
Ortiz, R., Sayre, K.D., Govaerts, B., Gupta, R., Subbarao, G.V., Ban, T., et al. (2008). Climate change:
can wheat beat the heat? Agr. Ecosyst. Environ. 126, 46-58.
Parra-Quijano, M., Iriondo, J.M., Torres, E. (2011). Improving representativeness of genebank
collections through species distribution models, gap analysis and ecogeographical maps. Biodivers.
Conserv. 21, 79-96.
Phalan, B., Onia, M., Balmford, A., Green, R.E. (2011). Reconciling food production and biodiversity
conservation: Land sharing and land sparing compared. Science 333 (6047), 1289–1291.
Phillips, S.J., Anderson, R.P., Schapire, R.E. (2006). Maximum entropy modeling of species geographic
distributions. Ecol. Model. 190, 231-259.
Phillips, S.J., Dudik, M. (2008). Modeling of species distributions with Maxent: new extensions and
comprehensive evaluation. Ecography 31, 161-175.
Pingali, P. (2007). Westernization of Asian diets and the transformation of food systems: implications
for research and policy. Food Pol. 32 (3), 281-298.
Plant Germplasm Operations Committee. (2010). Minutes of the Annual Meeting, Geneva, NY. 28–29
July 2010 (Beltsville: USDA National Germplasm Resources Laboratory).
Popkin, B.M. (2006). Technology, transport, globalization and the nutrition transition food policy. Food
14

General introduction
Pol. 31 (6), 554-569.
Prescott-Allen, R., Prescott-Allen, C. (1990). How many plants feed the world? Conserv. Biol. 4 (4),
365-374.
Ramírez-Villegas, J., Khoury, C., Jarvis, A., Debouck, D.G., Guarino, L. (2010). A gap analysis
methodology for collecting crop genepools: a case study with Phaseolus beans. PLoS One 5, e13497.
Ray, D.K., Mueller, N.D., West, P.C., Foley, J.A. (2013). Yield trends are insufficient to double global
crop production by 2050. PLoS One 8 (6), e66428.
Ray, D.K., Ramankutty, N., Mueller, N.D., West, P.C., Foley, J.A. (2012). Recent patterns of crop yield
growth and stagnation. Nat. Commun. 3, 1293.
Remans, R., Flynn, D.F.B., DeClerk, G., Diru, W., Fanzo, J., Gaynor, K., et al. (2011). Assessing
nutritional diversity of cropping systems in African villages. PLoS One 6 (6), e21235.
Rockström, J., Steffen, W., Noone, K., Persson, Å., Chapin, F.S., Lambin, E.F., et al. (2009). A safe
operating space for humanity. Nature 461 (7263), 472–475.
Ruel, M.T. (2003). Operationalizing dietary diversity: a review of measurement issues and research
priorities. J. Nutr. 133 (11), 3911S-3926S.
Stein, B.A., Kutner, L.S., Adams, J.S. (2000). Precious Heritage: The Status of Biodiversity in the
United States (New York: Oxford University Press).
Struik, P.C., Kuyper, T.W. (2014). Editorial overview: sustainable intensification to feed the world:
concepts, technologies and trade-offs. Cur. Op. Environ. Sustain. 8, vi – viii.
Tanksley, S.D., McCouch, S.R. (1997). Seed banks and molecular maps: unlocking genetic potential
from the wild. Science 277 (5329), 1063-1066.
Tapia, C., Torres, E., Parra-Quijano, M. (2014). Searching for adaptation to abiotic stress:
ecogeographical analysis of highland Ecuadorian maize. Crop Sci. 55 (1), 262-274.
Tester, M, Langridge, P. (2010). Breeding technologies to increase crop production in a changing world.
Science 327 (5967), 818–822.
The Plant List. (2010). Version 1. http://www.theplantlist.org/.
Thormann, I., Fiorino, E., Halewood, M., Engels, J.M.M. (2015). Plant genetic resources collections
and associated information as a baseline resource for genetic diversity studies: an assessment of the
IBPGR-supported collections. Genet. Resour. Crop Evol. doi 10.1007/s10722-015-0231-9.
Tilman, D., Balzer, C., Hill, J., Befort, B.L. (2011). Global food demand and the sustainable
intensification of agriculture. Proc. Natl. Acad. Sci. 108, 20260–20264.
Tilman, D., Clark, M. (2014). Global diets link environmental sustainability and human health. Nature
515 (7528), 518–522.
Ureta, C., Martinez-Meyer, E., Perales, H.R., Alvarez-Buylla, E.R. (2011). Projecting the effects of
climate change on the distribution of maize races and their wild relatives in Mexico. Glob. Change
Biol. 18, 1073-1082.
van de Wouw, M., Kik, C., van Hintum, T., van Treuren, R., Visser, B. (2009). Genetic erosion in crops:
concept, research results and challenges. Plant Genetic Resources 8, 1-15.
van de Wouw, M., van Hintum, T., Kik, C., van Treuren, R., Visser, B. (2010). Genetic diversity trends
in twentieth century crop cultivars: a meta analysis. Theor. Appl. Genet. 120 (6), 1241-1252.
Vavilov, N.I. (1926). Studies on the origin of cultivated plants. Bulletin of Applied Botany, Genetics and
Plant Breeding 16, 1-248.
Vincent, H., Wiersema, J., Kell, S., Fielder, H., Dobbie, S., Castañeda-Álvarez, N.P., et al. (2013). A
prioritized crop wild relative inventory to help underpin global food security. Biol. Conserv. 167,
265-275.
Vogt, W. (1948). Road to Survival (New York: W. Sloane Associates).
Volk, G.M., Richards, C.M. (2011). Integration of georeferencing, habitat, sampling, and genetic data
15

Chapter 1
for documentation of wild plant genetic resources. Hortic. Sci. 46 (11), 1446–1449.
West, P.C., Gerber, J.S., Engstrom, P.M., Mueller, N.D. Brauman, K.A., Carlson, K.M., et al. (2014).
Leverage points for improving global food security and the environment. Science 345 (6194), 325–
328.
Wheeler, T., von Braun, J. (2013). Climate change impacts on global food security. Science 341 (6145),
508-513.
WHO (World Health Organization). (2015). Micronutrient deficiencies. http://www.who.int/nutrition/
topics/vad/en/.
Wiersema, J.H., Leόn, B., Garvey, E.J. (2012). Identifying wild relatives of subtropical and temperate
fruit and nut crops. Acta Hortic. 948, 285-288.
Wilkes, G. (2007). Urgent notice to all maize researchers: disappearance and extinction of the last wild
teosinte population is more than half completed. A modest proposal for teosinte evolution and
conservation in situ: The Balsas, Guerrero, Mexico. Maydica 52, 49-58.
Xiao, J., Grandillo, S., Ahn, S.N., McCouch, S.R., Tanksley, S.D. (1996). Genes from wild rice improve
yield. Nature 384, 223-224.
Zamir, D. (2001). Improving plant breeding with exotic genetic libraries. Nat. Genet. 2, 983-989.
Zhu, Y., Chen, H., Fan, J., Wang, Y., Li, Y., Chen, J., et al. (2000). Genetic diversity and disease
control in rice. Nature 406 (6797), 718-722.

16

CHAPTER 2
Increasing homogeneity in global food supplies and the
implications for food security
Colin K. Khoury1,2, Anne D. Bjorkman3,4, Hannes Dempewolf 4,5,6,
Julian Ramirez-Villegas1,7,8, Luigi Guarino6, Andy Jarvis 1,7, Loren H. Rieseberg4,5,9, and
Paul C. Struik2
International Center for Tropical Agriculture (CIAT), Km 17, Recta Cali-Palmira, Apartado Aéreo
6713, Cali, Colombia.
2
Centre for Crop Systems Analysis, Wageningen University, Droevendaalsesteeg 1, 6708 PB
Wageningen, The Netherlands.
3
Department of Geography, University of British Columbia, Vancouver, BC, Canada V6T 1Z4.
4
Department of Botany, University of British Columbia, Vancouver, BC, Canada V6T 1Z4.
5
The Biodiversity Research Centre, University of British Columbia, Vancouver, BC, Canada V6T
1Z4.
6
Global Crop Diversity Trust, Platz der Vereinten Nationen 7, 53115 Bonn, Germany.
7
CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Cali,
Colombia.
8
School of Earth and Environment, University of Leeds, Leeds, LS2 9JT, United Kingdom.
9
Department of Biology, Indiana University, Bloomington, Indiana, 47405.
1

Published in Proceedings of the National Academy of Sciences of the United States of America
(PNAS) 111 (11), 4001-4006. doi: 10.1073/pnas.1313490111

Chapter 2
Abstract
The narrowing of diversity in crop species contributing to the world’s food supplies has been considered
a potential threat to food security. However, changes in this diversity have not been quantified globally.
We assess trends over the past 50 years in the richness, abundance, and composition of crop species in
national food supplies worldwide. Over this period national per capita food supplies expanded in total
quantities of food calories, protein, fat, and weight, with increased proportions of those quantities
sourcing from energy dense foods. At the same time, the number of measured crop commodities
contributing to national food supplies increased, the relative contribution of these commodities within
these supplies became more even, and the dominance of the most significant commodities decreased.
As a consequence, national food supplies worldwide became more similar in composition, correlated
particularly with an increased supply of a number of globally important cereal and oil crops, and a
decline of other cereal, oil, and starchy root species. The increase in homogeneity worldwide portends
the establishment of a global standard food supply, which is relatively species rich in regard to measured
crops at the national level, but species poor globally. These changes in food supplies heighten
interdependence among countries in regard to availability and access to these food sources and the
genetic resources supporting their production, and give further urgency to nutrition development
priorities aimed at bolstering food security.
Keywords: Crop diversity, Plant genetic resources, Agricultural development, Global analysis

18

Increasing homogeneity in global food supplies
INTRODUCTION
A shared axiom of ecology and nutrition is that, within certain ranges, diversity enhances the
health and function of complex biological systems. Species diversity has been shown to
stimulate productivity, stability, ecosystem services, and resilience in natural (Hooper et al.
2005, Cadotte et al. 2012, Hooper et al. 2012, Zhang et al. 2012, Gamfeldt et al. 2013) and in
agricultural ecosystems (Kirwan et al. 2007, Hajjar et al. 2008, Picasso et al. 2008, Bonin &
Tracy 2012, Cabell & Oelofse 2012, Davis et al. 2012, Kremen & Miles 2012, Mijatović et al.
2013). Likewise, variation in food species contributing to diet has been associated with
nutritional adequacy (Frei & Becker 2004, Kennedy et al. 2005, Graham et al. 2007, Negin et
al. 2009) and food security (Ruel 2003).
The development of sedentary agricultural societies and further rise of modern agriculture is
generally considered to have led to a decline in the total number of plant species upon which
humans depend for food (Harlan 1975, Gepts 2006), particularly the wild, semi-domesticated,
and cultivated vegetables and fruits, spices, and other food plants that supplemented staple
crops with the provision of micronutrients and that historically bolstered food security during
crop failures (Doughty 1979). Harlan (1975) warned that
“most of the food for mankind comes from a small number of crops and the total
number is decreasing steadily. In the United States in the past 40 years, many
vegetables and fruits have disappeared from the diet, and the trend is going on all
over the world. More and more people will be fed by fewer and fewer crops.”
(Harlan 1975).
More recent analyses of dietary transition in developing countries in association with
globalization have noted increases in the diversity of plants contributing to diets locally, along
with a Westernization transition in preference of energy dense foods (i.e. animal products, plant
oils and sugars) over cereals, pulses, and vegetables, and of particular major crop plants within
these food categories over traditional crops (Kearney 2010, Pingali 2007). The impact of such
changes on overall crop diversity worldwide has not been comprehensively documented,
although recent changes in varietal and allelic level diversity of some crops have been
investigated (van de Wouw et al. 2009, 2010, Bonneuil et al. 2012). Given the potential food
security implications of narrowing of the diversity of crop species both in production systems
19

Chapter 2
and in food supplies, an assessment of the global state of crop plant species diversity is
warranted.
Here we examine changes in the diversity of the portfolio of crop species upon which humans
primarily depend for food security in regard to calories, protein, fat, and food weight. Using
national per capita food supply data published by the Food and Agriculture Organization (FAO)
of the United Nations, we analysed trends in the richness, abundance, and composition of
measured crop commodities in the food supplies of 152 countries comprising 98% of the
world’s population from 1961-2009.
MATERIALS AND METHODS
We analyzed FAO national per capita food supply data for all available measurements [calories
(kcal/capita/day), protein (g/capita/day), fat (g/capita/day), and weight (g/capita/day)], from
1961-2009, utilizing the full set of commodity and country listings, standardized across all
years. Food supply from plants represents national production plus imports plus or minus stock
changes over the survey period; minus exports, quantities used for seed, animal feed, and in the
manufacture of non-food products, and losses during storage and transport (Prescott-Allen &
Prescott-Allen 1990). Plant commodities clearly comprised of the same crop species were
aggregated into single commodities representing the crop species, e.g. olive oil and olives. After
aggregation, 52 crop commodities remained. Animal foods were included within the crop
importance analysis as a single aggregated commodity.
In order to align all time periods and include as much of the world’s population as possible, the
current nations formerly comprising the USSR, Yugoslav SFR, Ethiopia PDR, and
Czechoslovakia were aggregated into their former states, with national data merged by
weighted average based upon population of the respective states at the respective reporting year.
Belgium and Luxembourg were reported together during 1961-1999 and therefore recent years
listing the countries separately were merged as above. Data for two countries (Occupied
Palestinian Territories and Netherlands Antilles) did not appear consistently reported over the
study period and were excluded from the analysis. The resulting 152 comparable countries
comprised 98% of the world’s population across the study period.
Changes in the spread of crop commodities [based on the presence (>0) or absence of a crop in
20

Increasing homogeneity in global food supplies
a given country in each year] were analyzed using generalized estimating equations with a
binomial error distribution, “country” as a grouping factor, and an autoregressive (AR-1)
correlation structure (i.e., time-lag correlation), using the “geeM” package in R. Slope
parameters from these models were extracted to show the direction and magnitude of change
for each commodity. Changes in globally aggregated absolute abundance were non-linear, and
were therefore analyzed by subtracting the 1961 abundance value from the 2009 value. Changes
in country-level absolute abundance were analyzed the same way, for consistency, and the percountry difference was then averaged across countries. We analyzed changes in relative
abundances of crops (i.e., relative to the total value of all crops within a given country) using a
generalized linear mixed model with a binomial error distribution, with fixed effects for “year”
and “crop” and a random effect of “country.” Comparisons between crops were performed by
estimating the slopes of the predicted values for each crop in the model. Error bars were
calculated by estimating the maximum and minimum slope values based on the standard errors
of the predicted values. Crops for which fat or protein make up less than 1% of their weight
were removed from the analysis for those respective measurements. All mixed-effects models
were performed using the “lme4” package in R.
Changes in crop commodity richness in national food supplies were analyzed with generalized
linear mixed effects models with a Poisson error distribution and “country” as a random effect,
with random slopes for “year”, to account for repeated measures over time. The significance of
the fixed-effect “year” term in predicting richness was ascertained by conducting a Chi-squared
likelihood ratio test on nested models with and without the term. In order to estimate the
differences between countries in change in richness we extracted slope coefficients from the
random effects of the mixed-effects model for one measurement type, calories. These
coefficients represent the magnitude and direction of the change in richness for each country
but do not provide estimates of error. One country (Namibia) displayed a markedly inconsistent
jump in richness. The slope for this country was therefore derived from data only from
consistent years, with slope parameters estimated by weighted averaging of the slope in
consistent time periods. Analyses were performed using the “lme4” package in R.
We used Pielou's evenness index to measure how equally crops contributed to a country’s food
supplies in a given year, with linear mixed effects models in the “lme4” package to measure
change in evenness over time, as described above. Because Pielou’s evenness index can also
21

Chapter 2
reflect changes in richness, we additionally analyzed changes in dominance, defined as the
change in the proportion of the most abundant crop commodity in each country in each year,
using generalized linear mixed models with a binomial error distribution with the “lme4”
package in R. For evenness and dominance, slope parameters were extracted from the random
effects of the mixed models to estimate the direction and magnitude of the change in each
country.
We quantified the homogenization of crop commodity composition using Bray-Curtis distance
(dissimilarity) from each country to the global centroid (comparable to the global mean
commodity composition, which is inclusive of abundance) in each year using the “betadisper”
function in the R package “vegan”, with similarity derived by subtracting dissimilarity from 1.
A mean increase in similarity-to-centroid therefore represents a decrease in country-to-country
variation, and a homogenization of commodity composition over time. We then used linear
mixed models with “country” as a random factor to assess the significance of the change in
similarity-to-centroid over 1961-2009. Slope parameters were extracted from the random
effects of the mixed-effects models to estimate the direction and magnitude of the change in
similarity in each country. We also assessed homogenization using a non-metric
multidimensional scaling analysis (NMDS) with Bray-Curtis dissimilarities to compare the
mean and variance of commodity composition between three measurement years (1961, 1985
and 2009). Circles for each year represent 95% confidence intervals around the mean; smaller
circles indicate lower country-to-country variation in composition.
The number of crop species currently important to food supply was estimated by listing for the
most recent year (2009) the plant and animal commodities by decreasing importance until the
total contribution equaled 90% of each country’s food supply for each measurement, a threshold
which is inclusive of major contributors to supply and exclusive of commodities contributing
very small quantities (Prescott-Allen & Prescott-Allen 1990) (Supplementary Figure 1). One
biological species per commodity was recognized, except in cases where additional species
were clearly likely to contribute to that commodity in at least one listed country. For general
commodities, the minimum number of species most likely to contribute was identified
(Supplementary Table 1).

22

Increasing homogeneity in global food supplies
RESULTS
As a global trend, national per capita food supplies from both plant and animal sources
consistently increased over the past 50 years for all variables, with animal foods becoming
increasingly important in contribution to protein and oil crops dominating fat food supplies
(Supplementary Figure 2).
From 1961-2009 all 52 measured crop commodities, which included both individual crops and
groups of crops producing similar products, with the exception of cottonseed oil, increased in
geographic spread (i.e. were counted as contributing to food supply in an increasing number of
countries) (Figure 1A, Supplementary Figure 3). The major cereals wheat, rice, maize and
barley, along with sugar, potatoes, and the general vegetables and fruits commodities
maintained their ubiquity in food supplies across the planet over the past 50 years. Crop
commodities with the greatest relative changes in spread over this period included oil crops
such as soybean, sunflower, palm oil, and rape & mustard. Commodities showing relatively
small changes in spread and not already globally ubiquitous included minor cereals such as
millets, sorghum, and rye, and starchy root crops such as yams, cassava, and sweet potatoes.
Wheat, rice, maize and other ubiquitous crop commodities were among those with the greatest
gains in both relative and absolute abundance in national per capita food supplies over the past
50 years (Figure 1B, Supplementary Figure 4-5). In addition, the degree of increase in spread
was generally a good predictor of change in the abundance of the crop commodities in food
supplies. For example, oil commodities such as soybean, sunflower, palm oil, and rape &
mustard were among the crops showing the greatest average increase in relative abundance in
national food supplies, whereas millets, rye, sorghum, yams, cassava, and sweet potatoes
showed the largest declines. Such global changes in abundance of measured crops in national
food supplies are similarly evident in global aggregated per capita food supply data, which
assesses global quantities in relation to total world population (Supplementary Figure 5).
The richness of national per capita food supplies in regard to the 52 measured crop commodities
increased consistently over the past 50 years for all variables (Figure 2A-B, Supplementary
Figure 6). National food supplies also increased slightly in evenness, indicating greater equality
in the relative abundance of each of the crop commodities contributing to per capita food supply
within each country, although mixed trends were evident worldwide (Figure 2C-D,
23

Figure 1. Global change in spread and abundance of crop commodities in national food supplies from 1961-2009. (A) Slope of the relative change in the
geographic spread of crop commodities, defined as the change over time in the presence (i.e., value >0) of a crop commodity in contribution to any
variable of food supply in a country in each year. Bars represent slope parameters (+/- 95% CI) from generalized estimation equations with a binomial
error distribution, country as a grouping factor, and an auto-regressive correlation structure. Sugar; vegetables, other; and fruits, other commodities are
not depicted as they were nearly ubiquitous in spread globally throughout the study period and therefore did not change significantly. (B) Slope of the
change in relative abundance of crop commodities in contribution to calories, as derived from the value contributed by a particular crop relative to the
sum of all crops within a given country within a given year. Bars represent slopes (+/- 95% CI) of the predicted values (1961-2009) for each crop from
generalized linear mixed models with “year” and “crop” as fixed effects and country as a random effect. Slopes for change in relative abundance for all
measurements are depicted in Supplementary Figure 3.

Chapter 2

24

Increasing homogeneity in global food supplies

Figure 2. Global change in richness, evenness, and dominance of crop commodities in national food
supplies from 1961-2009. (A) Global mean change in crop commodity richness (i.e., number of crops)
in national food supplies. Points represent actual data, and lines are 95% prediction intervals from
generalized linear mixed models with a Poisson error distribution. (B) World map displaying the slope
of change in crop commodity richness per country for calories. (C) Global mean change in evenness of
contributing crop commodities within national food supplies calculated by Pielou’s evenness index.
Points represent actual data, and lines are 95% prediction intervals for a linear mixed effects model. (D)
World map displaying the slope of change in evenness of national food supplies for calories. (E) Global
mean change in the dominance of the most abundant crop commodity in each country in each year.
Points represent actual data, and lines are 95% prediction intervals from a generalized linear mixed
model with a binomial error distribution. (F) World map displaying the slope of change in dominance
in national food supplies for calories.

Supplementary Figure 7). Dominance (estimated as the proportion of a country’s per capita
food supply comprised of the most abundant crop commodity) declined as a global trend over
the study period (Figure 2E-F, Supplementary Figure 8). Notable reductions in originally very
high levels of dominance were visible in rice in contribution to calories in Southeast Asia,
coconut for fat in Pacific Island countries, and groundnut for fat in Central African countries.
As national food supplies became richer for the 52 measured crop commodities and relative
abundance patterns of contributing commodities shifted, food supplies worldwide became
25

Chapter 2

Figure 3. Global change in similarity (homogeneity) of food supplies, as measured by Bray-Curtis
dissimilarity from each country to the global centroid (mean composition) in each year, converted to
similarity. (A) Global mean change in similarity-to-centroid of national food supplies. Points represent
actual data, and lines are 95% prediction intervals from linear mixed effects models. (B) Multivariate
ordination of crop commodity composition in contribution to calories in national food supplies in 1961,
1985 and 2009. Red points represent the multivariate commodity composition of each country in 1961,
blue points in 1985, and black points in 2009. Circles represent 95% confidence intervals around the
centroid in each year. Between 1961 and 2009, the area contained within these 95% confidence intervals
decreased by 68.8%, representing the decline in country-to-country variation of commodity composition
(i.e., homogenization) over time. (C) World map displaying the slope of change in similarity-to-centroid
of national food supplies for calories.

more similar in composition (inclusive of abundance) for all variables (Figure 3, Supplementary
Figure 9). This increase in similarity brought national food supplies around the planet closer to
a global standard composition. Between 1961 and 2009, homogeneity increased by 16.7%, as
measured by the mean change in similarity between each country and the global standard
composition, with a maximum (single-country) change of 59.7%. Likewise, mean amongcountry similarity increased by 35.7%. East and Southeast Asian as well as Sub-Saharan
African countries as regional groups displayed the greatest changes in composition toward the
global standard over 1961-2009, in association with the greatest increases in measured crop
commodity richness and decreases in dominance.
As a measure of the relative importance of crop commodities in total global food supply,
inclusive both of plant and animal food sources, we found that 50 of the measured crop
commodities currently contribute to the top 90% of calories, protein, fat and weight around the
world. We estimate that these crop commodities are composed of 94 crop species from 70
26

Increasing homogeneity in global food supplies
genera from 37 plant families (Supplementary Table 1).
DISCUSSION
The national-level data and multiple variables analyzed here represent the greatest degree of
disaggregation of globally comparable food supply data available and thus permit the highest
resolution analysis currently feasible for analyzing trends in food crop species utilization
worldwide. Food supply data is not directly equivalent to consumption, as food losses at the
household level are not measured, but represents a superior measure of the importance of food
crops to diets than does production data (Prescott-Allen & Prescott-Allen 1990). National level
food supply data both generalizes and underestimates total existing food crop species diversity,
as sub-national dietary variation, crops primarily encountered in home gardens and local
markets, seasonally important food plants, and culinary herbs, spices and other crops consumed
in relatively small quantities are not comprehensively reported in national statistics (FAO
2001). Furthermore, FAO food supply data does not specifically report statistics in regard to
micronutrients, where species richness may be particularly significant (Doughty 1979), nor
does it assess the indirect contribution to food supply of animal feed and forage plant species.
Along with listed crops such as maize and soybean that are major food sources for livestock,
global production statistics (FAO 2010a) indicate that crops such as pumpkins, alfalfa, and
clover should be recognized as critical indirect contributors to food supply.
More importantly, because of the limited number of individual crop commodities reported by
FAO combined with the aggregation of numerous crops into several general commodities, the
analyses lack the resolution necessary for elucidating trends in those geographically restricted
cereals, legumes, fruits, vegetables and other crops that may be especially sensitive to changes
in the global food system. Although some progress has been made in listing and proposing
monitoring strategies for such species (Hammer & Khoshbakht 2005, Padulosi & Dulloo 2012),
substantial improvements in the systems for reporting food production and supply are needed
to adequately monitor such crops globally. Ideally assessments of the state of crop diversity in
food supplies would integrate interspecific, varietal, and allelic level analyses of contributing
crop plants (Bonneuil et al. 2012) in combination with nutritional information (Remans et al.
2011) and recorded at subnational levels inclusive of regional, community, household, and
individual variation (Pingali 2007, Hawkesworth et al. 2010).

27

Chapter 2
The increased food energy, protein, fat, and weight from plants in food supplies worldwide over
the past 50 years appear to primarily have been sourced from globally dominant crops
specifically reported by FAO. The total number of important crop species we identified
remained relatively consistent in comparison to a previous point estimate based on nationallevel data from 1979-1981 (Prescott-Allen & Prescott-Allen 1990), but the spread and
abundance values of these crops have changed measurably. The rate of movement toward
homogeneity in food supply compositions globally continues with no indication of slowing.
This trend implies a likely deterioration in importance of un-reported minor and geographically
restricted food plants, along with the measured cereal, oil, starchy root and other crops that
displayed significant declines in abundance in national food supplies. Thus even as the number
of measured crops available to the consumer in a given country has increased over the past halfcentury as a global trend, the total diversity of crops contributing significantly worldwide has
narrowed.
A series of interrelated factors are considered to have driven change in food supply
compositions worldwide. A primary driver for both diversifying diets and shifting toward
increased consumption of energy dense foods entails wealth increase and associated gains in
purchasing power. Transitions in dietary preferences, particularly toward Western diets (e.g.
meat and dairy, wheat, temperate vegetables and fruit, and sugary beverages) have been
associated with many facets of globalization and urbanization, including trade liberalization
and the development of extensive commodity transport systems, multinational food industries,
food quality and safety standardization, mass media, labor changes, smaller family sizes,
supermarkets, fast food, processed foods, and human migration. These drivers are detailed
elsewhere (Popkin 2006, Pingali 2007, Hawkesworth et al. 2010, Kearney 2010). The
modernization and globalization of agriculture through the replacement of human labor with
machinery, investments in the breeding and distribution of high yielding major crops as a
development strategy, and subsidies dedicated to a narrow range of crop commodities, among
other factors, have further contributed to the increasing global availability of a limited number
of major crop plants, with lesser priority given to nutritional diversity (Davis et al. 2004, Negin
et al. 2009).
Increases in measured crop species richness and evenness in food supplies may be considered
as indicative of enhanced food security on the local level, particularly in regard to availability
28

Increasing homogeneity in global food supplies
and access. On the other hand, the increased ability globally to realize within diets the human
preference for energy dense foods (Kearney 2010) based upon a limited number of global crop
commodities and processed products has been associated with the rise in non-communicable
diseases such as adult onset diabetes, heart disease, and certain forms of cancer (Popkin 2006,
Pingali 2007, Kearney 2010), a trend which is impacting rapidly developing countries such as
China more quickly than projected (Kearney 2010). Such dietary changes have also contributed
to reductions in diversity in human oral and gut microbiota, which in return have negatively
impacted health (Lozupone et al. 2012, Adler et al. 2013). With the number of over-nourished
worldwide surpassing the number of under-nourished (Popkin 2006), such diseases are
becoming epidemics, including within countries still grappling with significant constraints in
food availability. Making available and accessible adequate nutritional diversity within and in
supplement to the major crops comprising an ever-greater proportion of global food supplies is
critical (Ruel 2003, Graham et al. 2007, Remans et al. 2011). Engendering consumption
patterns cognizant of the impact of food crop choice on health is equally pressing.
The increasing reliance on a suite of truly global crop commodities implies a narrowing in the
diversity present in global agricultural systems as a whole, necessitating an equivalent
expansion in attention to production stability for these crops (Alston et al. 2009). The
importance of this stability is exacerbated by the necessity of increasing yields of major crops
in order to keep pace with demand, the future success of which has been questioned (Ray et al.
2013). From a genetic diversity perspective, increasing homogeneity in global food supplies
highlights the importance of the breeding and cultivation of varieties of these crops with diverse
genetic backgrounds (National Research Council 1972, Zhu et al. 2000). While varietal
diversity particularly of major cereal crops can in some regions be relatively high, for other key
crops such as banana only a handful of varieties are widely cultivated, despite substantial
diversity in the gene pool as a whole (Perrier et al. 2011).
As crop development efforts rely upon the utilization of genetic resources (Gepts 2006), it is a
policy imperative to ensure the conservation of, and access to, as wide a range of genetic
diversity within these global crops as possible, along with the genotypic and phenotypic
information necessary to effectively utilize these resources (McCouch et al. 2012,
2013). Unfortunately, significant gaps remain in the conservation of crop genetic resources
(FAO 2010b) and access to this diversity requires improvement (Fowler & Hodgkin 2004). A
29

Chapter 2
number of crops considered here are not sufficiently covered under the pertinent international
treaty (FAO 2002), including plants significant to food supplies globally, such as sugar cane,
soybean and groundnut.
While our study identifies the major crop commodities critical to national food supplies
worldwide, current patterns of production of these crops are not guaranteed given ongoing and
predicted changes in climate (Lobell et al. 2008, Burke et al. 2009, Wheeler & von Braun 2013),
the decline in availability of non-renewable inputs (Cordell et al. 2009), and increasingly severe
impacts of agriculture on soil, water quality and biodiversity (Foley et al. 2005). Such trends
may impact food security in regard to crop commodity trade (Fader et al. 2013), decrease the
nutritional quality of major crops (Dwivedi et al. 2013), and enhance the attractiveness of
under-researched crop species (Naylor et al. 2004) that are productive and nutritious given
limited inputs, particularly under marginal or variable conditions (Jarvis et al. 2012).
Moreover the importance of crop commodities in food supplies, particularly in contribution to
protein and fat, may shift in response to health, natural resource, and climate pressures (Kearney
2010, Stehfest et al. 2009, Kastner et al. 2012, Pradhan et al. 2013), counteracting the trend
demonstrated over the past 50 years in increased animal as well as energy-dense plant food
commodity consumption. The trajectory of Northern European food supplies appears to be
demonstrating such a trend (Kearney 2010). Providing that alternative food crops may still be
encountered, a further diversification of food supplies with interesting and nutritious crop plants
may bolster this evolution.

30

Increasing homogeneity in global food supplies
SUPPLEMENTARY INFORMATION

Supplementary Figure 1. Global average additive contribution by ranked abundance of crop plant and
animal commodities to calories, protein, fat and weight in food supply, 2009. The dotted line shown at
90% displays the threshold utilized to determine the number of crops considered important to each
national food supply in 2009 (Supplementary Table 1), which is inclusive of major contributors to supply
and exclusive of commodities contributing very small quantities (Prescott-Allen & Prescott-Allen
1990).

31

Chapter 2

Supplementary Figure 2. Global change in national food supplies by food groups from plant and animal
sources, for (A) calories, (B) protein (C) fat and (D) weight, 1961-2009. Data displayed is mean per
capita values across all (152) countries.

32

Increasing homogeneity in global food supplies
Supplementary Figure 3. Change in the total number of countries reporting each crop commodity, per
crop, 1961-2009.

Supplementary Figure 3A. Change in the total number of countries reporting each crop commodity using
generalized additive modeling (GAM). Crop commodities are counted as present within a given country
in a given year when contributing to food supply (>0) for any variable. Crop commodities are listed in
ascending order from the value in 1961. The black line displays the mean value across crop commodities.

33

Supplementary Figure 3B. Change in the total number of countries reporting each crop commodity, per crop, for calories, protein, fat and
weight.

Chapter 2

34

Supplementary Figure 4. Slope of the change in relative abundance probability of crop commodities in contribution to (A) calories, (B)
protein, (C) fat, and (D) weight in national per capita food supplies from 1961-2009. Change in relative abundance probability was analyzed
using a generalized linear mixed model with a binomial error distribution, “year” and “crop” as fixed effects and “country” as a random
effect. Bars represent the slopes (+/- 95% confidence interval) of the predicted values derived from the model for each crop species.
Commodities contributing minimally to protein/fat (i.e. <1 gram of protein/fat per 100 g of the item) were not included in (B) and (C) here.

Increasing homogeneity in global food supplies

35

Supplementary Figure 5. Comparison of absolute abundance trends for crop commodities reported in contribution to (A) calories, (B) protein,
(C) fat, and (D) weight in global aggregate food supply data versus mean national food supply data, 1961-2009. In order to derive these
values, the absolute abundance of the crop in 1961 was subtracted from the 2009 value. For country-level changes, the between-years
difference was first calculated for each country and then averaged across all countries (+/- 95% CI). Global aggregate abundance is a single
global value and therefore no estimate of error is possible.

Chapter 2

36

Supplementary Figure 6. Change in the total number of crop commodities reported in contribution to national food supplies, per country, for calories,
protein, fat and weight, 1961-2009.

Increasing homogeneity in global food supplies

37

Supplementary Figure 7. Change in evenness in contributing crop commodities in national food supplies using Pielou’s evenness index, per country, for
calories, protein, fat and weight, 1961-2009.

Chapter 2

38

Supplementary Figure 8. Change in the proportion of the most abundant crop commodity in national food supplies, per country, for calories, protein, fat
and weight, 1961-2009.

Increasing homogeneity in global food supplies

39

Supplementary Figure 9. Per-country contribution to global homogenization, as measured by the change in similarity of the national food supply crop
commodity composition in comparison to the global mean composition (centroid), for calories, protein, fat and weight, 1961-2009.

Chapter 2

40

Countries
for
Calories
148
138

116
69
28
15

9

9
3
73
42
17

15
14

Commodity

Wheat (Triticum aestivum L., T. turgidum L. (or
Triticum durum Desf.))

Rice (Oryza sativa L., O. glaberrima Steud.)

Maize (Zea mays L.)
Barley (Hordeum vulgare L.)
Sorghum (Sorghum bicolor (L.) Moench)
Millets- Pearl millet (Pennisetum glaucum (L.) R.
Br.), Finger millet (Eleusine coracana (L.)
Gaertn., T eff (Eragrostis tef (Zucc.) Trotter))

Cereals, other- Fonio (Digitaria exilis (Kippist)
Stapf)

Rye (Secale cereale L.)
Oats (Avena sativa L.)
Potatoes (Solanum tuberosum L.)
Cassava (Manihot esculenta Crantz)
Roots, other- Taro (Colocasia esculenta (L.)
Schott), Yautia (Xanthosoma sagittifolium (L.)
Schott)

Sweet potatoes (Ipomoea batatas (L.) Poir.)
Yams (Dioscorea rotundata Poir, D. cayennensis
Lam, D. alata L.)
6
12

7
2
51
21
16

6

92
16
29
13

106

146

Countries
for
Protein

2
2

0
0
0
2
0

1

71
0
16
8

29

84

Countries
for Fat

24
15

2
0
110
52
27

6

83
103
23
14

93

144

Countries
for Food
Weight

T he countries in which this commodity is important are in West Africa, the West Indies and the
Pacific. We counted here the most common major species in these regions (International
Institute for Tropical Agriculture 2009).

We counted taro (Colocasia esculenta (L.) Schott) and yautia (Xanthosoma sagittifolium (L.)
Schott), the specific crops listed under this general commodity, as the commodity is listed as
important in the countries in which each is produced in the greatest quantity (e.g. Nigeria,
Ghana, and Cameroon for taro; and Peru, Dominica, and T rinidad and Tobago for yautia) (FAO
2010a). Both of these crops were listed as important in Prescott-Allen & Prescott-Allen (1990).

Countries listing the commodity as important are from West, Central, and East Africa, as well as
India and Nepal, therefore we counted both pearl millet (Pennisetum glaucum (L.) R. Br.) and
finger millet (Eleusine coracana (L.) Gaertn.) (Rachie 1975). Teff (Eragrostis tef (Zucc.)
T rotter) was also counted as the commodity is important in Ethiopia PDR (Zeid et al. 2011).
Fonio (Digitaria exilis (Kippist) Stapf, is listed under the Cereals, other commodity (FAO 2013)
and was therefore not counted here.
T his commodity encompasses Popcorn, Buckwheat, Quinoa, Fonio, Triticale, Canary seed, and a
variety of non-specific mixed grain and flour commodities (FAO 2013). We counted fonio as the
commodity is significant in Guinea (Dendy & Dobraszczyk 2001). The rest of the specific crops
included in this general commodityare not likely to be important in listed countries. Instead,
listed countries indicate that the general commodity is likely important due to the inclusion of
mixed cereals and flours (probably wheat and other cereals).

Commodities grouped included Maize and Maize Germ Oil.
Commodities grouped included Barley and Beer.

We counted both Asian rice (Oryza sativa L.) as well as African rice (O. glaberrima Steud), as
all West African countries in which this rice is consumed list the commodity as important, and
due to the increasing adoption of Asian-African hybrid (NERICA) rices in the same region
(Linares 2002). Commodities grouped included Rice (Milled Equivalent) and Ricebran Oil.

We counted the major species of bread wheat (Triticum aestivum L.) and pasta wheat (T.
turgidum L. or T. durum Desf.).

Comments

Supplementary Table 1. Relative importance of crop commodities worldwide, as measured by the number of countries in which each specific crop
commodity counted within the top 90% of national per capita food supply for that variable (total countries = 152) in the mostcurrent year (2009).

Increasing homogeneity in global food supplies

41

42
Countries
for
Calories
149

24

50

35
5
87
72

42
36
50
29

Commodity

Sugar- Sugar cane (Saccharum officinarum L.),
Sugar beet (Beta vulgaris L.)

Sweeteners, other

Pulses, Other; Cowpea (Vigna unguiculata (L.)
Walp.), Chickpea (Cicer arietinum L.), Lentil
(Lens culinaris Medik.), Faba bean (Vicia faba L.),
Pigeonpea (Cajanus cajan (L.) Millsp.)

Beans (Phaseolus vulgaris L., P. lunatus L.)

Peas (Pisum sativum L.)
Soybean (Glycine max (L.) Merr.)
Palm oil (Elaeis guineensis Jacq., E. oleifera
(Kunth) Cortés)

Groundnut (Arachis hypogaea L.)
Coconuts (Cocos nucifera L.)
Sunflower (Helianthus annuus L.)
Oilcrops, other; Shea nut (Vitellaria paradoxa C.
F. Gaertn.)
34
8
0
2

22
24
0

49

73

0

0

Countries
for
Protein

63
56
69
74

0
113
90

3

3

0

0

Countries
for Fat

8
23
2
1

0
16
12

16

12

9

137

Countries
for Food
Weight

Commodities grouped included Groundnut Oil and Groundnuts (Shelled Eq).
Commodities grouped included Coconut Oil and Coconuts - Incl Copra.
Commodities grouped included Sunflowerseed and Sunflowerseed Oil.
Commodities grouped included Oilcrops, Other and Oilcrops Oil, Other. The commodity
encompasses Karite Nuts (Sheanuts), Castor oil, Tung Nuts, Jojoba, Safflower, Poppy,
Melonseed, T allowtree, Kapok, Linseed, Stillingia, Cocoa Butter and Hempseed, along with a
number of non-specific oil commodities (FAO 2013). From the countries listed it was not
feasible to identify particular crop species. We followed Prescott-Allen & Prescott-Allen (1990)
in listing one crop- Shea nut (Vitellaria paradoxa C. F. Gaertn.), as it was listed as important in
Ghana, Mali, and Nigeria in their data, and these countries also list the commodity as important
in the current data. Melonseed, also listed by Prescott-Allen & Prescott-Allen (1990), was not
counted here, but the species is counted in Fruits, Other.

Commodities grouped included Soyabean Oil and Soyabeans.
Commodities grouped included Palm Oil, Palmkernel Oil, and Palmkernels. Given that the
spread of countries in which the commodity is important includes both African and Neotropical
countries, we listed both oil palm species (Elaeis guineensis Jacq., E. oleifera (Kunth) Cortés).

T he commodity includes sugar maple (Acer saccharum Marshall) as well as a number of nonspecific commodities (e.g. Fructose, Maltose, Sugar crops nes, Sugar nes, and Glucose and
Dextrose) (FAO 2013). As both Canada and the United States list the commodity as important, it
is possible that sugar maple should be considered, but the dominant presence of general
processed sweetener categories cause sufficient uncertainty such that we did not count the
species.
T his general commodity encompasses Broad beans, Chickpeas, Cowpeas, Pigeonpeas, Lentils,
Bambara beans, Vetches, and Lupins, along with a few non-specific commodities (FAO 2013).
Given the importance of this commodity throughout the Mediterranean, West and South Asia
and Africa, it is probable that at least Cow peas (Vigna unguiculata (L.) Walp.), Chick peas
(Cicer arietinum L.), Broad beans (Vicia faba L.), Pigeon peas (Cajanus cajan (L.) Millsp.), and
Lentils (Lens culinaris Medik.) can be counted here. Prescott-Allen & Prescott-Allen (1990)
listed each of these crops as important.
We counted the two most cultivated bean species, Common bean (Phaseolus vulgaris L.) and
Lima bean (P. lunatus L.) (Belay 2006).

Commodities grouped included Sugar (Raw Equivalent), Sugar Beet, Sugar Cane, and Sugar,
Non-Centrifugal. As the raw equivalent sugar commodity may derive from either species (FAO
2013), we counted both crops (Saccharum officinarum L. and Beta vulgaris L.).

Comments

Chapter 2

3

20

22
24
9
7
75

38

45

Olives (Olea europaea L.)
Rape & mustard (Brassica napus L., B. rapa L.)

Sesame (Sesamum indicum L.)
Cottonseed oil (Gossypium hirsutum L., G.
herbaceum L.)

Vegetables, other- Cabbage and other brassicas
(Brassica oleracea L.), Carrot (Daucus carota L.),
Cucumber (Cucumis sativus L.), Pumpkin and
squash (Cucurbita maxima Duchesne, Cucurbita
moschata Duchesne, Cucurbita pepo L.), Lettuce
(Lactuca sativa L.), Eggplant (Solanum melongena
L.), Garlic (Allium sativum L.), Spinach (Spinacia
oleracea L.), Artichoke (Cynara cardunculus L.)

Fruits, other- Watermelon (Citrullus lanatus
(T hunb.) Matsum. & Nakai), Avocado (Persea
americana Mill.), Mango (Mangifera indica L.),
Melon (Cucumis melo L.), Papaya (Carica papaya
L.), Pear (Pyrus communis L.), Peach (Prunus
persica (L.) Batsch), Plum (Prunus domestica L.),
Fig (Ficus carica L.), Strawberry (Fragaria ×
ananassa (Weston) Duchesne), Apricot (Prunus
armeniaca L.), Cherry (Prunus avium (L.) L.),
Currant (Ribes nigrum L. or Ribes rubrum L.)
Bananas & plantains (Musa acuminata Colla,
Musa balbisiana Colla)

90

6
0

0
0

8

19

T reenuts; Almond (Prunus dulcis (Mill.) D. A.
Webb), Walnut (Juglans regia L.), Pistachio
(Pistacia vera L.), Hazelnut (Corylus avellana L.),
Brazil nut (Bertholletia excelsa Humb. & Bonpl.)

Countries
for
Protein

Countries
for
Calories

Commodity

3

7

1

20
23

29
29

34

Countries
for Fat

78

127

149

1
0

5
2

2

Countries
for Food
Weight

Commodities grouped included Bananas and Plantains. As banana and plantain varieties are
either of these species or a hybrid, we counted both species (Musa acuminata Colla and Musa
balbisiana Colla) here.

T he commodity encompasses Cabbages, Cauliflowers, Broccoli, and other brassicas, Artichokes,
Asparagus, Lettuce and chicory, Spinach, Cassava leaves, Pumpkins, squash and gourds,
Cucumbers and gherkins, Eggplants (aubergines), Chillies and peppers green, Onions (inc.
shallots) green, Garlic, Leeks other alliaceous veg, Beans green, Peas green, String beans,
Carrots and turnips, Okra, Sweet Corn, Mushrooms, Chicory roots, Carobs, Watermelons, Other
melons (inc.cantaloupes), and Coffee Substitute, as well as a number of non-specific group
commodities (FAO 2013). The commodity is extremely important worldwide, especially for
weight, and it is likely that a large number of the crops contained within the general commodity
individually contribute significantly in at least one country. We followed Prescott-Allen &
Prescott-Allen (1990), counting 11 of the crop species now contained here. Chillies were
included instead in Pimento; and watermelons instead in Fruits, Other.
T he commodity encompasses Pears, Quinces, Apricots, Cherries, Peaches and nectarines, Plums
and sloes, Strawberries, Raspberries, Gooseberries, Currants, Blueberries, Cranberries,
Watermelons, Other melons (inc. cantaloupes), Figs, Mangoes, mangosteens, guavas, Avocados,
Persimmons, Cashewapple, Kiwi fruit, and Papayas, as well as a number of non-specific group
commodities (FAO 2013). The commodity is very important worldwide, especially for weight,
and it is likely that a large number of the crops contained within the general commodity
individually contribute significantly in at least one country. We followed Prescott-Allen &
Prescott-Allen (1990), listing 13 of the crop species now contained here.

Commodities grouped included Sesameseed and Sesameseed Oil.
We counted the common cotton species used for oil (Gossypium hirsutum L. and G. herbaceum
L.), which may be mixed in oil products.

T he commodity encompasses Brazil nuts, Cashew, Chestnuts, Almonds, Walnuts, Pistachios,
Kolanuts, Hazelnuts, Arecanuts, Nuts nes, and Prepared Nuts (Exc. Groundnuts). Countries
listing the commodity as important are spread worldwide, but concentrated in Europe, the
Mediterranean, and Western Asia. We counted the two crops with the largest production
worldwide- almond (Prunus dulcis (Mill.) D. A. Webb) and walnut (Juglans regia L.) (FAO
2010a). We also counted pistachio (Pistacia vera L.), hazelnut (Corylus avellana L.), and Brazil
nut (Bertholletia excelsa Humb. & Bonpl.), which were also listed as important in Prescott-Allen
& Prescott-Allen (1990). T axonomy of almond is per GRIN T axonomy (USDA, ARS, National
Genetic Resources Program 2011).
Commodities grouped included Olive and Olive Oil.
We counted two major oil species (Brassica napus L. and B. rapa L.), both common ingredients
in canola oil. Other species of mustard may also be important in individual regions.

Comments

Increasing homogeneity in global food supplies

43

Countries
for
Calories
2
4
31
1
0
0
7
2

0
0

11
2

0
2

Commodity

T omatoes (Solanum lycopersicum Lam.)
Oranges & Mandarines- Oranges (Citrus
sinensis (L.) Osbeck)

44

Grapes (Vitis vinifera L.)

Apples (Malus domestica Borkh.)
Onions (Allium cepa L.)

Pineapples (Ananas comosus (L.) Merr.)
Dates (Phoenix dactylifera L.)
Grapefruit (Citrus paradisi Macfad.)

Lemons & limes (Citrus limon (L.) Burm. f.,
C. aurantiifolia (Christm.) Swingle)

Citrus, other

Cocoa beans (Theobroma cacao L.)
Pimento- Chillies and peppers (Capsicum
annuum L., C. chinense Jacq.), Pimento
(Pimenta dioica (L.) Merr.)

Coffee (Coffea arabica L., C. canephora
Pierre ex A. Froehner)

Spices, other- Ginger (Zingiber officinale
Roscoe), Nutmeg/mace (Myristica fragrans
Houtt.), Cardamom (Amomum subulatum
Roxb.)
1

4

1
1

0

0

0
0
1

0
0

0

11
2

Countries
for
Protein

2

0

19
4

0

0

0
0
0

0
0

0

0
0

Countries
for Fat

0

2

0
0

6

8

13
5
8

54
44

48

77
82

Countries
for Food
Weight

T he commodity includes Vanilla; Cinnamon (canella); Nutmeg, mace and cardamoms; Anise,
badian, fennel, corian.; Ginger; and Spices, nes (FAO 2013). Countries where the commodity is
important include Grenada, and Nepal. Neither of these countries are significant producers of
vanilla, cinnamon, anise, badian, fennel, or coriander. Ginger (Zingiber officinale Roscoe) and
Cardamom (Amomum subulatum Roxb) are produced in significant quantifies in Nepal, and
Nutmeg/mace (Myristica fragrans Houtt) in Grenada (FAO 2010a) and were therefore included
here.

T he commodity is composed of Chillies and peppers, dry (FAO 2013), yet historically counted
pimenta spice/allspice (Pimenta dioica (L.) Merr.) (FAO 1984), which is not currently listed
under Spices, other (FAO 2013), and is therefore likely to still be included here. We counted both
chillies/peppers and allspice as Ethiopia and Peru list the commodity as important and are
significant consumers of allspice. Both the common pepper and bonnet pepper were included.
Capsicum frutescens L. is considered here as a synonym of C. annuum L. (The Plant List 2010).
We counted Coffea arabica L. and C. canephora Pierre ex A. Froehner, the major coffee species
globally. C. arabica is the source of 70% of coffee worldwide (Davis et al. 2012).

T he commodity includes Citrus fruit nes, Citrus juice single strength, and Citrus juice
concentrated (FAO 2013). Given the spread of countries in which the commodity is important it
is not possible to determine whether additional citrus species are represented here, or (more
likely), the commodity is important due to juice of common citrus species.

T he commodity includes Grapefruit (inc. pomelos), Juice of Grapefruit, and Grapefruit juice,
concentrated (FAO 2013). We counted only grapefruit (Citrus paradisi Macfad.), as countries in
which the commodity is important are generally from the West Indies. No South or Southeast
Asian countries, where pomelo (Citrus maxima (Burm.) Merr.) is consumed, list the commodity
as important.
We counted lemons (Citrus limon (L.) Burm. f.) and the major species of lime (C. aurantiifolia
(Christm.) Swingle), a crop which also includes a number of minor species and hybrids.

T he commodity includes Onions dry. Green onions/shallots are included under the Vegetables,
Other commodity (FAO 2013).

T he commodity includes Oranges, Orange juice single strength, Orange juice concentrated,
T angerines, mandarins, clem., and T angerine Juice (FAO 2013). We counted only orange (Citrus
sinensis (L.) Osbeck).
Commodities grouped included Grapes and Wine. We counted only Vitis vinifera L., source of
over 90% of grape and wine production worldwide (Winkler et al. 1974).

Comments

Chapter 2

0

35
10

13
152

T ea- Yerba Maté (Ilex paraguariensis A. St.-Hil.)

Beverages, alcoholic

Beverages, fermented

Miscellaneous

Animal Commodities total

152

11

3

0

2

Countries
for
Protein

152

0

0

0

0

Countries
for Fat

152

0

22

10

1

Countries
for Food
Weight

T he commodity includes all animal food commodities listed in FAOSTAT Food Balance Sheets,
a sum of all commodities in Meat; Offals; Animal Fats; Eggs; Milk, excluding butter; Fish,
Seafood; and Aquatic Products, other categories (FAO 2013).

T his plant general commodity is not further defined and therefore it is not possible to determine
specific crops.

T he commodity includes Wheat Fermented Beverage, Rice Fermented Beverages, Beer of
Maize, Beer of Millet, Beer of Sorghum, and Cider Etc (FAO 2013). These species are
represented in crop commodities above, and no species are counted here.

T he commodity includes Tea (Camellia sinensis (L.) Kuntze) and Maté (Ilex paraguariensis A.
St.-Hil.) (FAO 2013). As Paraguay and Uruguay, the main producers and consumers of Maté,
are the two countries that list the commodity as important, it is very likely that they refer to
Maté.
T he commodity is composed of Bever. Dist.Alc (FAO 2013). As it is not possible to determine
crops, no species are listed.

Comments

For any particular variable (e.g. calories) for each country, the contributing crop commodities were listed in descending order of importance until 90% of food supply
was reached. These commodities were then counted as significant to the food supply of the country, and the total number of countries finding each crop significant was
summed to derive a quantitative measure of importance worldwide. Commodities are listed in order of importance defined as sum of country count across the four
variables, within cereals, starchy roots, sugarcrops, pulses, oilcrops and nuts, fruits and vegetables, spices and stimulants , and alcoholic beverages categories. Taxonomy
followed The Plant List (2010).

Countries
for
Calories

Commodity

Increasing homogeneity in global food supplies

45

Chapter 2
REFERENCES
Adler, C.J., Dobney, K., Weyrich, L.S., Kaidonis, J., Walker, A.W., Haak, W., et al. (2013). Sequencing
ancient calcified dental plaque shows changes in oral microbiota with dietary shifts of the Neolithic
and Industrial Revolutions. Nat. Genetics 45, 450-455.
Alston, J.M., Beddow, J.M., Pardey, P.G. (2009). Agricultural research, productivity, and food prices in
the long run. Science 325 (5945), 1209-1210.
Belay, G. (2006). Cereals and Pulses (Wageningen: PROTA Foundation).
Bonin, C.L., Tracy, B.F. (2012). Diversity influences forage yield and stability in perennial prairie plant
mixtures. Agr. Ecosyst. Environ. 162, 1-7.
Bonneuil, C., Goffaux, R., Bonnin, I., Montalent, P., Hamon, C., Balfourier, F., et al. (2012). A new
integrative indicator to assess crop genetic diversity. Ecol. Indic. 23, 280-289.
Burke, M.B., Lobell, D.B., Guarino, L. (2009). Shifts in African crop climates by 2050, and the
implications for crop improvement and genetic resources conservation. Global Environ. Chang. 19
(3), 317-325.
Cabell, J.F., Oelofse, M. (2012). An indicator framework for assessing agroecosystem resilience. Ecol.
Soc. 17 (1), 18.
Cadotte, M.W., Dinnage, R., Tilman, D. (2012). Phylogenetic diversity promotes ecosystem stability.
Ecology 93 (sp8), S223-S233.
Cordell, D., Drangert, J-O., White, S. (2009). The story of phosphorus: global food security and food
for thought. Global Environ. Chang. 19 (2), 292–305.
Davis, A.P., Gole, T.W., Baena, S., Moat, J. (2012). The impact of climate change on indigenous arabica
coffee (Coffea arabica): predicting future trends and identifying priorities. PLoS One 7 (11), e47981.
Davis, A.S., Hill, J.D., Chase, C.A., Johanns, A.M., Liebman, M. (2012). Increasing cropping system
diversity balances productivity, profitability and environmental health. PLoS One 7(10), e47149.
Davis, D.R., Epp, M.D., Riordan, H.D. (2004). Changes in USDA food composition data for 43 garden
crops, 1950 to 1999. J. Am. Coll. Nutr. 23 (6), 669-682.
Dendy, D.A.V., Dobraszczyk, B.J. (2001). Cereals and Cereal Products: Chemistry and Technology
(Gaithersburg: Aspen Publishers, Inc.).
Doughty, J. (1979). Dangers of reducing the range of food choice in developing countries. Ecol. Food
Nutr. 8, 275-283.
Dwivedi, S., Sahrawat, K., Upadhyaya, H., Ortiz, R. (2013). Food, nutrition and agrobiodiversity under
global climate change. Adv. Agron. 120, 1-128.
Fader, M., Gerten, D., Krause, M., Lucht, W., Cramer, W. (2013). Spatial decoupling of agricultural
production and consumption: quantifying dependences of countries on food imports due to domestic
land and water constraints. Environ. Res. Lett. 8 (1), 014046.
FAO (Food and Agriculture Organization of the United Nations). (1984). Food Balance Sheets 19791981 Average (Rome: Food and Agriculture Organization of the United Nations).
FAO (Food and Agriculture Organization of the United Nations). (2001). Food Balance Sheets: A
handbook (Rome: Food and Agriculture Organization of the United Nations).
FAO (Food and Agriculture Organization of the United Nations). (2002). The International Treaty on
Plant Genetic Resources for Food and Agriculture (Rome: Food and Agriculture Organization of the
United Nations).
FAO (Food and Agriculture Organization of the United Nations). (2010a). FAOSTAT, Production.
http://faostat3.fao.org.
FAO (Food and Agriculture Organization of the United Nations). (2010b). Second Report on the State
of the World’s Plant Genetic Resources for Food and Agriculture (Rome: Food and Agriculture
46

Increasing homogeneity in global food supplies
Organization of the United Nations).
FAO (Food and Agriculture Organization of the United Nations). (2013). FAOSTAT, Methods and
standards, classification of crops primary equivalent. http://faostat3.fao.org/mes/classifications/E.
Foley, J.A., Defries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., et al. (2005). Global
consequences of land use. Science 309 (5734), 570-574.
Fowler, C., Hodgkin,T, (2004). Plant genetic resources for food and agriculture: assessing global
availability. Annu. Rev. Env. Resour. 29 (1), 143-179.
Frei, M., Becker, K. (2004). Agro-biodiversity in subsistence-oriented farming systems in a Philippine
upland region: nutritional considerations. Biodivers. Conserv. 13 (8), 1591-1610.
Gamfeldt, L., Snall, T., Bagchi, R., Jonsson, M., Gustafsson, L., Kjellander, P., et al. (2013). Higher
levels of multiple ecosystem services are found in forests with more tree species. Nat. Commun. 4,
1340.
Gepts, P. (2006). Plant genetic resources conservation and utilization: the accomplishments and future
of a societal insurance policy. Crop Sci. 46 (5), 2278-2292.
Graham, R.D., Welsh, R.M., Saunders, D.A., Ortiz-Monasterio, I., Bouis, H.E., Bonierbale, M., et al.
(2007), Nutritious subsistence food systems. Adv. Agron. 92, 1-74.
Hajjar, R., Jarvis, D.I., Gemmill-Herren, B. (2008). The utility of crop genetic diversity in maintaining
ecosystem services. Agr. Ecosyst. Environ. 123 (4), 261-270.
Hammer, K., Khoshbakht, K. (2005). Towards a “red list” for crop plant species. Genet. Resour. Crop
Ev. 52 (3), 249-265.
Harlan, J.R. (1975). Crops and Man (Madison: American Society of Agronomy).
Hawkesworth, S., Dangour, A.D., Johnston, D., Lock, K., Poole, N., Rushton, J., et al. (2010). Feeding
the world healthily: the challenge of measuring the effects of agriculture on health. Philos. T. R. Soc.
B. 365 (1554), 3083–3097.
Hooper, D.U., Adair, E.C., Cardinale, B.J., Byres, J.E.K., Hungate, B.A., Matulich, K.L., et al. (2012).
A global synthesis reveals biodiversity loss as a major driver of ecosystem change. Nature 486, 105108.
Hooper, D.U., Chapin III, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavore, S., et al. (2005). Effects of
biodiversity on ecosystem functioning: a consensus of current knowledge. Ecol. Monog. 75, 3-35.
International Institute for Tropical Agriculture. (2009). Yam. http://www.iita.org/yam.
Jarvis, A., Ramirez-Villegas, J., Herrera-Campo, B.V., Navarro-Racines, C. (2012). Is cassava the
answer to African climate change adaptation? Trop. Plant Biol. 5 (1), 9-29.
Kastner, T., Rivas, M.J.I., Koch, W., Nonhebel, S. (2012). Global changes in diets and the consequences
for land requirements for food. Proc. Natl. Acad. Sci. 109 (18), 6868-6872.
Kearney, J. (2010). Food consumption trends and drivers. Philos. T. R. Soc. B. 365 (1554), 2793-2807.
Kennedy, G., Islam, O., Eyzaguirre, P., Kennedy, S. (2005). Field testing of plant genetic diversity
indicators for nutrition surveys: rice-based diet of rural Bangladesh as a model. J. Food Compos.
Anal. 18 (4), 255-268.
Kirwan, L., Luscher, A., Sebastia, M.T., Finn, J.A., Collins, R.P., Porqueddu, C., et al. (2007). Evenness
drives consistent diversity effects in intensive grassland systems across 28 European sites. J. Ecol.
95 (3), 530-539.
Kremen, C., Miles, A. (2012). Ecosystem services in biologically diversified versus conventional
farming systems: benefits, externalities, and trade-offs. Ecol. Soc. 17 (4), 40.
Linares, O.F. (2002). African rice (Oryza glaberrima): history and future potential. Proc. Natl. Acad.
Sci. 99 (25), 16360-16365.
Lobell, D.B., Burke, M.B., Tebaldi, C., Mastrandrea, M.D., Falcon, W.P., Naylor, R.L. (2008).
Prioritizing climate change adaptation needs for food security in 2030. Science 319 (5863), 607-610.
47

Chapter 2
Lozupone, C.A., Stombaugh, J.I., Gordon, J.I., Jansson, J.K., Knight, R. (2012). Diversity, stability and
resilience of the human gut microbiota. Nature 489 (7415), 220-230.
McCouch, S., Baute, G.J., Bradeen, J., Bramel, P., Bretting, P.K., Buckler, E., et al. (2013). Agriculture:
feeding the future. Nature 499(7456), 23-24.
McCouch, S.R., McNally, K.L., Wang, W., Sackville Hamilton, R. (2012). Genomics of gene banks: A
case study in rice. Am. J. Bot. 99 (2), 407-423.
Mijatović, D., van Oudenhoven, F., Eyzaguirre, P., Hodgkin, T. (2013). The role of agricultural
biodiversity in strengthening resilience to climate change: Towards an analytical framework. Int. J.
Agric. Sustain. 11 (2), 95-107.
National Research Council, (1972). Genetic Vulnerability of Crops (Washington D.C.: National
Academy of Sciences).
Naylor, R.L., Falcon, W.P., Goodman, R.M., Jahn, M.M., Sengooba, T., Tefera, H., et al. (2004).
Biotechnology in the developing world: a case for increased investments in orphan crops. Food Pol.
29 (1), 15-44.
Negin, J., Remans, R., Karuti, S., Fanzo, J.C. (2009). Integrating a broader notion of food security and
gender empowerment into the African green revolution. Food Sec. 1 (3), 351-360.
Padulosi, S., Dulloo, E. (2012). “Towards a viable system for monitoring agrobiodiversity on-farm: a
proposed new approach for Red Listing of cultivated plant species,” in: On-farm Conservation of
Neglected and Underutilized Species: Status, Trends and Novel Approaches to Cope with Climate
Change, 14–16 June, 2011, eds. S. Padulosi, N. Bergamini, T. Lawrence T. (Friedrichsdorf,
Frankfurt: Proceedings of the International Conference), 171-199.
Perrier, X., De Langhe, E., Donohue, M., Lentfer, C., Vrydaghs, L., Bakry, F., et al. (2011).
Multidisciplinary perspectives on banana (Musa spp.) domestication. Proc. Natl. Acad. Sci. 108 (28),
11311-11318.
Picasso, V,D., Brummer, E.C., Liebman, M., Dixon, P.M., Wilsey, B.J. (2008). Crop species diversity
affects productivity and weed suppression in perennial polycultures under two management
strategies. Crop Sci. 48 (1), 331-342.
Pingali, P. (2007). Westernization of Asian diets and the transformation of food systems: implications
for research and policy. Food Pol. 32 (3), 281-298.
Popkin, B.M. (2006). Technology, transport, globalization and the nutrition transition food policy. Food
Pol. 31 (6), 554-569.
Pradhan, P., Reusser, D.E., Kropp, J.P. (2013). Embodied greenhouse gas emissions in diets. PLoS One
8 (5), e62228.
Prescott-Allen, R., Prescott-Allen, C. (1990). How many plants feed the world? Conserv. Biol. 4 (4),
365-374.
Rachie, K.O. (1975). The Millets: Importance, Utilization and Outlook. Research Report. (Hyderabad:
International Crops Research Institute for the Semi-Arid Tropics).
Ray, D.K., Mueller, N.D., West, P.C., Foley, J.A. (2013). Yield trends are insufficient to double global
crop production by 2050. PLoS One 8 (6), e66428.
Remans, R., Flynn, D.F.B., DeClerk, G., Diru, W., Fanzo, J., Gaynor, K., et al. (2011). Assessing
nutritional diversity of cropping systems in African villages. PLoS One 6 (6), e21235.
Ruel, M.T. (2003). Operationalizing dietary diversity: a review of measurement issues and research
priorities. J. Nutr. 133 (11), 3911S-3926S.
Stehfest, E., Bouwman, L., van Vuuren, D.P., den Elzen, M.G.J., Eickhout, B., Kabat, P. (2009). Climate
benefits of changing diet. Climatic Change 95 (1-2), 83-102.
The Plant List. (2010). Version 1. http://www.theplantlist.org/.
USDA, ARS, National Genetic Resources Program. (2011). Germplasm Resources Information
48

Increasing homogeneity in global food supplies
Network (GRIN). http://www.ars-grin.gov/cgi-bin/npgs/html/index.pl.
van de Wouw, M., Kik, C., van Hintum, T., van Treuren, R., Visser, B. (2009). Genetic erosion in crops:
concept, research results and challenges. Plant Genetic Resources 8, 1-15.
van de Wouw, M., van Hintum, T., Kik, C., van Treuren, R., Visser, B. (2010). Genetic diversity trends
in twentieth century crop cultivars: a meta analysis. Theor. Appl. Genet. 120 (6), 1241-1252.
Wheeler, T., von Braun, J. (2013). Climate change impacts on global food security. Science 341 (6145),
508-513.
Winkler, A.J., Cook, J.A., Kliewer, W.M., Lider, L.A. (1974). General Viticulture (Berkeley: University
of California Press).
Zeid, M., Echenique, V., Diaz, M., Pessino, S., Sorrells, M. (2011). “Chapter 8: Eragrostis,” in Wild
Crop Relatives: Genomic and Breeding Resources, Millets and Grasses, ed. C. Kole (Berlin
Heidelberg: Springer-Verlag).
Zhang, Y., Chen, H.Y.H., Reich, P.B. (2012). Forest productivity increases with evenness, species
richness and trait variation: a global meta-analysis. J. Ecol. 100 (3), 742-749.
Zhu, Y., Chen, H., Fan, J., Wang, Y., Li, Y., Chen, J., et al. (2000). Genetic diversity and disease control
in rice. Nature 406 (6797), 718-722.

49

CHAPTER 3
An inventory of crop wild relatives of the United States
Colin K. Khoury1,2, Stephanie Greene3, John Wiersema4, Nigel Maxted5, Andy Jarvis1,6, and
Paul C. Struik2
International Center for Tropical Agriculture (CIAT), Km 17, Recta Cali-Palmira, Apartado Aéreo
6713, Cali, Colombia.
2
Centre for Crop Systems Analysis, Wageningen University, Droevendaalsesteeg 1, 6708 PB
Wageningen, The Netherlands.
3
United States Department of Agriculture, Agricultural Research Service, Plant Germplasm
Introduction and Testing, 24106 North Bunn Rd., Prosser, Washington, 99350, USA.
4
United States Department of Agriculture, Agricultural Research Service, National Germplasm
Research Laboratory, Building 003, BARC-West, Beltsville, Maryland 20705-2350, USA.
5
School of Biosciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK.
6
CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Km 17,
Recta Cali-Palmira, Apartado Aéreo 6713, Cali, Colombia.
1

Published in Crop Science 53 (4), 1496. doi: 10.2135/cropsci2012.10.0585

Chapter 3
Abstract
The use of crop wild relatives (CWR) in breeding is likely to continue to intensify as utilization
techniques improve and crop adaptation to climate change becomes more pressing. Significant gaps
remain in the conservation of these genetic resources. As a first step toward a national strategy for the
conservation of CWR, we present an inventory of taxa occurring in the United States, with suggested
prioritization of species based upon potential value in crop improvement. We listed 4,600 taxa from 985
genera and 194 plant families, including CWR of potential value via breeding as well as wild species of
direct use for food, forage, medicine, herb, ornamental, and/or environmental restoration purposes. US
CWR are related to a broad range of important food, forage and feed, medicinal, ornamental, and
industrial crops. Some potentially valuable species are threatened in the wild, including relatives of
sunflower (Helianthus annuus L.), walnut (Juglans regia L.), pepo squash (Cucurbita pepo L.), wild
rice (Zizania L.), raspberry (Rubus idaeus L.), and plum (Prunus salicina Lindl.), and few accessions of
such taxa are currently conserved ex situ. We prioritize 821 taxa from 69 genera primarily related to
major food crops, particularly the approximately 285 native taxa from 30 genera that are most closely
related to such crops. Both the urgent collection for ex situ conservation and the management of such
taxa in protected areas are warranted, necessitating partnerships between concerned organizations,
aligned with regional and global initiatives to conserve and provide access to CWR diversity.
Keywords: Crop wild relatives, Plant genetic resources, Wild utilized species, National inventory

52

An inventory of crop wild relatives of the United States
INTRODUCTION
Nearly 40 years ago Jack Harlan outlined the major factors explaining the extent of use of wild
relatives of crops (CWR) in plant breeding. His list included the degree of domestication of the
crop, the perceived genetic vulnerability of the crop, the availability of CWR for utilization, the
degree of difficulty in using CWR in breeding, and the economic conditions and disposition of
breeders toward their use (Harlan 1976).
Utilization of CWR has steadily increased over the past decades, providing improved pest and
disease resistance, tolerance to abiotic stresses, increased yield, novel cytoplasms, and quality
traits to banana (Musa acuminata Colla), barley (Hordeum vulgare L.), bean (Phaseolus
vulgaris L.), cassava (Manihot esculenta Crantz), chickpea (Cicer arietinum L.), corn (Zea
mays L.), lettuce (Lactuca sativa L.), oat (Avena sativa L.), millet [Pennisetum glaucum (L.) R.
Br.], potato (Solanum tuberosum L.), rice (Oryza sativa L.), sugarcane (Saccharum officinarum
L.), sunflower, tomato (Solanum lycopersicum L.), and wheat (Triticum aestivum L.), among
others (Iltis 1988, Xiao et al. 1996, Gur & Zamir 2004, Hajjar & Hodgkin 2007, McCouch et
al. 2007, Maxted et al. 2012a). Advancements in breeding, particularly through novel molecular
approaches, have increased the efficiency of the use of wild germplasm substantially (Tanksley
& McCouch 1997, Zamir 2001, Ford-Lloyd et al. 2011, Volk & Richards 2011). Research for
adaptation to future climates is likely to increase the exploitation of the variation represented in
CWR (Ortiz et al. 2008, Guarino & Lobell 2011).
Despite substantial efforts over these years, the most significant bottleneck in utilization
identified by Harlan - the availability of CWR for research and breeding - continues to be of
concern, with substantial genetic variation yet to be conserved, and financial and political
constraints still to be resolved (Fowler & Hodgkin 2004, FAO 2010).
An estimated one out of five plant species is threatened worldwide by habitat loss or
modification, agricultural modernization, pollution, over-exploitation, invasive species, and/or
climate change (Brummitt & Bachman 2010), and nearly 30% of the native flora of the United
States is of conservation concern (CPC 2012). CWR are not exempt from these pressures
(Wilkes 2007, Jarvis et al. 2008, Bilz et al. 2011, Ureta et al. 2011).
The urgent collection and subsequent storage in ex situ facilities, where these genetic resources
53

Chapter 3
can be made available for research and breeding is thus warranted. The complementary
protection of CWR in situ is necessary to support the ongoing evolution of CWR populations
(Maxted et al. 1997, Meilleur & Hodgkin 2004, Heywood 2008, Maxted & Kell 2009) in both
wild areas and traditional agricultural systems (Rawal 1975, GSPC 2002, Zizumbo-Villarreal
et al. 2005).
It is becoming increasingly feasible to formulate comprehensive strategies for the conservation
of CWR diversity due to advancements in understanding the taxonomic relationships between
crops and their wild relatives (Andersson & de Vicente 2010, Wiersema et al. 2012), improved
availability of data on the distribution of these taxa (e.g. FNA 2008a, GBIF 2012) and increased
power of distribution modelling and conservation analysis (Hijmans & Spooner 2001, Jarvis et
al. 2005, Ramírez-Villegas et al. 2010, Parra-Quijano et al. 2011).
The starting point for CWR conservation planning typically involves the creation of a checklist
of included CWR taxon names, to which ancillary data (e.g. eco-geographic information,
conservation status, utilization potential, etc.) is added to generate an inventory for the target
area of research (Maxted et al. 2008). National inventories of CWR have been published for a
growing list of nations, particularly in Europe (Magos Brehm et al. 2007, Maxted et al. 2012b),
and targeted subsequent conservation efforts have been made in over 40 countries worldwide
(Meilleur & Hodgkin 2004). On the global level, a specialist group is active in listing CWR of
conservation concern (International Union for Conservation of Nature, Species Survival
Commission 2008), an initiative is underway to document, collect, conserve ex situ, and prebreed the CWR of major food and forage crops (Guarino and Lobell 2011), and progress is
being made in planning for an integrated system of genetic reserves for the CWR of highest
priority worldwide (Maxted & Kell 2009).
As a primary step in the process toward a national strategy for the conservation of CWR, we
report on an inventory of the CWR flora occurring in the US, and a prioritization of these taxa
based upon their potential value in agricultural crop research.
The United States Context
More than 20,000 species of plants, or about 7% of the world's flora, are native or naturalized
in North America north of Mexico (FNA 2008b), but the region has not been considered a major
54

An inventory of crop wild relatives of the United States
center of crop plant diversity (Vavilov 1926). Those indigenous species that were domesticated
before European contact in Eastern North America include pepo squash (Cucurbita pepo L.),
sunflower (Helianthus annuus L.), marsh-elder (Iva annua L.), and chenopod (Chenopodium
berlandieri Moq.) (Smith 2006), and in the arid southwestern US include Sonoran panic
(Panicum hirticaule J. Presl var. hirticaule) (Nabhan 1985) and devil’s-claw (Proboscidea
parviflora (Wooton) Wooton & Standl. subsp. parviflora) (Bretting & Nabhan 1986).
Blueberry (Vaccinium section Cyanococcus) and cranberry (Vaccinium section Oxycoccus)
(Ballington 2001), blackberry (Rubus fruticosus L., sensu lato and hybrids) (Finn, 2001), and
pecan [Carya illinoinensis (Wangenh.) K. Koch] (Flack 1970) may be counted as more recent
domestications. The number of CWR native to the US may thus at first glance be estimated to
be fairly small.
Three factors increase the number of potentially valuable CWR significantly. The relatives of
a complex of Mesoamerican crop species, including corn, a number of bean (Phaseolus L.) and
squash species, chili pepper (Capsicum L.), American cotton (Gossypium hirsutum L.), and
tobacco (Nicotiana rustica L.) (Nabhan 1985, Zizumbo-Villarreal & Colunga-GarcíaMarín
2010) are distributed in the southern regions of the US (USA-ARS National Genetic Resources
Program 2012). Second, a number of crops domesticated in other temperate regions of the world
are congeneric with species occurring in the US, for example strawberry (Fragaria L.)
(Hummer et al. 2011) and hops (Humulus L.) (Peredo et al. 2010). Finally, approximately 4,000
plant species have been introduced to the region since the Colombian Exchange (Guo et al.
2009), including weedy relatives of crop plants.
Several well-documented examples of use of native CWR in breeding exist. North American
wild grape (Vitis L.) germplasm proved critical in providing resistance to phylloxera
(Phylloxera vitifoliae Fitch) as a rootstock in European grape (Vitis vinifera L.) production in
the late 1800s, and these stocks continue to provide the basis for protection worldwide (Gale
2003). Genes for resistance to a range of diseases and pests, including rust (Puccinia helianthi
Schwein.), downy mildew [Plasmopara halstedii (Farl.) Berl. & De Toni], powdery mildew
[Golovinomyces cichoracearum (DC.) V.P. Heluta], broomrape (Orobanche cumana Wallr.),
sclerotinia head and stalk rot [Sclerotinia sclerotiorum (Lib.) de Bary], and sunflower moth
(Homoeosoma electellum Hulst), have been identified in native sunflowers and successfully
transferred into cultivars (Seiler & Gulya 2004).
55

Chapter 3
Several US government entities support activities focused on CWR conservation. The US
Department of Agriculture, Agricultural Research Service (ARS), National Plant Germplasm
System (NPGS) published in situ conservation guidelines for US CWR (Plant Germplasm
Operations Committee 1999), and recently formed a Subcommittee on CWR within the Plant
Germplasm Operations Committee (Plant Germplasm Operations Committee 2010). The NPGS
Germplasm Resources Information Network (GRIN) Taxonomy Section is preparing a database
of crop gene pools listing CWR based on an evaluation of breeding and crossability studies
(Wiersema et al. 2012).
Over the past decade the NPGS has supported 61 explorations for the US CWR of food, forage,
woody landscape and ornamental crop plants (K. Williams, personal communication 2012).
Genetic reserves for the wild relatives of grape (Pavek et al. 2001), chili pepper (Nabhan 1990),
and cranberry (K. Hummer, personal communication 2010) have been established. Explorations
regarding possible locations and feasibility of protected areas for CWR of pecan, potato, sweet
pea (Lathyrus L.), and edible alliums (Allium L.) were also completed (Plant Germplasm
Operations Committee 1999).
The US Forest Service (USFS) actively maintains a number of CWR populations in National
Forests (USFS 2010) and CWR are informally conserved across the US on these and other
public lands. The Bureau of Land Management, in partnership with the Millennium Seed Bank
of the Royal Botanic Gardens, Kew and local organizations across the country, has collected
CWR accessions within the “Seeds of Success” Program (Bureau of Land Management 2012).
Building upon a partnership between ARS and USFS to collaborate on the establishment of in
situ reserves for US CWR, the agencies are in the process of developing a coordinated strategy
for CWR management (L. Stritch, personal communication 2012).
MATERIALS AND METHODS
National Inventory of Crop Wild Relatives
In order to broadly cover the wild plant species occurring in the US that have potential value in
crop research, we compiled an Inventory of CWR that may be utilized in crop breeding as well
as wild utilized species (WUS) directly used for food, forage, medicine, herb, ornamental,
and/or environmental restoration purposes. Very few WUS are the central focus of plant
breeding programs, although some taxa may be semi-domesticated, and WUS may have a high
56

An inventory of crop wild relatives of the United States
potential for crop development.
For listed CWR we aimed to include the full range of taxa with the potential to contribute to
crop improvement, including both those species where gene exchange with the crop is relatively
straightforward, and more distant relatives requiring advanced techniques in order to produce
viable hybrid progeny. Our starting point for defining CWR followed Maxted et al. (2006, p.
2680): “A crop wild relative is a wild plant taxon that has an indirect use derived from its
relatively close genetic relationship to a crop; this relationship is defined in terms of the CWR
belonging to Gene Pools (GPs) 1 or 2, or taxon groups (TGs) 1 to 4 of the crop.” The definition
classifies taxa based on whether they occur in Harlan and de Wet’s (1971) crop GP 1 or 2,
encompassing closely related taxa that are relatively easy to cross using conventional methods
in breeding programs. If data from interspecific hybridization or genetic relatedness studies are
unavailable, Maxted et al. (2006) proposed a classification system based on taxonomic groups
equating to rank in relation to the crop species. Data on gene pool and taxon group concepts for
available crops was gathered from GRIN taxonomy (Wiersema et al. 2012) and from the
‘Harlan and de Wet Crop Wild Relative Inventory’ (Vincent et al. 2013).
A growing number of crops have benefitted from traits introgressed from distant gene pools
(Ballington 2001, Frese et al. 2001, Bradshaw et al. 2006, Mallikarjuna et al. 2006, Abberton
2007, Rygulla et al. 2007, Chuda & Adamus 2009, Mii 2009), and as breeding techniques
improve, taxa from such gene pools are increasingly likely to be of interest to crop improvement
programs. Such species are additionally useful for taxonomic and evolutionary research. We
therefore broadened our CWR definition to include species in the tertiary gene pool. In some
crops these may include taxa from related genera [e.g., Tripsacum L. for maize, Aegilops L.
and Amblyopyrum Eig for wheat (Wiersema et al. 2012)].
CWR and WUS taxa occurring in the US were compiled from the GRIN World Economic
Plants database, based on Wiersema and Leόn (1999), completed volumes of the Flora of North
America (FNA 2008a), McGuffin (2000), and the Native Seed Network Database (Native Seed
Network 2010). Both native and introduced taxa were included. In addition to listing taxa to the
infraspecific level, information on origin status, number of accessions (available plus
unavailable) in GRIN, noxious weed status, associated crop, crop gene pool and associated crop
use was obtained. Taxa with multiple uses were listed first by their primary use and thereafter
57

Chapter 3
by subsequent uses. For taxa with uses both as CWR and WUS, utilization as a genetic resource
was prioritized over direct uses. Additional data on occurrence and weed status was gathered
from the PLANTS Database (USDA NRCS 2010). The threat status of taxa was recorded from
NatureServe (NatureServe 2009) and the IUCN Red List of Threatened Species (IUCN 2012).
Taxonomic verification was performed via the ‘Taxonomic Name Resolution Service’ (Boyle
et al. 2013) and GRIN taxonomy (USDA-ARS National Genetic Resources Program 2012), the
latter of which served as the final authority. The Inventory was reviewed by NPGS curators,
members of the NPGS Crop Germplasm Committees, and USDA Agricultural Research
Service (ARS) crop experts, who submitted revisions and proposed additional taxa.
Prioritization of the Crop Wild Relatives of Agricultural Crops
When using an inclusive definition for crops together with a broad definition of CWR, national
and regional studies have resulted in the majority of flora being listed as CWR. Approximately
80% of the species in the European and Mediterranean floras were listed as CWR in an
inventory for that region (Kell et al. 2008), and 77% of the flora of Portugal similarly listed as
CWR or WUS (Magos Brehm et al. 2007). Given the extent of potentially useful plant taxa in
the US and general resource constraints in conservation and research funding, we further
prioritized taxa within the Inventory in order to focus subsequent conservation efforts on
species with the greatest potential impact on crop research.
We first compiled and prioritized crop species based on their contribution to global agricultural
production and food security, with the assumption that important crops are the focus of the most
active breeding programs with experience in the utilization of exotic germplasm. The crop list
was collated from the Food and Agriculture Organization of the United Nations statistical
database (FAOSTAT) production and food supply data (FAO 2011), Annex 1 of the
International Treaty on Plant Genetic Resources for Food and Agriculture (ITPGR) (FAO
2002), Appendix 2 (“Important Food Crops”) of Groombridge & Jenkins (2002), and PrescottAllen & Prescott-Allen 1990). These sources emphasize food crops, but some include fiber,
forage, and industrial crops.
Listed crops were further prioritized based upon the number of sources and importance
attributed within the sources: 1) Major crops (Priority 1) were assigned to crops listed in more
58

An inventory of crop wild relatives of the United States
than one source, among the specific crop commodities listed in FAOSTAT and in PrescottAllen & Prescott-Allen (1990), and all crops in Annex 1 of the ITPGR and in Groombridge &
Jenkins (2002); 2) Minor and non-food crops (Priority 2) were assigned to crops listed in only
one source, plus crops recorded in FAOSTAT general commodities, as well as Annex 1 forages
and the “Brassica complex” crops other than those in the genus Brassica L. itself. The resulting
compilation of the world’s major crops included 242 crops and 268 genera (101 crops and 119
genera in Priority 1, and 141 crops and 149 genera in Priority 2) (Supplementary Table 1). The
list included all agricultural crops recorded in FAOSTAT as important to production or to food
supply in the US (FAO 2011).
The National Inventory was compared to the world’s major crops list and crosschecked with
GRIN taxonomy to derive a list of CWR taxa occurring in the US that are within the gene pools
of priority crops. The resulting list of priority CWR was reviewed by NPGS curators, members
of the NPGS Crop Germplasm Committees, and ARS crop experts.
Priority 1 CWR taxa were further categorized based upon perceived value and ease of utility in
breeding programs. Native plant species were assigned a higher priority, as they have a long
history of adaptation in contrast to naturalized species, which may have limited variation due
to the founder effect (Amsellem et al. 2001). Closely related (defined here as within GP 1-2 or
TG 1-3) native taxa, plus any additional taxa recorded in the literature or identified by
researchers as potentially useful in crop breeding, were assigned the highest priority (Priority
1A). Distantly related and/or non-native taxa that were not specifically identified by the
research community as a target for utilization were listed as Priority 1B. The few gene pools
(notably blackberry and raspberry in Rubus L.) for which relatedness information was not
available were categorized based upon occurrence status.
RESULTS AND DISCUSSION
The National Inventory contains 4,596 taxa, representing 3,912 species from 985 genera and
194 plant families. CWR in the inventory are represented by 2,495 taxa representing 1,905
species from 160 genera and 56 families. WUS are represented by 2,101 taxa from 2,007
species from 833 genera and 182 families. Major families are listed in Table 1. The Inventory
is available online at: http://www.ars-grin.gov/misc/tax/ (accessed 1 Oct. 2012). Future plans
are to fully integrate these data into GRIN so that detailed information is available for each
59

Chapter 3
Table 1. Major families of US crop wild relatives (CWR) and wild utilized species (WUS).
CWR (genetic resource of)
WUS (direct use for)
Families
No. of genera No. of taxa Families
No. of genera No. of taxa
Fabaceae
19
693
Asteraceae
97
216
Poaceae
31
448
Poaceae
71
156
Asteraceae
12
182
Rosaceae
29
135
Rosaceae
6
163
Fabaceae
52
106
Amaranthaceae
6
137
Ericaceae
23
79
Brassicaceae
11
67
Pinaceae
6
71
Grossulariaceae
1
67
Cyperaceae
6
55
Solanaceae
4
63
Salicaceae
2
48
Cyperaceae
1
48
Ranunculaceae
14
45
Amaryllidaceae
1
47
Boraginaceae
17
43
Other (46)
68
580
Other (172)
516
1147

taxon and the Inventory can be queried by taxonomy, priority level, and geographic distribution.
CWR identified in the US are primarily related to food crops (Table 2). These include genetic
resources for globally important crops such as strawberry, sunflower, sweet potato [Ipomoea
batatas (L.) Lam.], bean, bean, stone fruits (Prunus L.), and grape, as well as regionally
important crops such as pecan, yerba maté (Ilex paraguariensis A. St.-Hil.), quinoa
(Chenopodium quinoa Willd.), and cranberry. Forage and feed CWR are also well represented,
particularly legumes (Trifolium L., Lupinus L., Lotus L., and Astragalus L.) and grasses
(Agrostis L., Bromus L., Festuca L., and Poa L.). Significant genetic resources of medicinal
crops include Echinacea (Echinacea Moench), tobacco (Nicotiana L.), St. John’s-wort
(Hypericum L.), poppy (Papaver L.), and artemisia (Artemisia L). CWR of ornamental crops
include Rosa L., Coreopsis L., Lilium L., Phlox L., Rudbeckia L. and Penstemon Schmidel.
CWR of material and industrial crops include relatives of flax (Linum L.), cotton (Gossypium
L.), and jatropha (Jatropha L).
The WUS species listed as distributed in the US are primarily utilized for ornamental,
restoration, and medicinal purposes. A number of food species of cultural and economic
significance are also identified, such as wild rice (Zizania L.), sugar maple (Acer saccharum
Marshall), and pawpaw (Asimina triloba (L.) Dunal).
Non-native species comprise 12.3% of the Inventory (14.7% of CWR and 9.6% of WUS), and
212 taxa (4.6% of total) are Federal and/or State listed noxious weeds. Genetic resource
priorities for listed taxa should take into account weed regulations and conservation priorities
60

An inventory of crop wild relatives of the United States
Table 2. Uses of US crop wild relatives (CWR) and wild utilized species (WUS).
Use
Food

Forage and feed
Medicinal
Ornamental
Material and
industrial
Herb
Soil conservation
Use
Ornamental

Restoration

Medicinal

Food
Forage and feed
Forestry
Material and
industrial
Soil conservation
Turf

Major families (and No. of taxa)

No. of
families

No. of
genera

No. of
taxa

47

103

1472

5

29

709

7

10

150

5

6

99

10

15

57

1
1

2
1

5
3

# of
families
149

# of
genera
440

# of
taxa
812

83

336

755

82

139

180

Rosaceae (72), Poaceae (6), Sapindaceae
(5)
Poaceae (72), Fabaceae (11), Araceae (5)
Pinaceae (49), Cupressaceae (7),
Betulaceae (7)
Fagaceae (9), Cupressaceae (6)

17

26

112

9
16

61
28

99
87

26

37

52

Fabaceae (3)
Poaceae (1)

1
1

3
1

3
1

CWR (genetic resource of)
Poaceae (304), Fabaceae (168), Rosaceae
(136), Amaranthaceae (95), Asteraceae
(90), Grossulariaceae (67), Brassicaceae
(61), Solanaceae (54), Cyperaceae (48),
Amaryllidaceae (47), Convolvulaceae
(40), Ericaceae (39), Asparagaceae (35)
Fabaceae (521), Poaceae (142),
Amaranthaceae (42)
Asteraceae (63), Hypericaceae (44),
Euphorbiaceae (17)
Plantaginaceae (39), Rosaceae (27),
Asteraceae (19)
Linaceae (21), Malvaceae (13),
Asteraceae (8)
Lamiaceae (5)
Fabaceae (3)
WUS (direct use for)
Major families and number of taxa
Ericaceae (65), Asteraceae (49), Fabaceae
(43), Rosaceae (41), Salicaceae (30),
Oleaceae (22), Poaceae (20),
Papaveraceae (17), Ranunculaceae (16)
Asteraceae (151), Poaceae (53),
Cyperaceae (48), Fabaceae (31),
Boraginaceae (29), Ranunculaceae (26),
Rhamnaceae (26), Apiaceae (23),
Onagraceae (22), Liliaceae (22),
Polygonaceae (18), Rosaceae (17)
Asteraceae (12), Lamiaceae (11),
Fabaceae (9)

for the species affected by these invasive plants.
Several introduced CWR taxa were identified as containing genetic resources of interest to
61

Chapter 3
breeders, including relatives of beet (Beta vulgaris L.) (L. Panella, personal communication
2011), lettuce (B. Hellier, personal communication 2011), and clover (W. Williams, personal
communication 1997). Recent alfalfa (Medicago sativa L.) breeding efforts have used coldadapted naturalized populations of Medicago sativa L. subsp. falcata (L.) Arcang from South
Dakota to develop rangeland varieties adapted to the Intermountain West (Peel et al. 2009).
These examples demonstrate the potential value of novel variation in naturalized species
(Bossdorf et al. 2005), which should not be neglected in inventories of useful plant species.
The threat status of 3,512 (76.4%) taxa in the Inventory has been recorded in NatureServe.
Eight (0.2%) taxa were assessed as known or presumed extinct in the wild; 115 (2.5%) as
globally critically imperiled and 111 (2.4%) as imperiled; 337 (7.3%) vulnerable; 798 (17.4%)
apparently secure; and 2,143 (46.6%) globally secure. Of the included taxa, the International
Union for Conservation of Nature Red List of Threatened Species assessed 16 taxa as extinct,
endangered, or vulnerable (IUCN 2012). Sixty-two taxa are listed as endangered under the US
Endangered Species Act (Endangered Species Act of 1973, 16 U.S.C. Sec 1531), 10 taxa as
threatened, and 11 taxa as candidates for listing (NatureServe 2009). Among the taxonomic
groups with the largest number of threatened taxa are members of the family Fabaceae,
particularly within the genera Astragalus, Lotus, Lupinus, and Trifolium (see Supplementary
Table 2 for a full listing of extinct, imperiled, endangered, and threatened taxa).
Threatened species with known or high potential value in crop breeding include the wild walnut
Juglans hindsii (Jeps.) R. E. Sm., which is used as a primary rootstock for English walnut
(Juglans regia L.) worldwide and is critically imperiled in its native California habitat (Phillips
& Meilleur 1998), and close relatives of sunflower, squash, cotton, gooseberry (Ribes uvacrispa L.), raspberry, onion (Allium cepa L.), wild rice, and plum (Table 3). The conservation
of these genetic resources should be prioritized urgently.
Over 96,000 gene bank accessions of 2,800 taxa listed in the Inventory are recorded in GRIN,
but a large proportion of this material is cultivated germplasm conspecific with wild taxa such
as American cotton (G. hirsutum) and chili pepper (C. annuum). Germplasm of Inventory taxa
listed as wild total 48,780 accessions, and that listed as both wild and from the US total 20,739
accessions from 2,135 taxa. These accessions are distributed unevenly within the Inventory,
with 51.8% of accessions comprised of 14 genera (Fraxinus L., Helianthus L., Pinus L., Avena
62

An inventory of crop wild relatives of the United States
Table 3. Threatened US crop wild relatives of major crops.
Taxon

Priority
category
1

US
ESA2

Nature
Serve3

No. of
accessions4

P1B
LE
G1
0
Allium munzii (Ownbey & Aase ex Traub) McNeal
T2 to 3 0
Allium obtusum Lemmon var. conspicuum Mortola & McNeal P1B
P1B
G2 to 3 0
Allium scilloides Douglas ex S. Watson
P1A
LE
G1
0
Cucurbita okeechobeensis (Small) L. H. Bailey
P1A
T2
2
Fragaria chiloensis (L.) Mill. subsp. sandwicensis (Decne.)
Staudt
P1A
G2 to 3 0
Gossypium tomentosum Nutt. ex Seem.
P1B
G1 to 2 2
Helianthus carnosus Small
P1A
G2*
10
Helianthus niveus (Benth.) Brandegee subsp. tephrodes (A.
Gray) Heiser
GX*
0
Helianthus nuttallii Torr. & A. Gray subsp. parishii (A. Gray) P1B
Heiser
P1A
G2
1
Helianthus paradoxus Heiser
P1B
G2
6
Helianthus smithii Heiser
P1B
G1
2
Helianthus verticillatus Small
P1B
G2 to 4 0
Hordeum arizonicum Covas
P1B
G2
1
Ipomoea microdactyla Griseb.
P1A
G1
16
Juglans hindsii (Jeps.) R. E. Sm.
P1B
G2
3
Lathyrus grimesii Barneby
P1B
G2
1
Lathyrus holochlorus (Piper) C. L. Hitchc.
P1B
G2 to 3 0
Leymus pacificus (Gould) D. R. Dewey
P1B
LE
G1
0
Manihot walkerae Croizat
P1B
G2
0
Phaseolus texensis A. Delgado & W. R. Carr
P1B
G1
0
Prunus eremophila Prigge
P1A
GX
0
Prunus murrayana E. J. Palmer
P1A
G2 to 3 3
Ribes binominatum A. Heller
P1B
LT
G1
3
Ribes echinellum (Coville) Rehder
P1B
G2
2
Ribes erythrocarpum Coville & Leiberg
P1A
GX
0
Rubus aliceae L. H. Bailey
P1A
G2 to 3 13
Rubus hawaiensis A. Gray
P1A
G2
1
Rubus macraei A. Gray
P1B
LE
G1
0
Solanum incompletum Dunal
P1B
C
G2
0
Solanum nelsonii Dunal
P1B
LE
G1
0
Solanum sandwicense Hook. & Arn.
P1B
G2
0
Solanum wallacei (A. Gray) Parish
P1A
G2
0
Tripsacum floridanum Porter ex Vasey
P1A
G2 to 4 0
Vanilla mexicana Mill.
P1B
LE
G1
0
Vicia menziesii Spreng.
P1B
G1
1
Vicia ocalensis R. K. Godfrey & Kral
P1A
LE
G1
0
Zizania texana Hitchc.
1
P1A = native taxa closely related to important crop plants; P1B = non-native and/or distantly related to important
crop plants. 2Taxa listed as endangered (LE), threatened (LT), or as a candidate for listing (C) under the US
Endangered Species Act of 1973, 16 U.S.C. Sec1531) (US ESA), and/or listed as known or presumed extinct in
the wild (GX), globally critically imperiled (G1), and globally imperiled (G2) in NatureServe (NatureServe 2009).
Note (G3) is categorized as globally vulnerable, and (G4) as apparently secure; and T denotes global listing at the
infraspecific level. *denotes threat assessment at the species level. 4Number of accessions denotes NPGS
germplasm listed as wild and collected in the US.

63

Chapter 3
L., Elymus L., Vaccinium L., Rubus L., Vitis L., Fragaria L., Lupinus L., Achnatherum P.
Beauv., Ribes L., Solanum L., and Trifolium L.). Of the 232 taxa listed as endangered,
threatened, or as a candidate for listing under the Endangered Species Act (Endangered Species
Act of 1973, 16 U.S.C. Sec 1531) as well as taxa listed as known or presumed extinct in the
wild, globally critically imperiled, and imperiled in NatureServe (2009), only 157 accessions
listed as wild and collected in the US are conserved in the NPGS.
Priority Crop Wild Relatives of Agricultural Crops
Priority species occurring in the US total 2,256 taxa within 176 genera. These include 821 taxa
from 69 genera related to 63 major agricultural crops (Priority 1 gene pools) (Table 4), and
1,435 taxa from 107 genera of minor food crops, forages, and other crops (Priority 2)
(Supplementary Table 3). Within Priority 1, 285 closely related, native taxa from 30 genera are
listed 1A, and 536 distantly related and/or non-native taxa within 57 genera in 1B.
A number of iconic US edible WUS were given priority for conservation considerations. Within
Priority 1, these include sugar maple, wild rice, and American chestnut (Castanea dentata
(Marshall) Borkh.), plants that have held important stature in traditional regional diets. In
addition, 148 food, medicinal, and ornamental WUS from 22 genera were assigned to Priority
2.
At least 17 major crops have benefited from traits contributed by 55 Priority 1 CWR taxa (Table
5). As this count is limited to published references, it is likely an underestimate of the taxa
occurring in the US that have been successfully utilized in breeding programs.
The NPGS conserves 8,195 accessions of wild native Priority 1 taxa (3,952 Priority 1A and
4,243 Priority 1B), and 4,020 accessions of Priority 2 taxa. Of Priority 1 CWR, 366 (44.6%)
taxa are completely absent from ex situ collections and another 307 (37.4%) are represented by
less than 10 germplasm accessions.
Far from possessing few genetic resources, the United States contains a wealth of native and
introduced plants related to a broad range of crops. Significant gaps in the ex situ collections of
these taxa remain to be filled, and a number of potentially valuable species are threatened in
the wild. Meanwhile, new populations of some species are still being discovered (Kraft et al.
64

An inventory of crop wild relatives of the United States
Table 4. Priority US crop wild relatives and wild utilized species.
Genus

Associated
crop name

Priority (and no.
of taxa)1

Genus

Associated crop
name

Priority (and no.
of taxa)1

P1A (6)

Juglans

Aegilops

Wheat

P1B (5)

Lactuca

Lettuce

P1A (10) and P1B (1)

Agropyron

Wheat

Allium

Asparagus

Onion, garlic, and
leek
Breadfruit and
jackfruit
Asparagus

P1B (2)

Lathyrus

Grasspea

P1B (31)

P1A (4) and P1B (43)

Leymus

Wheat

P1B (17)

P1B (1)

Malus

Apple

P1B (4)

Avena

Oat

P1B (3)

Manihot

Cassava

P1B (4)

P1B (3)

Medicago

Alfalfa

P1B (10)

Beta

P1A (4)

Nicotiana

Tobacco

P1B (9)

P1B (5)

Olea

Olive

P1B (1)

P1A (2)

Pennisetum

Pearl millet

P1B (10)

Carica

Beet, sugar beet
and chard
Cabbage, rapeseed,
etc.
Chili pepper and
sweet pepper
Papaya

P1B (1)

Persea

Avocado

P1B (3)

Carthamus

Safflower

P1B (1)

Phaseolus

Bean

P1A (4) and P1B (15)

Carya

Pecan

P1A (9) and P1B (4)

Piper

Pepper

P1B (1)

Castanea

Chestnut

P1A (3) and P1B (2)

Pistacia

Pistachio

P1A (1)

Cinnamomum

Cinnamon

P1B (1)

Prunus

Stone fruits

Cocos

Coconut

P1B (1)

Psathyrostachys

Wheat

P1A (17) and P1B
(26)
P1B (2)

Colocasia

Taro

P1B (1)

Pseudoroegneria

Wheat

P1B (1)

Corylus

Hazelnut

P1A (3)

Psidium

Guava

P1A (1) and P1B (1)

Cucumis

Melon

P1B (4)

Pyrus

Pear

P1B (1)

Cucurbita

P1A (8) and P1B (2)

Ribes

Cynara

Pumpkin and
squash
Artichoke

P1B (3)

Rubus

Daucus

Carrot

P1B (2)

Saccharum

Currant and
gooseberry
Raspberry and
blackberry
Sugar cane

P1A (27) and P1B
(40)
P1A (58) and P1B
(10)
P1B (9)

Dioscorea

Yam

P1B (3)

Solanum

Potato and tomato

P1A (1) and P1B (38)

Diospyros

Persimmon

P1A (2)

Sorghum

Sorghum

P1B (4)

Diplotaxis

P1B (2)

Syzygium

Clove

P1B (2)

Elymus

Cabbage, rapeseed,
etc.
Wheat

P1B (43)

Thinopyrum

Wheat

P1B (2)

Ficus

Fig

P1A (1) and P1B (3)

Tripsacum

Maize

P1A (4)

Foeniculum

Fennel

P1B (1)

Vaccinium

Fragaria

Strawberry

Vanilla

P1A (23) and P1B
(16)
P1A (2)

Gossypium

Cotton

P1A (11) and P1B
(10)
P1A (3)

Blueberry and
Cranberry
Vanilla

Vernicia

Tung nut

P1B (1)

Helianthus

Sunflower

Vicia

Fava bean and vetch

P1B (14)

Hordeum

Barley

P1A (23) and P1B
(49)
P1B (18)

Vigna

P1B (2)

Ilex

Maté

P1A (6) and P1B (15)

Vitis

Cowpea, bambara
groundnut, etc.
Grape

Illicium

Star-anise

P1A (1)

Zizania

Wild rice

P1A (6)

Ipomoea

Sweet potato

P1A (9) and P1B (31)

Acer

Artocarpus

Brassica
Capsicum

Sugar maple

Walnut

P1A (7) and P1B (2)

P1A (29)

P1A = native taxa closely related to important crop plants; P1B = non-native and/or distantly related to important
crop plants. Origin status from the Germplasm Resources Information Network (USDA-ARS National Genetic
Resources Program 2012). Contributing gene pool and taxon group concepts from Wiersema et al. (2012), and the
‘Harlan and de Wet Crop Wild Relative Inventory’ (Vincent et al. 2013).
1

65

Chapter 3
Table 5. Confirmed use of priority crop wild relatives.
Taxon
Trait1
Aegilops cylindrica Host
Salt tolerancea,b
Aegilops geniculata Roth
Hessian fly resistancec
Aegilops tauschii Coss.
Rust resistance, wheat soil-borne mosaic virus, and wheat
spindle-streak mosaic virusd; drought tolerancee; yellow rust and
leaf rust resistancef; glutenins improvementg; agronomic traits and
yield improvementh; hessian fly resistancei; karnal buntj; waterlogging tolerancek; and sprouting suppressionl
Allium fistulosum L.
Avena sterilis L.
Corylus americana Marshall
Helianthus anomalus S. F.
Blake
Helianthus argophyllus Torr.
& A. Gray

Disease resistancem
Crown rust resistancen and yield improvemento
Eastern filbert blight resistancep
Fertility restorationq

Helianthus bolanderi A. Gray
Helianthus debilis Nutt.
Helianthus deserticola Heiser
Helianthus divaricatus L.
Helianthus giganteus L.
Helianthus grosseserratus M.
Martens
Helianthus hirsutus Raf.
Helianthus maximilianii
Schrad.
Helianthus neglectus Heiser
Helianthus paradoxus Heiser
Helianthus pauciflorus Nutt.
Helianthus petiolaris Nutt.

Genetic stocku and fertility restorationq
Powdery mildew resistancev and fertility restorationq
Downy mildew resistancew
Broomrape resistancex
Fertility restorationy and cytoplasmic male sterilityz
Broomrape resistancex

Helianthus praecox Engelm. &
A. Gray
Helianthus resinosus Small
Helianthus strumosus L.
Helianthus tuberosus L.

Downy mildew resistancer,s, disease resistancet, and fertility
restorationq

Fertility restoration aay
Broomrape resistancex and cytoplasmic male sterilityz
Fertility restorationq
Salt toleranceab and fertility restorationq
Cytoplasmic male sterilityac and sclerotinia resistancead
Verticillium resistanceae, disease resistancet, cytoplasmic male
sterilityaf, sunflower moth resistanceag, and fertility restorationq
Downy mildew, rust, verticillium wilt and broomrape resistanceah;
fertility restorationq; and downy mildew resistancew
Fertility restorationaa
Fertility restorationy
Broomrape resistanceai, sunflower moth resistanceag, and fertility
restorationy

Hordeum bulbosum L.

Powdery mildew resistanceaj, mosaic virus resistanceak, septoria
resistanceal, and leaf rust resistanceam

Ipomoea trifida (Kunth) G.
Don
Juglans californica S. Watson
Juglans hindsii (Jepss) R. E.
Sm.
Juglans major (Torr.) A. Heller
and Juglans microcarpa
Berland
Juglans nigra L.
Lactuca serriola L.

Root knot nematode and root lesion nematode resistancean

66

Rootstockao
Rootstockap,ao
Rootstock for alkaline soilap
Anthracnose resistanceap and rootstockao
Downy mildew resistanceaq

An inventory of crop wild relatives of the United States
Taxon
Lactuca virosa L.
Malus fusca (Raf.) C. K.
Schneid.
Medicago sativa L. subsp.
falcata (L.) Arcang.
Pennisetum purpureum
Schumach.
Prunus andersonii A. Gray,
Prunus pumila L., Prunus
pumila L. var. besseyi (L. H.
Bailey) Gleason, and Prunus
rivularis Scheele
Pyrus calleryana Decne
Ribes nigrum L.
Ribes uva-crispa L.
Solanum stoloniferum Schltdl.
& Bouche

Trait1
Leaf aphid resistancear
Rootstockao

Tripsacum dactyloides (L.) L.

Corn leaf blight resistanceay and yield improvement and top firing
resistanceaf

Vitis acerifolia Raf., Vitis
aestivalis Michx., Vitis cinerea
(Engelm.) Engelm. ex
Millardet, Vitis cinerea var.
helleri (L. H. Bailey) M. O.
Moore, Vitis monticola
Buckley, Vitis mustangensis
Buckley, and Vitis vulpina L.
Vitis labrusca L.
Vitis riparia Michx. and Vitis
rupestris Scheele

Rootstockao

Winter hardinessas
Pest resistance, vigor and yieldat
Rootstockao

Rootstockao
Pest and disease resistanceau
Gall mite resistanceav
Late blight resistanceaw and potato Y virus resistanceax

Cold toleranceaz
Phylloxera vitifoliae resistanceaf and rootstockao

Published trait listing adapted from the ‘Harlan and de Wet Crop Wild Relative Inventory’ (Vincent et al.
2013). aFarooq et al. 1995, bWang et al. 2003, cEl Khlifi et al. 2004, dCox et al. 1995, eGororo et al. 2002, fMa et
al. 1995, gPena et al. 1995 hPestsova et al. 2006, iSuszkiw 2005, jVillareal et al. 1996, kVillareal et al. 2001, lXiuJin et al. 1997. mKhrustaleva and Kik 1998, nHoffman et al. 2006 , oTakeda & Frey 1976, pThompson et al.
1996, qSeiler 1991a, rHulke et al. 2010, sMiller & Gulya 1988, tJan et al. 2004, uJan 1992, vJan & Chandler
1988, wSeiler 1991b, xJan et al. 2002, ySeiler 2000, zWhelan & Dedio 1980, aaSeiler 1991c, abLexer et al. 2004, acJan
et al. 2006, adMiller & Gulya 1999, aeHoes et al. 1973, afPrescott-Allen & Prescott-Allen 1986, agRogers et al.
1984, ahHajjar & Hodgkin 2007, aiPutt 1978, ajPickering & Johnston 2005, akRuge-Wehling et al. 2006, alToubiaRahme et al. 2003, amZhang et al. 2001, anSakamoto 1976, aoUSDA-ARS National Genetic Resources Program
2012, apMcGranahan & Leslie 2009, aqHooftman et al. 2007, arEenink et al. 1982, asBarnes et al. 1977, atHanna
1997, auBarney & Hummer 2005, avBrennan 2008, awBradshaw et al. 2006, axRoss 1979, ayGoodman et al.
1987, azReisch & Pratt 1996.
1

2012). Crops that are nationally as well as globally important to food security could benefit
significantly from the long-term conservation and exploitation of these taxa. Following the
prioritization of such taxa based upon their potential use value, planning for conservation will
be facilitated through an analysis of the range of distribution of these taxa, and the subsequent
identification of hotspots of richness of CWR in the US as well as geographic and taxonomic
gaps in germplasm collections and in situ conservation.
67

Chapter 3
The focus on the gene pools of major agricultural crops during prioritization within the
Inventory resulted in a number of minor or locally important crops and WUS, forages and other
non-food crops holding secondary priority (Supplementary Table 3). Many of these taxa are
economically important and their native US genetic resources may have substantial use value.
The development and collation of information both on the utilization of these taxa in breeding
programs as well as the value of their associated crops will contribute significantly to their
potential for prioritization and subsequent conservation.
Given the considerable development pressures on wild plants in the US (Stein et al. 2000) and
projected increasing impacts from climate change (Loarie et al. 2009), both the urgent
collection for ex situ conservation and the management of taxa in conservation areas are
warranted. In order to achieve these goals for the diversity of prioritized taxa, broad partnerships
and networks between the federal, state, tribal and non-governmental organizations pursuing
conservation activities are needed. Because many of the taxa are distributed across national
borders and the genetic resources of such species are potentially valuable globally, such efforts
should be aligned with neighboring national strategies and with regional and global initiatives
to conserve and provide access to CWR diversity.

68

An inventory of crop wild relatives of the United States
SUPPLEMENTARY INFORMATION
Supplementary Table 1: Prioritization of agricultural crops worldwide.
Priority 1 Crops
Genus

Crop name

Genus

Crop name

Abelmoschus

Okra

Lablab

Lablab, hyacinth bean

Actinidia

Kiwi

Lactuca

Lettuce

Allium

Onion, garlic, leek

Lathyrus

Grass pea

Anacardium

Cashew

Lens

Lentil

Ananas

Pineapple

Linum

Flax, linseed

Arachis

Peanut

Malus

Apple

Artocarpus

Breadfruit, jackfruit

Mangifera

Mango

Asparagus

Asparagus

Manihot

Cassava

Avena

Oat

Medicago

Alfalfa

Bertholletia

Brazil nut

Musa, Ensete

Banana, plantain

Beta, Patellifolia

Beet, sugar beet, chard

Olea

Olive

Brassica, Diplotaxis

Cabbage, rapeseed, etc.

Oryza

Rice

Cajanus

Pigeonpea

Papaver

Poppy

Camellia

Tea

Pennisetum

Pearl millet

Capsicum

Chili pepper, sweet pepper

Persea

Avocado

Carica

Papaya, babaco

Phaseolus

Bean

Carthamus

Safflower

Phoenix

Date

Ceratonia

Carob

Pimenta

Pimento

Cicer

Chickpea

Pimpinella

Anise

Cicorium

Chicory

Piper

Pepper

Cinnamomum

Cinnamon

Pistacia

Pistachio

Citrullus

Watermelon

Pisum, Vavilovia

Pea

Citrus, Fortunella*, Poncirus*

Orange, grapefruit, lemon, etc.

Prunus

Stone fruits

Cocos

Coconut

Psidium

Guava

Coffea

Coffee

Pyrus

Pear

Colocasia

Taro

Ribes

Currant, gooseberry

Coriandrum

Coriander

Ricinus

Castor oil

Corylus

Hazelnut

Rubus

Raspberry, blackberry

Cucumis

Melon, cucumber

Saccharum

Sugar cane

Cucurbita

Pumpkin, squash

Secale

Rye

Cydonia

Quince

Sesamum

Sesame

Cynara

Artichoke

Sinapis

Mustard seed

Daucus, Tornabenea

Carrot

Solanum, Lycopersicon

Potato, tomato, eggplant

Dioscorea

Yam

Sorghum

Sorghum

Diospyros

Persimmon

Spinacia

Spinach

Elaeis

Oil palm

Syzygium

Clove

Elettaria

Cardamom

Theobroma

Cocoa bean

Eleusine

Finger millet

Trifolium

Clover

Fagopyrum

Buckwheat

Triticosecale

Triticale

69

Chapter 3
Ficus

Fig

Wheat

Fennel

Triticum, Aegilops,
Amblyopyrum, Dasypyrum,
Elymus, Leymus,
Psathyrostachys,
Pseudoroegneria, Thinopyrum
Vaccinium

Foeniculum
Fragaria

Strawberry

Vanilla

Vanilla

Garcinia

Mangosteen

Vernicia

Tung nut

Glycine

Soybean

Vicia

Fava bean, vetch

Gossypium

Cotton

Vigna

Helianthus

Sunflower

Vitellaria

Cowpea, bambara groundnut,
etc.
Karite nut, shea nut

Hordeum

Barley

Vitis

Grape

Ilex

Maté

Xanthosoma

Yautia, cocoyam

Illicium

Badian

Zea, Tripsacum

Maize, popcorn

Ipomoea

Sweet potato

Zingiber

Ginger

Juglans

Walnut

Blueberry, cranberry

Priority 2 Crops
Genus

Crop name

Genus

Crop name

Abroma

Indian flax

Lepidium

Cress

Abutilon

China jute

Lespedeza

Lespedeza

Achras

Sapodilla, chicle gum

Licania

Licania

Aframomum

Aframomum cardamom

Lolium

Lolium

Agrostis

Agrostis

Lotus

Lotus

Aleurites

Aleurites moluccana

Lupinus

Lupin

Alopecurus

Alopecurus

Lygeum

Alfa, esparto

Amaranthus

Quihuicha, Inca wheat

Macadamia

Macadamia nut

Andropogon

Andropogon

Majorana

Marjoram

Anethum

Dill seed

Mammea

Mammee

Annona

Cherimoya, custard apple

Manilkara

Balata

Anthriscus

Chervil

Maranta

Arrowroot

Apium

Celery

Melilotus

Melilotus

Arbutus

Strawberry tree

Mentha

Peppermint

Areca

Areca nut

Mespilus

Medlar

Arenga

Sugar palm

Metroxylon

Sago palm

Armoracia, Barbarea,
Camelina, Crambe, Eruca,
Isatis, Raphanobrassica,
Rorippa
Arracacoa

Brassica complex

Morus

Mulberry, loganberry

Arracacha

Myristica

Nutmeg, mace

Arrhenatherum

Arrhenatherum

Myrtus

Myrtleberry

Artemisia

Tarragon

Nasturtium

Watercress

Asimina

Pawpaw

Neoglaziovia

Caroa

Astrocaryum

Astrocaryum

Nephelium

Litchi, longan, rambutan

Atriplex

Atriplex

Nicotiana

Tobacco

Averrhoa

Carambola

Onobrychis

Onobrychis

Bambusa

Bamboo

Opuntia

Prickly pear

Bassia

Bassia

Ornithopus

Ornithopus

70

An inventory of crop wild relatives of the United States
Boehmeria

Ramie

Oxalis

Oca

Calocarpum

Sapote

Pachyrhizus

Jicama, yam bean

Canarium

Palachium

Gutta-percha

Canavalia

Pili nut, Java almond, Chinese
olives
jack bean

Panicum

Little millet, proso millet

Cannabis

Hemp

Parthenium

Guayule

Capparis

Caper

Paspalum

Kodo millet

Carapa

Carapa

Passiflora

Passion fruit

Caryocar

Butter nut

Pastinaca

Parsnip

Castanea

Chestnut

Perilla

Perilla

Ceiba

Kapok fruit

Petroselinum

Parsley

Chenopodium

Quinoa

Phalaris

Canary seed

Chrysanthemum

Pyrethrum

Phleum

Phleum

Chrysophyllum

Star apple

Phormium

New Zealand flax

Cochlearia

Horseradish

Poa

Poa

Coix

Adlay, Job's tears

Pongamia

Pongamia

Cola

Kola nut

Prosopis

Prosopis

Corchorus

Jute

Psophocarpus

Winged bean

Coronilla

Coronilla

Pueraria

Pueraria

Crataegus

Azarole

Punica

Pomegranate

Crocus

Saffron

Raphanus

Radish

Crotalaria

Sunn hemp

Rheum

Rhubarb

Croton

Croton

Rosa

Rose hips

Curcuma

Turmeric

Rumex

Sorrel

Cyamopsis

Guar bean

Salsola

Salsola

Cyperus

Chufa

Sambucus

Elderberry

Dactylis

Dactylis

Samuela

Palma ixtle

Dieva

Jelutong

Sansevieria

Bowstring hemp

Digitaria

Fonio

Satureja

Savory

Durio

Durian

Scorzonera

Scorzonera

Echinochloa

Setaria

Foxtail millet

Eragrostis

Barnyard Millet, Japanese
millet
Teff

Shorea

Shorea

Eriobotrya

Loquat

Sorbus

Rowanberry, service-apple

Fagus

Beech nut

Spondias

Hog plum, mombin

Feijoa

Feijoa

Stillingia

Stillingia

Festuca

Festuca

Stipa

Alfa, esparto

Furcraea

Fuque fibre, Mauritius hemp

Stizolobium

Velvet bean

Gaylussacia

Huckleberry, dangleberry

Tamarindus

Tamarind

Guizotia

Guizotia

Thymus

Thyme

Hedysarum

Hedysarum

Tragopogon

Oyster plant

Hevea

Rubber

Triadica

Tallowtree Seed

Hibiscus

Kenaf, meshta, rosella hemp

Trigonella

Fenugreek seed

Humulus

Hops

Tropaeolum

Mashua

Jatropha

Jatropha

Ullucus

Ullucu

Laurus

Bay leaves

Urena

Congo jute, malva, paka

Lecythis

Paradise nut

Zizyphus

Jujube

Priority 1 crops are generally major food crops; Priority 2 crops are minor food crops, forages, and other crops.
*denotes genera used for rootstock. Prioritization formed from an analysis of crops listed in FAO (2002), FAO
(2011), Groombridge & Jenkins (2002), and Prescott-Allen & Prescott-Allen (1990).

71

Chapter 3
Supplementary Table 2: Threatened US crop wild relatives and wild utilized species.
Taxon

Abies fraseri
Abutilon eremitopetalum
Abutilon menziesii
Abutilon parishii
Abutilon sandwicense
Agave murpheyi
Agave schottii var. treleasei
Agave shawii
Agrostis hendersonii
Allium munzii
Allium obtusum var. conspicuum
Allium scilloides
Alopecurus aequalis var. sonomensis
Amaranthus brownii
Amaranthus pumilus
Artemisia porteri
Asimina tetramera
Astragalus aequalis
Astragalus agnicidus
Astragalus albens
Astragalus ampullarioides
Astragalus applegatei
Astragalus bibullatus
Astragalus brauntonii
Astragalus calycosus var. monophyllidius
Astragalus clarianus
Astragalus cottonii
Astragalus cremnophylax
Astragalus cremnophylax var. cremnophylax
Astragalus cremnophylax var. myriorrhaphis
Astragalus cusickii var. packardiae
Astragalus desereticus
Astragalus diversifolius
Astragalus geyeri var. triquetrus
Astragalus holmgreniorum
Astragalus humillimus
Astragalus hypoxylus
Astragalus jaegerianus
Astragalus kentrophyta var. danaus
Astragalus lentiginosus var. coachellae
Astragalus lentiginosus var. micans
Astragalus lentiginosus var. piscinensis
Astragalus limnocharis
Astragalus limnocharis var. limnocharis
Astragalus limnocharis var. montii
Astragalus magdalenae var. peirsonii
Astragalus microcymbus
Astragalus mulfordiae
Astragalus nevinii
Astragalus osterhoutii

72

Priority
category

US
ESA

P2
P2
P2
P2

LE
LE

P2
P1B
P1B
P1B
P2
P2
P2
P2
P2

LE

LE

LE
LE
LT
LE

LE
LE
LE
LE
LE
LE

LE
C
LT

LE
LE
LE
LE
LT

LT
LT
C

LE

NatureServe

IUCN

G2
G1
G1
G2
G1
G2
T1
G2-3
G1
G1
T2-3
2-3
T1
G1
G2
G2
G1
G2
G2
G1
G1
G1
G1
G2
T2
G1
G2
G1
T1
T1
T2
G1
G2
T2-3
G1
G1
G1
G1
T2-3
T2
T1
T1
G2
T1
T1
T2
G1
G2
G2
G1

VU
CR
CR
CR

CR

EN

# of
accessions
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
0
5
2
0
0
4
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
1
1
0
0
0
0
0
0
0
0

An inventory of crop wild relatives of the United States
Taxon
Astragalus phoenix
Astragalus preussii var. laxiflorus
Astragalus pycnostachyus
Astragalus pycnostachyus var. lanosissimus
Astragalus pycnostachyus var. pycnostachyus
Astragalus rafaelensis
Astragalus rattanii var. jepsonianus
Astragalus ravenii
Astragalus robbinsii var. jesupii
Astragalus robbinsii var. robbinsii
Astragalus schmolliae
Astragalus sinuatus
Astragalus tener
Astragalus tener var. tener
Astragalus tener var. titi
Astragalus traskiae
Astragalus tricarinatus
Astragalus trichopodus var. trichopodus
Astrophytum asterias
Atriplex canescens var. gigantea
Atriplex coronata var. notatior
Atriplex joaquiniana
Betula uber
Canavalia galeata
Canavalia molokaiensis
Canavalia napaliensis
Capparis sandwichiana
Ceanothus cyaneus
Ceanothus impressus
Centrosema arenicola
Chenopodium foggii
Chenopodium incanum var. occidentale
Cneoridium dumosum
Crataegus beata
Crataegus harbisonii
Crotalaria avonensis
Croton alabamensis var. texensis
Cucurbita okeechobeensis
Cuphea aspera
Cupressus macrocarpa
Cyperus fauriei
Cyperus pennatiformis
Cyperus pennatiformis var. bryanii
Cyperus pennatiformis var. pennatiformis
Cyperus rockii
Cyperus trachysanthos
Desmodium humifusum
Digitaria floridana
Digitaria pauciflora
Echinacea angustifolia var. strigosa
Echinacea laevigata

Priority
category

US
ESA
LT

LE

LE
C

LE
LE

P2
P2
P2

LE

P2
P2
P2
P2

LE
LE

LT

P2
P2
P2
P2
P2
P2
P1A
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2

LE
LE

LE
LE

LE

C
LE

NatureServe
G2
T2
G2
T1
T2
G2-3
T2
G1
T1
TX
G1
G1
G1
T1
T1
G2
G1
T2-3
G1
T1
T1
G2
G1
G2
G1
G1
G2
G2
G2-3
G2
G2-3
T2-4
G2-3
G2-4
G1
G1
T2
G1
G2
G1
G1
G1
T1
TX
GX
G1
G1-2
G1
G1
TX
G2-3

IUCN

VU

CR
CR
VU

# of
accessions
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
1
0
0
0
0
0
0
0
0
1
3
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
2
9

73

Chapter 3
Taxon
Echinacea paradoxa
Echinacea paradoxa var. neglecta
Echinacea paradoxa var. paradoxa
Echinacea tennesseensis
Eragrostis fosbergii
Ericameria pinifolia
Eriogonum cinereum
Eriogonum crocatum
Eriogonum giganteum
Eugenia koolauensis
Festuca hawaiiensis
Festuca ligulata
Fragaria chiloensis subsp. sandwicensis
Franklinia alatamaha
Fraxinus papillosa
Fremontodendron mexicanum
Gossypium tomentosum
Guaiacum sanctum
Hazardia cana
Hedysarum boreale var. gremiale
Helianthus carnosus
Helianthus niveus subsp. tephrodes
Helianthus nuttallii subsp. parishii
Helianthus paradoxus
Helianthus smithii
Helianthus verticillatus
Hibiscus brackenridgei
Hibiscus clayi
Hibiscus dasycalyx
Hibiscus waimeae
Hordeum arizonicum
Hypericum cumulicola
Hypericum lissophloeus
Ipomoea microdactyla
Juglans hindsii
Lathyrus grimesii
Lathyrus holochlorus
Lepidium arbuscula
Lepidium barnebyanum
Lepidium crenatum
Lepidium jaredii
Lepidium papilliferum
Lespedeza leptostachya
Leymus pacificus
Lindera melissifolia
Linum carteri
Linum carteri var. carteri
Linum carteri var. smallii
Linum lewisii var. alpicola
Lotus argophyllus var. adsurgens
Lotus argyraeus var. multicaulis

74

Priority
category
P2
P2
P2
P2
P2

US
ESA

P2
P2
P2
P1A

LE
C
C

LE

P1A

P2
P1B
P1A
P1B
P1A
P1B
P1B
P2
P2
P2
P2
P1B

P1B
P1A
P1B
P1B
P2
P2
P2
P2
P2
P2
P1B
P2
P2
P2
P2
P2
P2

LE
LE
C

LE

LE
LE

LT
LT
LE
C

NatureServe
G2
T1
T2
G2
G1
G2-3
G2
G2
G2
G1
G1
G1
T2
GX
G2
G2
G2-3
G2
G2
T1
G1-2
G2*
GX*
G2
G2
G1
G1
G1
G1
G2
G2-4
G2
G2
G2
G1
G2
G2
G1
G1
G2
G1
G2
G3
G2-3
G2-3
G2
T1
T2
T2-4
T1
T1

IUCN

EN

EW

EN

CR

# of
accessions
1
4
5
4
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
2
10
0
1
6
2
0
0
0
0
0
0
1
1
16
3
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0

An inventory of crop wild relatives of the United States
Taxon

Priority
category
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2
P2

Lotus argyraeus var. notitius
Lotus crassifolius var. otayensis
Lotus dendroideus var. traskiae
Lotus nuttallianus
Lotus oblongifolius var. cupreus
Lotus procumbens var. jepsonii
Lotus rubriflorus
Lupinus angustifolius
Lupinus antoninus
Lupinus arboreus
Lupinus citrinus
Lupinus nipomensis
Lupinus rivularis
Lupinus tidestromii
Lupinus westianus var. aridorum
Lyonothamnus floribundus subsp. aspleniifolius
Malva assurgentiflora
Manihot walkerae
P1B
Mespilus canescens
P2
Opuntia basilaris var. longiareolata
P2
Opuntia basilaris var. treleasei
P2
Opuntia engelmannii var. flexospina
P2
Opuntia engelmannii var. linguiformis
P2
Panicum fauriei var. carteri
P2
Panicum niihauense
P2
Paxistima canbyi
Penstemon barrettiae
P2
Penstemon haydenii
P2
Phaseolus texensis
P1B
Pinus radiata
Pinus torreyana
Poa atropurpurea
Poa mannii
Poa napensis
Poa sandvicensis
Poa siphonoglossa
Portulaca molokiniensis
Portulaca sclerocarpa
Portulaca umbraticola subsp. coronata
Prunus eremophila
Prunus murrayana
Quercus dumosa
Ribes binominatum
Ribes echinellum
Ribes erythrocarpum
Robinia hispida var. fertilis
Robinia viscosa var. hartwigii
Rorippa subumbellata
Roystonea regia
Rubus aliceae

P2
P2
P2
P2
P2
P2
P2
P2
P1B
P1A

US
ESA
LE

LE
LE
LE

LE

LE

LE
LE

LE

LE
LE
LE
LE
LE
LE

P1A
P1B
P1B

LT

P2

C

P1A

NatureServe
T1
T1
T2
G1
T2
T1-2
G1
G1-5
G1
G1-5
G2
G1
G2-4
G2
T1
T2
G2
G1
G1
T2
T2
T1
T1
T1
G1
G2
G2
G1-2
G2
G1
G1
G2
G1
G1
G1
G1
G1
G2
T2
G1
GX
G1-2
G2-3
G1
G2
T1
T2
G1
G2-3
GX

IUCN

LR/
cd
VU

EN

# of
accessions
0
0
0
0
0
0
0
0
0
7
1
0
6
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0

0
0
0
3
3
2

0
0
0

75

Chapter 3
Taxon

Priority
category
P1A
P1A
P2

US
ESA

NatureServe

IUCN

# of
accessions
13
1
0
0
0
0
0
0
0

Rubus hawaiensis
G2-3
Rubus macraei
G2
Rumex giganteus
G2-3
Santalum ellipticum
G2-3
Setaria arizonica
P2
G2-4
Solanum incompletum
P1B
LE
G1
Solanum nelsonii
P1B
C
G2
Solanum sandwicense
P1B
LE
G1
Solanum wallacei
P1B
G2
Spiraea virginiana
LT
G2
Suaeda californica
G1
0
Trifolium amoenum
P2
LE
G1
1
Trifolium andinum var. podocephalum
P2
T1
0
Trifolium barnebyi
P2
G1-2
0
Trifolium buckwestiorum
P2
G1
1
Trifolium calcaricum
P2
G1
0
Trifolium douglasii
P2
G2
3
Trifolium jokerstii
P2
G1
1
Trifolium leibergii
P2
G2
1
Trifolium neurophyllum
P2
G2
0
Trifolium owyheense
P2
G2
2
Trifolium stoloniferum
P2
LE
G3
3
Trifolium thompsonii
P2
G2
4
Trifolium trichocalyx
P2
LE
G1
1
Tripsacum floridanum
P1A
G2
0
Vanilla mexicana
P1A
G2-4
0
Vicia menziesii
P1B
LE
G1
0
Vicia ocalensis
P1B
G1
1
Wikstroemia uva-ursi
G2
0
Zizania texana
P1A
LE
G1
0
Taxa listed as endangered (LE), threatened (LT), or as a candidate for listing (C) under the US Endangered Species
Act (US ESA), and/or listed as known or presumed extinct in the wild (GX), globally critically imperiled (G1),
and globally imperiled (G2) in NatureServe (NatureServe, 2012). Note (G3) is categorized as globally vulnerable,
(G4) as apparently secure, and (G5) as globally secure; and T denotes global listing at the infraspecific level.
*denotes threat assessment at the species level. IUCN Red Listing categories include extinct in the wild (EW),
critically endangered (CR), endangered (EN), vulnerable (VU), and lower risk/conservation dependent (LR/cd)
(IUCN 2012). P1A = native taxa closely related to important crop plants; P1B = non-native and/or distantly related
to important crop plants. # of accessions denotes NPGS germplasm listed as wild and collected in the U.S.

76

An inventory of crop wild relatives of the United States
Supplementary Table 3: Additional priority US crop wild relatives and wild utilized species.
Genus
Associated crop
# of
Genus
Associated crop
name
taxa
name
Abutilon
Chinese bell flower, 8
China jute
Actaea
Black cohosh
1
Licania
Oiticica
Agave
Agave
4
Lilium
Lily
Agrostis
Bentgrass
15
Limnanthes
Meadowfoam
Alopecurus
Foxtail grass
4
Linum
Flax
Amaranthus
Amaranth
40
Lolium
Annual ryegrass
Andropogon
Andropogon
13
Lotus
Lotus
Annona
Cherimoya, custard 1
Lupinus
Lupine
apple
Apios
American
1
Manilkara
Balata
groundnut, potato
bean
Apium
Celery
1
Melilotus
Sweet clover
Arbutus
Strawberry tree
3
Mentha
Mint
Armoracia
Horseradish
1
Mespilus
Medlar
Aronia
Chokeberry
3
Morus
Mulberry,
loganberry
Arrhenatherum
Oat-grass
2
Nasturtium
Watercress
Artemisia
Artemisia,
50
Oplopanax
Devil's club
wormwood,
tarragon
Asimina
Pawpaw
9
Opuntia
Prickly pear
Atriplex
Saltbush
37
Oxalis
Oca
Bassia
Bassia
1
Panax
Ginseng
Boehmeria
Ramie
1
Panicum
Little millet, proso
millet
Bromus
Brome
35
Papaver
Poppy
Camelina
Camelina
1
Parthenium
Guayule
Canavalia
Jack bean
6
Paspalum
Kodo millet, ditch
millet
Capparis
Caper
2
Passiflora
Passion fruit
Chenopodium
Quinoa
51
Pastinaca
Parsnip
Chrysanthemum Pyrethrum
1
Penstemon
Penstemon
Chrysophyllum
Star apple
2
Phalaris
Canary seed,
Phalaris
Cochlearia
Scurvy-grass
1
Phleum
Phleum
Coix
Adlay, Job's tears
1
Phlox
Phlox
Corchorus
Corchorus
2
Physalis
Groundcherry,
tomatillo
Coreopsis
Coreopsis
8
Physaria
Physaria
(Lesquerella)
Crataegus
Hawthorn azarole
70
Pinus
Pine nut
Crotalaria
Sunn hemp
6
Poa
Kentucky blue
grass
Croton
Croton
15
Portulaca
Portulaca
Cuphea
Cuphea
5
Prosopis
Mesquite
Cyperus
Chufa
48
Pueraria
Pueraria
Dactylis
Dactylis
1
Raphanus
Radish
Digitaria
Fonio
20
Rhododendron
Rhododendron,
azalea

# of
taxa
1
5
1
21
3
77
95
1
3
4
1
2
4
1
31
8
1
37
14
7
42
13
1
39
6
2
9
13
4
4
42
10
9
3
1
30
77

Chapter 3
Genus

# of
taxa
13
15

Genus

Humulus

Associated crop
name
Echinacea
Barnyard millet,
Japanese millet
Teff
Rocket
Pitanga, Suriname
Cherry
Beech nut
Festuca
Huckleberry,
dangleberry
Liquorice
Hedysarum
Kenaf, meshta,
rosella hemp
Hop

Hydrastis
Hypericum
Jatropha
Lepidium
Lespedeza

Goldenseal
St John's wort
Jatropha
Cress
Lespedeza

Echinacea
Echinochloa
Eragrostis
Eruca
Eugenia
Fagus
Festuca
Gaylussacia
Glycyrrhiza
Hedysarum
Hibiscus

78

Rorippa
Rosa

Associated crop
name
Cress
Rose

# of
taxa
9
27

27
2
3

Rudbeckia
Ruellia
Rumex

Rudbeckia
Ruellia
Sorrel

11
2
19

2
36
8

Salsola
Sambucus
Satureja

Salsola
Elderberry
Savory

4
11
1

1
7
18

Scorzonera
Setaria
Simmondsia

Scorzonera
Foxtail millet
Jojoba

1
27
1

6

Sorbus

11

1
1
4
37
11

Stillingia
Thlaspi
Tragopogon
Triadica
Trifolium

Rowanberry,
service-apple
Stillingia
Field penny-cress
Oyster plant
Tallowtree Seed
Clover

2
1
4
1
96

An inventory of crop wild relatives of the United States
REFERENCES
Aberton, M. (2007). Interspecific hybridization in the genus Trifolium. Plant Breed. 126, 337-342.
Amsellem, L., Noyer, J.L., Le Bourgeois, T., Hossaert-Mckey, M. (2001). Comparison of genetic
diversity of the invasive weed Rubus alceifolius Poir. (Rosaceae) in its native range and in areas of
introduction, using amplified fragment length polymorphism (AFLP) markers. Mol. Ecol. 9, 443455.
Andersson, M.S., de Vicente, M.C. (2010). Gene Flow Between Crops and their Wild Relatives
(Baltimore: The Johns Hopkins University Press).
Ballington, J.R. (2001). Collection, utilization and preservation of genetic resources of Vaccinium.
Hortic. Sci. 36, 213-220.
Barnes, D.K., Bingham, E.T., Murphy, R.P., Hunt, O.J., Beard, D.F., Skrdla, W.H., et al. (1977). Alfalfa
Germplasm in the United States: Genetic Vulnerability, Use, Improvement and Maintenance.
Agricultural Research Service Tech. Bull. No. 1571 (Hyattsville: USDA).
Barney, D., Hummer, K. (2005). Currants, Gooseberries and Jostaberries - A Guide for Growers,
Marketers and Researchers in North America (Binghampton: Haworth Press).
Bilz, M., Kell, S.P., Maxted, N, Lansdown, R.V. (2011). European Red List of Vascular Plants
(Luxembourg: Publications Office of the European Union).
Bossdorf, O., Auge, H., Lafuma, L., Rogers, W.E., Siemann, E., Prati, D. (2005). Phenotypic and genetic
differentiation between native and introduced plant populations. Oecologia 144, 1-11.
Boyle, B., Hopkins, N., Lu, Z., Raygoza Garay, J.A., Mozzherin, D., Rees, T., et al. (2013). The
taxonomic name resolution service: An online tool for automated standardization of plant names.
BMC Bioinfor. 14, 16.
Bradshaw, J.E., Bryan, G.J., Ramsay, G. (2006). Genetic resources (including wild and cultivated
Solanum species) and progress in their utilization in potato breeding. Potato Res. 49, 49-65.
Brennan, R. (2008). “Currants and gooseberries,” in: Temperate Fruit Crop Breeding: Germplasm to
Genomics, ed. J. Hancock (New York: Springer Science and Business Media B.V.), 177-196.
Bretting, P.K., Nabhan, G.P. (1986). Ethnobotany of devil's claw (Proboscidea parviflora ssp.
parviflora: Martyniaceae) in the Greater Southwest. J. California and Great Basin Anthropology 8,
226-237.
Brummitt, N., Bachman, S. (2010). Plants Under Pressure: A Global Assessment. The First Report of
the IUCN Sampled Red List Index for Plants (London: Natural History Museum, UK).
Bureau of Land Management. (2012). Seeds of success. U.S. Department of the Interior, Bureau of Land
Management, Washington, DC. http://www.nps.gov/plants/sos/ (accessed 1 Aug. 2012).
Center for Plant Conservation (CPC). (2012). Center for Plant Conservation: Conserving and restoring
America’s native plants. Center for Plant Conservation, St. Louis, MO. http://
www.centerforplantconservation.org/ (accessed 1 Aug. 2012).
Chuda, A., Adamus, A. (2009). Aspects of interspecific hybridization within edible Alliaceae. Acta
Physiologiae Plantarum 31, 223-227.
Cox, T.S., Sears, R.G., Bequette, R.K., Martin, T.J. (1995). Germplasm enhancement in winter wheat x
Triticum tauschii backcross populations. Crop Sci. 35, 913-919.
Eenink, A.H., Groenwold, R., Dieleman, F.L. (1982). Resistance of lettuce (Lactuca) to the leaf aphid
Nasonovia ribisnigri: transfer of resistance from L. virosa to L. sativa by interspecific crosses and
selection of resistant breeding lines. Euphytica 31, 291-300.
El Khlifi, O.K., Sharma, H., Benlhabib, O. (2004). Transfer of Hessian fly resistance through
interspecific crosses between cultivated wheat and four Aegilops species. Plant Gen. Resour.
Newsletter 138, 43-49.
79

Chapter 3
FAO (Food and Agriculture Organization of the United Nations). (2002). International Treaty on Plant
Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2010). Second Report on the State of
the World’s Plant Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2011). FAOSTAT (Rome: FAO).
http://faostat.fao.org/ (accessed 1 May 2011).
Farooq, S., Asghar, M., Iqbal, N., Asian, E., Arif, M., Shah, T.M. (1995). Production of salt tolerant
wheat germplasm through crossing cultivated wheat with Aegilops cylindrica, I. Field evaluation of
salt tolerant germplasm. Cereal Res. Commun. 23, 275-282.
Finn, C. (2001). Trailing blackberries: from clear-cuts to your table. Hortic. Sci. 36, 236-238.
Flack, J.R. (1970). The spread and domestication of the pecan (Carya illinoensis) in the United States
(Madison: University of Wisconsin Press).
Flora of North America Association (FNA). (2008a). Flora of North America (Cambridge: Flora of
North America Association, Harvard University Herbaria) http:// floranorthamerica.org/ (accessed 1
Jan. 2011).
Flora of North America Association (FNA). (2008b). Flora of North America – Introduction.
(Cambridge: Flora of North America Association, Harvard University Herbaria). http://
floranorthamerica.org/introduction (accessed 1 Jan. 2011).
Ford-Lloyd, B., Schmidt, M., Armstrong, S.J., Barazani, O., Engels, J., Hadas, R., et al. (2011). Crop
wild relatives- undervalued, underutilized and under threat? BioScience 61, 559-565.
Fowler, C., Hodgkin, T. (2004). Plant genetic resources for food and agriculture: assessing global
availability. Annu. Rev. Env. Resour. 29, 10.1-10.37.
Frese, L., Desprez, B., Ziegler, D. (2001). “Potential of genetic resources and breeding strategies for
base-broadening in Beta,” in: Broadening the Genetic Base of Crop Production, eds. H.D. Cooper,
C. Spillane, T. Hodgkin (Wallingford: CABI), 295-309.
Gale, G. (2003). “Saving the vine from Phylloxera: a never-ending battle,” in: Wine: A Scientific
Exploration, eds. M. Sandler, R. Pinder (London & New York: Taylor & Francis).
GBIF (Global Biodiversity Information Facility). (2012). GBIF Data Portal (Copenhagen: GBIF).
http://data.gbif.org/ welcome.htm (accessed 1 Jan. 2011).
Global Strategy for Plant Conservation (GSPC). (2002). Global Strategy for Plant Conservation
(GSPC), Convention on Biological Diversity (Montreal: Secretariat of the Convention on Biological
Diversity).
Goodman, R.M., Hauptli, H., Crossway, A., Knauf, V.C. (1987). Gene transfer in crop improvement.
Science 236, 48-54.
Gororo, N.N., Eagles, H.A., Eastwood, R.F., Nicolas, M.E., Flood, R.G. (2002). Use of Triticum tauschii
to improve yield of wheat in low-yielding environments. Euphytica 123, 241-254.
GRIN (Germplasm Resources Information Network), USDA, ARS, National Genetic Resources
Program. (2012). National Germplasm Resources Laboratory, Beltsville, Maryland. http://www.arsgrin.gov/~sbmljw/cgi-bin/index.pl?language=en (Accessed January-August 2012)
GRIN (Germplasm Resources Information Network), USDA, ARS, National Genetic Resources
Program. (2012). World Economic Plants Database. http://www.ars-grin.gov/cgibin/npgs/html/wep.pl (accessed Jan. 2011)
Groombridge, B., Jenkins, M.D. (2002). World Atlas of Biodiversity. Prepared by the UNEP World
Conservation Monitoring Centre (Berkeley: University of California Press).
Guarino, L., Lobell, D.B. (2011). A walk on the wild side. Nat. Clim. Chang. 1 (8), 374-375.
Guo, Q., Falcone, J., Brownsmith, J. (2009). “Building the database for introduced plants in the United
States,” in: Proceedings 20th U.S. Department of Agriculture Interagency Research Forum on
80

An inventory of crop wild relatives of the United States
Invasive Species 2009, Gen. Tech. Rep. NRS-P-51, 2009 January 13-16, eds. K.A. McManus. K.W.
Gottschalk (Annapolis: U.S. Department of Agriculture, Forest Service, Northern Research Station),
73.
Gur, A., Zamir, D. (2004). Unused natural variation can lift yield barriers in plant breeding. PLoS Biol.
2(10), e245.
Hajjar, R., Hodgkin, T. (2007). The use of wild relatives in crop improvement: a survey of developments
over the last 20 years. Euphytica 156, 1-13.
Hanna, W.W. (1997). Influence of cytoplasms from a wild grassy subspecies on dry matter yields in
pearl millet. Crop Sci. 37, 614-616.
Harlan, J.R. (1976). Genetic resources in wild relatives of crops. Crop Sci. 16, 329-333.
Harlan, J.R., de Wet, J.M.J. (1971). Toward a rational classification of cultivated plants. Taxon 20, 509517.
Heywood, V.H. (2008). Challenges of in situ conservation of crop wild relatives. Turk. J. Bot. 32, 421432.
Hijmans, R.J., Spooner, D. (2001). Geographic distribution of wild potato species. Am. J. Bot. 88, 21012112.
Hoes, J.A., Putt, E.D., Enns, H. (1973). Resistance to Verticillium wilt in collections of wild Helianthus
in North America. Phytopathology 63, 1517-1520.
Hoffman D.L, Chong, J., Jackson, E.W., Obert, D.E. (2006). Characterization and mapping of a crown
rust resistance gene complex (Pc58) in TAM O-301. Crop Sci. 46, 2630-2635.
Hooftman, D.A.P., De Jong, M.J., Oostermeijer, G.B., den Nijs, H.J.C.M. (2007). TI: Modelling the
long-term consequences of crop x wild relative hybridization: a case study using four generations of
hybrids. J. Appl. Ecol. 44(5), 1035-1045.
Hulke, B.S., Miller, J.F., Gulya, T.J., Vick, B.A. (2010). Registration of the oilseed sunflower genetic
stocks HA 458, HA 459, and HA 460 possessing genes for resistance to downy mildew. J. Plant
Registrations 4, 1-5.
Hummer, K.E., Bassil, N., Njuguna, W. (2011). “Fragaria,” in: Wild Crop Relatives: Genomic and
Breeding Resources, ed. C. Kole (Berlin: Springer).
Iltis, H.H. (1988). “Serendipity in the exploration of biodiversity. what good are weedy tomatoes?,” in:
Biodiversity, ed. E.O. Wilson (Washington D.C.: National Academy Press), 98-105.
International Union for Conservation of Nature (IUCN), Species Survival Commission. (2008). Crop
wild relative specialist group. Bioversity International, Rome, Italy. http://www.cwrsg.org/index. asp
(accessed 1 Jan. 2012).
International Union for Conservation of Nature (IUCN). (2012). The IUCN Red List of Threatened
Species. Version 2012.1 (Cambridge: International Union for Conservation of Nature and Natural
Resources). http://www.iucnredlist.org/ (accessed 1 Oct. 2012).
Jan C.C, Fernandez-Martinez, J.M., Ruso, J., Muñoz-Ruz, J. (2002). Registration of four sunflower
germplasms with resistance to Orobanche cumana race F. Crop Sci. 42, 2217-2218.
Jan, C.C. (1992). Registration of sunflower amphiploid germplasm line, ANN-BOL-AMP1. Crop Sci.
32, 1513.
Jan, C.C., Chandler, J.M. (1988). Registration of powdery mildew resistant sunflower germplasm pool,
PM 1. Crop Sci. 28, 1040.
Jan, C.C., Miller, J.F., Seiler, G.J., Fick, G.N. (2006). Registration of one cytoplasmic male sterile and
two fertility restoration sunflower genetic stocks. Crop Sci. 46, 1835.
Jan, C.C., Quresh, Z., Gulya, T.J. (2004). Registration of seven rust resistant sunflower germplasms.
Crop Sci. 44, 1887-1888.
Jarvis, A., Lane, A., Hijmans, R.J. (2008). The effect of climate change on crop wild relatives. Agr.
81

Chapter 3
Ecosyst. Environ. 126, 13-23.
Jarvis, A., Williams, K., Williams, D.E., Guarino, L., Caballero, P.J., Mottram, G. (2005). Use of GIS
for optimizing a collecting mission for a rare wild pepper (Capsicum flexuosum Sendtn.) in Paraguay.
Genet. Resour. Crop Ev. 52, 671-682.
Kell, S.P., Knüpffer, H., Jury, S.L., Ford-Lloyd, B.V., Maxted, N. (2008). “Crops and wild relatives of
the Euro-Mediterranean region: making and using a conservation catalogue,” in: Crop Wild Relative
Conservation and Use, eds. N. Maxted, B.V. Ford-Lloyd, S.P. Kell, J.M. Iriondo, M.E. Dulloo, J.
Turok (Wallingford: CAB International), 69-109.
Khrustaleva, L., Kik, C. (1998). Cytogenetical studies in the bridge cross Allium cepa x (A. fistulosum
x A. roylei). Theor. Appl. Genet. 96, 8-14.
Kraft, K.H., Luna-Ruíz, J.J., Gepts, P. (2012). A new collection of wild populations of Capsicum in
Mexico and the southern United States. Genet. Resour. Crop Ev. 60, 225-232.
Lexer, C., Lai, Z., Rieseberg, L.H. (2004). Candidate gene polymorphisms associated with salt tolerance
in wild sunflower hybrids: implications for the origin of Helianthus paradoxus, a diploid hybrid
species. New Phytol. 161, 225-233.
Loarie, S.R., Duffy, P.B., Hamilton, H., Asner, G.P., Field, C.B., Ackerly, D.D. (2009). The velocity of
climate change. Nature 462 (7276), 1052-1055.
Ma, H., Singh, R.P., Mujeeb-Kazi, A. (1995). Resistance to stripe rust in O. rufipogon, T. tauschii and
their synthetic hexaploids. Euphytica 82, 117-124.
Magos Brehm, J., Maxted, N., Ford-Lloyd, B.V., Martins-Loução, M.A. (2007). National inventories of
crop wild relatives and wild harvested plants: case-study for Portugal. Genet. Resour. Crop Ev. 55,
779-796.
Mallikarjuna, N., Deepak, J., Prabhakar, R. (2006). Introgression of Cajanus platycarpus genome into
cultivated pigeonpea, C. cajan. Euphytica 149, 161-167.
Maxted, N., Akparov, Z.I., Aronsson, M., Asdal, Å., Avagyan, A., et al. (2012b). “Current and future
threats and opportunities facing European crop wild relative and landrace diversity,” in:
Agrobiodiversity Conservation: Securing the Diversity of Crop Wild Relatives and Landraces, eds.
N. Maxted, M.E. Dulloo, B.V. Ford-Lloyd, L. Frese, J.M. Iriondo, et al. (Wallingford: CAB
International), 333-354.
Maxted, N., Dulloo, E., Ford-Lloyd, B.V., Iriondo, J.M., Jarvis, A. (2008). Gap analysis: a tool for
complementary genetic conservation assessment. Divers. Distrib. 14, 1018-1030.
Maxted N., Ford-Lloyd, B.V., Hawkes, J.G. (1997). “Complementary conservation strategies,” in: Plant
Genetic Conservation: The In situ Approach, eds. N. Maxted, B.V. Ford-Lloyd, J.G. Hawkes
(London: Chapman and Hall), 20-55.
Maxted, N., Ford-Lloyd, B.V., Jury, S.L., Kell, S.P., Scholten, M.A. (2006). Towards a definition of a
crop wild relative. Biodivers. Conserv. 15, 2673-2685.
Maxted, N., Kell, S.P. (2009). Establishment of a Global Network for the In Situ Conservation of Crop
Wild Relatives: Status and Needs (Rome: FAO Commission on Genetic Resources for Food and
Agriculture).
Maxted, N., Kell, S.P., Ford-Lloyd, B.V., Dulloo, E., Toledo, Á. (2012a). Toward the systematic
conservation of global crop wild relative diversity. Crop Sci. 52, 774-785.
McCouch, S., Sweeney, M., Li, J., Jiang, H., Thomson, M., Septiningsih, E., et al. (2007). Through the
genetic bottleneck: O. rufipogon as a source of trait enhancing alleles for O. sativa. Euphytica 154,
317-339.
McGranahan, G., Leslie, C. (2009). “Breeding walnuts (Juglans regia),” in: Breeding Plantation Tree
Crops: Temperate Species, eds. S.M. Jain, P.M. Priyadarshan (New York: Springer), 249-273.
McGuffin, M., American Herbal Products Association. (2000). Herbs of Commerce. 2nd ed (Silver
82

An inventory of crop wild relatives of the United States
Spring: American Herbal Products Association).
Meilleur, B.A., Hodgkin, T. (2004). In situ conservation of crop wild relatives: status and trends.
Biodivers. Conserv. 13, 663-684.
Mii, M. (2009). Breeding ornamental plants through interspecific hybridization using advanced
techniques with a special focus on Dianthus, Primula, Cosmos, and Kalanchoe. Acta Hortic. 836,
63-72.
Miller, J. F., Gulya, T.J. (1988). Registration of six downy mildew resistant sunflower germplasm lines.
Crop Sci. 28, 1040-1041.
Miller, J.F., Gulya, T.J. (1999). Registration of eight Sclerotinia-tolerant sunflower germplasm lines.
Crop Sci. 39, 301-302.
Nabhan, G.P. (1985). Native crop diversity in Aridoamerica: conservation of regional gene pools. Econ.
Bot. 39, 387-399.
Nabhan, G.P. (1990). Conservationist and forest service join forces to save wild chiles. Diversity
Magazine 6, 47-48.
Native Seed Network. (2010). Native Seed Network Database (Corvallis: Native Seed Network).
http://www.nativeseednetwork.org (accessed 1 Jan. 2011).
NatureServe. (2009). NatureServe Explorer: An Online Encyclopedia of Life. Version 7.1 (Arlington:
NatureServe). http://www.natureserve.org/explorer/ (accessed 1 Apr. 2010).
Ortiz, R., Sayre, K.D., Govaerts, B., Gupta, R., Subbarao, G.V., Ban, T., et al. (2008). Climate change:
can wheat beat the heat? Agr. Ecosyst. Environ. 126, 46-58.
Parra-Quijano, M., Iriondo, J.M., Torres, E. (2011). Improving representativeness of genebank
collections through species distribution models, gap analysis and ecogeographical maps. Biodivers.
Conserv. 21, 79-96.
Pavek, D.S., Lamboy, W.F., Garvey, E.J. (2001). In situ conservation of America’s wild grapes. Hortic.
Sci. 36, 232-235.
Peel, M.D., Asay, K.H., Waldron, B.L., Jensen, K.B., Robins, J.G., Mott, I.W. (2009). 'Don', a diploid
falcata alfalfa for western U.S. rangelands. J. Plant Registrations 3, 115-118.
Pena, R.J., Zarco-Hernandez, J., Mujeeb-Kazi, A. (1995). Glutenin subunit compositions and
breadmaking quality characteristics of synthetic hexaploid wheats derived from O. rufipogon x
Triticum tauschii (Coss.) Schmal crosses. J. Cereal Sci. 21, 15-23.
Peredo, E.L., Ángeles Revilla, M., Reed, B.M., Javornik, B., Cires, E., Fernández Prieto, J.A., et al.
(2010). The influence of European and American wild germplasm in hop (Humulus lupulus L.)
cultivars. Genet. Resour. Crop Ev. 57, 575-586.
Pestsova, E.G., Borner, A., Roder, M.S. (2006). Development and QTL assessment of Triticum
aestivum-Aegilops tauschii introgression lines. Theor. Appl. Genet. 112, 634-647.
Phillips, O.L., Meilleur, B.A. (1998). Usefulness and economic potential of the rare plants of the United
States: a statistical survey. Economic Bot. 52(1), 57-67.
Pickering, R., Johnston, P.A. (2005). Recent progress in barley improvement using wild species of
Hordeum. Cytogenet. Genome Res. 109, 344-349.
Plant Germplasm Operations Committee. (1999). The American Wild Relatives of Crops: In Situ
Conservation Guidelines (Beltsville: USDA National Plant Germplasm System, In situ
subcommittee).
Plant Germplasm Operations Committee. (2010). Minutes of the Annual Meeting, Geneva, NY. 28–29
July 2010 (Beltsville: USDA National Germplasm Resources Laboratory).
Prescott-Allen, R., Prescott-Allen, C. (1986). The First Resource: Wild Species in the North American
Economy (New Haven: Yale University Press).
Prescott-Allen, R., Prescott-Allen, C. (1990). How many plants feed the world? Conserv. Biol. 4, 36583

Chapter 3
374.
Putt, E.D. (1978). “History and present world status,” in: Sunflower Science and Technology, ed. J.P.
Carter (Madison: American Society of Agronomy), 1-29.
Ramírez-Villegas J., Khoury, C., Jarvis, A., Debouck, D.G., Guarino, L. (2010). A gap analysis
methodology for collecting crop genepools: a case study with Phaseolus beans. PLoS One 5(10),
e1349.
Rawal, K.M. (1975). Natural hybridization among wild, weedy and cultivated Vigna unguiculata (L.)
Walp. Euphytica 24, 699-707.
Reisch, B., Pratt, C. (1996). “Grapes,” in: Fruit Breeding Volume II: Vine and Small Fruit Crops, eds.
J. Janick, J. Moore (New York: John Wiley & Sons), 297-369.
Rogers, C.E., Thompson, T.E., Seiler, G.J. (1984). Registration of three Helianthus germplasms for
resistance to the sunflower moth. Crop Sci. 24, 212-213.
Ross, H. (1979). “Wild species and primitive cultivars as ancestors of potato varieties,” in: Proceedings
of the Conference Broadening the Genetic Base of Crops, 3–7 July 1978, eds. A.C. Zeven, A.M. van
Harten (Wageningen, Centre for Agricultural Publishing and Documentation), 237-245.
Ruge-Wehling, B., Linz, A., Habeku, A., Wehling, P. (2006). Mapping of RYMl6Hb, the second
soilborne virus resistance gene introgressed from Hordeum bulbosum. Theor. Appl. Genet. 113, 867673.
Rygulla, W., Snowdon, R., Eynck, C., Koopmann, B., Von Tiedemann, A., Lühs, W., et al. (2007).
Broadening the genetic basis of Verticillium longisporum resistance in Brassica napus by
interspecific hybridization. Phytopathology 97, 1391-1396.
Sakamoto, S. (1976). Breeding of a new sweet potato variety, Minamiyutaka, by the use of wild
relatives. Japanese Agric. Res. Quest. 10(4), 183-186.
Seiler, G.J. (1991a). Registration of 15 interspecific sunflower germplasm lines derived from wild
annual species. Crop Sci. 31, 1389-1390.
Seiler, G.J. (1991b). Registration of 13 downy mildew tolerant interspecific sunflower germplasm lines
derived from wild annual species. Crop Sci. 31, 1714-1716.
Seiler, G.J. (1991c). Registration of six interspecific sunflower germplasm lines derived from wild
perennial species. Crop Sci. 31, 1097-1098.
Seiler, G.J. (2000). Registration of 10 interspecific germplasms derived from wild perennial sunflower.
Crop Sci. 40, 587-588.
Seiler, G.J., Gulya, T.J. (2004). “Exploration for wild Helianthus species in North America: challenges
and opportunities in the search for global treasures,” in: International Sunflower Conference
Proceedings, v.1, Fargo, ND. 29 Aug. – 2 Sept. 2004 (Paris: International Sunflower Association),
43-68.
Smith, B.D. (2006). Eastern North America as an independent center of plant domestication. Proc. Natl.
Acad. Sci. 103, 12223-12228.
Stein, B.A., Kutner, L.S., Adams, J.S. (2000). Precious Heritage: The Status of Biodiversity in the
United States (New York: Oxford University Press).
Suszkiw, J. (2005). Hessian Fly Resistant Wheat Germplasm Available (Washington D.C.: USDAARS). http://www.ars.usda.gov/is/pr/2005/050211.htm (accessed 1 Aug. 2012).
Takeda, K., Frey, K.J. (1976). Contributions of vegetative growth rate and harvest index to grain yield
of progenies from Avena sativa X A. sterilis Crosses. Crop Sci. 16, 817-821.
Tanksley, S.D., McCouch, S.R. (1997). Seed banks and molecular maps: unlocking genetic potential
from the wild. Science 277, 1063-1066.
Thompson, M., Lagerstedt, H., Mehlenbacher, S. (1996). “Hazelnuts,” in: Fruit Breeding, Volume III:
Nuts, eds. J. Janick, J. Moore (New York: John Wiley & Sons), 125-184.
84

An inventory of crop wild relatives of the United States
Toubia-Rahme, H., Johnston, P.A., Pickering, R.A., Steffenson, B.J. (2003). Inheritance and
chromosomal location of Septoria passerinii resistance introgressed from Hordeum bulbosum into
Hordeum vulgare. Plant Breeding 122, 405-409.
Ureta, C., Martinez-Meyer, E., Perales, H.R., Alvarez-Buylla, E.R. (2011). Projecting the effects of
climate change on the distribution of maize races and their wild relatives in Mexico. Glob. Change
Biol. 18, 1073-1082.
USDA-ARS National Genetic Resources Program. (2011). World Economic Plants: A Standard
Reference (Beltsville: National Germplasm Resources Laboratory). http://www.ars-grin.gov/cgibin/npgs/html/wep.pl (accessed 1 Jan. 2011).
USDA-ARS National Genetic Resources Program. (2012). GRIN (Germplasm Resources Information
Network) Taxonomy for Plants (Beltsville: National Germplasm Resources Laboratory).
http://www.ars-grin.gov/~sbmljw/cgi-bin/index.pl?language=en (accessed 1 Jan. 2012).
USDA-Natural Resources Conservation Service (USDA-NRCS). (2010). The PLANTS Database
(Baton Rouge: National Plant Data Center). http://plants.usda.gov (accessed 23 Apr. 2010).
US Forest Service (USFS). (2010). Celebrating Wildflowers – Ethnobotany – Wild Crop Progenitors
(Washington D.C.: USFS). http://www.fs.fed.us/wildflowers/ethnobotany/progenitors.shtml
(accessed 1 Aug. 2012).
Vavilov, N.I. (1926). Studies on the origin of cultivated plants. Bulletin of Applied Botany, Genetics and
Plant Breeding 16, 1-248.
Villareal, R.L., Mujeeb-Kazi, A., Fuentes-Davila, G., Rajaram, S. (1996). Registration of four synthetic
hexaploid wheat germplasm lines derived from O. rufipogon x T. tauschii crosses and resistance to
kernal bunt. Crop Sci. 36, 218.
Villareal, R.L., Sayre, K., Banuelos, O., Mujeeb-Kazi, A. (2001). Registration of four synthetic
hexaploid wheat (O. rufipogon/Aegilops tauschii) germplasm lines tolerant to waterlogging. Crop
Sci. 41, 274.
Vincent, H., Wiersema, J., Kell, S., Fielder, H., Dobbie, S., Castañeda-Álvarez, N.P., et al. (2013). A
prioritized crop wild relative inventory to help underpin global food security. Biol. Conserv. 167,
265-75.
Volk, G.M., Richards, C.M. (2011). Integration of georeferencing, habitat, sampling, and genetic data
for documentation of wild plant genetic resources. Hortic. Sci. 46 (11), 1446-1449.
Wang, W., Vinocur, B., Altaian, A. (2003). Plant responses to drought, salinity and extreme
temperatures: towards genetic engineering for stress tolerance. Planta 218, 1-14.
Whelan, E.D.P., Dedio, W. (1980). Registration of sunflower germplasm composite crosses CMG-1,
CMG-2, and CMG-3. Crop Sci. 20, 832.
Wiersema, J.H., Leόn, B. (1999). World Economic Plants: A Standard Reference (Boca Raton: CRC
Press LLC).
Wiersema, J.H., Leόn, B., Garvey, E.J. (2012). Identifying wild relatives of subtropical and temperate
fruit and nut crops. Acta Hortic. 948, 285-288.
Wilkes, G. (2007). Urgent notice to all maize researchers: disappearance and extinction of the last wild
teosinte population is more than half completed. A modest proposal for teosinte evolution and
conservation in situ: The Balsas, Guerrero, Mexico. Maydica 52, 49-58.
Xiao J., Grandillo, S., Ahn, S.N., McCouch, S.R., Tanksley, S.D. (1996). Genes from wild rice improve
yield. Nature 384, 223-224.
Xiu-Jin, L., Deng-Cai, L., Zhi-Rong, W. (1997). Inheritance in synthetic hexaploid wheat ‘RSP’ of
sprouting tolerance derived from Aegilops tauschii Coss. Euphytica 95, 321-323.
Zamir, D. (2001). Improving plant breeding with exotic genetic libraries. Nat. Genet. 2, 983-989.
Zhang, L., Pickering, R.A., Murray, B.G. (2001). Hordeum vulgare x H. bulbosum tetraploid hybrid
85

Chapter 3
provides useful agronomic introgression lines for breeders. New Zeal. J. Crop Hort. 29, 239-246.
Zizumbo-Villarreal, D., Colunga-GarcíaMarín, P. (2010). Origin of agriculture and plant domestication
in west Mesoamerica. Genet. Resour. Crop Ev. 57, 813-825.
Zizumbo-Villarreal, D., Colunga-GarciaMarin, P., Payro de la Cruz, E., Delgado-Valerio, P., Gepts, P.
(2005). Population structure and evolutionary dynamics of wild-weedy-domesticated complexes of
common bean in a Mesoamerican region. Crop Sci. 45, 1073-1083.

86

CHAPTER 4
A gap analysis methodology for collecting crop genepools:
a case study with Phaseolus beans
Julián Ramírez-Villegas1, Colin Khoury2, Andy Jarvis1,3,4, Daniel G. Debouck5, and
Luigi Guarino2
Decision and Policy Analysis Program, International Center for Tropical Agriculture (CIAT), Cali,
Colombia.
2
Global Crop Diversity Trust, Rome, Italy.
3
Adaptation to Progressive Climate Change, Challenge Program on Climate Change, Agriculture and
Food Security (CCAFS), Copenhagen, Denmark.
4
Bioversity International, Regional Office for the Americas, Cali, Colombia.
5
Genetic Resources Unit, International Center for Tropical Agriculture (CIAT), Cali, Colombia.
1

Published in PLoS One 5 (10), e13497. doi:10.1371/journal.pone.0013497

Chapter 4
Abstract
The wild relatives of crops represent a major source of valuable traits for crop improvement. These
resources are threatened by habitat destruction, land use changes, and other factors, requiring their
urgent collection and long-term availability for research and breeding from ex situ collections. We
propose a method to identify gaps in ex situ collections (i.e., gap analysis) of crop wild relatives as a
means to guide efficient and effective collecting activities. The methodology prioritizes among taxa
based on a combination of sampling, geographic, and environmental gaps. We apply the gap analysis
methodology to wild taxa of the Phaseolus genepool. Of 85 taxa, 48 (56.5%) are assigned high priority
for collecting due to lack of, or under-representation, in genebanks, 17 taxa are given medium priority
for collecting, 15 low priority, and 5 species are assessed as adequately represented in ex situ collections.
Gap "hotspots", representing priority target areas for collecting, are concentrated in central Mexico,
although the narrow endemic nature of a suite of priority species adds a number of specific additional
regions to spatial collecting priorities. Results of the gap analysis method mostly align very well with
expert opinion of gaps in ex situ collections, with only a few exceptions. A more detailed prioritization
of taxa and geographic areas for collection can be achieved by including in the analysis predictive threat
factors, such as climate change or habitat destruction, or by adding additional prioritization filters, such
as the degree of relatedness to cultivated species (i.e., ease of use in crop breeding). Furthermore, results
for multiple crop genepools may be overlaid, which would allow a global analysis of gaps in ex situ
collections of the world's plant genetic resources.
Keywords: Crop wild relative, Ex situ conservation, Plant Genetic Resources, Germplasm exploration,
Tropical legumes

88

A gap analysis methodology for collecting crop genepools
INTRODUCTION
Crop wild relatives (CWR) are wild plant species sharing relatively recent common ancestry
with cultivated plants. CWR typically possess wide diversity, much of it not found in the crop,
and this diversity may be introgressed into the crop by plant breeders, with the ease of transfer
of genes generally dependent on the degree of relatedness between the wild species and the
domesticate (Harlan & de Wet 1971, Singh 2001). Wild relatives have provided to crops traits
such as pest and disease resistance, tolerance to abiotic stresses, increased yield, male sterility,
and quality, increasing the value and sustainability of banana, barley, beans, cassava, chickpea,
lettuce, maize, oats, pearl millet, potatoes, rice, sugar cane, sunflower, tomato, and wheat
production, among others. In the past 20 years, there has been a steady increase in the rate of
release of cultivars containing genes from CWR, and their contribution should only increase as
the development of molecular technologies makes identification and utilization of diverse
germplasm more efficient (Prescott-Allen & Prescott-Allen 1986, Tanksley & McCouch 1997,
Singh 2001, Hajjar & Hodgkin 2007).
Plant breeders obtain CWR material from genebanks. However, major gaps in the genetic
diversity of important crop genepools remain to be filled in ex situ germplasm collections.
These gaps are particularly evident for non-cereal crops (e.g., legumes, roots and tubers,
vegetables), and for wild and weedy forms (FAO 1997, Maxted & Kell 2009, Khoury et al.
2010). Maxted and Kell (2009) estimated that 94% of European CWR species are completely
missing from ex situ collections. At the same time, habitat destruction, invasive species,
urbanization, and the shift from traditional to industrial agricultural practices, among other
factors, continue to threaten PGR, and climate change is projected to impose further pressures
on both wild and agricultural ecosystems (Meilleur & Hodgkin 2004, Thuiller et al. 2005,
Brooks et al. 2006, Challinor et al. 2007, Intergovernmental Panel on Climate Change 2007,
Hawkins et al. 2008, Burke et al. 2009).
Clearly, much collecting of CWR diversity is still required. Unfortunately international efforts
in collecting plant genetic resources in general have been in decline in recent decades (FAO
2009b). The recent coming into force of the International Treaty on Plant Genetic Resources
for Food and Agriculture is, however, expected to provide impetus for the development of an
integrated, effective, efficient, global approach to conserving PGR. The development of
strategic planning approaches will be necessary to prioritize PGR for collecting as part of such
89

Chapter 4
a rational global system.
Gap analysis refers to a systematic method of analyzing the degree of conservation of taxa, in
order to identify those locations, taxa, and particular traits (adaptations) un- or under- secured
in conservation systems (Maxted et al. 2008). Nabhan (1990) identified four ways by which
gap analysis techniques may lead to better collecting and conservation: targeting localities
where sets of species absent from existing collections can be obtained with least effort and cost;
determining which areas are ‘under-collected’ or ‘over-collected’ for germplasm relative to the
known distribution of a taxon; locating which regions have the greatest or most dissimilar
species richness compared with other regions; and outlining the ecological amplitudes of each
species so that a wider representation of the ecotypes or genetically adapted populations of each
can be sampled.
Geographic Information Systems (GIS) technologies have enabled a better understanding of
species distributions and of the representativeness of germplasm collections, and have
contributed to conservation planning of wild species, CWR, and domesticates (Nabhan 1990,
Jones et al. 1997, Hijmans & Spooner 2001, Hijmans et al. 2002, Jarvis et al. 2002, Jarvis et
al. 2003, Lobo Burle et al. 2003, Maxted et al. 2004, Jarvis et al. 2005, Graham & Hijmans
2006, Parthasarathy et al. 2006, Maxted et al. 2008). Pioneering the use of these tools in
conservation, Jones et al. (1997) successfully predicted the location of populations of wild
common bean (Phaseolus vulgaris), based on climatic suitability. Significant developments
have occurred in recent years in the application of GIS to PGR conservation planning, including
the development and validation of various approaches to niche modeling, new analysis tools
and extensions, and better access to geographic information, results and approaches (Hijmans
et al. 2001).
We propose here a gap analysis method designed to inform planning of germplasm collecting
for ex situ conservation, based upon available information resources, using GIS. The
distributions of ex situ collections are compared to GIS-modeled taxon distributions based on
both herbarium and genebank data. The gross total number of germplasm accessions, as well
as the distribution (geographical and environmental) of those accessions, are compared against
modeled distributions in order to identify gaps in ex situ conservation coverage. These results
form the basis for a prioritization of taxa across the genepool for collecting, and the
90

A gap analysis methodology for collecting crop genepools
identification of the highest priority locations (i.e., diverse and under-represented areas) for the
most efficient and effective collecting, in order to further enhance ex situ holdings. Our model
genepool is Phaseolus.
The genus Phaseolus originated in the tropics and subtropics of the New World, and contains
up to 81 species and 34 infraspecific taxa (Gepts 2001, Singh 2001, Freytag & Debouck 2002,
Debouck 2009), having undergone a series of revisions, notably in association with members
of Vigna, which have included splitting some species into new genera (e.g., Strophostyles,
Dysolobium, Macroptilium, Minkelersia and Alepidocalyx) (Maxted et al. 2004). The main
centers of diversity for the genus are in wide Mesoamerica (from southern USA, Mexico, and
Central America down to Panama), the northern Andean region (Colombia to northern Peru),
and the central Andes (northern Peru, Bolivia to northwest Argentina). Of these, the
Mesoamerican centre is the richest in species (Delgado-Salinas 1985, Debouck 2000, Freytag
& Debouck 2002, Debouck 2009).
Phaseolus has five domesticated species, each a result of an independent domestication process:
P. vulgaris L.- common bean; P. lunatus L.- lima bean; P. coccineus L.- runner bean; P.
acutifolius A. Gray - tepary bean; and P. dumosus Macfady - year bean. The genus has been
cultivated for over 7000 years, and each of the cultivated species has distinct ecological
adaptations (Debouck & Smartt 1995). Common bean is the world’s most important legume for
food production and security, and represents 50% of the grain legumes consumed worldwide,
reaching primary importance in the staple diet of over 500 million people, especially for its
protein content (Gepts 2001, McClean et al. 2004). Common bean is now grown on over 27
million hectares globally, producing over 20 million tons (FAO 2009a).
Diversity in Phaseolus in relation to the cultivated species is organized into genepools based
on phylogenetic relationships (Smartt 1981, Singh & Jauhar 2005). The primary genepool of
cultivated species includes both cultivars and wild populations, hybrids of which are generally
fully fertile with no major reproductive barriers. P. vulgaris also allows a measure of
interspecific hybridization with species in its secondary genepool. P. lunatus and P. acutifolius
appear less capable of gene exchange with related species (Debouck 1999).
Like many important food crops, cultivars of common bean have a narrow genetic base,
91

Chapter 4
attributable to the genetic bottleneck accompanying the domestication process, stringent quality
requirements in the market, limited past use of exotic germplasm in breeding, and conservative
breeding programs for the crop (Singh 2001). Interspecific and wide intraspecific crossing have
been useful strategies for crop improvement, but given the still limited genetic base, more along
these lines is needed. Useful alleles for many agronomic traits deficient in common bean
cultivars, including resistance to storage insects, leafhoppers, ascochyta blight, common
bacterial blight, white mold, bean common mosaic virus, rust, drought, and soil fertility
problems, as well as early maturity, adaptation to higher latitudes, upright plant type, pod
quality, and seed yield have been identified in wild common bean and species in the secondary
and tertiary genepools, and utilized in breeding programs (Kornegay & Cardona 1991, DelgadoSalinas et al. 1999, Singh 2001, Acosta-Gallegos et al. 2007). Wild common bean has also
contributed high protein digestibility (Shelley-Dessert & Bliss 1991) and nodulation (Kipe-Nolt
et al. 1992) traits. Despite the increasing utilization of CWR in common bean breeding, Singh
(2001) estimated as much as 90% of the genetic variability available in the primary genepool
and related species as under- or not utilized. Widening of genetic diversity in the other
Phaseolus crop species may also prove important. The domestication of tepary bean involved
a severe genetic bottleneck event, leading to a particularly low level of genetic diversity in the
crop (Schinkel & Gepts 1988, Garvin & Weeken 1994, Munoz et al. 2006).
Close to 250 ex situ germplasm collections of Phaseolus, holding approximately 260,000
accessions, have been established worldwide (FAO 2009b). The vast majority of these
accessions are of common bean, with much smaller collections of the other cultivated species,
and a small percentage of wild species. The largest collections of CWR of Phaseolus are held
in the international collection managed by the Consultative Group on International Agricultural
Research (CGIAR), with close to 2000 accessions (The Consultative Group on International
Agricultural Research [CGIAR], System-wide Information network for Genetic Resources
[SINGER] 2009) and in the United States National Genetic Resources Program, with close to
500 accessions) (USDA, ARS, National Genetic Resources Program [NGRP] 2009).
MATERIALS AND METHODS
An eight-step gap analysis process is presented, which attempts to evaluate conservation
deficiencies at three different levels: (1) taxonomic, (2) geographic and (3) environmental. The
aim is to define the extent to which current genebank holdings represent total genetic diversity
92

A gap analysis methodology for collecting crop genepools
within a genepool. We apply the protocol to all the wild members of the genus Phaseolus.
Based upon the average of overall taxonomic, geographic, and environmental coverage factors,
the method produces a table outlining the high, medium and low priority species for collecting.
From this table, potential collecting areas for high priority species may be highlighted, and
overlapping high priority regions for the collection of multiple taxa identified. In detail, the
method is as follows:
Determination of target taxa, delineation of target area and harvesting of occurrence data
This involves five steps:
1. Identification of the target cultivated species.
2. Taxonomical review of all CWR related to the cultivated species, and analysis of
relatedness to the domesticated species using the concept established by Maxted et al.
(2006).
3. Creation of a database containing as many records as possible both of genebank
accessions and herbarium specimens, along with (when available) their respective
passport data, specifically the names of the places of collection and coordinates (i.e.,
latitude and longitude). Samples listed as weedy or cultivated are not included in the
database.
4. Cross-check, verification, and correction of geographic references (coordinates) through
thorough review of data and use of verification tools such as BioGeomancer
(www.biogeomancer.org) (Guralnick et al. 2006), Google Earth, and high detail physical
maps of localities, and strict selection only of verified geo-referenced samples for
distribution modeling, as the quality of location data strongly affects the performance of
niche modeling techniques (Graham et al. 2008).
5. Determine target area for the gap analysis: based upon the native (wild) distribution of
the target taxa. Depending on the genepool, the area can range from a small region within
a country to the entire world.
Determination of sampling deficiencies at the taxon level
A gross representativeness of genebank accessions for each taxon is calculated using the
‘sampling representativeness score’ (SRS, Equation 1), comparing total germplasm accessions
to herbarium records.
93

Chapter 4

SRS =

GS
*10
GS + HS

(Equation 1)

SRS is calculated as the number of germplasm samples (GS) divided by the total number of
samples, i.e., the sum of germplasm plus herbarium samples (HS), regardless of whether
samples contain location data. SRS therefore permits a general estimation of adequacy of
germplasm holdings of each taxon based upon all available data. In the case that a taxon has no
genebank samples, it is listed as a “high priority species” for collecting by setting the FPS (see
step 7 below) to 0.
In the rare case that for a particular taxon there is obviously deficient herbarium sample data in
comparison to germplasm records, the analysis should eliminate SRS as an input for that taxon,
as its inclusion would overestimate adequacy of conservation. Mapping of herbarium samples
and genebank accessions can be performed (e.g., using DIVA-GIS (version 7.1.70) (Hijmans
et al. 2001, Hijmans et al. 2005b) in order to provide a general geographic assessment of the
available data.
Create potential distribution models for taxa
Potential distributions of taxa are calculated using the maximum entropy (Maxent) model
(Phillips et al. 2006), with a set of bioclimatic variables and species presence data as inputs.
We do not consider the total number of samples with coordinates but the number of different
populations represented by those samples (unique locations) (Hernandez et al. 2006, Phillips et
al. 2006, Loiselle et al. 2008, Costa et al. 2010). We use Maxent due to its precision and
confidence when predicting species distributions (Dormann 2006, Elith et al. 2006, Hernandez
et al. 2006, Hijmans & Graham 2006, Phillips et al. 2006, Loiselle et al. 2008, Phillips & Dudik
2008, Costa et al. 2010). Default features are used in Maxent, in which complexity of the
models (represented by the number of terms and the type of interactions between environmental
variables) depend upon the number of input data points (Phillips et al. 2006, Phillips & Dudik
2008). Background points for model training equal 10,000 random points over the distributional
range of the genepool in order to avoid overfitting (Phillips 2008, VanDerWal et al. 2009).
As the Maxent distribution is generally broader than the real distribution of the species, the
modeled distribution is further refined by selecting only known native areas and high
94

A gap analysis methodology for collecting crop genepools
probability zones, which generally are defined as the most climatically suitable for the taxon,
thus avoiding over-estimation of the realized niche (VanDerWal et al. 2009, Smolik et al.
2010). The potential distribution is limited to the native area reported in the literature and then
thresholded using the ROC (receiver operating characteristic) curve plot-based approach (point
on the ROC curve [sensitivity vs. 1-specificity] which has the shortest distance to the top-left
corner [0,1] in the ROC plot) (Liu et al. 2005, Phillips et al. 2006, Phillips & Dudik 2008). We
use this threshold as it provides a decent omission rate, is taxon-specific and shows better
performance than other thresholds when predicting potential presence (Liu et al. 2005). We call
this thresholded modeled distribution the “potential distribution coverage”.
Based on the above, for each taxon, we report three model performance metrics: (1) the 25-fold
average area under the ROC curve (AUC) (Liu et al. 2005, Elith et al. 2006, Phillips et al. 2006,
Phillips 2008) of test data (ATAUC), (2) the standard deviation of the test AUC of the 25
different folds (STAUC), and (3) the proportion of the potential distribution coverage with
standard deviation above 0.15 (ASD15). Maxent models with ATAUC above 0.7 (Smolik et al.
2010), STAUC below 0.15, and ASD15 below 10% can be considered “accurate and stable”
and are thus used in further calculations. We use three measures of model accuracy as the use
of AUC alone might mislead the interpretation given the sensitivity of this measure to spatial
autocorrelation (Lobo et al. 2008, Veloz 2009).
For those taxa for which the Maxent model training fails or is inaccurate or unstable, we assign
a priority to the taxa using the following criteria:
1. As with step (2), taxa with no genebank samples are listed as “high priority species” for
collecting by setting the FPS (see step 5 below) to 0.
2. Taxa with genebank samples but no herbarium samples with verified location data are
listed as “high priority species” for collecting, as more data are needed in order to perform
the analysis. Taxa with such paucity of herbarium records are likely to also have limited
germplasm conserved, and are therefore very likely to be “high priority species”.
However, these taxa might differ from taxa in (3a) since they already have at least one
genebank accession, which certainly permits some type of analyses (e.g., genetic
diversity). These taxa are thus differentiated from taxa in (3a) by a flag in the final
priorities table (see results).
3. Taxa with genebank samples and one or more herbarium samples with verified location
95

Chapter 4
data are assessed using the area of the convex hull around all known populations (unique
locations) of the taxon in lieu of potential distribution coverage. We use the convex hull
since, particularly for taxa with very limited occurrence data, it provides a polygon
resembling the type of area produced by the Maxent distribution model.
At this point, the potential distribution coverage for all taxa (for which a niche model is
possible) may be mapped together in order to display the distribution of the genus, and a
richness map along with an uncertainty map (i.e., maximum standard deviation of probabilities
among the species that are present in each pixel) for the genepool may be calculated from the
results.
Geographic coverage assessment
The adequacy of geographic coverage of genebank accessions is calculated as a ‘geographic
representativeness score’ (GRS, Equation 2), assessed by comparing the taxon potential
distribution coverage with the genebank samples geographic coverage, modeled using the
‘circular area statistic with a 50 km radius’ (CA50) value (Hijmans et al. 2001).

GRS =

GCG
*10
PDC

(Equation 2)

GRS is thus the geographic coverage of germplasm collections (GCG) divided by the potential
distribution coverage of the taxon under analysis (PDC). The higher the GRS, the higher the
representativeness of genebank collections in relation to the potential distribution of the taxon.
Determination of environmental gaps
The adequacy of environmental coverage of genebank accessions is calculated as an
‘environmental representativeness score’ (ERS, Equation 3), assessed by comparing the
germplasm samples in relation to the full environmental range of the modeled taxon
distribution. The same set of climatic layers used for developing the potential distribution
coverage are standardized to have an average of zero and a standard deviation of 1 in order to
perform a principal components analysis. The first two of these spatially explicit components
(which normally account for more than 70% of the spatial variability) are reclassified into
twenty equal classes.
96

A gap analysis methodology for collecting crop genepools
2

ERS = ∑ wi
i =1

ECi
*10
PEDi

(Equation 3)

For these two principal components (i = 2), ERS is calculated as the environmental coverage
(i.e., number of different classes) of germplasm collections (EC) divided by the potential
environmental coverage of the taxon under analysis (PED), times the weight (w) of the principal
component (weights of the two components are re-scaled so that the sum of their weights is 1).
If the total variation explained by the first two components is too small (i.e., less than 70%),
additional components can be included in the analysis, and should be weighted accordingly.
Rarity of each species based on environmental variables determination
All records for the genepool (i.e., GS + HS for all taxa combined) are plotted against a specific
environmental variable or linear combination of variables (i.e., vector or principal component)
to identify taxa with records falling in rare environmental classes (i.e., extremes of the
distribution). We assume that the frequency of the data presents a normal distribution and
‘environmentally rare’ taxa are those located in sites where extreme environmental conditions
are found (tails of the distribution - 5th [NS P5 ] and 95th [NS P95 ] percentiles). A numeric value
(environmentally rare taxa score, ERTS, Equation 4) is calculated for each taxon as the number
of populations in rare environments divided by the total number of populations of that taxon.

ERTS =

NS P 5 + NS P 95
*10
GS + HS

(Equation 4)

As this step of the gap analysis should be conducted only when there is sufficient data for all
the taxa under analysis in order to avoid bias in the results (an abundant number of populations
so that a histogram can be calculated), usually it will not be included in the overall assessment.
We suggest that this step can be usefully included for the assessment of a specific subset of
well-sampled species.
Numeric assessment to determine the priority of collecting for ex situ conservation
All level-specific representativeness scores (SRS, GRS, ERS, and if possible ERTS) are
averaged with equal weight to obtain a final score of prioritization of species. The ‘final priority
score’ (FPS), is then used to classify taxa according to the following ranges: (1) as high priority
97

Chapter 4
species if the FPS is between 0 and 3, (2) as medium priority species if the FPS is between 3.01
and 5, (3) as low priority species if the FPS is between 5.01 and 7.5, and (4) as well conserved
species (no need for further collection) if the FPS is between 7.51 and 10. All taxa flagged as
high priority in steps (2) and (3) are included in the list of high priority taxa to be further
collected.
Prioritization of geographic areas for collecting germplasm
The potential collection zones for each high priority species are identified separately and then
combined to highlight those zones where gaps for multiple species overlap (“collection gap
richness”). This is done through the following steps:
1. Identify un-collected zones for each taxon by comparing the potential distribution
coverage with the current geographic coverage of germplasm collections (CA50). Areas
where the taxon is potentially present but already sampled are dismissed at this stage; the
remaining areas are highlighted as uncollected.
2. Four products treating all mappable high priority taxa are finally produced: (1) individual
maps showing potential collecting zones of all high priority taxa, (2) a map of collection
gap richness: the number of different taxa that can be collected in each 2.5 arc-minutes
(~5 km at the Equator) grid cell, (3) a map showing the maximum standard deviation of
high priority taxa (derived from the 25-fold Maxent model training procedure) in each
pixel, and (4) a map of the maximum distance of each pixel to the nearest accession (this
calculation is done taxon-by-taxon and then aggregated into a single map output, by
calculating the maximum of all ‘high priority taxa’.
Testing the gap analysis methodology
The methodology relies on available data and utilizes modeling tools, and is therefore
vulnerable to the quantity and quality of input data and the limitations of the modeling applied.
In order to test the quality of the results, we have compared them to expert opinion, as following:
1. Identify one or more experts on the target taxa (i.e., genepool)
2. Query the selected expert(s) to provide
a. A ranking of taxa for importance for conservation: To achieve this, the list of taxon
names under analysis is sent to the expert(s), who is asked to provide a rating from 1
to 10 for each taxon (where 1 corresponds to a very high priority [i.e., an incomplete
collection], and 10 corresponds to the lowest priority [i.e., a complete collection]),
98

A gap analysis methodology for collecting crop genepools
without having seen the results of the gap analysis. The expert is requested to rate taxa
strictly on the basis of adequacy of ex situ holdings for the taxon.
b. The expert is then shown the results of the analysis and is asked to give general
comments on the validity of the taxa and geographic prioritizations.
3. Compare the expert and method-based prioritization of each taxon using the relative
difference (RD) between the expert priority score (EPS) and the gap analysis FPS, with
respect to the total maximum possible difference (Equation 5).

RD =

FPS − EPS
* 100
10

(Equation 5)

RD is calculated for each taxon and the number of taxa with very similar ratings
(-30% < RD < 30%), the number of taxa somewhat similar ratings (-50% < RD < 50%), and
the number of taxa with very different ratings (RD < -70% and RD > 70%) are then counted.
We also plot the FPS and the EPS in a scattergram and calculate both the Spearman correlation
coefficient and the P-value of the Spearman correlation coefficient. With these metrics, we aim
to provide a general evaluation of the gap analysis method in identifying high priority taxa in
comparison to best available expert knowledge.
RESULTS
Determination of target taxa, delineation of the target area and harvesting of occurrence
data
We conducted a literature review for the Phaseolus genus (Freytag & Debouck 2002, Salcedo
et al. 2006, Delgado-Salinas & Carr 2007, Debouck 2009, Salcedo et al. 2009), checked against
genepool experts (Debouck) and created a complete list of taxonomically verified species. We
used the concept established by Maxted et al. (2006), including with equal weight all taxa
belonging to taxon groups 1 to 4 of the genepool.
According to a recent revision of the Phaseolus genepool (Debouck 2009), there are 81 species
and 34 infra-specific taxa, totaling 115 taxa within the genepool. With various species
synonyms and historical revisions (Maxted et al. 2004, Freytag & Debouck 2002, Debouck
2009), specimen identification and data availability issues persist. Although taxonomically
verified herbarium specimens provided the bulk of the data used in the analysis, we also rely
99

Chapter 4
on the specimen identification performed by the individual holding institutions. Based on the
recent history of Phaseolus taxonomy, we made the following changes to the determination of
specimens used in the data: Any variant within P. polymorphus Wats. was considered as P.
polymorphus, and the same was done for P. coccineus L. and P. leptostachyus Benth. (Debouck
2000). The variants P. polystachyus subsp. smilacifolius (Pollard) Freytag and P. polystachyus
subsp. sinuatus (Nuttall ex Torrey & Gray) Freytag were considered as separated species (P.
smilacifolius and P. sinuatus, respectively), and the species P. pyramidalis Freytag and P.
palmeri Piper were merged into P. grayanus Woot. & Standl. The only infraspecific taxa that
were considered were those of wild teparies (P. acutifolius) and those of P. maculatus, for
which there was not enough evidence for merging into single species. For taxa with ongoing
taxonomic uncertainty (e.g., P. neglectus Hermann), we followed Debouck (2009) and CIAT’s
Genetic Resources Unit genebank practice. After these modifications, a total of 85 taxa were
finally listed, including 81 species and 4 infraspecific taxa.
We gathered data from all known available sources, including primary datasets accessed
directly from herbaria and genebanks, as well as online global databases, such as the Global
Biodiversity Information Facility (GBIF, www.gbif.org), the System-wide Information
Network for Genetic Resources (SINGER, www.singer.cgiar.org) database held by the CGIAR,
and the United States Department of Agriculture (USDA) Germplasm Resources Information
Network (GRIN, www.ars-grin.gov) database (Table 1).
Data were available for all taxa, including the 81 species, 2 subspecies and 2 varieties. The
entire dataset was carefully geographically verified and corrected using BioGeomancer, and,
when possible, new geographic references (coordinates) were added to the passport data. The
final dataset contained 11,442 records, of which 6,926 (60.5%) had coordinates or enough
location data to obtain coordinates, and 4,516 (39.5%) samples had no location data or
coordinates.
The analysis was based on the native range for the genus throughout the Americas (northeastern
United States to northern Argentina, including the Caribbean and the Galapagos Islands)
(Freytag & Debouck 2002, USDA, ARS, National Genetic Resources Program [NGRP] 2009).
Records outside the boundaries of the Americas, as well as those listed as weedy or cultivated,
were deleted and a final dataset was produced for analysis. The average total number of samples
100

A gap analysis methodology for collecting crop genepools
Table 1. List of institutions from which data was harvested.
Institution
Genebank accessions
Bioversity International
CIAT-Genetic Resources Unit (via SINGER)
German National Resource Centre for Biological Material (DSMZ)
International Livestock Research Institute (ILRI)
Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)
National Vegetable Germplasm Bank, Mexico (BANGEV)
Native Seeds/SEARCH (NSS)
Plant Breeding and Acclimatization Institute (IHAR)
US National Plant Germplasm System (NPGS-GRIN)
Sub-total
Herbarium samples
A Database System for Systematics and Taxonomy (SysTax)
Arizona State University Vascular Plant Herbarium
Bernice Pauahi Bishop Museum
Botanic Garden and Botanical Museum Berlin-Dahlem
Cahiers de Phaseologie (DGD)
Canadian Biodiversity Information Facility
Colorado State University Herbarium (CSU)
Comision nacional para el conocimiento y uso de la biodiversidad (CONABIO)
Dutch national node of the Global Biodiversity Information Facility (NLBIF)
Fairchild Tropical Botanic Garden Virtual Herbarium
GBIF-Spain
GBIF-Sweden
Harvard University Herbaria
Herbarium of the University of Aarhus
Instituto de Biologia, Universidad Nacional de Mexico, (IBUNAM)
Instituto de Ciencias Naturales
Instituto Nacional de Biodiversidad (Costa Rica)
Integrated Taxonomic Information System (ITIS)
Louisiana State University Herbarium
Missouri Botanical Garden
Museo Nacional de Costa Rica
Muséum national d'histoire naturelle et Réseau des Herbiers de France
National Botanic Garden of Belgium (NBGB)
National Museum of Natural History
NatureServe
NavNat, GE, FR
New Mexico Biodiversity Collections Consortium
New York Botanical Garden (NYBG)
Royal Botanic Gardens, Kew
The Deaver Herbarium, Northern Arizona University
University of Alabama Biodiversity and Systematics
University of California, Davis
University of Connecticut
University of Kansas Biodiversity Research Center
USDA Plants
Utah Valley State College (UVSC)
Sub-total
Total (genebank accessions and herbarium samples)

Number of
records
with
coordinates

Number of
records
without
coordinates

7
2278
0
0
0
7
37
0
1081
3410

51
250
2
271
21
0
1
17
771
1384

2
829
0
0
1486
0
33
1049
0
2
0
0
2
8
0
22
78
8
0
713
100
4
70
28
0
2
0
7
1
8
6
0
1
1
402
1
4863
8273

49
172
1
1
182
1
4
360
25
18
5
6
86
0
2
68
0
0
9
621
45
0
20
64
134
0
112
4
2
0
0
7
0
3
65
3
2069
3453

101

Chapter 4

Figure 1. Density of sampling (sampling richness) for (A) genebank assessions and (B) herbarium
records for Phaseolus.

per taxon was 144.8, but data was unevenly distributed. Samples were predominantly
concentrated in wild progenitors of domesticated species (i.e., P. acutifolius, P. coccineus, P.
dumosus, P. lunatus, P. vulgaris), comprising about 55% of the total records.
Germplasm collections of the Phaseolus genepool are not distributed equally in relation to total
herbarium collections (Figure 1). The number of genebank accessions in a 200 km cell ranged
from 1 to 273, while that of herbarium collections ranged from 1 to 373. Observable differences
in the two maps (gaps) are present in the eastern United States, Costa Rica, Nicaragua, and in
the north of Mexico and along its border with United States. Most of the areas in central Mexico
are however well sampled and it is possible that species occurring in those areas are adequately
conserved. This was also observed in some areas in South America (particularly in the
Colombian, Ecuadorian and Peruvian Andes), where a greater proportion of genebank
accessions have been collected, potentially indicating a better coverage of taxa in genebanks
for populations from these regions.

102

A gap analysis methodology for collecting crop genepools

Figure 2. Number of genebank accessions versus all samples (genebank accessions plus herbarium
specimen records).

Sampling deficiencies at the taxa level
Of 85 taxa, 35 (41.2%) had no germplasm accessions, 26 taxa (30.6%) had 1-9 accessions, and
24 taxa (28.2%) had 10 or more accessions. From the 85 taxa, 61 (71.8%) taxa presented a SRS
below 3, indicating poor representativeness of the number of genebank accessions in relation
to herbarium collections, whilst 16 taxa (18.8%) showed SRS between 3.01 and 5, 4 (4.7%)
between 5.01 and 7.5, and 4 (4.7%) greater than 7.5.
The total representativeness (only in terms of the total number of samples, Figure 2 –
intermittent line) is above the average representativeness of germplasm collections (continuous
line), signifying that on average, species are likely to have fewer genebank accessions than
herbarium specimens. P. vulgaris, P. acutifolius and P. lunatus appear well conserved in
relation to both the gross number of accessions (compared to other taxa), and in proportion to
their respective number of herbarium records.
Potential distribution models for taxa
We used high-resolution global climatic datasets developed by Hijmans et al. (2005).
WorldClim includes monthly data at 30 arc-seconds resolution (approximately 1 km near the
Equator) for total precipitation, and mean, maximum and minimum temperatures. Using such
103

Chapter 4
Table 2. List of derived bioclimatic variables used in the analysis.
ID
Variable name
P1
Annual mean temperature
P2
Mean diurnal temperature range
P3
Isothermality
P4
Temperature seasonality (standard deviation)
P5
Maximum temperature of warmest month
P6
Minimum temperature of coldest month
P7
Temperature annual range
P8
Mean temperature of wettest quarter
P9
Mean temperature of driest quarter
P10
Mean temperature of warmest quarter
P11
Mean temperature of coldest quarter
P12
Annual precipitation
P13
Precipitation of wettest month
P14
Precipitation of driest month
P15
Precipitation seasonality (coefficient of variation)
P16
Precipitation of wettest quarter
P17
Precipitation of driest quarter
P18
Precipitation of warmest quarter
P19
Precipitation of coldest quarter

Units
ºC
ºC
N/A
ºC
ºC
ºC
ºC
ºC
ºC
ºC
ºC
mm
mm
mm
%
mm
mm
mm
mm

monthly datasets, 19 bioclimatic variables have been derived (Busby 1991), representing
average yearly climates, stressful and extreme conditions, and interannual seasonality (Table
2).
We downloaded WorldClim data at 30 arc-seconds, calculated the bioclimatic indices and
aggregated the 30 arc-seconds datasets to 2.5 arc-minutes using a bilinear interpolation in order
to reduce the computational time and data storage needs. Although most of the bioclimatic
indices used to develop the niche models are highly correlated (particularly in the tropics), we
used the complete set of 19 bioclimatic variables in Table 2 because (1) they are useful to
provide the best possible description of the climatic requirements of species during a single
average year, (2) these correlations might not hold in space and time, (3) the alternative
approach of dropping some variables leads to underestimation of distributions and poor
performance of Maxent (Hijmans & Graham 2006), (4) the alternative approach of reducing the
set of variables to a subset of orthogonal vectors (Dormann 2006) might lead to loss of valuable
climatic information and tends to complicate the interpretation of results of the application of
the niche model, and (5) the Maxent model prevents over-fitting due to the use of a set of
correlated environmental predictors by assigning weights based on the relative importance of
the variable to the model (Elith et al. 2006, Phillips et al. 2006, Phillips & Dudik 2008).

104

A gap analysis methodology for collecting crop genepools

Figure 3. (A) Taxon richness for Phaseolus based upon potential area of distribution of all taxa. (B)
modeling uncertainties as maximum standard deviations among all modeled taxa.

The geographic distributions of 51 out of the 85 taxa were considered sufficiently accurate and
stable to be mapped. Potential distribution coverage was estimated via the convex hull method
for 3 additional taxa (P. marechalii, P. salicifolius, and P. rotundatus). Therefore, a total of 54
taxa were assessed further.
The genus was modeled to occur from the northern border of the United States through Central
America, and along the Andean chain into northern Argentina (Figure 3A). Potential taxon
richness ranged from 1 to 23 taxa per grid cell. Taxon diversity hotspots were mainly found in
southern and western Mexico and in the southern United States, as well as some highland areas
of Guatemala, Honduras and Costa Rica, where 6 to 11 taxa are potentially distributed in a
single 5 km pixel.
Uncertainties in modeling distributional range calculated by the maximum standard deviation
among any possible class (i.e., taxon) varied from 0 to 0.32 (Figure 3B), with the vast majority
of the potential distribution coverage of the genus presenting a modeling uncertainty below
10%, and only very few areas presenting more than 15% variation in predicted probabilities.
High uncertainty areas do not coincide with high diversity areas, confirming the
105

Chapter 4

Figure 4. Geographic coverage of genebank accessions against total potential distribution coverage of
taxa.

reliability of the Maxent algorithm in predicting the geographic distributions of our set of taxa.
These small spots are located in southwestern Mexico along the very western edge of Nayarit
(municipalities of El Nayar, Rosamorada, Tepic), along the borders of Guerrero and Oaxaca, in
northern Oaxaca, and in northeastern Michoacán. Despite the observed uncertainties, these
areas with more than 15% variability among predictions account to less than 10% of the total
potential distribution coverage of the genus.
Geographic coverage assessment
The comparison between the CA50 and the size of the potential distribution showed that there
are 30 taxa out of the 54 assessed (55.6%) with GRS below 3.01 (less than 30% of
representativity in terms of geographic coverage), 12 taxa (22.2%) with GRS between 3.01 and
5, 4 taxa (7.1%) with GRS between 5.01 and 7.5, and 8 taxa (14.8%) with GRS greater than
7.5. The great majority of taxa have germplasm collections covering a geographic range
considerably smaller than the potential geographic area in which the taxon is distributed (Figure
4), thus indicating the need for further collecting in order to fill geographic gaps.
The average representativeness line (intermittent line) is above the complete representativeness
line (continuous line), indicating that the representativeness of germplasm collections in
106

A gap analysis methodology for collecting crop genepools
comparison to the total potential distribution coverage is low on average, and relatively high
only for a few species (namely the wild progenitors P. vulgaris, P. coccineus, P. acutifolius and
P. lunatus).
Determination of environmental gaps
The principal components analysis showed that the first two components explained up to 81.5%
of the total spatial variability among the Phaseolus genepool target area (61.2 and 20.3% for
PC1 and PC2 respectively). Re-scaling of these two components’ weights resulted in a weight
of 75.03% for PC1 and 24.97% for PC2. Out of the 54 modeled taxa, 10 (18.5%) presented
ERS below 3.01, indicating a significantly low environmental representativeness (i.e., less than
30%) in germplasm collections; 7 (13%) taxa presented an ERS between 3.01 and 5; 7 taxa
(13%) between 5.01 and 7.5; and 30 taxa (55.6%) above 7.5. Notably, environmental
representativeness of genebank accessions was found to predominantly fit in the two extreme
classes (below 30% and above 75%) for most of the taxa.
P. vulgaris and P. lunatus showed the highest coverage of potential environmental range, with
8 and 14 respectively out of the 20 classes along PC1, and 8 and 16 classes along PC2 (Figure
5). Germplasm representativeness of these environmental classes is for both species
significantly high (90% or more representativeness in both classes). For wild P. vulgaris, among
other cases (Figure 5), we found the environmental distribution of genebank accessions to be
broader than the environmental distribution of the potential distribution coverage, which may
be explained as an artifact given the use of the ROC-plot based threshold for binning the species
distributions (i.e., the omission rate), the native area (i.e., one or two small localities where the
taxon occurs might not be reported in literature), or the use of the CA50 around germplasm
locations, which might enlarge the range towards unsuitable habitats, particularly where the
landscape changes rapidly (e.g., topographically diverse regions, such as the Andes). A broad
range of adaptation to climatic conditions may be covered by current germplasm collections,
but it should be noted that small environmental gaps remain even for these well-sampled
species.
Rarity of each species
Rarity of species was not included in the analysis since there were significant sampling biases
that would lead to inaccurate results. In order to produce accurate results, the weight of the
107

Chapter 4

Figure 5. Coverage of genebank accessions versus potential environmental area for modeled species for
the first (left) and second (right) principal components.

108

A gap analysis methodology for collecting crop genepools
Table 3. List of taxa and data included in the analysis (see Supplementary Table 1 for full details).
Taxon
Sect. Acutifolii
P. acutifolius
P. acutifolius var. acutifolius
P. acutifolius var. tenuifolius
P. parvifolius
Sect. Bracteati
P. macrolepis
P. talamancensis
Sect. Brevilegumeni
P. campanulatus
P. oligospermus
P. tuerckheimii
Sect. Chiapasana
P. chiapasanus
Sect. Coccinei
P. coccineus
Sect. Coriacei
P. maculatus
P. maculatus ssp. maculatus
P. maculatus ssp. ritensis
P. novoleonensis
P. reticulatus
P. venosus++
Sect. Digitati
P. albiflorus
P. albiviolaceus+
P. altimontanus#
P. neglectus
P. trifidus++
Sect. Falcati
P. leptostachyus
P. macvaughii
P. micranthus
P. opacus++
P. persistentus+
Sect. Minkelersia
P. amabilis++
P. amblyosepalus
P. anisophyllus++
P. nelsonii
P. parvulus
P. pauciflorus
P. perplexus
P. plagiocylix++
P. pluriflorus
P. tenellus
Sect. Paniculati
P. acinaciformis+
P. albinervus++
P. augusti
P. jaliscanus
P. juquilensis+
P. lignosus+
P. longiplacentifer+
P. lunatus
P. maculatifolius++
P. marechalii
P. mollis++
P. nodosus++
P. pachyrrhizoides
P. polystachyus
P. rotundatus++
P. salicifolius
P. scrobiculatifolius+
P. sinuatus#
P. smilacifolius++
P. sonorensis++
P. xolocotzii++
Sect. Pedicellati

HS1

HS
(RP)3

GA2

GA
(RP)3

Total

Total
(RP)3

FPS
(GAP)

Class
(GAP)4

EPS
(DGD)

Class
(DGD)4

219
87
177
62

119
75
103
56

396
211
232
37

67
81
93
22

615
298
409
99

186
154
188
74

5.7
6.8
6.1
4.5

LPS
LPS
LPS
MPS

NA
4
5
4

NA
MPS
MPS
MPS

24
13

6
4

3
2

3
1

27
15

6
4

8.3
7.5

NFCR
LPS

4
6

MPS
LPS

4
26
43

4
22
24

0
13
3

0
11
2

4
39
46

4
33
26

0.0
5.8
3.5

HPS
LPS
MPS

0
3
3

HPS
HPS
HPS

53

8

3

3

56

8

4.1

MPS

2

HPS

1041

356

417

206

1458

560

7.3

LPS

4

MPS

106
203
190
4
6
10

62
138
120
3
4
6

39
30
68
2
2
0

17
18
30
1
2
0

145
233
258
6
8
10

79
151
150
3
6
6

4.0
4.5
4.6
3.6
2.3
0.0

MPS
MPS
MPS
MPS
HPS
HPS

NA
4
2
2
3
0

NA
MPS
HPS
HPS
HPS
HPS

49
1
2
9
1

4
1
2
6
1

1
0
2
0
0

1
0
2
0
0

50
1
4
15
1

4
1
2
11
1

4.3
0.0
NA
0.0
0.0

MPS
HPS
HPS
HPS
HPS

6
2
4
2
NA

LPS
HPS
MPS
HPS
NA

308
11
21
4
1

170
7
9
1
1

115
1
2
0
0

102
1
1
0
0

423
11
23
4
1

270
7
10
1
1

6.7
1.4
2.1
0.0
0.0

LPS
HPS
HPS
HPS
HPS

4
2
4
NA
0

MPS
HPS
MPS
NA
HPS

8
10
2
38
168
234
11
4
86
21

1
10
2
32
101
161
7
2
56
9

0
0
0
0
29
4
2
0
10
2

0
0
0
0
17
2
1
0
7
1

8
10
2
38
197
238
13
4
96
22

1
10
2
32
118
163
8
2
63
9

0.0
0.0
0.0
0.0
3.2
4.4
1.7
0.0
4.0
1.0

HPS
HPS
HPS
HPS
MPS
MPS
HPS
HPS
MPS
HPS

0
0
0
2
2
2
3
0
3
0

HPS
HPS
HPS
HPS
HPS
HPS
HPS
HPS
HPS
HPS

1
3
27
66
1
2
1
575
2
10
14
2
5
580
3
10
1
76
13
16
1

1
1
15
12
1
2
1
275
1
4
6
2
2
344
2
3
1
12
2
3
1

0
0
43
2
0
0
0
742
0
5
0
0
21
6
1
1
0
1
0
0
0

0
0
39
1
0
0
0
342
0
2
0
0
20
2
1
1
0
1
0
0
0

1
3
70
68
1
2
1
1317
2
15
14
2
26
586
3
11
1
77
2
16
1

1
1
54
12
1
2
1
616
1
4
6
2
22
346
2
4
1
12
1
3
1

0.0
0.0
7.4
1.8
0.0
0.0
0.0
6.9
0.0
8.3
0.0
0.0
7.8
0.9
6.7
7.2
0.0
NA
0.0
0.0
0.0

HPS
HPS
LPS
HPS
HPS
HPS
HPS
LPS
HPS
NFCR
HPS
HPS
NFCR
HPS
LPS
LPS
HPS
HPS
HPS
HPS
HPS

0
0
7
2
0
0
0
4
0
3
0
2
8
2
5
0
0
2
0
0
0

HPS
HPS
LPS
HPS
HPS
HPS
HPS
MPS
HPS
HPS
HPS
HPS
NFCR
HPS
MPS
HPS
HPS
HPS
HPS
HPS
HPS

109

Chapter 4
P. dasycarpus++
P. esperanzae
P. grayanus
P. laxiflorus++
P. oaxacanus++
P. pedicellatus
P. polymorphus
P. purpusii++
P. scabrellus+
P. teulensis+
P. texensis
Sect. Phaseoli
P. albescens
P. costaricensis
P. dumosus
P. vulgaris
Sect. Revoluti
P. leptophyllus+
Sect. Rugosi
P. angustissimus
P. carteri
P. filiformis
Sect. Xanthotricha
P. esquincensis++
P. gladiolatus++
P. hintonii
P. magnilobatus
P. xanthotrichus
P. zimapanensis
Not classified
P. glabellus
P. microcarpus

5
26
184
5
6
129
23
5
4
1
7

5
15
77
1
2
71
5
1
4
1
6

0
7
49
0
0
8
1
0
0
0
0

0
7
36
0
0
8
1
0
0
0
0

5
33
233
5
6
137
24
5
4
1
7

5
15
113
1
2
79
6
1
4
1
6

0.0
4.4
5.0
0.0
0.0
2.9
1.4
0.0
0.0
0.0
0.0

HPS
MPS
MPS
HPS
HPS
HPS
HPS
HPS
HPS
HPS
HPS

0
2
3
0
3
4
3
0
2
0
3

HPS
HPS
HPS
HPS
HPS
MPS
HPS
HPS
HPS
HPS
HPS

8
64
52
284

8
44
14
209

0
4
9
1674

0
3
7
452

8
68
61
1958

8
46
14
661

0.0
6.6
6.5
8.9

HPS
LPS
LPS
NFCR

2
6
5
7

HPS
LPS
MPS
LPS

6

1

0

0

6

1

0.0

HPS

0

HPS

617
8
682

269
3
397

17
5
98

8
2
46

634
13
780

275
4
441

2.8
3.9
4.6

HPS
MPS
MPS

2
2
2

HPS
HPS
HPS

4
1
12
16
11
10

3
1
7
7
8
5

0
0
11
2
38
16

0
0
7
1
30
13

4
1
23
18
49
26

3
1
14
8
38
17

0.0
0.0
4.3
1.6
9.0
7.3

HPS
HPS
MPS
HPS
NFCR
LPS

0
3
2
2
5
6

HPS
HPS
HPS
HPS
MPS
LPS

128
223

42
161

15
51

10
35

160
274

42
193

3.7
5.1

MPS
LPS

5
4

MPS
MPS

Number of herbarium specimens. 2Number of genebank accessions. 3Refers to the number of populations (unique
locations identified) represented by the set of samples. 4Prioritization of taxa is done as follows: HPS: High priority
species, MPS: Medium priority species, LPS: Low priority species, NFCR: No further urgent conservation
required. FPS indicates the result of the method proposed in this paper, and EPS indicates the prioritization given
by expert knowledge (based on Daniel G. Debouck’s expertise in Phaseolus). +Indicates that the taxon had no
genebank accessions and no herbarium samples with coordinates or location data; ++indicates a taxon for which
a Maxent model was not possible and for which 0-few genebank accessions were available; #indicates a taxon
with some genebank accessions but no or limited herbarium samples with coordinates or location data. These taxa
are listed as HPS for further collecting in order to inform the gap analysis.
1

ERTS was finally established at 0.05, which is practically irrelevant and thus the step was
dropped. If a subset of species with reliable sampling were to be analyzed separately (e.g., the
five wild progenitors of the domesticated species), however, the ERTS could be calculated and
weighted equally with the other scores when calculating the FPS.
Numeric assessment to determine the priority of collecting for ex situ conservation for
each taxon
Out of the 85 taxa under analysis, 48 (56.5%) are either under-represented or not represented
in any way in genebanks and therefore flagged as HPS for collecting (Table 3). Of these taxa,
35 had no germplasm accessions, and 11 are listed as HPS due to the average of gross
representativeness, geographic, and environmental gaps (FPS below 3.01). A further 2 taxa (P.
sinuatus and P. altimontanus) couldn’t be assessed due to uncertainties in the modeling and the
110

A gap analysis methodology for collecting crop genepools
data, and are included as HPS due to the need for collecting in order to provide adequate data
for a gap analysis.
Medium priority for further collecting was given to 17 taxa (20%), 15 taxa (17.7%) were given
low priority, and only 5 taxa (P. macrolepis, P. marechalii, P. pachyrrhizoides, P.
xanthotrichus and P. vulgaris) were assessed as well represented in ex situ collections.
Prioritization of geographic areas for collecting germplasm
Thirty-six priority taxa (i.e., those flagged as high priority and with sufficient location data)
were mapped together, along with standard deviations on predicted Maxent probabilities
(aggregated for all the taxa using the maximum value) and distances to the nearest population
(also aggregated) (Figure 6). Potential collection sites have a richness of up to 7 taxa per grid
(Figure 6A). Zones where gaps in ex situ collections for many Phaseolus taxa overlap are
concentrated in central-western Mexico, with an extension along the Sierra Madre Occidental
north to Sonora.
Andean environments where Phaseolus species are likely distributed appear in general to be
adequately represented in genebanks for most of the species. Note that the narrow endemic
nature of many of the under- or un-sampled taxa results in a need for very finely targeted
collection trips to specific regions outside of the gap richness areas identified, for example to
collect from populations of P. carteri, P. novoleonensis, and P. plagiocylix in isolated regions
of Mexico, and P. mollis in South America.
The maximum modeling uncertainty (given by the maximum standard deviation of the 25 folds
per taxon) was slightly greater than 15% in a very small area (dark blue spot in western Nayarit,
Figure 6B). Interestingly, modeling uncertainties of high priority taxa had a maximum of 19%,
significantly lower than uncertainties of the whole set of taxa under analysis (Figure 3B),
strengthening confidence in results regarding high priority taxa. The distance to verified
populations (Figure 6C) was greatest (i.e., uncertainty) in northwestern Mexico (southern
Sonora, northern Sinaloa, and southwestern Chihuahua). The areas identified in these
uncertainty analyses are least likely to contain target species.

111

Chapter 4

Figure 6. (A) Zones where gaps in ex situ collections for multiple taxa overlap (collecting gap richness)
for high priority species. (B) modeling uncertainties as standard deviations among high priority modeled
taxa. (C) collecting uncertainties as maximum geographic distance to nearest known population.

Comparison with expert opinion
The expert authority for Phaseolus was Daniel G. Debouck (DGD), head of the Genetic
Resources Unit at the International Center for Tropical Agriculture (CIAT), author and coauthor of numerous publications on Phaseolus, including a survey of the Phaseolus genepool
in North and Central America (Freytag & Debouck 2002), who has participated in many
collecting missions for the genus throughout the Americas and has extensive expertise in
taxonomy (including research at 67 different herbaria in the last 32 years), ecogeographic
distributions, and level of in situ and ex situ conservation of the genepool.
DGD did not assess 4 taxa: P. maculatus, and P. acutifolius since he considered it enough to
assess the subspecies and/or variants, and P. trifidus and P. opacus, since he considered them
as doubtful taxa. All figures below are thus based on the total number of taxa assessed by DGD
(81). Further taxonomic analyses of these species are needed in order to inform conservation
priorities.
In comparison to expert opinion, the gap analysis approach tended to underestimate priority for
collecting in a considerable number of cases (30.9% of the taxa); however, scores for 28 taxa
(34.6%) did align with expert opinion (with 0 as score for 24 of these). For 51 taxa (63%), the
method and DGD agreed on the priority class, and from the remaining proportion, the difference
was of one single class. In addition, the relative difference (RD) varied from -50% to 72.2%
and the maximum difference between our approach and the expert’s concept was around 7
112

A gap analysis methodology for collecting crop genepools

Figure 7. (A) Frequency distribution of the relative difference [RD] and (B) linear trend between Final
Priority Score (FPS) and Expert Priority Score (EPS) (the red dotted line indicates a 95% confidence
interval).

units in the priority scale of 10 units. Moreover, 87.7% of the validated taxa
(81) presented differences lower than 30% or greater than -30%, and only 2 taxa presented more
than a 50% or less than -50% difference (P. salicifolius with 7.2 in EPS and 0 in FPS, P.
marechalii with 8.3 in EPS and 3 in FPS). Only P. salicifolius was found to have more than
70% difference between EPS and FPS (Figure 7A).
The linear trend between EPS and FPS has a Spearman correlation coefficient of 0.79 (p <
0.0001, n=79). However, as previously stated, the gap analysis approach tends to underestimate
the priorities compared to expert opinion (average underestimation is -10.7%, Figure 7B).
A number of taxa fall far from the linear trend (i.e., P. neglectus, P. albiflorus, P. salicifolius
and P. pachyrrhizoides). Whilst for P. pachyrrhizoides this is due to a very high accuracy (ERS
and FPS are equal) in comparison with the propagating error in the regression line (i.e., the
underestimation error), differences for other taxa generally result from lack of geographic data
for a robust gap analysis, likely taxonomic misidentifications in records, and/or difficulty in
eliminating duplicates in records (e.g., P. neglectus, P. albiflorus).
For species such as P. xanthotrichus and P. oligospermus, the gap analysis approach indicated
little need for further collection, as germplasm has been collected throughout most of the region
of recorded herbarium collections and environments occupied by those collections. However,
expert knowledge on other areas of distribution of the species, under-recorded in online
herbarium data, gave the species higher priority on the EPS.
113

Chapter 4
DISCUSSION
Success of the gap analysis method in identifying priority taxonomic, geographic, and
environmental gaps is directly dependent on the quality of input data and robustness of the
modeling based upon the data. In this section we discuss uncertainties and limitations
concerning the method:
Input data availability, bias and certainty
The quality of the input geographic information (i.e., climatic and occurrence data) directly
affects the performance of species distributions models (Dormann 2006, Hijmans & Graham
2006, Graham et al. 2008, Loiselle et al. 2008, Wisz et al. 2008, Costa et al. 2010). Geographic
data for specimens is generally less than optimal and is unevenly distributed across taxa, due to
the bias of collecting activities toward particular species or locations, a historically insufficient
prioritization of recording and maintaining of geographic data, lack of high quality absence data
for species, and limited accessibility of stored data for some collections. Many regions of the
world remain un- or under-sampled, particularly highly inaccessible areas, and those
chronically affected by war or civil strife.
Recently described and/or under-studied taxa, such as P. acinaciformis, P. juquilensis, P.
longiplacentifer, P. persistentus, P. scrobiculatifolius, P. teulensis, P. albiviolaceus, P.
leptophyllus, P. lignosus, P. scabrellus, and P. sinuatus, may require further taxonomic
clarification, and are generally in need of further collecting, and characterization of the
collected populations, in order to clarify identification and facilitate accurate prioritization.
Infraspecific taxa (variants and subspecies), such as those of P. maculatus and P. acutifolius,
may also be incompletely treated in the analysis due to data constraints. There are several
records of these species that remain undetermined at the infraspecific level. Due to overlapping
ranges of distribution for various infraspecific taxa, unassigned records cannot be easily
differentiated based on collection location. In the gap analysis we have therefore assessed both
the species level and the infraspecific taxa.
More germplasm of Phaseolus may be conserved worldwide than the accession data used in
this analysis indicate, as the data from some genebanks was not accessible. We assume that,
with few exceptions, the accessions whose data was not accessible are also generally
114

A gap analysis methodology for collecting crop genepools
inaccessible to crop breeders and researchers worldwide. Areas where these collections were
made may not represent a gap for the particular holding collection, yet they are effectively a
very real gap for rest of the global community.
Duplication between and within institutes might inflate the numbers of unique records for some
of the taxa, leading to bias in the prioritization results. The use of different numbering systems,
and lack of tracking of former records, leads to an overestimation of samples held, and difficulty
in identifying duplicates, perhaps especially for the most commonly exchanged species (e.g.,
wild progenitors). For Phaseolus, we found that large differences can exist between the number
of records and of actual populations both for genebanks (up to 83.1%) and herbaria (up to
87.5%). The data preparation phase of the analysis involved a thorough identification of
duplicates in order to avoid inflation of numbers of records and therefore biases in prioritization.
Further, the geographic representativeness score (GRS) takes distinctness/uniqueness of
populations into account indirectly, and the environmental representativeness score (ERS)
addresses the issue by illuminating gaps in the abiotic adaptations of the sampled material (i.e.,
number of different climatic environments covered by the conserved material).
Location data constraints may also limit the taxa for which the method may be applied, as well
as lead to an underestimation of taxon distributions. From the 45 different data sources, 24
(53.3%) had more records without location data than with location data, and only 9 (20%) of
the sources presented all of their records with coordinates or with detailed location data (Table
1). For genebanks, 71.1% of the data presented reliable location data and 28.9% had either no
location data or location data were unreliable, whilst for herbaria, 70.2% of the data presented
coordinates and 29.8% did not present any useful location data.
Additional data, such as absence of the taxon, would certainly improve model-training by
increasing the model’s ability to discriminate between presence and absence areas. These data
are unfortunately not available in conventional genetic resources databases (Phillips & Dudik
2008, Lobo et al. 2010). Future collecting should be planned with an eye to the improvement
of gap analysis approaches and should thus consider a more systematic recording of absences,
geo-referencing all records, and making widely accessible data from all available germplasm
and herbarium samples. These actions will improve the performance of species distribution
models and any conclusions drawn from them.
115

Chapter 4
Ability of the species distribution model used (i.e., Maxent) to adequately predict the
potential and realized niche of taxa
The Maxent modeling technique was chosen for its ability to handle sample bias and spatial
autocorrelation of species data (Loiselle et al. 2008, Phillips 2008, Costa et al. 2010) so as to
provide high confidence species distributions models even given limited or biased location data.
Maxent is an algorithm known to reliably predict the potential niches of species, and has been
tested by several authors under a wide range of conditions and configurations (e.g., Elith et al.
2006, Phillips et al. 2006, Loiselle et al. 2008, Phillips 2008, VanDerWal et al. 2009, Costa et
al. 2010, Smolik et al. 2010, among others); although we note that some reports (JimenezValverde et al. 2008, Lobo et al. 2008, Fitzpatrick & Hargrove 2009) consider niche modeling
techniques misleading and of limited use in certain contexts. As the robustness of Maxent is
considered in the publications listed above, we do not provide a full analysis here.
We used the average test-data AUC, which showed that 52 species distribution models were
reliable (i.e., accurate and stable). Using the current configuration, the AUC statistic is not
likely to be biased by the pseudo-absences range (VanDerWal et al. 2009). Discrimination
between presence and absence sites was therefore considerably positive for most of the taxa
(~70%). Particularly good was the performance of taxon distribution models with more than 40
data points.
Moreover, the uncertainties associated with the application of a probabilistic model such as
Maxent and depicted by the 25-fold cross-validated models for each of the taxa indicated that
standard deviations among predictions ranged from 0 to 0.19. Collecting priorities are more
uncertain in limited areas (e.g., along the western coast of Mexico), but are relatively robust
across the vast majority of the distributional range of the genepool.
However, there was a set of taxa (those marked with + in Table 3) for which we were not able
to develop species distributions models due to either lack of samples or to the distribution of
those samples. These species could benefit from other approaches, such as Bayesian techniques
(Termansen et al. 2006), which are able to develop probability surfaces even from a single
point. Here we did not include these additional approaches, given the uncertainties involved
with these models. We rather use specimen data (i.e., herbarium sampling points) to depict areas
where these species can be potentially collected.
116

A gap analysis methodology for collecting crop genepools
Geographic collecting priorities
To analyze the validity of geographic gap results, we have calculated the stability (standard
deviation) of the Maxent models and have also provided the distance to the nearest population
within the collection zone (Figure 6).
Additional analyses, including threat level, can be incorporated into the methodology in order
to refine conservation priorities. Possible threats that could lead to genetic erosion in wild
species populations include fires, grazing pressure, invasive species, deforestation, habitat
modification and degradation, urbanization, and climate change, among others (Tanksley &
McCouch 1997). Accession-level genetic data may also serve as an input in order to identify
gaps in genetic diversity. Additional environmental data, such as soil type, may further define
potential distributions of species. These additional inputs are currently only rarely available at
high detail over large geographic areas or for all taxa in a genepool, but this may improve with
the ongoing development of GIS and decreasing costs of genotyping. Taxon-specific
knowledge may also be used to refine or weight priorities, giving some species higher
importance in the final result (e.g., focusing on specific traits of interest, adjusting to
phytosanitary/noxious weed constraints, recognizing legal constraints to access, prioritizing in
order to capitalize on appropriate seasonal collecting windows, etc.)
In our approach, we include all wild relatives of the crop without regard to relatedness to
cultivated species, weighting them equally, with the assumption that a wide range of taxa are
potentially useful to provide genes for crop improvement (Challinor et al. 2007), recognizing
the lack of data on relatedness. Information on relatedness and threat level can be added to the
prioritization exercise by experts with specific interests or familiar with local conditions.
When this is done for Phaseolus the following gaps are highlighted. Collecting a few (1-5)
populations is needed for 35 taxa that currently have no genebank samples conserved. Out of
the five wild progenitors of the domesticated species, P. vulgaris and P. dumosus have been
relatively well sampled, and only small gaps remain to be filled. Briefly, gaps for wild P.
vulgaris are present in: Oaxaca, El Salvador, Panama, western Andes of Venezuela, northern
central Bolivia, and San Luis in Argentina. For wild P. dumosus: eastern Chiapas and Alta
Verapaz in Guatemala. For the remaining three progenitors, the gaps are substantially more
important. For P. acutifolius: Sonora, Chihuahua, many spots in western Mexico and in
117

Chapter 4
Guerrero. For P. coccineus: Chihuahua down to Guatemala. For P. lunatus, gaps exist
throughout the very large range (from the Revillagigedo Islands, Baja California Sur and
Sinaloa to Puerto Rico, and down to Salta and Formosa in Argentina.
Regarding the secondary genepool of each of the five cultigens: for common bean, runner bean,
and year bean, additional collecting is needed for P. albescens, P. costaricensis, and P.
persistentus. For tepary, collecting is needed for P. parvifolius (all across its range from
Chihuahua down into Guatemala). For Lima bean, concerted effort is required because few (if
any) accessions are available for taxa within Section Paniculati, as well as P. maculatus, P.
novoleonensis, P. reticulatus, P. ritensis, and P. venosus within Section Coriacei.
For the remaining Phaseolus species (not highly related to any cultigen given molecular
evidence available today), a few accessions exist for taxa such as P. chiapasanus, P.
esperanzae, P. pluriflorus, and P. micranthus. Remaining species are in need of further
collecting in order to secure germplasm ex situ.
Comparison with expert knowledge
The method performed well as compared with expert knowledge on the Phaseolus genepool,
81.2% of the taxa presenting differences between -30% and 30%, and only one taxon with a
difference of more than 70% between EPS and FPS. We note that although the expert will often
refine the analysis by adding further insight and by qualifying data, the gap analysis also holds
the potential to highlight taxonomic, geographic, and environmental gaps previously unknown
to the expert.
In order to provide a more robust test, multiple experts could be consulted. As GIS approaches
continue to expand and improve, a more comprehensive validation procedure may be performed
with a network of experts, facilitated through an online portal.
Expert intervention within the gap analysis method is especially critical during (1) thorough
taxonomical review of the genepool, including variants and/or subspecies changes according to
the latest studies, (2) the full evaluation and georeferencing of locality names in the dataset, and
(3) the further refining and correction of priorities when a data availability issue is detected.

118

A gap analysis methodology for collecting crop genepools
Expert taxonomic knowledge will of course also be vital in the actual field collecting, especially
for understudied species (e.g., P. albinervus, P. leptophyllus, and P. purpusii). This has proven
to be important in this genus, as numerous new species have been identified only during
germplasm collecting missions (e.g., P. altimontanus, P. costaricensis, P. novoleonensis, P.
persistentus, P. rotundatus, and P. talamancensis).
CONCLUSION
This study proposes a method for the rational prioritization of taxa within a genepool for
collection for ex situ conservation, using Phaseolus as a model. The method builds upon the
standard comparison of herbarium samples with genebank accessions via gap analysis (Maxted
et al. 2008), yet aims to address sampling biases by modeling species distributions with a robust
algorithm, and refining these distributions using two different criteria. Furthermore, the method
identifies priorities based not only on taxonomic and geographic gaps, but also environmental
gaps. Priority locations for sampling of gaps result, as well as gap richness models contributing
to the identification of collection locations for maximum efficiency. The results cover the four
target outcomes of gap analysis identified by Nabhan (1990). Collecting for ex situ conservation
should prioritize the resulting taxa, including those not or under-sampled ex situ, as well as
geographic and environmental gaps in the distribution of taxa with some degree of germplasm
currently conserved.
We found 48 high priority taxa (56.5%) (Table 3), 35 (41.1% of total) of these not recorded as
represented ex situ by even a single accession. Acknowledging that the results for a number of
these species may potentially be affected by data availability constraints, in the most optimistic
case, around half of the taxa in the genepool are highly under-represented in ex situ
conservation. There is therefore a clear need for further collecting in order to cover the full
range of taxonomic, geographic and environmental diversity.
The greatest priority regions for further collecting are located in northern Central America (i.e.,
Mexico and Guatemala), with a maximum potential sampling richness of 7 species per 5 km
cell. However, there are a number of species that require individually targeted efforts in other
areas (e.g., P. mollis, in the Galapagos Islands).
Additional criteria, such as threats to taxa, and degree of relatedness of taxa to cultivated
119

Chapter 4
species, may also be included in the analysis, when data is sufficiently available. In order to
include a more complete picture of conservation, the method should ideally be coupled with in
situ gap analysis results (e.g., Meilleur & Hodgkin 2004), i.e., comparison of distributions with
the extent of protected areas. In general, the high priority taxa identified in the analysis are
likely to be those also most highly prioritized for in situ conservation, although this was not
explored in the current analysis.
The method is applicable to any set of related taxa, given adequate geographic data and a
thorough taxonomic and geographic referencing process. Genepools whose taxonomy has not
received sufficient attention (e.g., Oryza in the Americas), or which have not been well sampled
for herbarium specimens, will present particular challenges in producing reliable results. As
each genepool is different, the analysis must be adapted according to data availability, and
tested against expert knowledge, preferably repeatedly. Once the method has been applied to a
number of crop genepools, the prioritization of taxa and “gap richness” mapping may be applied
for these genepools together, potentially facilitating the identification of priority regions (“plant
genetic resource gap megacenters”) for the efficient and effective collecting of CWR diversity
on a global scale.

120

87

177

62

P. acutifolius var. acutifolius

P. acutifolius var. tenuifolius

P. parvifolius

13

P. talamancensis

43

P. tuerckheimii

203

190

4

6

10

P. maculatus ssp. ritensis

P. novoleonensis

P. reticulatus

P. venosus++

49

1

2

9

1

P. albiflorus

P. albiviolaceus+

P. altimontanus#

P. neglectus

P. trifidus++

Sect. Digitati

106

P. maculatus ssp. maculatus

1041

P. maculatus

Sect. Coriacei

P. coccineus

Sect. Coccinei

P. chiapasanus

53

26

P. oligospermus

Sect. Chiapasana

4

P. campanulatus

Sect. Brevilegumeni

24

P. macrolepis

Sect. Bracteati

219

P. acutifolius

Sect. Acutifolii

Taxon

Number
of
herbarium
specimens

1

6

2

1

4

6

4

3

120

138

62

356

8

24

22

4

4

6

56

103

75

119

Number of
herbarium
specimens
(represented
populations)

0

0

2

0

1

0

2

2

68

30

39

417

3

3

13

0

2

3

37

232

211

396

Number
of
genebank
accessions

0

0

2

0

1

0

2

1

30

18

17

206

3

2

11

0

1

3

22

93

81

67

Number of
genebank
accessions
(represented
populations)

1

15

4

1

50

10

8

6

258

233

145

1458

56

46

39

4

15

27

99

409

298

615

Total

Supplementary Table 1. List of taxa and data included in the analysis.

1

11

2

1

4

6

6

3

150

151

79

560

8

26

33

4

4

6

74

188

154

186

Total
(represented
populations)
ATAUC
NA

0.997

0.5

NA

0.995

0.957

0.959

0.995

0.987

0.979

0.986

0.951

0.993

0.996

0.979

0.995

0.999

0.998

0.972

0.971

0.965

0.962

STAUC
NA

0.005

NA

NA

0.0131

0.0371

0.0309

0.002

0.0055

0.0096

0.0193

0.0207

0.0055

0.0066

0.0657

0.002

0.001

0.0055

0.0479

0.0308

0.0193

0.0324

ASD15
NA

0

NA

NA

0

12.9

0.2

0

0

0

0

0

0

0

0

0.2

0.4

0.8

0

0

0

0

SRS
0

0

10

0

2.5

0

3.3

3.3

2

1.2

2.2

3.7

3.8

0.8

3.3

0

2.5

5

3

4.9

5.3

3.6

GRS
NA

0

NA

NA

0.5

NA

0.4

2.4

3.2

2.3

0.8

8.1

2.3

2.1

3.9

0

10

10

2.2

3.4

5.3

3.5

ERS
NA

0

NA

NA

10

NA

3.2

5

8.5

10

9.2

10

6.3

7.7

10

0

10

10

8.5

10

10

10

FPS (GAP)
0

0

NA

0

4.3

0

2.3

3.6

4.6

4.5

4

7.3

4.1

3.5

5.8

0

7.5

8.3

4.5

6.1

6.8

5.7

Class (GAP)
HPS

HPS

HPS

HPS

MPS

HPS

HPS

MPS

MPS

MPS

MPS

LPS

MPS

MPS

LPS

HPS

LPS

NFCR

MPS

LPS

LPS

LPS

EPS (DGD)
NA

2

4

2

6

0

3

2

2

4

NA

4

2

3

3

0

6

4

4

5

4

NA

Class (DGD)
NA

HPS

MPS

HPS

LPS

HPS

HPS

HPS

HPS

MPS

NA

MPS

HPS

HPS

HPS

HPS

LPS

MPS

MPS

MPS

MPS

NA

A gap analysis methodology for collecting crop genepools

SUPPLEMENTARY INFORMATION

121

122

11

21

4

1

P. macvaughii

P. micranthus

P. opacus++

P. persistentus+

10

2

38

168

234

11

4

86

21

P. amblyosepalus

P. anisophyllus++

P. nelsonii

P. parvulus

P. pauciflorus

P. perplexus

P. plagiocylix++

P. pluriflorus

P. tenellus

1

3

27

66

1

2

1

575

2

10

14

2

P. acinaciformis+

P. albinervus++

P. augusti

P. jaliscanus

P. juquilensis+

P. lignosus+

P. longiplacentifer+

P. lunatus

P. maculatifolius++

P. marechalii

P. mollis++

P. nodosus++

Sect. Paniculati

8

P. amabilis++

Sect. Minkelersia

308

P. leptostachyus

Sect. Falcati

Taxon

Number
of
herbarium
specimens

2

6

4

1

275

1

2

1

12

15

1

1

9

56

2

7

161

101

32

2

10

1

1

1

9

7

170

Number of
herbarium
specimens
(represented
populations)

0

0

5

0

742

0

0

0

2

43

0

0

2

10

0

2

4

29

0

0

0

0

0

0

2

1

115

Number
of
genebank
accessions

0

0

2

0

342

0

0

0

1

39

0

0

1

7

0

1

2

17

0

0

0

0

0

0

1

1

102

Number of
genebank
accessions
(represented
populations)

2

14

15

2

1317

1

2

1

68

70

3

1

22

96

4

13

238

197

38

2

10

8

1

4

23

11

423

Total

2

6

4

1

616

1

2

1

12

54

1

1

9

63

2

8

163

118

32

2

10

1

1

1

10

7

270

Total
(represented
populations)
ATAUC
0.5

NA

0.973

NA

0.93

NA

NA

NA

0.99

0.987

NA

NA

0.994

0.992

0.5

0.964

0.986

0.986

0.995

0.5

0.982

NA

NA

NA

0.979

0.994

0.969

STAUC
NA

NA

0.0271

NA

0.0191

NA

NA

NA

0.0138

0.0102

NA

NA

0.0077

0.008

NA

0.0419

0.0069

0.0193

0.0046

NA

0.0475

NA

NA

NA

0.0264

0.0045

0.0082

ASD15
NA

NA

40.2

NA

0

NA

NA

NA

0

0

NA

NA

0.1

0

NA

0

0

0

0

NA

0

NA

NA

NA

0

0

0

SRS
0

0

5

0

5.6

0

0

0

0.8

7.2

0

0

1.1

1.1

0

1.3

0.1

1.4

0

0

0

0

0

0

1

0.9

3.8

GRS
NA

NA

9.8

NA

5.2

NA

NA

NA

0.3

4.9

NA

NA

0.3

1

NA

0.2

3.2

1.2

0

NA

0

NA

NA

NA

0.1

0.6

6.2

ERS
NA

NA

10

NA

10

NA

NA

NA

4.2

10

NA

NA

1.7

10

NA

3.7

10

7

0

NA

0

NA

NA

NA

5.2

2.1

10

FPS (GAP)
0

0

8.3

0

6.9

0

0

0

1.8

7.4

0

0

1

4

0

1.7

4.4

3.2

0

0

0

0

0

0

2.1

1.4

6.7

Class (GAP)
HPS

HPS

NFCR

HPS

LPS

HPS

HPS

HPS

HPS

LPS

HPS

HPS

HPS

MPS

HPS

HPS

MPS

MPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

LPS

EPS (DGD)
2

0

3

0

4

0

0

0

2

7

0

0

0

3

0

3

2

2

2

0

0

0

0

NA

4

2

4

Class (DGD)
HPS

HPS

HPS

HPS

MPS

HPS

HPS

HPS

HPS

LPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

NA

MPS

HPS

MPS

Chapter 4

5

580

3

10

1

76

13

16

1

Taxon

P. pachyrrhizoides

P. polystachyus

P. rotundatus++

P. salicifolius

P. scrobiculatifolius+

P. sinuatus#

P. smilacifolius++

P. sonorensis++

P. xolocotzii++

26

184

5

6

129

23

5

4

1

7

P. esperanzae

P. grayanus

P. laxiflorus++

P. oaxacanus++

P. pedicellatus

P. polymorphus

P. purpusii++

P. scabrellus+

P. teulensis+

P. texensis

52

284

P. dumosus

P. vulgaris

P. leptophyllus+

6

64

P. costaricensis

Sect. Revoluti

8

P. albescens

Sect. Phaseoli

5

P. dasycarpus++

Sect. Pedicellati

Number
of
herbarium
specimens

1

209

14

44

8

6

1

4

1

5

71

2

1

77

15

5

1

3

2

12

1

3

2

344

2

Number of
herbarium
specimens
(represented
populations)

0

1674

9

4

0

0

0

0

0

1

8

0

0

49

7

0

0

0

0

1

0

1

1

6

21

Number
of
genebank
accessions

0

452

7

3

0

0

0

0

0

1

8

0

0

36

7

0

0

0

0

1

0

1

1

2

20

Number of
genebank
accessions
(represented
populations)

6

1958

61

68

8

7

1

4

5

24

137

6

5

233

33

5

1

16

2

77

1

11

3

586

26

Total

1

661

14

46

8

6

1

4

1

6

79

2

1

113

15

5

1

3

1

12

1

4

2

346

22

Total
(represented
populations)
ATAUC
NA

0.949

0.988

0.995

0.995

0.99

NA

NA

NA

0.903

0.982

0.5

NA

0.985

0.992

0.903

NA

0.992

NA

NA

NA

0.99

NA

0.95

0.996

STAUC
NA

0.0096

0.0144

0.0094

0.0045

0.0069

NA

NA

NA

0.0498

0.0161

NA

NA

0.0164

0.0237

0.1941

NA

0.0037

NA

NA

NA

0.0076

NA

0.0117

0.0058

ASD15
NA

0

2.5

0

0.2

0

NA

NA

NA

0

0

NA

NA

0

2

83.1

NA

37.4

NA

NA

NA

12.5

NA

0

0

SRS
0

6.8

5

0.7

0

0

0

0

0

1.7

1

0

0

3.2

4.7

0

0

0

0

0.8

0

2.5

5

0.1

9.1

GRS
NA

10

4.4

9

0

0

NA

NA

NA

0.1

2.3

NA

NA

1.8

4.8

NA

NA

NA

NA

NA

NA

10

5.1

0.1

4.2

ERS
NA

10

10

10

0

0

NA

NA

NA

2.5

5.2

NA

NA

10

3.7

NA

NA

NA

NA

NA

NA

9.2

10

2.7

10

FPS (GAP)
0

8.9

6.5

6.6

0

0

0

0

0

1.4

2.9

0

0

5

4.4

0

0

0

0

NA

0

7.2

6.7

0.9

7.8

Class (GAP)
HPS

NFCR

LPS

LPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

MPS

MPS

HPS

HPS

HPS

HPS

HPS

HPS

LPS

LPS

HPS

NFCR

EPS (DGD)
0

7

5

6

2

3

0

2

0

3

4

3

0

3

2

0

0

0

0

2

0

0

5

2

8

Class (DGD)
HPS

LPS

MPS

LPS

HPS

HPS

HPS

HPS

HPS

HPS

MPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

MPS

HPS

NFCR

A gap analysis methodology for collecting crop genepools

123

124

8

682

P. carteri

P. filiformis

1

12

16

11

10

P. gladiolatus++

P. hintonii

P. magnilobatus

P. xanthotrichus

P. zimapanensis

223

P. microcarpus

161

42

5

8

7

7

1

3

397

3

269

Number of
herbarium
specimens
(represented
populations)

51

15

16

38

2

11

0

0

98

5

17

Number
of
genebank
accessions

35

10

13

30

1

7

0

0

46

2

8

3

Number of
genebank
accessions
(represented
populations)

274

160

26

49

18

23

1

4

780

13

634

Total

193

42

17

38

8

14

1

3

441

4

275

Total
(represented
populations)
ATAUC
0.981

0.995

0.985

0.993

0.985

0.967

NA

0.97

0.971

0.998

0.98

STAUC
0.007

0.0042

0.0375

0.014

0.0099

0.0545

NA

0.0105

0.0082

0.0014

0.0049

ASD15
0

0

0

0

0

1.2

NA

61.4

0

6

0

SRS
1.8

2.4

7.6

7.9

1.3

5

0

0

1

5

0.3

GRS
3.4

1.9

4.4

9

0.4

0.8

NA

NA

2.6

3

0.7

ERS
10

6.7

10

10

3.2

7

NA

NA

10

3.6

7.5

FPS (GAP)
5.1

3.7

7.3

9

1.6

4.3

0

0

4.6

3.9

2.8

Class (GAP)
LPS

MPS

LPS

NFCR

HPS

MPS

HPS

HPS

MPS

MPS

HPS

EPS (DGD)
4

5

6

5

2

2

3

0

2

2

2

MPS

MPS

LPS

MPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

Class (DGD)

1

2

Number of herbarium specimens. Number of genebank accessions. Refers to the number of populations (unique locations identified) represented by the set of
samples. 4Prioritization of taxa is done as follows: HPS: High priority species, MPS: Medium priority species, LPS: Low priority species, NFCR: No further urgent
conservation required. FPS indicates the result of the method proposed in this paper, and EPS indicates the prioritization given by expert knowledge (based on
Daniel G. Debouck’s expertise in Phaseolus). +Indicates that the taxon had no genebank accessions and no herbarium samples with coordinates or location data;
++indicates a taxon for which a Maxent model was not possible and for which 0-few genebank accessions were available; #indicates a taxon with some genebank
accessions but no or limited herbarium samples with coordinates or location data. These taxa are listed as HPS for further collecting in order to inform the gap
analysis.

128

P. glabellus

Not classified

4

P. esquincensis++

Sect. Xanthotricha

617

P. angustissimus

Sect. Rugosi

Taxon

Number
of
herbarium
specimens

Chapter 4

A gap analysis methodology for collecting crop genepools
REFERENCES
Acosta-Gallegos, J.A., Kelly, J.D., Gepts, P. (2007). Prebreeding in common bean and use of genetic
diversity from wild germplasm. Crop Sci. 47 (3-12), S-44-S-59.
Brooks, T.M., Mittermeier, R.A., da Fonseca, G.A.B., Gerlach, J., Hoffmann, M., Lamoreux, J.F., et al.
(2006). Global biodiversity conservation priorities. Science 313 (5783), 58-61.
Burke, M.B., Lobell, D.B., Guarino, L. (2009). Shifts in African crop climates by 2050, and the
implications for crop improvement and genetic resources conservation. Global Environ. Chang. 19
(3) (8), 317-325.
Busby, J.R. (1991). “BIOCLIM – a bioclimatic analysis and prediction system,” in: Nature
Conservation: Cost Effective Biological Surveys and Data Analysis, eds. C.R. Margules, M.P. Austin
(Canberra: Commonwealth Scientific and Industrial Research Organisation [CSIRO]). 64-68.
Challinor, A.J., Wheeler, T.R., Craufurd, P.Q., Ferro, C.A.T., Stephenson, D.B. (2007). Adaptation of
crops to climate change through genotypic responses to mean and extreme temperatures. Agr.
Ecosyst. Environ. 119, 190-204.
Costa, G.C., Nogueira, C., Machado, R.B., Colli, G.R. (2010). Sampling bias and the use of ecological
niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodiv.
Conserv. 19, 883-899.
Debouck, D.G. (1999). “Diversity in Phaseolus species in relation to the common bean, Chapter 2,” in:
Common Bean Improvement in the Twenty-first Century, ed. S.P. Singh (Dordrecht: Kluwer
Academic Publishers), 25-52.
Debouck, D.G. (2000). “Biodiversity, ecology, and genetic resources of Phaseolus beans- seven
answered and unanswered questions,” in: Wild Legumes, ed. K. Oono (Tsukuba, Ibaraki: Ministry
of Agriculture, Forestry and Fisheries, and National Institute of Agrobiological Resources), 95-123.
Debouck, D.G. (2009). Cahiers de Phaséologie – Synonymie (Cali : International Center for Tropical
Agriculture
[CIAT]).
http://isa.ciat.cgiar.org/urg/urgweb_folder/files/unitfiles/Cahiersde
PhaseologieSynonymie.pdf [Accessed: 10 November 2009]
Debouck, D.G., Smartt, J. (1995). “Beans,” in: Evolution of Crop Plants, eds. J. Smartt, N.W. Simmonds
(Singapore: Longman Scientific and Technical), 287-294.
Delgado-Salinas, A. (1985). Systematics of the Genus Phaseolus (Leguminosae) in North and Central
America. Ph.D. Thesis (Austin: Univ. of Texas).
Delgado-Salinas, A., Carr, W.R. (2007). Phaseolus texensis (Leguminosae: Phaseolinae): a new species
from the Edwards Plateau of Central Texas. Lundellia 10, 11-17.
Delgado-Salinas, A., Turley, T., Richman, A., Lavin, M. (1999). Phylogenetic analysis of the cultivated
and wild species of Phaseolus (Fabaceae). Syst. Bot. 24, 438-460.
Dormann, C.F. (2006). Promising the future? Global change projections of species distributions. Basic
Appl. Ecol. 8, 387-397.
Elith, J., Graham, C.H., Anderson, R.P., Dudik, M., Ferrier, S., Guisan, A., et al. (2006). Novel methods
improve prediction of species’ distributions from occurrence data. Ecography 29, 129-151.
FAO (Food and Agriculture Organization of the United Nations). (1997). State of the World’s Plant
Genetic Resources for Food and Agriculture (Rome: FAO). http://apps3.fao.org/wiews/
wiewspage.jsp?i_l=EN&show=SOW1.
FAO (Food and Agriculture Organization of the United Nations). (2009a). FAOSTAT.
http://faostat.fao.org/site/339/default.aspx.
FAO (Food and Agriculture Organization of the United Nations) (2009b). Second Report on the State
of the World’s Plant Genetic Resources for Food and Agriculture (draft) (Rome: FAO).
ftp://ftp.fao.org/docrep/fao/meeting/017/ak528e.pdf.
125

Chapter 4
Fitzpatrick, M.C., Hargrove, W.W. (2009). The projection of species distribution models and the
problem of non-analog climate. Biodiv. Conserv. 18, 2255-2261.
Freytag, G.F., Debouck, D.G. (2002). Taxonomy, Distribution, and Ecology of the Genus Phaseolus in
North America, Mexico and Central America (Ft. Worth: Botanical Research Institute).
Garvin, F.D., Weeden, N. (1994). Isozyme evidence supporting a simple geographic origin for
domesticated tepary bean. Crop Sci. 34, 1390-1395.
Gepts, P. (2001). “Phaseolus vulgaris (Beans),” in: Encyclopedia of Genetics (Academic Press), 14441445.
Graham, C., Hijmans, R.J. (2006). A comparison of methods for mapping species range and species
richness. Global Ecol. Biogeog. 15, 578-587.
Graham, C., Elith, J., Hijmans, R.J., Guisan, A., Peterson, A.T., Loiselle, B.A., et al. (2008). The
influence of spatial errors in species occurrence data used in distribution models. J. Appl. Ecol. 45,
239-247.
Guralnick, R.P., Wieczorek, J., Beaman, R., Hijmans, R.J., and the BioGeomancer Working Group.
(2006). Biogeomancer: automated georreferencing to map the world’s biodiversity data. PLoS
Biology 4(11), 1908-1909.
Hajjar, R., Hodgkin, T. (2007). The use of wild relatives in crop improvement: a survey of developments
over the last 20 years. Euphytica 156, 1-13.
Harlan, J.R., de Wet, J.M.J. (1971). Toward a rational classification of cultivated plants. Taxon 20 (4),
509-517.
Hawkins, B., Sharrock, S., Havens, K. (2008). Plants and Climate Change: Which Future? (Richmond:
Botanic Gardens and Conservation International).
Hernandez, P.A., Graham, C.H., Master, L.L., Albert, D.L. (2006). The effect of sample size and species
characteristics on performance of different species distribution modeling methods. Ecography 29,
773-785.
Hijmans, R.J., Graham, C. (2006). The ability of climate envelope models to predict the effect of climate
change on species distributions. Glob. Change Biol. 12, 2272-2281.
Hijmans, R.J., Spooner, D. (2001). Geographic distribution of wild potato species. Am. J. Bot. 88, 21012112.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A. (2005). Very high resolution
interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965-1978.
Hijmans, R.J., Guarino, L., Cruz, M., Rojas, E. (2001). Computer tools for spatial analysis of plant
genetic resources data: 1. DIVA-GIS. Plant Genet. Res. Newsl. 127, 15-19.
Hijmans, R.J., Guarino, L., Jarvis, A., O’Brien, R., Mathur, P., Bussink, C., et al. (2005b). DIVA-GIS
version 5.2 manual. http://www.diva-gis.org/Materials.htm.
Hijmans, R.J., Spooner, D.M., Salas, A.R., Guarino, L., de la Cruz. J, (2002). Atlas of Wild Potatoes
(Rome: International Plant Genetic Resources Institute).
Intergovernmental Panel on Climate Change. (2007). IPCC Fourth Assessment Report: Climate Change
2007 (Geneva: IPCC).
Jarvis, A., Ferguson, M., Williams, D., Guarino, L., Jones, P., Stalker, H., et al. (2003). Biogeography
of wild Arachis: Assessing conservation status and setting future priorities. Crop Sci. 43, 1100-1108.
Jarvis, A., Guarino, L., Williams, D., Williams, K., Hyman, G. (2002). Spatial analysis of wild peanut
distributions and the implications for plant genetic resource conservation. Plant Genet. Res. Newsl.
131, 29-35.
Jarvis, A., Yeaman, S., Guarino, L., Tohme, J. (2005). “The role of geographic analysis in locating,
understanding, and using plant genetic diversity,” in: Molecular Evolution: Producing the
Biochemical Data, Part B, ed. E. Zimmer (New York: Elsevier), 279-298.
126

A gap analysis methodology for collecting crop genepools
Jimenez-Valverde, A., Lobo, J.M., Hortal, J. (2008). Not as good as they seem: the importance of
concepts in species distribution modelling. Divers. Distrib. 14, 885-890.
Jones, P.B., Beebe, S.E., Tohme, J., Galwey, N.W. (1997). The use of geographical information systems
in biodiversity exploration and conservation. Biodiv. Conserv. 6, 947-958.
Khoury, C., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic resources:
a review of global crop and regional conservation strategies. Genet. Resour. Crop Ev. 57(4), 625639.
Kipe-Nolt, J.A., Montealegre, C.M., Tohme, J. (1992). Restriction of nodulation by the broad host range
Rhizobium tropici strain CIAT899 in wild accessions of Phaseolus vulgaris. New phytol. 120, 48994.
Kornegay, J., Cardona, C. (1991). “Breeding for insect resistance,” in Common Beans: Research for
Crop Improvement, eds. A. van Schoonhaven, O. Voysest (Wallingford: CAB International), 61941.
Liu, C., Berry, P.M., Dawson, T.P., Pearson, R.G. (2005). Selecting thresholds of occurrence in the
prediction of species distributions. Ecography 28 (3), 385-393.
Lobo Burle, M., Torres Cordeiro, C.M., Fonseca, J.R., Martins Palhares de Melo, L.A,, de Belem das
Neves Alves, R., Abadie, T. (2003). Characterization of germplasm according to environmental
conditions at the collecting site using GIS- two case studies from Brazil. Plant Gen. Resour.
Newsletter 135, 1-11.
Lobo, J.M., Jimenez-Valverde, A., Hortal, J. (2010). The uncertain nature of absences and their
importance in species distribution modeling. Ecography 33, 103-114.
Lobo, J.M., Jiménez-Valverde, A., Real, R. (2008). AUC: a misleading measure of the performance of
predictive distribution models. Global Ecol. Biogeogr. 17(2), 145-151.
Loiselle, B.A., Jorgensen, P.M., Consiglio, T., Jimenez, I., Blake, J.G., Lohmann, L.G., et al. (2008).
Predicting species distributions from herbarium collections, does climate bias in collection sampling
influence model outcomes? J. Biogeogr. 35(1), 105-116.
Maxted, N., Kell, S. (2009). Establishment of a global network for the in situ conservation of crop wild
relatives: status and needs (Rome: FAO Commission on Genetic Resources for Food and
Agriculture).
Maxted, N., Dulloo, E., Ford-Lloyd, B.V., Iriondo, J.M., Jarvis, A. (2008). Gap analysis: a tool for
complementary genetic conservation assessment. Divers. Distrib. 14 (6), 1018-1030.
Maxted, N., Ford-Lloyd, B.V., Jury, S.L., Kell, S.P., Scholten, M.A. (2006). Towards a definition of a
crop wild relative. Biodiv. Conserv. 15 (8), 2673-2685.
Maxted, N., Mabuza-Dlamini, P., Moss, H., Padulosi, S., Jarvis, A., Guarino, L. (2004). An
Ecogeographic Study. African Vigna (Rome: International Plant Genetic Resources Institute).
McClean, P., Gepts, P., Kami, J. (2004). “Genomics and genetic diversity in common bean,” in: Legume
Crop Genomics, eds. E. Wilson, H.T. Stalker, E.C. Brummer (Champaign: AOCS), 60-82.
Meilleur, B.A., Hodgkin, T. (2004). In situ conservation of crop wild relatives: status and trends. Biodiv.
Conserv. 13 (4), 663-684.
Munoz, L.C., Duque, M.C., Debouck, D.G., Blair, M.W. (2006). Taxonomy of tepary bean and wild
relatives as determined by amplified fragment length polymorphism (AFLP) markers. Crop Sci. 46,
1744-1754.
Nabhan, G.P. (1990). Wild Phaseolus Ecogeography in the Sierra Madre Occidental, Mexico:
Areographic Techniques for Targeting and Conserving Species Diversity. Systematic and
Ecogeographic Studies on Crop Genepools 5 (Rome: International Board of Plant Genetic
Resources).
Parthasarathy, U., Saji, K.V., Jayarajan, K., Parthasarathy, V.A. (2006). Biodiversity of Piper in South
127

Chapter 4
India – application of GIS and cluster analysis. Current Science 91 (5).
Phillips, S.J. (2008). Transferability, sample selection bias and background data in presence-only
modeling: a response to Peterson et al. (2007). Ecography 31, 272-278.
Phillips, S.J., Anderson, R.P., Schapire, R.E. (2006). Maximum entropy modeling of species geographic
distributions. Ecol. Model. 190, 231-259.
Phillips, S.J., Dudik, M. (2008). Modeling of species distributions with Maxent: new extensions and
comprehensive evaluation. Ecography 31, 161-175.
Prescott-Allen, C., Prescott-Allen, R. (1986). The First Resource: Wild Species in the North American
Economy (New Haven: Yale University).
Salcedo, C.J., Arroyave, J.A., Toro, O., Debouck, D.G. (2006). Phaseolus novoleonensis, a new species
(Leguminosae, Phaseolinae) from the Sierra Madre Oriental, Nuevo León, Mexico. Novon 16 (1),
106-112.
Salcedo, C.J., Lépiz, R.I., Castañeda, N.A., Ocampo, C.N., Debouck, D.G. (2009). Additional
observations about Phaseolus rotundatus (Fabaceae), an endemic bean species from western Mexico.
J. Bot. Res. Inst. Texas 3(2), 751-762.
Schinkel, C., Gepts, P. (1988). Phaseolin diversity in the tepary bean Phaseolus acutifolius A Gray.
Plant Breeding 101, 292-301.
Shelley-Dessert, K.C., Bliss, F.A. (1991). “Genetic improvement in food quality factors,” in: Common
Beans: Research for Crop Improvement, eds. A. van Schoonhaven, O. Voysest (Wallingford: CAB
International), 649-671.
Singh, R.J., Jauhar, P. (2005). Genetic Resources, Chromosome Engineering, and Crop Improvement:
Grain Legumes, Volume 1 (CRC Press).
Singh, S.P. (2001). Broadening the genetic base of common bean cultivars: a review. Crop Sci. 41, 16591675.
Smartt, J. (1981). Gene pools in Phaseolus and Vigna cultigens. Euphytica 30, 445-449.
Smolik, M.G., Dullinger, S., Essl, F., Kleinbauer, I., Leitner, M., Peterseil, J., et al. (2010). Integrating
species distribution models and interacting particle systems to predict the spread of an invasive alien
plant. J. Biogeogr. 37(3), 411-422.
Tanksley, S.D., McCouch, S.R. (1997). Seed banks and molecular maps: unlocking genetic potential
from the wild. Science 277 (5329), 1063-1066.
Termansen, M., McClean, C.J., Preston, C.D. (2006). The use of genetic algorithms and Bayesian
classification to model species distributions. Ecol. Model. 192, 410-414.
The Consultative Group on International Agricultural Research (CGIAR), System-wide Information
network for Genetic Resources (SINGER). (2009). http://singer.grinfo.net/
Thuiller, W., Lavorel, S., Araújo, M.B., Sykes, M.T., Prentice, I.C. (2005). Climate change threats to
plant diversity in Europe. Proc. Natl. Acad. Sci. 102 (23), 8245.
USDA, ARS, National Genetic Resources Program (NGRP). (2009). Germplasm Resources Information
Network - (GRIN) [Online Database]. National Germplasm Resources Laboratory, Beltsville,
Maryland. http://www.ars-grin.gov/cgi-bin/npgs/html/index.pl.
VanDerWal, J., Shoo, L.P., Graham, C.H., Williams, S.E. (2009). Selecting pseudo-absence data for
presence-only distribution modeling: How far should you stray from what you know? Ecol. Model.
220, 589-594.
Veloz, S.D. (2009). Spatially autocorrelated sampling falsely inflates measures of accuracy for presenceonly niche models. J. Biogeogr. 36, 2290-2299.
Wisz, M.S., Hijmans, R.J., Li, J., Peterson, A.T., Graham, C.H., Guisan, A., and NCEAS Predicting
Species Distributions Working Group. (2008). Effects of sample size on the performance of species
distribution models. Divers. Distrib. 14, 763-77.
128

CHAPTER 5
Distributions, ex situ conservation priorities, and
genetic resource potential of crop wild relatives of
sweetpotato [Ipomoea batatas (L.) Lam., I. series Batatas]
Colin K. Khoury1,2, Bettina Heider3, Nora P. Castañeda-Álvarez1,4, Harold A. Achicanoy1,
Chrystian C. Sosa1, Richard E. Miller5, Robert W. Scotland6, John R.I. Wood6,
Genoveva Rossel3, Lauren A. Eserman7, Robert L. Jarret8, G. Craig Yencho9,
Vivian Bernau1,10, Henry Juarez3, Steven Sotelo1, Stef de Haan1,3, and Paul C. Struik2
International Center for Tropical Agriculture (CIAT), Cali, Colombia.
Centre for Crop Systems Analysis, Wageningen University, Wageningen, Netherlands.
3
CGIAR Research Program on Roots, Tubers and Bananas (RTB), International Potato Center, Lima,
Peru.
4
School of Biosciences, University of Birmingham, Birmingham, UK.
5
Department of Biological Sciences, Southeastern Louisiana University, Hammond, LA, USA.
6
Department of Plant Sciences, University of Oxford, Oxford, UK.
7
Plant Biology Department, University of Georgia, Athens, GA, USA.
8
Plant Genetic Resources Conservation Unit, United States Department of Agriculture- Agricultural
Research Service, Griffin, GA, USA.
9
Department of Horticultural Science, North Carolina State University, Raleigh, NC, USA.
10
Department of Horticulture and Crop Science, The Ohio State University, Columbus, Ohio, USA.
1
2

Published in Frontiers in Plant Science 6, 251. doi: 10.3389/fpls.2015.00251

Chapter 5
Abstract
Crop wild relatives of sweetpotato [Ipomoea batatas (L.) Lam., I. series Batatas] have the potential to
contribute to breeding objectives for this important root crop. Uncertainty in regard to species
boundaries and their phylogenetic relationships, the limited availability of germplasm with which to
perform crosses, and the difficulty of introgression of genes from wild species has constrained their
utilization. Here we compile geographic occurrence data on relevant sweetpotato wild relatives and
produce potential distribution models for the species. We then assess the comprehensiveness of ex situ
germplasm collections, contextualize these results with research and breeding priorities, and use
ecogeographic information to identify species with the potential to contribute desirable agronomic traits.
The fourteen species that are considered the closest wild relatives of sweetpotato generally occur from
the central United States to Argentina, with richness concentrated in Mesoamerica and in the extreme
southeastern United States. Currently designated species differ among themselves and in comparison to
the crop in their adaptations to temperature, precipitation, and edaphic characteristics and most species
also show considerable intraspecific variation. With 79% of species identified as high priority for further
collecting, we find that these crop genetic resources are highly under-represented in ex situ conservation
systems and thus their availability to breeders and researchers is inadequate. We prioritize taxa and
specific geographic locations for further collecting in order to improve the completeness of germplasm
collections. In concert with enhanced conservation of sweetpotato wild relatives, further taxonomic
research, characterization and evaluation of germplasm, and improving the techniques to overcome
barriers to introgression with wild species are needed in order to mobilize these genetic resources for
crop breeding.
Keywords: Crop diversity, Crop improvement, Crop wild relatives, Food security, Gap analysis, Plant
genetic resources

130

Crop wild relatives of sweetpotato
INTRODUCTION
Sweetpotato [Ipomoea batatas (L.) Lam.] counts among the world’s most important root crops,
grown on at least eight million hectares in 114 countries worldwide, with particular significance
to food supplies in the tropics and subtropics of East and Southeast Asia, and Sub-Saharan
Africa. The success of the crop in these regions is due to its adaptability to a wide range of agroecological conditions, ease of propagation from cuttings, low input cultivation requirements,
and high productivity as well as nutritive value (Woolfe 1992). The crop can be cultivated from
humid to semi-arid conditions, from sea level to 3000 m.a.s.l. (Huaman 1987), and can
translocate photosynthetic products to the storage roots throughout the growing season, thereby
mitigating negative effects of temporary adverse conditions (Kays 1985). Sweetpotato produces
among the highest amounts of edible energy per hectare of all major food crops (de Vries et al.
1967), and is an important source of vitamin A and C, calcium, iron, and a number of essential
amino acids (Kays & Kays 1998, Tumwegamire et al. 2011). In addition to human consumption
of the storage roots and young leaves, the crop is utilized for animal feed, fuel, and starch
production.
In Sub-Saharan Africa, sweetpotato is predominantly cultivated in small plots characterized by
low fertility and drought-prone soils, producing relatively good yields with low inputs and
minimal labor costs. The crop has recently become the focus of targeted bio-fortification for
enhanced vitamin A. Orange-fleshed varieties have been bred with 50-fold more β-carotene
than standard varieties and these newly-released varieties rank first among roots and tubers in
Sub-Saharan Africa for their nutritional quality (Low et al. 2007, Hotz et al. 2012). Given its
adaptability, low-external input requirements, nutritional quality, and improvement potential, it
is not surprising that sweetpotato has become a priority in crop based strategies for enhancing
food security in the tropics (Pfeiffer & McClafferty 2007, Bill and Melinda Gates Foundation
2011, Bouis & Islam 2012).
The full potential of sweetpotato is far from realized, with particularly large yield gaps (ca. 20
t ha-1) remaining across rain-fed Sub-Saharan Africa due to a range of biotic and abiotic
constraints, especially sweetpotato virus disease (SPVD) and sweetpotato weevils (SPW), as
well as susceptibility to drought (Sutherland 1986, Valverde et al. 2007, Ngailo et al. 2013,
Kivuva et al. 2015a, b). SPVD is a severe constraint on the continent, caused by the synergistic
interaction of two viruses transmitted by whiteflies and aphids, causing yield losses of up to
131

Chapter 5
98% under severe infections (Ngailo et al. 2013). SPW (Cylas spp.) are the most devastating
insect pests of the crop. Cylas formicarius elegantulus Summers, C. puncticollis Boheman and
C. brunneus Fabricius can cause yield losses of between 67% and 100% in Sub-Saharan Africa
(Smit 1997). The concealed feeding behavior, oviposition, and larval development of the
weevils in the storage roots make their control very difficult, necessitating the development of
improved management options, in particular via enhanced genetic resistance. Drought, and the
compounding effect of increasing heat on drought, is a rising concern particularly in regions
undergoing significant climatic change, both due to its direct effect on productivity (Low et al.
2009, Schafleitner et al. 2010) as well as to its association with increased severity of damage
from SPW and SPVD (Munyiza et al. 2007, Mwololo et al. 2007). Lack of drought tolerance
in high β-carotene sweetpotato varieties has led to constraints in their adoption (Mwanga &
Ssemakula, 2011).
Crop wild relatives (CWR) are increasingly being recognized for their potential to contribute
valuable traits to breeding programs (Feuillet et al. 2008, Guarino & Lobell 2011, Dempewolf
et al. 2014). CWR have provided breeders with genes for pest and disease resistance, abiotic
stress tolerance, and quality traits in an ever increasing number of food crops, such as banana,
barley, bean, cassava, chickpea, maize, lettuce, oat, potato, rice, sugarcane, sunflower, tomato,
and wheat, among others (Xiao et al. 1996, Hajjar & Hodgkin 2007, McCouch et al. 2007,
Khoury et al. 2010). Yet despite the successful history of contribution to the improvement of
major crops, systematic approaches to the use of CWR in the breeding programs of a number
of important staples, including sweetpotato, remain underdeveloped.
The morning glory tribe Ipomoeeae contains approximately 650-900 species and includes the
genus Ipomoea and nine other related genera (Wilkin 1999, Mabberley 2008). Although many
genera, subgenera, and sections of the Ipomoeeae are not monophyletic in phylogenetic
analyses, Ipomoea series Batatas (Choisy) D.F. Austin, which contains sweetpotato and 14
closely related CWR (Austin 1978, McDonald & Austin 1990), does form a monophyletic
lineage (Miller et al. 1999, McDonald et al. 2011, Eserman et al. 2014). These species include
wild I. batatas (L.) Lam. [including Ipomoea batatas var. apiculata (Martens and Galeotti)
McDonald and Austin], Ipomoea cordatotriloba Dennstedt, Ipomoea cynanchifolia Meisn.,
Ipomoea grandiflora (Dammer) O’Donell, Ipomoea lacunosa L., Ipomoea leucantha Jacquin,
Ipomoea littoralis Blume, Ipomoea ramosissima (Poir.) Choisy, Ipomoea splendor-sylvae
132

Crop wild relatives of sweetpotato
House, Ipomoea tabascana McDonald and Austin, Ipomoea tenuissima Choisy, Ipomoea
tiliacea (Willd.) Choisy in D.C., Ipomoea trifida (H.B.K.) G. Don, and Ipomoea triloba L.
Many sweetpotato CWR can be hybridized with the crop through controlled pollinations,
somatic cell, and/or ovule culture techniques (Diaz et al. 1996). Crosses involving I. tabascana,
I. trifida, I. triloba, I. littoralis, I. grandifolia, I. lacunosa, I. leucantha, and wild I. batatas in
particular have resulted in relatively viable progeny (Nimmakayala et al. 2011). The wild
conspecific as well as I. trifida have been documented for their contribution to increases in
protein and starch content, and nematode and SPW resistance (Iwanaga 1988, Shiotani et al.
1994), although there is uncertainty for some material as to whether they may actually have
been feral forms of the cultivar (Nimmakayala et al. 2011). Species that have been explored for
potential traits of use in crop improvement include I. trifida and I. littoralis for yield and SPW,
scab [Elsinoë batatas (Saw.) Viegas et Jenkins], and black rot disease (Ceratocystis fimbriata
Ell. et Halst.) resistance; I. grandifolia for sweetpotato stem nematode and SPVD resistance;
and I. triloba for drought tolerance, root rot resistance, and foliar fungal disease resistances
(Iwanaga 1988, Jarret et al. 1992, Komaki 2004, Zhang & Liu 2005, Nimmakayala et al. 2011).
Challenges in the creation of viable progeny between the CWR and the cultivated species are
not insignificant, though, due to differences in ploidy and interspecific incompatibility (Martin
1970, 1982, Teramura 1979, Shiotani et al. 1994, Lu & Li 1992, Diaz et al. 1996, Komaki 2004,
Nimmakayala et al. 2011).
A lack of basic knowledge about boundaries between species within I. series Batatas and a
dearth of diagnostic characters enabling differentiation of taxa - to facilitate reliable and
accurate species identification - is a fundamental stumbling block constraining the utilization
of sweetpotato CWR (Austin 1978, Austin 1988, Jarret et al. 1992, Diaz et al. 1996, McDonald
et al. 2011, Nimmakayala et al. 2011, Eserman et al. 2014). Needed studies have been delayed
in part due to the absence of plant materials for research, in particular the availability of
specimens with flowers and ripe fruits. Studies that have been performed have often been based
upon limited sampling (e.g., single accessions for I. littoralis and I. tabascana).
Many unanswered questions regarding the relationships of CWR to sweetpotato potentially
impact the efficiency of breeding strategies for the crop. For example, do species such as I.
tabascana and I. tenuissima represent distinct taxa (hybrid or otherwise) or rather rare variants
133

Chapter 5
of the crop? What is the range and genetic diversity present in truly wild forms of the crop
conspecific, compared to feral escapees? How accurate are the classifications of species with
highly disjunct distributions (e.g., I. cordatotriloba)? Are there as yet unrecognized cryptic
species within I. series Batatas [e.g., I. ‘austinii’ (Duncan & Rausher 2013)]? What are the
lineages and genetic resources potential of purported hybrid species (i.e., I. leucantha and I.
grandifolia)? What are the geographic locations of new variation being generated through
hybridization among sweetpotato CWR?
The investigation, conservation, and availability of genetic resources of sweetpotato provide a
foundation for the crop’s long term viability and for its potential for improvement. To contribute
to these objectives, we analyzed the comprehensiveness of ex situ conservation of sweetpotato
CWR through a series of questions: (a) what constitutes a potentially useful wild relative of
sweetpotato?, (b) where are these species encountered?, (c) what is the state of conservation
and availability of these species to researchers, and what are the highest taxonomic and
ecogeographic priorities for further collecting? And finally, (d) what traits do sweetpotato CWR
possess that may be valuable to crop improvement?
MATERIALS AND METHODS
Identification of target CWR species and occurrence data compilation
The CWR of sweetpotato analyzed in this study were selected based upon recent and historical
taxonomic and molecular phylogenetic research (Austin 1978, Austin & Huaman 1996, Austin
1997, Miller et al. 1999, McDonald et al. 2011, GRIN 2013, Eserman et al. 2014), identifying
those wild species with a relatively close phylogenetic relationship to the crop (i.e., members
of I. series Batatas). We included all 14 wild species comprising the series in the analysis.
Domesticated sweetpotato I. batatas (6x) has been proposed as originating from interspecific
hybridization involving I. trifida (2x, 4x, 6x), I. littoralis (2x), and/or I. leucantha (2x)
(Nishiyama 1971, Nishiyama 1982, Austin 1988). The species most closely related to the crop
have been posited to be I. trifida followed by I. tabascana (4x) (Srisuwan et al. 2006,
Nimmakayala et al. 2011). Following the genepool concept of Harlan and de Wet (1971) and
due to ploidy incompatibility with the cultivated species, the putative closest related taxa to
sweetpotato are placed in the secondary genepool: wild forms of I. batatas (4x), I. trifida, I.
littoralis, and I. tabascana (Jarret et al. 1992, Jarret & Austin 1994, Rajapakse et al. 2004,
134

Crop wild relatives of sweetpotato
GRIN 2013). Species classified as tertiary wild relatives include: I. cordatotriloba (syn.
Ipomoea trichocarpa Elliott), I. cynanchifolia, I. grandifolia, I. lacunosa, I. leucantha, I.
ramosissima, I. splendor-sylvae (syn. Ipomoea umbraticola House), I. tenuissima, I. tiliacea,
and I. triloba (Jarret & Austin 1994, Huang et al. 2002, Rjapakse et al. 2004, GRIN 2013).
Occurrence records for these species were acquired from online biodiversity, herbaria, and
germplasm databases; through communications with herbaria and genebank managers, and
other crop researchers; and via direct recording of provenance data during visits to selected
herbaria (Supplementary Table 1). Germplasm data were obtained from repositories that
provide straightforward access to genetic resources and associated data to the global research
community through online information systems. The occurrence data were then compiled in a
standardized format, nomenclature was checked against GRIN Taxonomy for Plants (2013) and
The Plant List (2010), and duplicate records were eliminated. Existing coordinates were crosschecked to country and being on land (Hijmans et al. 1999), and records with locality
information but no coordinates were geo-referenced using the Google Maps Geocoder v.3
(2013) application programming interface. Occurrence data were mapped, iteratively evaluated
for correctness, and further processed in order to form a final dataset of improved taxonomic
and spatial accuracy.
Challenges in using and in improving the large quantities of occurrence data now available from
online resources such as the Global Biodiversity Information Facility (GBIF) have been noted
(Gaiji et al. 2013), including geographic and nomenclatural data quality and the slow speed
with which aggregated datasets are updated (Mesibov 2013, Otegui et al. 2013, Hjarding et al.
2014). In addition, particular caution must be applied to the occurrence records used in the
current paper as ongoing work (unpublished data) indicates that many Ipomoea occurrence
records in such online databases are identified as synonyms, excluded or invalid names, and
that many valid names were applied to specimens well outside of species known ranges. We
have identified some of these obvious errors but until all specimen records are correctly
identified and checked against an accurate taxonomy these data must be treated with caution.
A total of 5,614 occurrence records for the 14 taxa were included in potential distribution
modeling and/or in the conservation analysis, including 749 germplasm records sourced from
four genebanks, and 4,865 herbarium and other occurrence reference records sourced from 42
135

Chapter 5
providers. Records per species ranged from eight (I. tabascana) to 1,409 (I. trifida). Of these,
3,650 records containing unique cross-checked coordinates were used to model species
potential distributions and to locate the original collecting site of existing germplasm
accessions.
Species potential distribution modeling
A potential distribution model for each species was calculated using the maximum entropy
(Maxent) algorithm (Phillips et al. 2006), with a set of ecogeographic variables and unique
species presence records as inputs. We chose Maxent due to its extensive application in
predicting species distributions (Elith et al. 2006; Phillips & Dudik 2008, Costa et al. 2010),
including those for wild relatives (Ramírez-Villegas et al. 2010, Conolly et al. 2012, Khoury et
al. 2015). We performed modeling at a resolution of 2.5 arc-minutes (~ 5 km2 cell size at the
equator), employing 10,000 background points for model training over the combined
distributional range of the sweetpotato CWR. Ecogeographic inputs included altitude and 19
bioclimatic variables from the WorldClim database (Hijmans et al. 2005), and seven major
edaphic drivers of plant species distributions with consistent data coverage throughout the range
of the sweetpotato CWR species, obtained from ISRIC-World Soil Information (Hengl et al.
2014) (Supplementary Table 2). For the edaphic variables we calculated a weighted mean
across 0-5, 5-15, 15-30, 30-60, and 60-100 cm soil depth values in order to derive a single data
value for 0-100 cm. We then resampled the 30 arc-seconds resolution data to form 2.5 arcminutes inputs aligned with the WorldClim datasets.
Potential distribution models were produced by calculating the mean of replicates (k = 5), and
clipped by measuring the shortest distance between the receiver operating characteristic curve
(ROC-curve) and the top-left corner of the plot (Liu et al. 2005). Models were constrained per
species by a native range defined at the country level as given in GRIN Taxonomy for Plants
(GRIN 2013), in order to focus prioritization recommendations on those regions with
populations exhibiting long-term adaptations to local ecogeographic conditions. We further
cross-validated and refined occurrence data based upon our knowledge of native distributions,
constraining localities for wild I. batatas to Mexico south to Peru, and not in the Caribbean; I.
leucantha to the USA and Mexico, and I. littoralis to points within 100 km of the ocean.
In order to derive robust distribution models for each species, we analyzed Maxent results
136

Crop wild relatives of sweetpotato
across three groups of ecogeographic inputs: a) the full set of 19 bioclimatic variables (RamírezVillegas et al. 2010); b) the bioclimatic variables, altitude, and the additional set of seven
edaphic variables, totaling 27 input variables; and c) a species-specific derivation of the most
important drivers of distribution based upon presence data, further refined by removing highly
correlated variables. For the ecogeographic variables in the species-specific subset method, we
utilized a non-linear iterative partial least squares (NIPALS) algorithm to perform a principalcomponent analysis (PCA), as NIPALS has the potential to handle data arrays in which the
number of observations is less than the number of input variables. We then identified those
variables with the greatest contribution (> 0.7 or < - 0.7) to the first two principal components
per species. Finally, we used a variance inflation factor (VIF) to identify the variables with a
low degree of collinearity (see Supplementary Table 3 for a list of variables identified per
species). A calibrated area under the ROC curve (cAUC) was obtained to assess the predictive
performance of each model (Hijmans 2012, Khoury et al. 2015).
The three modeling methods were compared against null models, and the species-specific
subset variables method showed the least overall spatial sorting bias among methods
(Spearman’s p for the 19 variables was 0.65, for 27 variables it was 0.72, and for the subset
method it was 0.25), although the differences in median AUC distributions across species for
each method were not significant (p = 0.095) through a Kruskal-Wallis non-parametric analysis
of variance test. Maxent models were subjected to a four-fold assessment process including: a)
the 5-fold average area under the ROC curve of test data (ATAUC), b) the standard deviation
of the test AUC of the 5 different folds (STAUC), c) the proportion of the potential distribution
coverage with standard deviation above 0.15 (ASD15), and d) the cAUC value. Models with
ATAUC above 0.7, STAUC below 0.15, ASD15 below 10%, and cAUC exceeding 0.40 were
considered accurate and stable (Ramírez-Villegas et al. 2010, Khoury et al. 2015).
The potential distribution models of the sweetpotato CWR generally performed well in regard
to the modeling assessment process. Four species demonstrated low cAUC values and one of
these an elevated ASD15 value, indicating greater uncertainty in the models (Supplementary
Table 3). Species-specific subset model outputs for taxa with relatively few distinct occurrence
points (< 20) (I. tabascana, I. tenuissima, and I. cynanchifolia) lacked sufficient discriminatory
power, leading to highly inflated spatial models in comparison to recorded distributions.
Potential distribution models for these species were resolved by deriving an ensemble (i.e.,
137

Chapter 5
overlap) of the three input variation methods, verified by researchers knowledgeable in the
distribution of the species as more accurately representing the true known distributions.
Potential distribution models based upon the species-specific subset variables method were
therefore utilized in subsequent analyses for all species with adequate distinct occurrence points
(> 20). The ensemble method was used for the three species with limited occurrence data.
Analysis of current ex situ conservation and further collecting needs for CWR
We adapted a gap analysis methodology proposed by Ramírez-Villegas et al. (2010), combining
three metrics used to assess the urgency of further collecting in order to fill gaps in ex situ
conservation of CWR. The total sample representation of each species in genebank collections
was estimated via a sampling representativeness score (SRS), calculated as the number of
germplasm samples (G) divided by the total number of samples (G + herbarium samples (H))
(i.e., all other records aside from available genebank accessions).
The sufficiency of geographic coverage of germplasm collections of each species was estimated
through a geographic representativeness score (GRS), calculated as the share of the combined
total area of circular buffers of 50 km (CA50) placed around existing germplasm collection
points compared with the overall potential geographic distribution of the species.
The comprehensiveness of ecological coverage of germplasm collections of each species was
estimated through an ecological representativeness score (ERS), calculated by estimating the
distinct ecosystem classifications (Olson et al. 2001) represented in the CA50 of existing
germplasm collection points compared with the diversity of ecosystems in which the overall
potential geographic distribution model of the species occurs.
A final priority score (FPS) for further collecting for ex situ conservation was assigned to each
species by averaging the three gap analysis metrics (SRS, GRS, and ERS). FPS scores were
further classified into four categories of urgency for collecting: high priority species (HPS) for
taxa whose 0 < FPS ≤ 2.5 or when no germplasm accessions are currently conserved, medium
priority species (MPS) when 2.5 < FPS ≤ 5, low priority species (LPS) when 5 < FPS ≤ 7.5,
and ‘no further collecting recommended’ (NFCR) when 7.5 < FPS ≤ 10. We produced
collecting priorities maps for all species, displaying the geographic areas that have not yet been
collected from within the potential distributions of taxa.
138

Crop wild relatives of sweetpotato
The ecogeographic data preparation, species distribution modeling, and gap analysis were
written and performed in R v2.15.1 (R Core Team 2013), utilizing packages maptools (Bivand
& Lewin-Koh 2014), rgdal (Bivand et al. 2014), SDMTools (van der Wal et al. 2014), raster
(Hijmans 2014), sp (Pebesma & Bivand 2005, Bivand et al. 2013), dismo (Hijmans et al. 2013),
and plsdepot (Sanchez 2012). Resulting spatial files were mapped in ArcMap v.10 (ESRI 2011).
Collecting priorities spatial files were analyzed using the Zonal Statistics tool in ArcMap to list
the countries prioritized for further collecting for ex situ conservation.
Expert evaluation of conservation assessment results
In order to validate and/or expose deficiencies in our findings, we subjected the gap analysis
numerical and spatial results to an evaluation performed by seven crop experts with experience
in the systematics, distribution, and/or conservation status of CWR of sweetpotato: Richard E.
Miller, Southeastern Louisiana University; Robert W. Scotland and John R.I. Wood, University
of Oxford; Genoveva Rossel, International Potato Center; Lauren A. Eserman, University of
Georgia; Robert L. Jarret, USDA-ARS Griffin; and G. Craig Yencho, North Carolina State
University. These experts were first asked to provide an evaluation of the sufficiency of
germplasm collections per species based only upon their knowledge of the total number of
accessions conserved, and geographic as well as environmental gaps. Such an assessment
[comparable expert priority score (EPS)] was considered directly comparable to the FPS of the
gap analysis results.
A second evaluation score (contextual EPS) based on the entirety of expert knowledge,
including additional factors such as threats to species in situ and prioritization by usefulness in
crop breeding, was collected in order to provide additional information to collecting
prioritization efforts. In both cases, an EPS between 0 and 10, aligned with the gap analysis
results prioritization scale, was requested. After these steps, experts were shown the gap
analysis data and asked to comment on assessed quantitative results, occurrence data, potential
distribution models, and maps of collecting priorities. Following these contributions by experts,
input occurrence data were further refined by eliminating clearly incorrect points and adjusting
country-level native areas, and the potential distribution modeling and gap analyses were rerun using the refined datasets in order to improve the quantitative and spatial outputs. Expert
evaluation metrics displayed in the results pertain to a final evaluation of improved gap analysis
outputs, performed by five of the experts.
139

Chapter 5
A multiple factor analysis was used in order to compare the various forms of expert evaluation
inputs (i.e., comparable expert priority score, contextual expert priority score, evaluation of gap
analysis results score, evaluation of occurrence data, evaluation of potential species distribution
models, and evaluation of collecting priorities map) with the gap analysis results. An expert
evaluation index was created, which estimated the degree of accord between all experts and the
gap analysis results for each species, with a scale from 0 (disagreement) to 100 (agreement).
Analyses were performed using R package FactoMineR (Husson et al. 2009).
Identification of ecogeographic characteristics of CWR
In order to evaluate the pairwise degree of geographic overlap between sweetpotato CWR
distribution models, we calculated an overlap measure equal to the frequency of shared 2.5 arcminute geographic cells between taxa divided by the sum of the total number of cells in which
either of the species are present (Kernohan et al. 2001, Fieberg & Kochanny 2005). To assess
the pairwise degree of ecogeographic niche overlap between species, we used Schoener’s index
for niche similarity (D) and the adjusted similarity index (I) as outlined in Warren et al. (2008),
using species distribution models and the species-specific subset of the 27 ecogeographic
layers/ensemble models as inputs. Overlap indices range from 0 (no overlap) to 1 (complete
overlap) and were implemented in the R package Phyloclim (Heibl 2011).
We utilized ecogeographic information in combination with species presence data in order to
identify populations of species with the potential for outstanding adaptations to climatic and/or
edaphic conditions of interest to sweetpotato breeding objectives. We assessed the relative
importance of the 27 ecogeographic variables (Supplementary Table 2) in explaining the total
variation through a PCA after confirming the validity of its use through a Bartlett’s test
performed in R package psych (Revelle 2015). We created a hierarchical cluster of principal
components (HCPC) in order to identify ecogeographic clusters for the sweetpotato wild
relatives, driven by those variables demonstrating ≥ 15% difference (±) from the mean and a
reduction of ≥ 20% from the mean standard deviation exhibited across all species, using R
package FactoMineR. Boxplots for each of the 27 ecogeographic variables were created based
upon CWR species occurrence data points, displaying the median and variance parameters per
species per variable. Comparable ecogeographic variable data for the sweetpotato crop was
extracted from area of cultivation maps (Monfreda et al. 2008) at a resolution of 5 arc-minutes,
with a random sample of 1,000 points weighted by harvested area, taken from the major
140

Crop wild relatives of sweetpotato
cultivation areas in Asia, Africa, and Latin America.
RESULTS
Distributions of the wild relatives of sweetpotato
Sweetpotato CWR were modeled to occur from the central USA to northern Argentina in the
Americas, including the Caribbean (Supplementary Figure 1). Species richness was greatest in
central Mexico through Central America to the northern Andean region, and in the southeastern
USA, with up to nine species potentially overlapping in Mexico from the states of Veracruz
through the Yucatan peninsula (Figure 1). The Mexican and Central American regions of
distribution represent one of the posited centers of origin and primary diversity of cultivated
sweetpotato (Austin 1988, Austin & Huamán 1996, Zhang et al. 2000, Gichuki et al. 2003,
Roullier et al. 2013). Northwestern South America, with archeological remains of cultivated
sweetpotato from Peru dating back to 8,000 years BP, which are among the oldest recorded
domestication events on the continent (Piperno & Pearsall 1998, Shady Solis et al. 2001),
displayed a considerably lesser but still notable degree of CWR species richness. One Old
World species (I. littoralis) (Austin 1991) was also modeled to occur in coastal areas of
Madagascar, South and Southeast Asia, Australia, and the Pacific region.
Analysis of current ex situ conservation and further collecting needs for CWR
Eleven out of the fourteen CWR species were assigned high priority for further collecting due
to insufficient genebank accessions in comparison to the total number of occurrence samples
(SRS), and to large geographic (GRS) and ecological (ERS) gaps in ex situ germplasm
collections in comparison to the full potential distributions of the species (Figure 2, Table 1,
Supplementary Figure 2). Six of these taxa are currently represented by ≤ 10 accessions
conserved ex situ, moreover, with few exceptions these accessions lack associated geographic
occurrence information (Supplementary Table 3). Total sampling representativeness and
geographic coverage of species in germplasm collections were particularly lacking for taxa
assessed high priority, while gaps in ecological representativeness were less extreme for some
species (e.g., I. cordatotriloba, I. triloba, and I. splendor-sylvae). Ipomoea grandifolia and I.
trifida were assessed as relatively well covered in regard to ecosystems represented ex situ,
which raised their final priority score into the medium priority for further collecting category.
Ipomoea tabascana was assessed as of low priority for further collecting due to existing
germplasm collections largely covering its very restricted distribution in southern Mexico,
141

Chapter 5

Figure 1. Species richness map for assessed sweetpotato crop wild relative potential distribution models
worldwide (A), with concentration of species in the neotropics (B).

142

Crop wild relatives of sweetpotato
resulting in a high GRS score. The mean final priority score across all CWR was 1.75 ± 1.82.
Paralleling the distribution of species richness of sweetpotato CWR, the regions identified for
further collecting of the greatest number of species included central and southern Mexico and
the southeastern USA, with up to seven species prioritized for further collecting occurring in
the same area (Figure 3). Further collecting priorities were recognized in a total of 50 countries
throughout the range of the genepool (Supplementary Figure 3, Supplementary Table 4).
Occurrence data, potential distribution models, and collecting priorities maps for all modeled
species are available in an interactive map format at http://www.cwrdiversity.org/ distributionmap/.
Expert evaluation of conservation assessment results
The average of the directly comparable expert priority scores (EPS) across experts was 2.65
(±1.10) as a mean across species, varying 1.95 points on average from the FPS, with seven taxa
designated by the experts as HPS, and seven as MPS (Table 1, Figure 2). For most species, this
mean was highly influenced by one or two experts’ giving species considerably less priority
than the other experts. Species with closest accord between the gap analysis results and the
comparable expert analysis included I. splendor-sylvae, I. trifida, I. tenuissima, I. littoralis, I.
tileacea, I. ramosissima, and I. grandifolia. Taxa with the greatest difference between gap
analysis and comparable expert values included I. tabascana and I. leucantha.
Two species were assessed by the experts as higher priority for further collecting than the results
of the gap analysis. Although ex situ collections for the highly restricted distribution of I.
tabascana were determined in the gap analysis to be fairly comprehensive (LPS), the experts
assigned the species high priority (HPS) for further collecting due to its very limited overall
number of germplasm holdings, and to the difficulty in producing viable seed in ex situ
conservation. Ipomoea grandifolia was assessed in the gap analysis as reasonably
comprehensively represented in regard to ecosystem diversity, and thus assigned medium
priority for further collecting, while the experts gave moderately higher priority to the species.
The contextual expert priority score per species, which also included the expert’s knowledge
of threats in situ as well as usefulness for crop improvement, did not vary widely from the

143

Chapter 5

Figure 2. Gap analysis results and comparable expert assessments per sweetpotato crop wild relative.
Species are listed by descending priority for further collecting by priority categories [high priority
species, HPS (red); medium priority species, MPS (orange); low priority species, LPS (yellow); and no
further collecting recommended, NFCR (green)]. The black circle represents the final priority score
(FPS) for the species, which is the mean of the sampling representativeness score (SRS), geographic
representativeness score (GRS), and ecological representativeness score (ERS).

144

Crop wild relatives of sweetpotato

Figure 3. Further collecting priorities hotspots map for high priority (HPS) sweetpotato crop wild
relatives. The map displays areas worldwide (A) within the potential distributions of HPS species that
have not been previously collected for ex situ conservation, with concentration of species in the
neotropics (B).

145

Chapter 5
comparable score [mean across all experts and species was 2.76 (± 0.85); mean difference
between comparable and contextual expert scores across all species and experts was 0.11]. Due
largely to differences between the perceptions of relative sufficiency in regard to the overall
number of germplasm accessions by the experts versus gap analysis results (e.g., for I.
tabascana, I. triloba, and I. leucantha), the comparable and contextual assessments did not
correlate well with the gap analysis results for the genepool as a whole (Supplementary Figure
4A-B).
The multiple factor analysis combining the comparable expert priority score, contextual expert
priority score, evaluation of gap analysis results score, evaluation of occurrence data, evaluation
of potential species distribution models, and evaluation of collecting priorities map, revealed
sufficient agreement among experts and variables to justify the use of the expert evaluation
index, although variation in expert option was notable for many species. Data inputs and
resulting distribution models were generally assessed positively as a mean across experts, with
eight species receiving very positive index values. Those species with the highest accord
between all variables and experts and the gap analysis results included I. littoralis, I. spendorsylvae, I. ramosissima, I. tenuissima, and I. tiliacea. Those species with the lowest accord
included I. tabascana, I. triloba, and I. grandifolia (Supplementary Figure 4). FPS results were
particularly influenced by spotty occurrence records for the majority of species, with gaps
recognized by the experts.
Identification of ecogeographic characteristics of CWR
The analysis of geographic and ecogeographic accord between pairwise potential species
distribution models segregated species well into temperate North American (e.g., I. lacunosa,
I. tenuissima), Mesoamerican (e.g., I. splendor-sylvae, I. tabascana), widely distributed tropical
(e.g., I. triloba, I. trifida), and South American (e.g., I. grandifolia, I. cynanchifolia) areas
(Supplementary Figure 5).
Strong linear relationships were found between bioclimatic variables within the study area,
justifying the application of the PCA, with 75.6% of variance explained through four principal
components (Supplementary Figure 6). The first principal component (32% of variation) was
correlated with temperature extremes and fluctuation. The second component (21.6% of
variation) was most occupied by precipitation variables related to drought. The third component
146

Crop wild relatives of sweetpotato
(13.9% of variation), was related with altitude, and the final component (8.1%) with soil texture
characteristics.
Species occurrence data segregated into temperate and tropical ecogeographic clusters, with the
great majority of species’ distributions strongly occurring within a single cluster.
Ecogeographic variables most strongly influencing the definition of the temperate cluster (1)
included those associated with temperature variation and relatively low precipitation, elevation,
and soil organic matter. The most determinant variables in the tropical cluster (2) were related
to relatively high and consistent temperatures. Those species displaying a notable proportion of
occurrences within both clusters included I. cynanchifolia and I. triloba, and to a lesser degree
I. leucantha and I. ramosissima (Table 1, Supplementary Figure 6).
Ecogeographic niche assessments of sweetpotato CWR based upon occurrence data points
revealed large differences in potential adaptation to temperature, precipitation, and edaphic
characteristics (Table 1, Figure 4, Supplementary Figure 7). Such adaptation for many species
fell well outside the modeled niche of the cultivated species, particularly for high temperatures
both in wet and dry conditions, as well as high precipitation. Species of notable adaptation to
high mean annual, monthly, and quarterly temperatures included I. littoralis, I. tabascana, I.
trifida, I. leucantha, I. tiliacea, I. tenuissima, I. triloba, I. splendor-sylvae, and wild I. batatas.
USA species I. lacunosa stood out for adaptation to low temperatures, with I. grandifolia, I.
cordatotriloba, I. tenuissima, and I. ramossissima also demonstrating cold tolerance. These
same species were among those displaying the widest adaptation to temperature fluctuation.
CWR of sweetpotato occurring in areas of high precipitation included I. ramossissima, I.
littoralis, I. splendor-sylvae, I. tabascana, I. tiliacea, I. trifida, and wild I. batatas. Ipomoea
littoralis, I. trifida, I. splendor-sylvae, I. leucantha, I. cynanchifolia, I. triloba, I. lacunosa, and
wild I. batatas were distributed in regions with low precipitation. These species were also
among those displaying the widest adaptation to precipitation seasonality. Sweetpotato CWR
also displayed variation in adaptation to edaphic characteristics. Ipomoea tabascana, I.
grandifolia, I. tiliacea, I. splendor-sylvae, I. ramosissima, I. cordatotriloba, and wild I. batatas
occurred in clay soils, while I. tenuissima, I. littoralis, I. cynanchifolia, I. grandifolia, I.
leucantha, and I. cordatotriloba were distributed in sandy soils.

147

Chapter 5
Table 1. Utilization characteristics, number of germplasm accessions conserved ex situ, further
collecting priorities, and potential adaptive traits associated with ecogeographic niches of sweetpotato
crop wild relatives.
Taxon

Gene
pool

Ploidy

Germplasm
accessions

Gap
analysis
priority

Mean
expert
priority

Eco
geographic
cluster

Potential adaptation to

I. batatas

2

4x=60

4 (0)

HPS

HPS

2

Heat, high precipitation,
drought, precipitation
seasonality, clay soils

I. cordatotriloba

3

2x, 4x

103 (67)

HPS

MPS

1

Cold, temperature
variation, clay soils,
sandy soils

I. cynanchifolia

3

2x=30

1 (0)

HPS

HPS

1,2

Drought, precipitation
seasonality, sandy soils

I. grandifolia

3

2x=30

124 (83)

MPS

MPS

1

Cold, temperature
variation, clay soils,
sandy soils

I. lacunosa

3

2x=30

10 (1)

HPS

MPS

1

Cold, temperature
variation, drought

I. leucantha

3

2x=30

18 (15)

HPS

MPS

1,2

I. littoralis

2

2x=30

2 (2)

HPS

HPS

2

Heat, drought,
precipitation seasonality,
sandy soils
Heat, high precipitation,
drought, precipitation
seasonality, sandy soils

I. ramosissima

3

2x=30

34 (30)

HPS

MPS

2,1

Cold, high precipitation,
clay soils

I. splendor-sylvae

3

2x=30

16 (9)

HPS

HPS

2

Heat, high precipitation,
drought, precipitation
seasonality, clay soils

I. tabascana

2

4x=60

4 (2)

LPS

HPS

2

Heat, high precipitation,
clay soils

I. tenuissima

3

2x=30

3 (1)

HPS

HPS

1

Heat , cold, temperature
variation, sandy soils

I. tiliacea

3

4x=60

61 (44)

HPS

HPS

2

Heat, high precipitation,
clay soils

I. trifida

2

2x,3x,4x,6x

248 (159)

MPS

MPS

2

Heat, high precipitation,
drought, precipitation
seasonality

I. triloba

3

2x=30

121 (74)

HPS

MPS

2,1

Heat, drought

Ploidy data adapted from Nimmakayala et al. (2011). Germplasm accessions displays both the total number of
accessions recorded in genebanks, as well as the number of accessions with unique geographic coordinates (i.e.,
unique populations) in parenthesis. HPS = high, MPS = medium, and LPS = low priority species for further
collecting.

148

Crop wild relatives of sweetpotato

Figure 4. Ecogeographic niches of crop wild relative species based upon their occurrence data presence
locations, and the sweetpotato crop, for (A) annual mean temperature, (B) mean temperature of the driest
quarter of the year, (C) annual precipitation, and (D) precipitation of the driest quarter of the year. The
thick grey line represents median values, boxplots between 25-75% variation, and open circles outliers
within 90% of total variation. For a principal component analysis of variables see Supplementary Figure
6 and for ecogeographic niches displaying total variation for all variables per species see Supplementary
Figure 7.

DISCUSSION
This article utilizes the most current knowledge on species concepts within I. series Batatas.
Due to taxonomic uncertainties and to the notable dearth of study material for sweetpotato
CWR, the results represent a preliminary understanding of the geography and conservation
status of the series, to be further refined following increased levels of collecting sufficient to
support the needed taxonomic and crossability research. Further collecting of germplasm thus
serves two purposes: a) conserves genetic resources for the long-term and makes these
resources available to breeders; and b) provides the basic materials needed by researchers to
understand the diversity present in the CWR of sweetpotato.
A total of 78.6% of the wild relatives of sweetpotato considered in this study were assessed as
high priority for further collecting for ex situ conservation. With general agreement from expert
evaluators of medium to high importance for all species, it is clear that much remains to be done
149

Chapter 5
to safeguard the wild genetic resources of this critically important root crop. Included in this
list of priorities are species with very few germplasm accessions accessible to the global
community in genebank information systems, including I. cynanchifolia, I. littoralis, I.
tenuissima, I. tabascana, I. lacunosa, I. leucantha, I. splendor-sylvae, and clearly designated
wild forms of the crop conspecific I. batatas. Such species represent the highest level of priority
for further collecting for use in systematic analyses as well as genetic resources conservation.
As the species diversity gaps in genebank collections largely align with the geographic
distribution of species richness of sweetpotato CWR, hotspots in central Mexico to Central
America, and in the extreme southeastern USA, represent particularly high priority regions for
efficient collecting of the sweetpotato genepool (Figure 3). Additional unrepresented
populations of high priority species such as I. littoralis and I. cynanchifolia occur outside those
regions, thus targeted collecting throughout the geographic distribution of the genepool is
necessary in order to form comprehensive germplasm collections.
Due to having relatively large potential distributions occupying a diversity of ecosystems,
species such as I. triloba, I. cordatotriloba, and I. tiliacea were categorized as high priority,
and I. trifida and I. grandifolia as medium priority for further collecting despite having sizable
currently-existing germplasm collections. As the cost of conserving and investigating
germplasm ex situ is significant, a further assessment of what constitutes sufficient
representation of these species in germplasm collections is warranted. Given adequate
resources, further collecting may be worthwhile, as extremely valuable traits sourced from
CWR of some crops have been found in only a few populations despite screening of a large
number of accessions (Brar & Khush 1997), and accessions of individual CWR species such as
I. triloba have been shown to possess notable variation in traits such as tolerance to precipitation
(Martin & Jones 1973, Nimmakayala et al. 2011).
As Maxent models are based upon known presence points for species and are thus subject to
sampling bias, they may not fully capture the possibility of occurrence of populations of CWR
species in unique climates (Araújo & Guisan 2006, Loiselle et al. 2008, Kramer-Schadt et al.
2013). Further field exploration of climatic extremes beyond the edges of the distributions
created through these methods may therefore lead to the discovery of new populations with
particularly valuable adaptations to abiotic stress (Williams et al. 2009). Investigation of nonnative populations (e.g., I. trifida in Asia) may also yield novel adaptations of use in crop
150

Crop wild relatives of sweetpotato
improvement. As techniques for the utilization of distantly related germplasm improve, the
exploration of other Ipomoea species outside of I. series Batatas may also result in the
identification of beneficial traits [e.g., Ipomoea purpurea (L.) Roth, for stem nematode and
SPVD resistance (Cao et al. 2009)].
Analysis of geographic overlap and ecogeographic similarities between species, as well as
ecogeographic clusters among all species, can supplement morphological and genetic analyses
in differentiating useful genetic resources, and can serve as a point of departure for identifying
taxonomically problematic populations for further investigation. These analyses may also
indicate geographic areas of particular interest in regard to high rates of hybridization, as in the
case of I. cordatotriloba and I. lacunosa (Duncan & Rausher 2013), which indeed were
identified as sharing similar geographies and ecogeographic niches. The purported hybrid
descendent of these species, I. leucantha, was modeled as containing a much more extended
latitudinal gradient in the northern hemisphere than its parents, as well as a differing
ecogeographic niche, including potential adaptation to high heat and to drought.
Genetic resistance is essential to efforts to overcome major biotic and abiotic constraints in
sweetpotato production. As these constraints are often interrelated, e.g., drought stress with
SPW and SPVD damage, enhancement of broad resistance for traits such as drought may
improve yield across relatively large geographic areas, without the need to breed for resistance
to localized viral strains (Ngailo et al. 2013). Such broad tolerance may also improve adoption
rates for sweetpotato varieties with other desirable characteristics, such as high β-carotene
content.
Reliable funding for germplasm collections is paramount in order to safeguard sweetpotato
CWR genetic resources in the long-term and to continue to make ex situ collections available
to the global community. Further investment in genebank information systems, ex situ
conservation technologies (i.e., storage, testing, and regeneration), safety duplication of unique
germplasm, and ensuring facilitated access to genetic diversity is equally essential (FAO 2002,
FAO 2010, Khoury et al. 2010). In order to maximize the usefulness of conserved germplasm,
characterization and evaluation for traits of interest, and increased breeding research, which
have been limited for CWR of sweetpotato, are also needed. Further research combining
morphological studies, trait evaluations, and genetic diversity analyses is likewise critically
151

Chapter 5
needed for elucidating species boundaries and highlighting accessions of particular value for
use in breeding. Recent focused research has produced quick gains, including the identification
of new species (Duncan & Rausher 2013, Wood et al. 2015). Through these actions the crop
research community will contribute to ensuring the long term viability of this important root
crop.

152

Crop wild relatives of sweetpotato
SUPPLEMENTARY INFORMATION

Supplementary Figure 1(A-D). Sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative potential
distribution models.

153

Chapter 5

Supplementary Figure 1(E-H). Sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative potential
distribution models.

154

Crop wild relatives of sweetpotato

Supplementary Figure 1(I-L). Sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative potential
distribution models.

155

Chapter 5

Supplementary Figure 1(M-N). Sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative potential
distribution models.

156

Supplementary Figure 2. Distribution of priority scores per sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative species for (A) sampling
representativeness score (SRS), (B) geographic representativeness score (GRS), and (C) ecological representativeness score (ERS).

Crop wild relatives of sweetpotato

157

Chapter 5

Supplementary Figure 3. Number of sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relatives
prioritized for further collecting for ex situ conservation per country. HPS = high, MPS = medium, and
LPS = low priority species for further collecting.

158

Crop wild relatives of sweetpotato

Supplementary Figure 4. Expert evaluation accordance with gap analysis results: (A) correlation
between gap analysis results and comparable expert evaluation scores. (B) correlation between gap
analysis results and contextual expert evaluation scores. (C) correlation circle of all evaluation variables
[comparable expert priority score (Comparable), contextual expert priority score (Contextual),
evaluation of gap analysis results score (Evaluation), evaluation of occurrence data (Occ_data),
evaluation of potential species distribution models (SDM_map), and evaluation of collecting priorities
map (Gap_map)]. (D) combined expert evaluation index score per sweetpotato [Ipomoea batatas (L.)
Lam.] crop wild relative.

159

Chapter 5

Supplementary Figure 5. Geographic overlap and ecogeographic similarity of potential distribution
models between sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative species: (A) geographic
overlap of potential distribution models. (B) ecogeographic similarity index (I) of potential distribution
models.
160

Supplementary Figure 6. Principal component analysis of bioclimatic variables associated with occurrence data for sweetpotato [Ipomoea batatas
(L.) Lam.] crop wild relatives: (A) variation represented in the first four principal components. (B) determining bioclimatic variables for the first
four principle components.

Crop wild relatives of sweetpotato

161

Supplementary Figure 6. Principal component analysis of bioclimatic variables associated with occurrence data for sweetpotato [Ipomoea batatas (L.)
Lam.] crop wild relatives: (C) contribution of cluster identified occurrence data points per species. (D) geographic origin of clustered occurrence data
points.

Chapter 5

162

Crop wild relatives of sweetpotato
Supplementary Figure 7. Ecogeographic niches of crop wild relative (CWR) species based upon their
occurrence data presence locations, and the sweetpotato [Ipomoea batatas (L.) Lam.] crop, per
bioclimatic and edaphic variable. The bold grey line represents median values, boxplots between 2575% variation, and circles outliers.

Supplementary Figure 7A. Niches of CWR species and the sweetpotato crop for altitude.

Supplementary Figure 7B. Climatic niches of CWR species and the sweetpotato crop for annual mean
temperature.
163

Chapter 5

Supplementary Figure 7C. Climatic niches of CWR species and the sweetpotato crop for mean diurnal
range.

Supplementary Figure 7D. Climatic niches of CWR species and the sweetpotato crop for isothermality.
164

Crop wild relatives of sweetpotato

Supplementary Figure 7E. Climatic niches of CWR species and the sweetpotato crop for temperature
seasonality.

Supplementary Figure 7F. Climatic niches of CWR species and the sweetpotato crop for maximum
temperature of the warmest month of the year.
165

Chapter 5

Supplementary Figure 7G. Climatic niches of CWR species and the sweetpotato crop for minimum
temperature of the coldest month of the year.

Supplementary Figure 7H. Climatic niches of CWR species and the sweetpotato crop for annual
temperature range.
166

Crop wild relatives of sweetpotato

Supplementary Figure 7I. Climatic niches of CWR species and the sweetpotato crop for mean
temperature of the wettest quarter of the year.

Supplementary Figure 7J. Climatic niches of CWR species and the sweetpotato crop for mean
temperature of the driest quarter of the year.
167

Chapter 5

Supplementary Figure 7K. Climatic niches of CWR species and the sweetpotato crop for mean
temperature of the warmest quarter of the year.

Supplementary Figure 7L. Climatic niches of CWR species and the sweetpotato crop for mean
temperature of the coldest quarter of the year.
168

Crop wild relatives of sweetpotato

Supplementary Figure 7M. Climatic niches of CWR species and the sweetpotato crop for annual
precipitation.

Supplementary Figure 7N. Climatic niches of CWR species and the sweetpotato crop for precipitation
of the wettest month of the year.
169

Chapter 5

Supplementary Figure 7O. Climatic niches of CWR species and the sweetpotato crop for precipitation
of the driest month of the year.

Supplementary Figure 7P. Climatic niches of CWR species and the sweetpotato crop for precipitation
seasonality.
170

Crop wild relatives of sweetpotato

Supplementary Figure 7Q. Climatic niches of CWR species and the sweetpotato crop for precipitation
of the wettest quarter of the year.

Supplementary Figure 7R. Climatic niches of CWR species and the sweetpotato crop for precipitation
of the driest quarter of the year.
171

Chapter 5

Supplementary Figure 7S. Climatic niches of CWR species and the sweetpotato crop for precipitation
of the warmest quarter of the year.

Supplementary Figure 7T. Climatic niches of CWR species and the sweetpotato crop for precipitation
of the coldest quarter of the year.
172

Crop wild relatives of sweetpotato

Supplementary Figure 7U. Edaphic niches of CWR species and the sweetpotato crop for bulk density.

Supplementary Figure 7V. Edaphic niches of CWR species and the sweetpotato crop for cation exchange
capacity.
173

Chapter 5

Supplementary Figure 7W. Edaphic niches of CWR species and the sweetpotato crop for percent clay.

Supplementary Figure 7X. Edaphic niches of CWR species and the sweetpotato crop for organic carbon.
174

Crop wild relatives of sweetpotato

Supplementary Figure 7Y. Edaphic niches of CWR species and the sweetpotato crop for pH.

Supplementary Figure 7Z. Edaphic niches of CWR species and the sweetpotato crop for percent silt.
175

Chapter 5

Supplementary Figure 7AA. Edaphic niches of CWR species and the sweetpotato crop for percent sand.

176

Crop wild relatives of sweetpotato
Supplementary Table 1. Sources of occurrence data for assessed sweetpotato [Ipomoea batatas (L.)
Lam.] crop wild relative species.
Data provider

Record type

International Potato Center (CIP)

Germplasm accession

Number of
records
585

Leibniz Institute of Plant Genetics and Crop Plant Research

Germplasm accession

1

Millennium Seed Bank Partnership, Royal Botanic Gardens, Kew

Germplasm accession

3

USDA, National Plant Germplasm System, Germplasm Resources Information Network
(USDA NPGS GRIN)
Atlas of Living Australia (Flora Atlas NT) (via GBIF)

Germplasm accession

160

Herbarium or other record

356

Australia's Virtual Herbarium (AD, BRI, CANB, CBG, DNA, HO, MEL, NSW, PERTH)

Herbarium or other record

39

Bioversity International

Herbarium or other record

208

Brazil Virtual Herbarium (CRIA)

Herbarium or other record

61

California Academy of Sciences Herbarium (CAS)

Herbarium or other record

128

Comisión Nacional para el Conocimiento y Uso de la Biodiversidad (CONABIO) (via GBIF)

Herbarium or other record

97

Consortium of Pacific Northwest Herbaria (CPNWH)

Herbarium or other record

5

DIVEA, DEP, FEEMA Herbário Alberto Castellanos (GUA)

Herbarium or other record

18

Fairchild T ropical Botanical Garden (FTG) (via GBIF)

Herbarium or other record

712

Florida State University Robert K. Godfrey Herbarium (FSU)

Herbarium or other record

6

Harvard University Herbarium (HUH)

Herbarium or other record

8

Herbarium of National Taiwan University (TAI) (via GBIF)

Herbarium or other record

63

Herbarium of T aiwan Forestry Research Institute (TAIF) (via GBIF)

Herbarium or other record

56

Instituto Nacional de Biodiversidad, Costa Rica (INB) (via GBIF)

Herbarium or other record

87

International Center for Tropical Agriculture (CIAT)- Bioversity International, USDA. Atlas
of Paraguayan Crop Wild Relatives.
International Center for Tropical Agriculture (CIAT)- C. Azurdia, K.A. Williams, D.E.
Williams, V. Van Damme, A. Jarvis and S.E. Castaño. 2011. Atlas of Guatemalan Crop Wild
Relatives.
International Potato Center (CIP)

Herbarium or other record

155

Herbarium or other record

139

Herbarium or other record

231

Jardim Botanico de Rio de Janeiro (JABOT/R)

Herbarium or other record

106

Jardim Botânico do Rio de Janeiro (JBRJ) (via GBIF)

Herbarium or other record

119

Louisiana State University Herbarium (LSU) (via GBIF)

Herbarium or other record

116

Manchester University Herbarium (MANCH)

Herbarium or other record

2

McDonald & Austin (1990)

Herbarium or other record

1

Missouri Botanical Garden Herbarium (MO)

Herbarium or other record

44

Museo Nacional de Costa Rica (MNCR) (via GBIF)

Herbarium or other record

80

Museum national d'Histoire naturelle (MNHN) (via GBIF)

Herbarium or other record

509

Nationaal Herbarium Nederland (NHN)

Herbarium or other record

112

Natural History Museum UK Herbarium (BM)

Herbarium or other record

19

New York Botanical Garden Herbarium (NY)

Herbarium or other record

454

Plants of Taiwan

Herbarium or other record

56

Real Jardín Botánico de Madrid (MA)

Herbarium or other record

25

Royal Botanic Gardens, Kew (K)

Herbarium or other record

20

Scotland & Wood, personal communication

Herbarium or other record

8

Smithsonian Institution, National Herbarium (US)

Herbarium or other record

357

T he Field Museum (F)

Herbarium or other record

58

Universidad del Valle Colombia Herbarium (CUVC)

Herbarium or other record

4

University of California Jepson Herbarium (UC/Jeps)

Herbarium or other record

63

University of California Riverside Herbarium (UCR)

Herbarium or other record

23

University of Kansas McGregor Herbarium (KU) (via GBIF)

Herbarium or other record

79

University of Oklahoma, Robert Bebb Herbarium (OKL) (via GBIF)

Herbarium or other record

53

US Academy of Natural Sciences Philadelphia Herbarium (PH)

Herbarium or other record

45

USDA, National Plant Germplasm System, Germplasm Resources Information Network
(USDA NPGS GRIN)
West Virginia University Herbarium (WVU)

Herbarium or other record

131

Herbarium or other record

12

177

Chapter 5
Supplementary Table 2. Ecogeographic variables utilized in sweetpotato [Ipomoea batatas (L.) Lam.]
crop wild relative potential distribution modeling and climatic niche analyses.
Variable Variable name
Units
number
0
Altitude
m
1
Annual mean temperature
ºC
2
Mean diurnal temperature range
ºC
3
Isothermality
N/A
4
Temperature seasonality (standard deviation)
ºC
5
Maximum temperature of warmest month
ºC
6
Minimum temperature of coldest month
ºC
7
Temperature annual range
ºC
8
Mean temperature of wettest quarter
ºC
9
Mean temperature of driest quarter
ºC
10
Mean temperature of warmest quarter
ºC
11
Mean temperature of coldest quarter
ºC
12
Annual precipitation
mm
13
Precipitation of wettest month
mm
14
Precipitation of driest month
mm
15
Precipitation seasonality (coefficient of variation)
%
16
Precipitation of wettest quarter
mm
17
Precipitation of driest quarter
mm
18
Precipitation of warmest quarter
mm
19
Precipitation of coldest quarter
mm
20
Bulk density
kg/m3
21
Cation exchange capacity
cmol/kg
22
Percent clay
%
23
Organic carbon
g/kg
24
pH in H2 O
pH
25
Percent silt
%
26
Percent sand
%

178

104

424

37

460

413

134

468

341

177

8

34

740

1409

865

Taxon

I. cordatotriloba

I. cynanchifolia

I. grandifolia

I. lacunosa

I. leucantha

I. littoralis

I. ramosissima

I. splendor-sylvae

I. tabascana

I. tenuissima

I. tiliacea

I. trifida

I. triloba

Total Records- Distinct

446

1023

441

12

6

135

217

340

93

252

334

17

290

44

H Records- Total

744

1161

679

31

4

161

307

466

116

403

336

36

321

100

H Records- Distinct

376

888

400

12

4

130

187

338

79

251

254

17

229

44

G Records- Total

121

248

61

3

4

16

34

2

18

10

124

1

103

4

G Records- Distinct
74

159

44

1

2

9

30

2

15

1

83

0

67

0

SRS

1.40

1.76

0.82

0.88

5.00

0.90

1.00

0.04

1.34

0.24

2.70

0.27

2.43

0.38

Ecogeographic
Variables

0, 2, 3, 9, 12, 16, 24

0, 5, 11, 12, 13, 24

0, 5, 11, 12, 17, 19, 20, 24

10, 13, 14, 21(*)

22, 23, 24, 26(*)

0, 5, 11, 19, 20

0, 7, 11, 13, 19, 20, 23, 24

1, 2, 4, 15, 18, 19, 20, 23

1, 2, 3, 8, 12, 16, 20, 24

0, 3, 5, 7, 9, 12, 13, 25

3, 7, 8, 9, 10, 15, 16

0, 15, 16, 18, 19, 22, 23, 26(*)

1, 2, 3, 9, 13, 15, 18

2, 4, 8, 13, 14, 19, 24

ATAUC

0.93

0.92

0.94

0.95

0.87

0.90

0.85

0.98

0.91

0.97

0.95

0.76

0.97

0.78

STAUC

0.01

0.01

0.01

NA

NA

0.04

0.02

0.00

0.04

0.00

0.01

NA

0.01

0.06

ASD15

0.12

0.01

0.37

NA

NA

2.76

1.07

0.62

0.19

0.49

0.02

NA

0.93

11.19

cAUC

0.43

0.42

0.44

0.45

0.37

0.40

0.35

0.48

0.41

0.47

0.45

0.26

0.47

0.28

GRS

1.07

2.24

0.59

0.32

10.00

0.97

0.50

0.02

0.06

0.07

2.54

0.00

1.24

0.00

ERS

4.12

4.95

1.58

1.05

4.29

4.88

2.34

0.07

0.57

0.91

7.22

0.00

2.86

0.00

FPS
2.20

2.98

1.00

0.75

6.43

2.25

1.28

0.05

0.66

0.41

4.15

0.09

2.18

0.13

FPCAT
HPS

MPS

HPS

HPS

LPS

HPS

HPS

HPS

HPS

HPS

MPS

HPS

HPS

HPS

EPS 1

6

3

4

0

0

1

4

0

4

4

3

0

6

1

EPS 2

1

1

1

5

1

7

3

3

7

8

3

7

7

1

EPS 3

4

4

5

3

2

3

4

2

3

2

5

2

5

7

EPS 4
4

4

1

0

0

1

1

0

NA

1

1

NA

1

1

6

6

1

0

0

0

2

0

2

1

1

0

1

2

EPS 5

4.2

3.6

2.4

1.6

0.6

2.4

2.8

1

4

3.2

2.6

2.3

4

2.4

Distinct records denote occurrence records with unique coordinates. H records denote herbarium and other locality data used to inform species distribution models. G records denote germplasm accessions
accessible to the global community. Ecogeographic variables lists those species-specific bioclimatic and edaphic variables used in Maxent modeling per taxon (T able S2); (*) denotes species for which an
ensemble model was derived. Maxent assessment techniques include the 5-fold average area under the ROC curve of test data (AT AUC), the standard deviation of the test AUCofthe5 different folds(STAUC),
the proportion of the potential distribution coverage with standard deviation above 0.15 (ASD15), and the calibrated AUC ( cAUC). T he final priority score (FPS) for the species isthemean of thesampling
representativeness score (SRS), geographic representativeness score (GRS), and ecological representativeness score (ERS). Final priority categories (FPCAT ) for the CWR of sweetpotato includedhigh (HPS),
medium (MPS), and low (LPS) priority species for further collecting for ex situ conservation. EPS denotes comparable expert priority scores for 5 experts, along with their mean value.

Total Records

I. batatas

EPS mean

Supplementary Table 3. Gap analysis and comparable expert evaluation priorities results per sweetpotato [Ipomoea batatas (L.) Lam.] crop wild relative.

Crop wild relatives of sweetpotato

179

Chapter 5
Supplementary Table 4. Countries identified for further collecting per sweetpotato [Ipomoea batatas
(L.) Lam.] crop wild relative.
Taxon
Priority Countries identified for further collecting
I. batatas
HPS
Belize, Bolivia, Brazil, Colombia, Costa Rica, Ecuador, El Salvador,
French Guiana, Guatemala, Guyana, Honduras, Mexico, Nicaragua,
Panama, Peru, Suriname, Venezuela
I. cordatotriloba HPS
Argentina, Bolivia, Brazil, Mexico, Paraguay, USA
I. cynanchifolia
HPS
Bolivia, Brazil
I. grandifolia
MPS
Argentina, Bolivia, Brazil, Paraguay, Uruguay
I. lacunosa
HPS
USA
I. leucantha
HPS
Mexico, USA
I. littoralis
HPS
Australia, Fiji, French Polynesia, Guam, India, Indonesia, Kiribati,
Madagascar, Malaysia, Mauritius, Micronesia, N. Mariana Islands,
New Caledonia, Papua New Guinea, Philippines, Seychelles, Sri
Lanka, Taiwan, Thailand, Timor-Leste, USA
I. ramosissima
HPS
Belize, Bolivia, Brazil, Colombia, Costa Rica, Ecuador, El Salvador,
Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru,
Venezuela
I. splendor-sylvae HPS
Belize, Costa Rica, El Salvador, Guatemala, Honduras, Mexico,
Nicaragua, Panama
I. tabascana
LPS
Mexico
I. tenuissima
HPS
Cuba, Dominican Republic, Haiti, Puerto Rico, USA
I. tiliacea
HPS
Bahamas, Belize, Brazil, Colombia, Costa Rica, Cuba, Dominican
Republic, El Salvador, French Guiana, Grenada, Guatemala, Guyana,
Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, Puerto Rico,
Sao Tome and Principe, St. Vincent and the Grenadines, Suriname,
USA, Venezuela
I. trifida
MPS
Belize, Colombia, Costa Rica, Cuba, Ecuador, El Salvador,
Guatemala, Honduras, Mexico, Nicaragua, Panama, USA, Venezuela
I. triloba

180

HPS

Bahamas, Belize, Colombia, Costa Rica, Cuba, Dominican Republic,
Ecuador, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Mexico,
Nicaragua, Panama, Peru, USA, Venezuela

Crop wild relatives of sweetpotato
REFERENCES
Araújo, M.B., Guisan, A. (2006). Five (or so) challenges for species distribution modelling. J. Biogeogr.
33 (10), 1677-1688.
Austin, D.F. (1978). The Ipomoea batatas complex - 1. Taxonomy. Bull. Torrey Bot. Club 105, 114129.
Austin, D.F. (1988). “The taxonomy, evolution and genetic diversity of sweetpotatoes and related wild
species,” in: Exploration, Maintenance, and Utilization of Sweetpotato Genetic Resources, ed. P.
Gregory (Lima, Peru: International Potato Center), 27-60.
Austin, D.F. (1991). Ipomoea littoralis (Convolvulaceae) - Taxonomy, distribution, and ethnobotany.
Econ. Bot. 45, 251-256.
Austin, D.F., (1997). Dissolution of Ipomoea series Anisomerae (Convolvulaceae). Taxon 28, 359-361.
Austin, D.F., Huamán, Z. (1996). A synopsis of Ipomoea (Convolvulaceae) in the Americas. Taxon 45,
3-38.
Bill and Melinda Gates Foundation. (2011). Agricultural Development Strategy Overview. Available at:
https://docs.gatesfoundation.org/Documents/agricultural-development-strategy-overview.pdf
Bivand, R., Lewin-Koh, N. (2014). Maptools: tools for reading and handling spatial objects. R package
version 0.8-30. http://CRAN.R-project.org/package=maptools.
Bivand, R., Keitt, T., Rowlingson, B. (2014). Rgdal: bindings for the geospatial data abstraction library.
R package version 0.8-16. http://CRAN.R-project.org/package=rgdal.
Bivand, R., Pebesma, E., Gomez-Rubio, V. (2013). Applied Spatial Data Analysis with R, Second
edition. NY, Springer. http://www.asdar-book.org/.
Bouis, H., Islam, Y. (2012). Delivering nutrients widely through biofortification: building on orange
sweet potato - scaling up in agriculture, rural development and nutrition. Focus 19, Brief 11.
(Washington DC: International Food Policy Research Institute).
Brar, D.S., Khush, G.S. (1997). Alien introgression in rice. Plant Mol. Biol. 35 (1-2), 35-47.
Cao, Q., Zhang, A., Ma, D., Li, H., Li, Q., Li, P. (2009). Novel interspecific hybridization between
sweetpotato (Ipomoea batatas (L.) Lam.) and its two diploid wild relatives. Euphytica 169, 345-352.
Conolly, J., Manning, K., Colledge, S., Dobney, K., Shennan, S. (2012). Species distribution modelling
of ancient cattle from early neolithic sites in SW Asia and Europe. Holocene 22, 997-1010.
Costa, G.C., Nogueira, C., Machado, R.B., Colli, G.R. (2010). Sampling bias and the use of ecological
niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodivers.
Conserv. 19, 883-899.
De Vries, C.A., Ferwerda, J.D., Flach, M. (1967). Choice of food crops in relation to actual and potential
production in the tropics. Neth. J. Agr. Sci. 15, 241-248.
Dempewolf, H., Eastwood, R.J., Guarino, L., Khoury, C.K., Müller, J.V., Toll, J. (2014). Adapting
agriculture to climate change: a global initiative to collect, conserve, and use crop wild relatives.
Agroecol. Sust. Food 38, 369-377.
Diaz, J., Schmiediche, P., Austin, D.F. (1996). Polygon of crossability between eleven species of
Ipomoea: section Batatas (Convolvulaceae). Euphytica 88, 189-200.
Duncan, T.M., Rausher, M.D. (2013). Evolution of the selfing syndrome in Ipomoea. Front. Plant Sci.
4, 301.
Elith, J., Graham, C.H., Anderson, R.P., Dudik, M., Ferrier, S., Guisan, A., et al. (2006). Novel methods
improve prediction of species’ distributions from occurrence data. Ecography 29, 129-151.
Eserman, L.A., Tiley, G.P., Jarret, R.L., Leebens-Mack, J.H., Miller, R.E. (2014). Phylogenetics and
diversification of morning glories (tribe Ipomoeeae, Convolvulaceae) based on whole plastome
sequences. Am. J. Bot. 101, 92-103.
181

Chapter 5
ESRI. (2011). ArcGIS Desktop: Release 10.1. Available at: http://www.esri.com/software/arcgis/
arcgis-for-desktop.
FAO (Food and Agriculture Organization of the United Nations). (2002). International Treaty on Plant
Genetic Resources for Food and Agriculture. Rome, Italy: FAO.
FAO (Food and Agriculture Organization of the United Nations). (2010). The Second Report on the
State of the World’s Plant Genetic Resources for Food and Agriculture. Rome, Italy: FAO.
Feuillet, C., Langridge, P., Waugh, R. (2008). Cereal breeding takes a walk on the wild side. Trends
Genet. 24, 24-32.
Fieberg, J., Kochanny, C.O. (2005). Quantifying home-range overlap: the importance of the utilization
distribution. J. Wildlife Manage. 69, 1346-1359.
Gaiji, S., Chavan, V., Ariño, A.H., Otegui, J., Hobern, D., Sood, R., et al. (2013). Content assessment
of the primary biodiversity data published through GBIF network: status, challenges and potentials.
Biodiversity Informatics 8, 94-172.
Gichuki, S.T., Berenyi, M., Zhang, D., Hermann, M., Schmidt, J., Glössl, J., et al. (2003). Genetic
diversity in sweetpotato [Ipomoea batatas (L.) Lam.] in relationship to geographic sources as
assessed with RAPD markers. Genet. Resour. Crop Ev. 50, 429-437.
Google Maps Geocoder. (2013). Version 3. https://developers.google.com/maps/documentation
/geocoding/?hl=pl.
GRIN (Germplasm Resources Information Network), USDA, ARS, National Genetic Resources
Program. (2013). USDA, ARS, National Genetic Resources Program. http://www.ars-grin.gov/cgibin/npgs/html/wep.pl.
Guarino, L., Lobell, D.B. (2011). A walk on the wild side. Nature Clim. Change 1 (8), 374-375.
Hajjar, R., Hodgkin, T. (2007). The use of wild relatives for crop improvement: a survey of
developments over the past 20 years. Euphytica 156, 1-13.
Harlan, J.R., de Wet, J.M.J. (1971). Toward a rational classification of cultivated plants. Taxon 20: 509517.
Heibl, C. (2011). Phyloclim: integrating phylogenetics and climatic niche modeling. v.0.9-4.
http://cran.r-project.org/web/packages/phyloclim/index.html.
Hengl, T., de Jesus, J.M., MacMillan, R.A., Batjes, N.H., Heuvelink, G.B.M., Ribeiro, E., et al. (2014).
SoilGrids1km - global soil information based on automated mapping. PLoS One 9 (8), e105992.
Hijmans, R.J. (2012). Cross-validation of species distribution models: removing spatial sorting bias and
calibration with a null model. Ecology 93: 679-688.
Hijmans, R.J. (2014). Raster: geographic data analysis and modeling. R package version 2.2-31.
http://CRAN.R-project.org/package=raster.
Hijmans, R.J., Cameron, S., Parra, J., Jones, P. G., Jarvis, A. (2005). WorldClim, Version 1.3 (Berkeley,
University of California).
Hijmans, R.J., Phillips, S., Leathwick, J., Elith, J. (2013). Dismo: Species distribution modeling. R
package version 0.9-3. http://CRAN.R-project.org/package=dismo.
Hijmans, R.J., Schreuder, M., De La Cruz, J., Guarino, L. (1999). Using GIS to check co-ordinates of
genebank accessions. Genet. Resour. Crop Ev. 46, 291-296.
Hjarding, A., Tolley, K.A., Burgessa, N.D. (2014). Red List assessments of East African chameleons: a
case study of why we need experts. Oryx First View 1-7. doi: 10.1017/S0030605313001427.
Hotz, C., Loechl, C., Lubowa, A., Tumwine, J.K., Ndeezi, G., Nandutu Masawi, A., et al. (2012).
Introduction of β-carotene-rich orange sweet potato in rural Uganda resulted in increased vitamin A
intakes among children and women and improved vitamin A status among children. J. Nutr. 142,
1871-1880.
Huaman, Z. (1987). “Current status on maintenance of sweetpotato genetic resources at CIP,” in:
182

Crop wild relatives of sweetpotato
Exploration, maintenance and utilization of sweetpotato genetic resources. Report of the First
Sweetpotato Planning Conference 1987 (Lima: International Potato Center [CIP]), 101-120.
Huang, J., Corke, H., Sun, M. (2002). Highly polymorphic AFLP markers as a complementary tool to
ITS sequences in assessing genetic diversity and phylogenetic relationships of sweetpotato (Ipomoea
batatas (L.) Lam.) and its wild relatives. Genet. Resour. Crop Ev. 49, 541-550.
Husson, F., Josse, J., Le, S., Mazet, J. (2009). FactoMineR: multivariate exploratory data analysis and
data mining with R. v.1.26. http://factominer.free.fr.
Iwanaga, M. (1988). “Use of wild germplasm for sweetpotato breeding,” in: Exploration, Maintenance,
and Utilization of Sweetpotato Genetic Resources, ed. P. Gregory (Lima: International Potato Center
[CIP]), 199-210.
Jarret, R.L., Austin, D.F. (1994). Genetic diversity and systematic relationships in sweetpotato (Ipomoea
batatas (L.) Lam.) and related species as revealed by RAPD analysis. Genet. Resour. Crop Ev. 41,
165-173.
Jarret, R.L., Gawel, N., Whittemore, A. (1992). Relationships of the sweetpotato [Ipomoea batatas (L.)
Lam.]. J. Am. Soc. Hortic. Sci. 117, 633-637.
Kays, S.J. (1985). “The physiology of yield in sweet potato,” in: Sweet Potato Products: A Natural
Resource of the Tropics, ed. J. Bouwkamp (Boca Raton, Florida: CRC Press), 19-133.
Kays, S.J., Kays, S.E. (1998). “Sweet potato chemistry in relation to health,” in: Sweet Potato
Production System Towards the 21st Century (Miyakonojo: Kyushu National Agricultural
Experimental Station), 231-272.
Kernohan, B.J., Girzen, R.A., Millspaugh, J.J. (2001). “Analysis of animal space use and movements,”
Radio tracking animal populations. eds. J.J. Millspaugh, and J.M. Marzluff (San Diego: Academic
Press), 125-166.
Khoury, C.K., Castañeda Álvarez, N.P., Achicanoy, H., Sosa, C.C., Bernau, V., Kassa, M.T., et al.,
(2015). Crop wild relatives of pigeonpea [Cajanus cajan (L.) Millsp.]: distributions, ex situ
conservation status, and potential genetic resources for abiotic stress tolerance. Biol. Conserv. 184,
259-270.
Khoury, C., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic resources:
a review of global crop and regional conservation strategies. Genet. Resour. Crop Ev. 57, 625-639.
Kivuva, B.M., Githiri, S.M., Yencho, C., Sibiya, J. (2015a). Combining ability and heterosis for yield
and drought tolerance traits under managed drought stress in sweetpotato. Euphytica 201, 423-40.
Kivuva, B.M., Githiri, S.M., Yencho, C., Sibiya, J. (2015b). Screening sweetpotato genotypes for
tolerance to drought stress. Field Crop Res. 171, 11-22.
Komaki, K. (2004). “Breeding value of wild species closely related to sweetpotato,” in: Proceedings of
international workshop on production, utilization and development of sweetpotato, Korea, 164-172.
Kramer-Schadt, S., Niedballa, J., Pilgrim, J.D., Schroder, B., Lindenborn, J., Reinfelder, V., et al.
(2013). The importance of correcting for sampling bias in MaxEnt species distribution models.
Divers. Distrib. 19, 1366-1379.
Liu, C., Berry, P.M., Dawson, T.P., Pearson, R. G. (2005). Selecting thresholds of occurrence in the
prediction of species distributions. Ecography 28 (3), 385-393.
Loiselle, B., Jørgensen, P.M., Consiglio, T., Jiménez, I., Blake, J.G., Lohmann, L.G., et al. (2008).
Predicting species distributions from herbarium collections: does climate bias in collection sampling
influence model outcomes? J. Biogeogr. 35, 105-116.
Low, J.W., Arimond, M., Osman, N., Cunguara, B., Zano, F., Tschirley, D. (2007). A food-based
approach introducing orange-fleshed sweet potatoes increased vitamin A intake and serum retinol
concentrations in young children in rural Mozambique. J. Nutr. 137, 1320-1327.
Low, J., Lynam, J., Lemaga, B., Crissman, C., Barker, I., Thiele, G., et al. (2009). “Sweetpotato in Sub183

Chapter 5
Saharan Africa,” in: The Sweet Potato, eds. G. Loebenstein, G. Thottappilly (Springer Netherlands),
359-390.
Lu, S., Li, T. (1992). Study on the characterization of intra and inter-specific incompatibility of Batatas
section. Acta Agron. Sin. 18, 161-168.
Mabberley, D.J. (2008). Mabberley’s Plant Book: a portable dictionary of plants, their classifications,
and uses, 3rd Edition (Cambridge: Cambridge University Press).
Martin, F.W. (1970). Self- and interspecific incompatibility in the Convolvulaceae. Bot. Gaz. 131, 139144.
Martin, F.W. (1982). “Analysis of the incompatibility and sterility of sweetpotato,” in: Sweetpotato,
Proceedings of 1st international symposium, No. 82-172, eds. R.L. Villareal, T.D. Griggs (Tainan:
AVRDC), 275-283.
Martin, F.W., Jones, A. (1973). The species of Ipomoea closely related to the sweetpotato. Econ. Bot.
26, 201-215.
McCouch, S.R., Sweeney, M., Li, J., Jiang, H., Thomson, M., Septinginsih, E., et al. (2007). Through
the genetic bottleneck: O. rufipogon as a source of trait-enhancing alleles for O. sativa. Euphytica
154, 317-339.
McDonald, J.A., Austin, D.F. (1990). Changes and additions in Ipomoea section Batatas
(Convolvulaceae). Brittonia 42, 116-120.
McDonald, J.A., Hansen, D.R., McDill, J.R., Simpson, B.B. (2011). A phylogenetic assessment of
breeding systems and floral morphology of North American Ipomoea (Convolvulaceae). J. Bot. Res.
Inst. Texas 5, 159-177.
Mesibov, R. (2013). A specialist’s audit of aggregated occurrence records. ZooKeys 293, 1-18.
Miller, R.E., Rausher, M.D., Manos, P.S. (1999). Phylogenetic systematics of Ipomoea
(Convolvulaceae) based on ITS and waxy sequences. Syst. Bot. 24, 209-227.
Monfreda, C., Ramankutty, N., Foley, J.A. (2008). Farming the planet: 2. Geographic distribution of
crop areas, yields, physiological types, and net primary production in the year 2000. Global
Biogeochemical Cycles 22: GB1022. Data available at http://www.geog.mcgill.ca/landuse/
pub/Data/175crops2000/.
Munyiza, H., Stevenson, P., Mwanga, R., Talwana, H., Murumu, J., Odongo, B. (2007). The relationship
between stem base and root damage by Cylas spp. on sweet potato. Afr. Crop Sci. J. 8, 955-957.
Mwanga, R., Ssemakula, G. (2011). Orange-fleshed sweet potatoes for food, health and wealth in
Uganda. Int. J. Agric. Sustain. 9, 42-49.
Mwololo, J.K., Mburu, M.W.K., Njeru, R.W., Ateka, E.M., Kiarie, N., Munyua, J.K., et al. (2007).
Resistance of sweet potato genotypes to sweet potato virus disease in Coastal Kenya. Proc. ACSS
Conf. 8, 2083-2086.
Ngailo, S., Shimelis, H., Sibiya, J., Mtunda, K. (2013). Sweet potato breeding for resistance to sweet
potato virus disease and improved yield: Progress and challenges. Af. J. Ag. Resear. 8, 3202-3215.
Nimmakayala, P., Vajja, G., Reddy, U.K. (2011). “Ipomoea,” in: Wild Crop Relatives: Genomic and
Breeding Resources, ed. C. Kole (Berlin: Springer), 123-132.
Nishiyama, I. (1971). Evaluation and domestication of sweetpotato. Botanical Magazine Tokyo 84, 377387.
Nishiyama, I. (1982). Autohexaploid evolution of the sweetpotato. In: Sweetpotato. Proceedings of 1st
international symposium, No. 82-172, eds. R.L. Villareal, T.D. Griggs (Tainan: AVRDC), 263-274.
Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V.N., Underwood, E. C.,
et al. (2001). Terrestrial ecoregions of the world: a new map of life on earth. BioScience 51, 933938.
Otegui, J., Ariño, A.H., Encinas, M.A., Pando, F. (2013). Assessing the primary data hosted by the
184

Crop wild relatives of sweetpotato
Spanish node of the Global Biodiversity Information Facility (GBIF). PLoS One 8: e55144.
Pebesma, E.J., Bivand, R.S. (2005). Classes and methods for spatial data in R. R News 5 (2). Available
at: http://cran.r-project.org/doc/Rnews/.
Pfeiffer, W.H., McClafferty, B. (2007). HarvestPlus: breeding crops for better nutrition. Crop Sci. 47,
88-105.
Phillips, S.J., Dudik, M. (2008). Modeling of species distributions with Maxent: new extensions and
comprehensive evaluation. Ecography 31, 161-175.
Phillips, S.J., Anderson, R.P., Schapire, R.E. (2006). Maximum entropy modeling of species geographic
distributions. Ecol. Model. 190, 231-59.
Piperno, D.R., Pearsall, D.M. (1998). The Origins of Agriculture in the Lowland Neotropics. San Diego:
Academic Press.
R Core Team. (2013). R: A Language and Environment for Statistical Computing. http://www.rproject.org/.
Rajapakse, S., Nilmalgoda, S.D., Molnar, M., Ballard, R.E., Austin, D.F., Bohac, J.R. (2004).
Phylogenetic relationships of the sweetpotato in Ipomoea series Batatas (Convolvulaceae) based on
nuclear β-amylase gene sequences. Mol. Phylogenet. Evol. 30, 623-632.
Ramírez-Villegas, J., Khoury, C., Jarvis, A., Debouck, D.G., Guarino, L. (2010). A gap analysis
methodology for collecting crop genepools: a case study with Phaseolus beans. PloS One 5, e13497.
Revelle, W. (2015). Psych: Procedures for Psychological, Psychometric, and Personality Research, v.
1.5.1. http://cran.r-project.org/web/packages/psych/index.html
Roullier, C., Duputié, A., Wennekes, P., Benoit, L., Fernández Bringas, V. M., Rossel, G., et
al. (2013). Disentangling the origins of cultivated sweet potato (Ipomoea batatas (L.) Lam.). PLoS
One 8, e62707.
Sanchez. G. (2012). Package ‘plsdepot’. Partial Least Squares (PLS) Data Analysis Methods, v. 0.1.17.
http://cran.r-project.org/web/packages/plsdepot/index.html.
Schafleitner, R., Tincopa, L.R., Palomino, O., Rossel, G., Robles, R.F., Alagon, R., et al. (2010). A
sweet potato gene index established by de novo assembly of pyrosequencing and Sanger sequences
and mining for gene-based microsatellite markers. BMC Genomics 11, 604.
Shady Solis, R., Haas, J., Creamer, W. (2001). Dating caral, a preceramic site in the Supe valley on the
central coast of Peru. Science 292, 723-726.
Shiotani, I., Huang, Z.Z., Sakamoto, S., Miyazaki, T. (1994). The role of the wild Ipomoea trifida
germplasm in sweet potato breeding. Acta Hort. (ISHS) 380, 388-398.
Smit, N.E.J.M. (1997). Integrated Pest Management for Sweetpotato in Eastern Africa. PhD thesis,
(Wageningen: Wageningen University).
Srisuwan, S., Sihachakr, D., Siljak-Yakovlev, S. (2006). The origin and evolution of sweet potato
(Ipomoea batatas Lam.) and its wild relatives through the cytogenetic approaches. Plant Sci. 171,
424-433.
Sutherland, J.A. (1986). A review of the biology and control of the sweetpotato weevil, Cylas
formicarius elegaluntus (F). Trop. Pest Manag. 32, 304-315.
Teramura, T. (1979). Phylogenetic study of Ipomoea species in the section Batatas. Mem. Coll. Agric.,
Kyoto Univ. 114, 29-48.
The Plant List. (2010). Version 1. http://www.theplantlist.org/.
Tumwegamire, S., Kapinga, R., Rubaihayo, P.R., LaBonte, D.R., Grüneberg, W.J., Burgos, G., et al.
(2011). Evaluation of dry matter: protein, starch, β-carotene, iron, zinc, calcium, and magnesium in
East African sweetpotato [Ipomoea batatas (L.) Lam] germplasm. Hort. Science 46, 348-357.
Valverde, R.A., Clark, C.A., Valkonen, J.P.T. (2007). Viruses and virus disease complexes of
sweetpotato. Plant Viruses 1, 116-126.
185

Chapter 5
van der Wal, J., Falconi, L., Januchowski, S., Shoo, L., Storlie, C. (2014). SDMTools: Species
Distribution Modelling Tools: Tools for processing data associated with species distribution
modelling exercises. R package version 1.1-20. http://CRAN.R-project.org/package=SDMTools.
Warren, D.L., Glor, R.E., Turelli, M. (2008). Environmental niche equivalency versus conservatism:
quantitative approaches to niche evolution. Evolution 62, 2868-2883.
Wilkin, P. (1999). A morphological cladistic analysis of the Ipomoeeae (Convolvulaceae). Kew Bull.
54, 853-876.
Williams, J.N., Seo, C., Thorne, J., Nelson, J.K., Erwin, S., O’Brian, J.M., et al. (2009). Using species
distribution models to predict new occurrences for rare plants. Divers. Distrib. 15, 565-576.
Wood, J.R.I., Carine, M.A., Harris, D., Wilkin, P., Williams, B., Scotland, R.W. (2015). Ipomoea
(Convolvulaceae) in Bolivia. Kew Bull. In press.
Woolfe, J.A. (1992). Sweetpotato, an untapped food source (New York: Cambridge University Press).
Xiao, J., Grandillo, S., Ahn, S.N., McCouch, S.R., Tanksley, S.D. (1996). Genes from wild rice improve
yield. Nature 384, 223-224.
Zhang, D.P., Cervantes, J., Huamán, Z., Carey, E., Ghislain, M. (2000). Assessing the genetic diversity
of sweetpotato (Ipomoea batatas (L.) La.) cultivars from tropical America using AFLP. Genet.
Resour. Crop Ev. 47, 659-665.
Zhang, S.S., Liu, L.F. (2005). “Utilization of the wild relatives in sweetpotato breeding,” in: Sweetpotato
Breeding and Industrialization in China, ed. Q.C. Liu. (Beijing: China Agricultural University
Press), 29-32.

186

CHAPTER 6
Crop wild relatives of pigeonpea [Cajanus cajan (L.) Millsp.]:
distributions, ex situ conservation status, and
potential genetic resources for abiotic stress tolerance
Colin K. Khoury1,2,, Nora P. Castañeda-Álvarez1,3, Harold A. Achicanoy1, Chrystian C. Sosa1,
Vivian Bernau1,4, Mulualem T. Kassa5, Sally L. Norton6, L. Jos G. van der Maesen7,
Hari D. Upadhyaya8,9,10, Julian Ramírez-Villegas1,11,12, Andy Jarvis1,11 and Paul C. Struik2
International Center for Tropical Agriculture (CIAT), Km 17, Recta Cali-Palmira, Apartado Aéreo
6713, Cali, Colombia.
2
Centre for Crop Systems Analysis, Wageningen University, Droevendaalsesteeg 1, 6708 PB
Wageningen, The Netherlands.
3
School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK.
4
Department of Horticulture and Crop Science, The Ohio State University, 202 Kottman Hall, 2021
Coffey Rd, Columbus, Ohio 43210, USA.
5
Cereal Research Centre, Agriculture and Agri-Food Canada, 195 Dafoe Road, Winnipeg, Manitoba,
T3T 2M9, Canada.
6
Australian Grains Genebank, Department of Environment and Primary Industries, Private Bag 260,
Horsham, Victoria 3401, Australia.
7
Naturalis Biodiversity Center, Darwinweg 2, 2333 CR, Leiden, the Netherlands.
8
International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Patancheru 502 324,
Telangana, India.
9
Department of Agronomy, Kansas State University, Manhattan, Kansas 66506, USA.
10
Institute of Agriculture, University of Western Australia, Crawley WA 6009, Australia.
11
CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Km 17,
Recta Cali-Palmira, Apartado Aéreo 6713, Cali, Colombia.
12
School of Earth and Environment, University of Leeds, Leeds, UK.
1

Published in Biological Conservation 184, 259-270. doi: 10.1016/j.biocon.2015.01.032

Chapter 6
Abstract
Pigeonpea [Cajanus cajan (L.) Millsp.] is a versatile, stress-tolerant, and nutritious grain legume,
possessing traits of value for enhancing the sustainability of dry sub-tropical and tropical agricultural
systems. The use of crop wild relatives (CWR) in pigeonpea breeding has been successful in providing
important resistance, quality, and breeding efficiency traits to the crop. Current breeding objectives for
pigeonpea include increasing its tolerance to abiotic stresses, including heat, cold, drought, and
waterlogging. Here we assess the potential for pigeonpea CWR to be further employed in crop
improvement by compiling wild species occurrence and ex situ conservation information, producing
geographic distribution models for the species, identifying gaps in the comprehensiveness of current
germplasm collections, and using ecogeographic information to identify CWR populations with the
potential to contribute agronomic traits of priority to breeders. The fifteen prioritized relatives of
pigeonpea generally occur in South and Southeast Asia to Australia, with the highest concentrations of
species in southern India and northern Australia. These taxa differ considerably among themselves and
in comparison to the crop in their adaptations to temperature, precipitation and edaphic conditions. We
find that these wild genetic resources are broadly under-represented in ex situ conservation systems,
with 80% of species assessed as high priority for further collecting, thus their availability to plant
breeders is insufficient. We identify species and highlight geographic locations for further collecting in
order to improve the completeness of pigeonpea CWR germplasm collections, with particular emphasis
on potential traits for abiotic stress tolerance.
Keywords: Crop diversity, Crop improvement, Crop wild relatives, Food security, Germplasm
conservation, Plant genetic resources

188

Crop wild relatives of pigeonpea
INTRODUCTION
Challenges to global food production are compounding. Our growing population and dietary
expectations are projected to increase demand on food systems for at least the next four decades,
outpacing current yield trends (Ray et al. 2013). Limitations in land, water, and natural resource
inputs, competition for arable soils with non-food crops and other land uses, the need to
minimize harmful impacts on biodiversity and other ecosystem services, and greater climatic
variability further constrain production potential (Lobell et al. 2008, Cordell et al. 2009,
Rosenzweig, et al. 2013). Although gains in food availability may partially be obtained through
dietary change and waste reduction (Tilman & Clark 2014, West et al. 2014), a transition toward
more sustainable, yet highly productive, agricultural systems is necessary. This transformation
must be achieved through improved agronomic practices combined with the use of varieties of
crops with reliable yields under more adverse conditions (Foley et al. 2011).
One such crop is pigeonpea [Cajanus cajan (L.) Millsp.], a sub-tropical and tropical grain
legume that originated in the northern region of the Indian sub-continent, spreading to East
Africa at least 4000 years BCE, and then to Southeast Asia, West Africa, Latin America, and
the Caribbean. The seed is eaten as a green vegetable and dry pulse and is an important source
of protein, vitamin B, carotene, and ascorbic acid (Odeny 2007, Choudhary et al. 2013). The
pods and foliage of the plant are used as livestock forage and fodder, the crop is cultivated as a
green manure, and its woody stem is used as fuel and construction material (Mallikarjuna et al.
2011). Pigeonpea is an important income generator, particularly in Tanzania, Malawi, and
Myanmar as an export crop to India (Odeny 2007).
Pigeonpea is generally planted by smallholder farmers in low input, rain-fed conditions. The
crop is well suited to a wide range of agricultural systems, including intercropping and no-till.
Cultivation improves soil fertility through biological nitrogen fixation as well as through the
solubilization of soil-bound phosphorus (Mallikarjuna et al. 2011, Choudhary et al. 2013),
increasing the yield of intercropped cereals, other pulses, and vegetables (Saxena 2005, Odeny
2007), and has been shown to enhance the control of Striga (Odeny 2007). Pigeonpea is more
heat tolerant than the majority of grain legume crops (Figure 1) and is regarded as droughtresistant. These traits are associated with the ability to maintain or regulate transpiration under
high temperatures and/or low soil moisture, for example through adjustment of leaf osmotic
pressure (Subbarao et al. 2000), maintenance of photosynthetic function under stress
189

Chapter 6

Figure 1. Climatic niches for temperature and precipitation for major grain legumes. (A) optimal range
and (B) minimum and maximum observed range (Bogdan 1977, van der Maesen 1989, Valenzuela and
Smith 2002, Odeny 2007, FAO 2010a, Sardana et al. 2010, Saxena et al. 2010).

(Lopez et al. 1987), and deep root systems (Flower & Ludlow 1987).
Due to its high nutritive value and agronomic traits, pigeonpea can play an increasing role in
low input production systems in India, East Africa, and elsewhere in the dry sub-tropics and
tropics (Saxena 2005, Odeny 2007). Concerted breeding efforts for this purpose have resulted
in a number of promising advances, particularly the creation of early maturing varieties, and
developments toward diverse high yielding hybrids (Saxena 2005, Saxena et al. 2013, Saxena
& Sawarganokar 2014). However, crop yield in most production regions is well below its
potential and has been stagnant for a number of decades, with increased production during this
time largely due to an expansion of harvested area (Jones et al. 2002, Saxena 2005, Odeny
2007). In order to increase pigeonpea yield and adaptability, current breeding priorities include
photoperiod insensitivity, resistance to biotic pressures, and tolerance to abiotic stresses,
notably waterlogged and mineral deficient soils, cold and heat stress, salinity, and drought
(Mligo & Craufurd 2005, Saxena 2005, Odeny 2007, Upadhyaya et al. 2007, Choudhary et al.
2011).
The long-term viability of major food crops, particularly in light of the increasing need for
sustainable production techniques, is dependent upon the use of diverse genetic resources to
maintain productivity and adapt to changing climatic conditions and emerging pest and disease
190

Crop wild relatives of pigeonpea
Table 1. Published confirmed (C) and potential (P) uses of pigeonpea CWR in crop improvement.
Taxon

Trait

C. acutifolius

Cytoplasmic male sterility (Ca), Pod borer resistance (Cb), High seed weight
(Cc), Sterility mosaic disease resistance (Pc), Pod fly resistance (Pd)

C. albicans

High seed protein (Cc,Pe), Pod borer resistance (Pb), Pod fly resistance (Pd), Pod
wasp resistance (Pd), Broad pods (Pe), More seeds per pod (Pe), Good forage
source (Pe), Sterility mosaic disease resistance (Pf), Salt tolerance (Pg)

C. cajanifolius

Nuclear male sterility (Ca), Cytoplasmic male sterility (Ch), High seed protein
(Pe)
More seeds per pod (Pe)

C. cinereus
C. crassus

High seed protein (Pe), Good forage source (Pe), Sterility mosaic disease
resistance (Pf)

C. lineatus

Cleistogamy (Ci), Cytoplasmic male sterility (Cj), Pod fly resistance (Pd),
Sterility mosaic disease resistance (Pf)

C. mollis

More seeds per pod (Pe), High seed protein (Pe), Good forage source (Pe)

C. platycarpus

Phytophthora blight resistance (Cf,k), Sterility mosaic disease resistance (Pc),
Pod borer resistance (Pc), Early flowering (Pe), High seed protein (Pe), Cyst
nematode resistance (pf), Salt tolerance (Pg,l), Aluminum toxicity resistance (Pg),
Annuality (Pm), Photoperiod insensitivity (pn), High flower and pod setting (Pn)

C. scarabaeoides

Pod borer resistance (Cb,f), Sterility Mosaic Disease Resistance (Cc), Protein
improvement (Co), Dwarfism (Cp), Cytoplasmic male sterility (Cq), Pod fly
resistance (Pd), Pod wasp resistance (Pd), Early flowering (Pe,r), Salt tolerance
(Pg), Aluminum toxicity resistance (Pg), Drought tolerance (Pr)

C. sericeus

High seed protein (Cc), Cytoplasmic male sterility (Cs), Pod borer resistance
(Pb), Pod fly resistance (Pd), Salt tolerance (Pg), Sterility mosaic virus resistance
(Pm,f), Phytophthora blight resistance (Pm)

Mallikarjuna and Saxena 2005, bMallikarjuna et al. 2007, cMallikarjuna et al. 2011, dSharma et al.
2003, eUpadhyaya et al. 2013b, fSaxena 2005, gChoudhary et al. 2011, hSaxena et al. 2005, iSaxena et al.
1998, jSaxena et al. 2010, kMallikarjuna et al. 2005, lSubbarao et al. 1990, mobservation by authors, nMudaraddi
et al. 2013, oReddy et al. 1979, pReddy 1990, qSaxena & Kumar 2003, rUpadhyaya 2006, sAriyanayagam et al.
1995.
a

pressures (Xiao et al. 1996, Guarino & Lobell 2011, McCouch et al. 2013). Due to the genetic
bottleneck effect caused by domestication and crop improvement, pigeonpea cultivars possess
only a small portion of the overall genetic diversity present within the genepool (Kassa et al.
2012), which also includes traditional farmer varieties and wild related species (Vincent et al.
2013). Crop wild relatives (CWR) of pigeonpea have contributed valuable genetic resources for
pest and disease resistance, improved nutritional quality, desirable plant architecture, and
breeding efficiency. They are considered to possess superior levels of resistance to diseases
such as Fusarium wilt and Phytophthora blight, insect pests such as pod borer, pod fly, and pod
wasp, and tolerance to abiotic stress, in comparison to the cultivated species (Table 1).

191

Chapter 6
Increasing awareness of the extent of habitat destruction, invasive species, and other threats to
the habitats of the CWR of major crops has given urgency to efforts to identify important
species, determine their distributions, and to ensure their conservation for the long term and
thus their availability to plant breeders (Jarvis et al. 2008, FAO 2010b, Khoury et al. 2010).
Genetic resource conservation planning efforts have benefitted from advancements in
geographic information systems technologies, which have enabled high resolution species
distribution modeling in order to inform collecting priorities (Jarvis et al. 2005), recognition of
important gaps in ex situ collections (Ramírez-Villegas et al. 2010), and the identification of
populations that may possess particularly valuable traits for crop improvement (Tapia et al.
2014).
Given the importance of pigeonpea in low input production systems in regions facing food and
nutritional insecurity and the capacity for enhancement of the crop through breeding, the aim
of this research is to contribute to ensuring the conservation and availability of a broad range
of diversity of CWR genetic resources of potential value to present and future crop
improvement objectives. Therefore, we analyzed the comprehensiveness of ex situ conservation
of pigeonpea CWR through a series of questions: (a) what constitutes a potentially useful wild
relative of pigeonpea?, (b) where are these species encountered in the wild?, (c) what is the
state of conservation and availability of these species to plant breeders? If suboptimal, what are
the highest taxonomic and ecogeographic priorities for further collecting? And finally, (e) what
CWR resources possess high potential for contribution of traits of value for crop breeding
objectives?
MATERIALS AND METHODS
Identification of target CWR species and occurrence data compilation
We identified potentially useful CWR at the species level based upon a genepool concept
(Harlan and de Wet 1971) for pigeonpea, which focused on those wild species capable of
hybridization with the crop (i.e., members of the primary or secondary genepools), as these
species possess the greatest potential for successful introgression of traits (Vincent et al. 2013).
Taxa in the tertiary genepool with published evidence of confirmed or potential use in crop
improvement (Table 1) were also included.
Occurrence records for pigeonpea CWR were acquired from online biodiversity, herbarium,
192

Crop wild relatives of pigeonpea
and germplasm databases; via communications with herbarium and genebank managers, and
other crop researchers; and through direct recording of provenance data during visits to selected
herbaria (Supplementary Table 1). Germplasm data were obtained from repositories that
provide straightforward access to genetic resources and associated data to the global research
community through online information systems. Occurrence data were compiled in a
standardized format and taxonomically verified following GRIN Taxonomy for Plants (2012)
and The Plant List (2010) as references. Existing coordinates were cross-checked to country
and verified as occurring on land (Hijmans et al. 1999), and records with locality information
but no coordinates were geo-referenced using the Google Maps Geocoder (2013) application
programming interface. Occurrence data were mapped, iteratively evaluated for correctness
with pigeonpea CWR experts, and subsequently further processed in order to form a final
dataset of maximized taxonomic and spatial accuracy.
Species potential distribution modeling
A potential distribution model for each species was calculated using the maximum entropy
(Maxent) algorithm (Phillips et al. 2006), with a set of ecogeographic variables and unique
species presence records as inputs. We chose Maxent due to its wide application in predicting
species distributions (Elith et al. 2006, Phillips & Dudik, 2008, Costa et al. 2010). We
performed modeling at a resolution of 2.5 arc- minutes (~ 5km × 5km cell size at the equator),
employing 10,000 background points for model training over the combined distributional range
of the pigeonpea CWR. Ecogeographic inputs included altitude and nineteen bioclimatic
variables from the WorldClim database (Hijmans et al. 2005), and seven major edaphic drivers
of plant species distributions with consistent data coverage throughout the range of the
pigeonpea CWR species, obtained from ISRIC- World Soil Information (Hengl et al. 2014)
(Supplementary Table 2). For the edaphic variables we calculated a weighted mean across 0-5,
5-15, 15-30, 30-60, and 60-100 cm soil depth values in order to derive a single data value for 0
-100 cm. We then resampled the 1 km resolution data to form 2.5 arc-minutes resolution inputs
aligned with the WorldClim datasets.
In order to refine and test the stability of the distribution models for each species, we analyzed
Maxent results across three ecogeographic input variations: (a) the full set of nineteen
bioclimatic variables (Ramírez-Villegas et al. 2010); (b) the bioclimatic variables, altitude, and
the additional set of seven edaphic variables, totaling 27 input variables; and (c) a species193

Chapter 6
specific derivation of the most important drivers of distribution based upon presence data,
further refined by removing highly correlated variables. For the ecogeographic variables in the
species-specific method, we utilized a non-linear iterative partial least squares (NIPALS)
algorithm to perform a principal-component analysis (PCA), as NIPALS can handle data arrays
in which the number of observations is less than the number of input variables, and identified
those variables with the greatest contribution (>0.7 or <-0.7) to the first two principal
components per species based upon occurrence data points. We then used a variance inflation
factor (VIF) to identify the variables with a low degree of collinearity (see Supplementary Table
3 for a list of variables utilized per species). A calibrated area under the ROC curve (cAUC)
was obtained to assess the predictive performance of each model (Hijmans 2012). The three
modeling methods were evaluated with a correlation coefficient against a null model, and the
species-specific variables method showed the least spatial sorting bias among methods
(spearman’s rho for the 19 variables was 0.53; for 27 variables was 0.56; and for the speciesspecific method was 0.37), and the differences in median AUC distributions across species for
each method were found to be statistically significant (p = 0.0002) through a Kruskal-Wallis
non-parametric analysis of variance test. Potential distribution models based upon the speciesspecific variables method were therefore utilized in subsequent analyses.
Potential distribution models were constrained per species by a native range defined at the
country level as listed in GRIN (2012) and van der Maesen (1986), and were clipped by
measuring the shortest distance between the receiver operating characteristic curve (ROCcurve) and the top-left corner of the plot (Liu et al. 2005). We limited the spatial analysis to the
native distributions of taxa in order to focus prioritization recommendations on those regions
with species with long-term adaptation to specific ecogeographic conditions.
Adapted from Ramírez-Villegas et al. (2010), Maxent models were produced using the crossvalidation option (k=5) and were subjected to a four-fold assessment process including: (a) the
5-fold average area under the ROC curve of test data (ATAUC), (b) the standard deviation of
the test AUC of the 5 different folds (STAUC), (c) the proportion of the potential distribution
coverage with standard deviation above 0.15 (ASD15), and (d) the cAUC value. Models with
ATAUC above 0.7, STAUC below 0.15, ASD15 below 10%, and cAUC exceeding 0.40 were
considered accurate and stable. For species where the Maxent model did not pass the crossvalidation, potential distributions were mapped with a circular buffer of 50 km (CA50)
194

Crop wild relatives of pigeonpea
surrounding each geo-referenced record (Hijmans et al. 2001).
Analysis of current ex situ conservation and further collecting needs for CWR
We adapted a gap analysis methodology proposed by Ramírez-Villegas et al. (2010), combining
three metrics used to assess the urgency of further collecting in order to fill gaps in ex situ
conservation of CWR. The total sample representation of each species in genebank collections
was estimated via a sampling representativeness score (SRS), calculated as the number of
germplasm samples (G) divided by the total number of samples (G + herbarium samples (H))
(i.e., all other records aside from available genebank accessions).
The sufficiency of geographic coverage of germplasm collections of each species was estimated
through a geographic representativeness score (GRS), calculated as the share of the combined
total area of CA50 placed around each existing germplasm collection point compared to the
overall potential geographic distribution of the species.
The comprehensiveness of ecological coverage of germplasm collections of each species was
estimated through an ecological representativeness score (ERS), calculated by estimating the
distinct ecosystem classifications (Olson et al. 2001) represented in the CA50 of existing
germplasm collection points compared to the diversity of ecosystems in which the overall
potential geographic distribution model of the species occurs.
A final priority score (FPS) for further collecting for ex situ conservation was assigned to each
species by averaging the three gap analysis metrics (SRS, GRS, and ERS). FPS scores were
further classified into four categories of urgency for collecting: high priority species (HPS) for
taxa whose 0 < FPS ≤ 2.5 or when no germplasm accessions currently exist; medium priority
species (MPS) when 2.5 < FPS ≤ 5; low priority species (LPS) when 5 < FPS ≤ 7.5; and ‘no
further collecting recommended’ (NFCR) when 7.5 < FPS ≤ 10. We produced collecting
priorities maps for all species, displaying the geographic areas that have not yet been collected
from within the potential distributions of taxa.
The ecogeographic data preparation, species distribution modeling, and gap analysis were
written and performed in R v2.15.1 (R Core Team 2013), utilizing packages maptools (Bivand
& Lewin-Koh 2014), rgdal (Bivand et al. 2014), SDMTools (van der Wal et al. 2014), raster
195

Chapter 6
(Hijmans 2014), sp (Pebesma & Bivand 2005, Bivand et al. 2013), dismo (Hijmans et al. 2013),
and plsdepot (Sanchez 2012). Resulting spatial files were mapped in ArcMap v.10 (ESRI 2011).
Collecting priorities spatial files were analyzed using the Zonal Statistics tool in ArcMap to list
the countries prioritized for further collecting for ex situ conservation.
In order to validate and/or expose deficiencies in our findings, we subjected the gap analysis
numerical and spatial results to an evaluation performed by four crop experts with experience
in the distribution and/or conservation status of CWR of pigeonpea: Mulualem Kassa, Cereal
Research Centre, Agriculture and Agri-Food Canada; Sally Norton, Australian Grains
Genebank, Australia; Hari Upadhyaya, International Crops Research Institute for the Semi-Arid
Tropics (ICRISAT); and Jos van der Maesen, Naturalis Biodiversity Center, the Netherlands).
These experts were first asked to provide an evaluation of the sufficiency of germplasm
collections per species based only upon their knowledge of total accessions, and geographic
and environmental gaps. Such an assessment [comparable expert priority score (EPS)] was
considered directly comparable to the FPS of the gap analysis results. A second evaluation score
(contextual EPS) based on the entirety of expert knowledge, including threats to species in situ
and prioritization by usefulness in crop improvement, was collected in order to provide
additional information to collecting prioritization efforts. In both cases, an EPS between 0 and
10, aligned with the gap analysis results prioritization scale, was requested. After these steps,
experts were shown the gap analysis results and asked to comment on assessed quantitative
results, occurrence data, potential distribution models, and maps of collecting priorities.
Following these contributions by experts, input occurrence data were further refined by
eliminating clearly incorrect points and adjusting country-level native areas, and the potential
distribution modeling and gap analyses were re-run in order to improve the quantitative and
spatial outputs. Expert metrics displayed in the results pertain to the final evaluation of
improved gap analysis outputs.
A multiple factor analysis (MFA) was used in order to compare the various forms of expert
evaluation inputs with the gap analysis results, and an expert evaluation index was created,
which estimated the degree of accord between all experts and the gap analysis results for each
species, with a scale from 0 (disagreement) to 100 (agreement). Analyses were performed using
R package FactoMineR (Husson et al. 2009).

196

Crop wild relatives of pigeonpea

Identification of CWR with potential traits of use in breeding for abiotic stress tolerance
We utilized ecogeographic information in combination with species presence data in order to
identify populations of species with outstanding adaptations to climatic and/or edaphic
conditions of interest to pigeonpea breeding objectives. We assessed the relative importance of
the 27 ecogeographic variables (Supplementary Table 2) in explaining the total variation among
pigeonpea CWR through a PCA, utilizing all occurrence data points found within the native
areas of the species. We created a hierarchical cluster of principal components (HCPC) in order
to identify ecogeographic clusters for the species using R package FactoMineR.
Boxplots for each of the 27 ecogeographic variables were created based upon CWR species
occurrence data points, displaying the median and variance parameters per species per variable.
Comparable ecogeographic variable data for the pigeonpea crop was extracted from area of
cultivation maps (Monfreda et al. 2008) at a resolution of 5 arc-minutes, with a random sample
of 1000 points weighted by harvested area, taken from the major cultivation areas in Asia,
Africa, and Latin America. As both the CWR and the crop displayed outlier occurrence points
that could potentially contribute to an overinflated ecogeographic niche concept, for further
comparative analyses focused on breeding objectives for the crop we restricted the
ecogeographic niche per species to the central 90% of variation (i.e., 10% outliers were
excluded). Ecogeographic niches for CWR and the pigeonpea crop were mapped in R package
ggplot2 (Wickham 2009).
RESULTS
Wild relatives of pigeonpea
The genus Cajanus Adans. is composed of 32-34 taxa divided into three clades: Indian,
Australian, and Scarabaeoides (van der Maesen 1986, Kassa et al. 2012). No wild conspecific
to the cultivated species exists, and thus there are no wild taxa falling within the primary
genepool of pigeonpea as defined by Harlan and de Wet (1971). The secondary genepool is
comprised of Cajanus cajanifolius (Haines) Maesen, in the Indian clade, the putative progenitor
of the crop (Kassa et al. 2012), C. acutifolius (F.Muell.) Maesen, C. albicans (Wight & Arn.)
Maesen, C. cinereus (F.Muell.) F.Muell., C. confertiflorus F.Muell., C. lanceolatus (W.Fitzg.)
Maesen, C. latisepalus (Reynolds & Pedley) Maesen, C. lineatus (Wight & Arn.) Maesen, C.
reticulatus (Dryand) F.Muell., C. scarabaeoides (L.) Thouars, C. sericeus (Baker) Maesen, and
197

Chapter 6
C. trinervius (DC.) Maesen (Saxena et al. 2005, Mallikarjuna et al. 2011). Three additional
species from the tertiary gene pool [C. crassus (King) Maesen, C. mollis (Benth.) Maesen, and
C. platycarpus (Benth.) Maesen] have been the subject of publications of confirmed or potential
uses in crop improvement and were therefore also included in the analysis (Table 1,
Supplementary Table 3). Cajanus volubilis (Blanco) Blanco was recorded in Wanjari et al.
(1999) as contributing sterility traits, but we believe that the material studied in this reference
was actually C. crassus, therefore C. volubilis was not included in this analysis.
A total of 3171 occurrence records for the fifteen CWR were gathered for use in potential
distribution modeling and in the gap analysis, including 377 germplasm accession records
sourced from six genebanks, and 2794 herbarium and other occurrence reference records
sourced from 17 providers (Supplementary Table 1). Records per species ranged from 15 (C.
mollis) to 594 (C. acutifolius). Of these, 1068 records containing distinct cross-checked
coordinates were used to model species potential distributions and to locate the original
collecting site of existing germplasm accessions (Supplementary Table 3).
CWR species distributions
Potential distribution models performed in Maxent passed the four-fold cross-validation for
eleven out of the fifteen CWR. Models for C. albicans, C. cajanifolius, C. mollis and C.
platycarpus failed the cross-validation due generally to insufficient and dispersed presence
records, and were instead mapped by creating CA50 buffers around their occurrences. Native
distributions of pigeonpea CWR occur from South Asia through Southeast Asia into northern
Australia, as well as on the eastern coast of Madagascar. Species diversity is richest in southern
India and in northern Australia, with up to six modeled potential species distributions
overlapping in a single area (Figure 2).
Analysis of current ex situ conservation and further collecting needs for CWR
Twelve out of fifteen species were assigned high priority for further collecting due to the
average of total samples, geographic, and ecological gaps in their ex situ germplasm collections
(Figure 3, Supplementary Table 3, Supplementary Figure 1). These high priority species
included taxa with narrow distributions (C. cajanifolius) as well as those with large ranges
(e.g., C. cinereus, C. crassus, and C. scarabaeoides). Cajanus albicans was assessed as medium
priority due to being modeled as relatively well represented ex situ in regard to ecosystem
198

Crop wild relatives of pigeonpea

Figure 2. Richness map for assessed pigeonpea CWR potential distribution models, including high
species richness areas in (A) southern India and (B) northern Australia.

coverage, and C. mollis and C. platycarpus as low priority for the same reason, plus high
sampling representativeness scores due to having a disproportionately large number of
germplasm samples in comparison to herbarium records. The failure of cross-validation of the
Maxent models for these species as well as for C. cajanifolius resulted in CA50 buffer potential
distributions that are likely to be underestimates of the full range of the taxa, especially given
the relatively dispersed distributions of available presence records. Thus, the gap analysis
assessments for these species likely overestimated the comprehensiveness of their coverage in
ex situ repositories. The mean final priority score across all CWR was 2.05 ± 1.94.
Further collecting priorities for the pigeonpea CWR were identified in 20 countries, all of which
contained gaps for high priority species (Supplementary Figure 2, Supplementary Table 4). As
with species richness, the regions identified for further collecting of the greatest number of
species occurred in southern India and in northern Australia (Figure 4). Occurrence data,
potential distribution models, and collecting priorities maps for all species are available in an
interactive map format at http://www.cwrdiversity.org/distribution-map/.
199

Chapter 6

Figure 3. Gap analysis results and comparable expert assessments per species. CWR are listed by
descending priority for further collecting by priority categories [high priority species, HPS (red);
medium priority species, MPS (orange); low priority species, LPS (yellow); and no further collecting
recommended, NFCR (green)]. The black circle represents the final priority score (FPS) for the species,
which is the mean of the sampling representativeness score (SRS), geographic representativeness score
(GRS), and ecological representativeness score (ERS).

200

Crop wild relatives of pigeonpea

Figure 4. Further collecting priorities hotspots map for high priority (HPS) pigeonpea CWR. The map
displays areas within the potential distributions of HPS species that have not been previously collected
for ex situ conservation, including areas of high species richness in (A) southern India and (B) northern
Australia.

The average of the directly comparable expert evaluation priority scores (EPS) across the four
experts correlated with the gap analysis results for pigeonpea CWR, with a mean EPS across
all experts and all CWR of 2.67 ± 0.9. The assignment of lower priority for further collecting
in the mean score across experts in comparison to the gap analysis, with seven species assigned
as HPS and eight as MPS, was highly influenced by one expert’s determination of species at a
lower priority level than the other three experts. This trend was consistent across species (Figure
3; Supplementary Figure 3A, Supplementary Table 3). The contextual expert priority score per
species did not vary widely from the comparable score, with a mean across all experts and
species of 2.3 ± 0.89. The mean contextual score gave slightly higher priority to species for
further collecting than did the comparable score, due to knowledge of threats to taxa in situ
and/or to the generally high value given to pigeonpea CWR in regard to their potential
contributions to crop improvement (Supplementary Figure 3B).
The multiple factor analysis revealed relatively strong agreement among the experts and the
201

Chapter 6
quantitative and spatial evaluation variables and thus confidence in the expert evaluation index
(Supplementary Figure 3C). Those taxa with the highest accord between all experts and
variables and the gap analysis results included Indian species C. trinervius and C. sericeus, and
most of the Australian species (C. cinereus, C. acutifolius, C. confertiflorus, C. lanceolatus,
and C. latisepalus). Cajanus scarabaeoides was given a very low index score, and assigned
least priority of all species for further collecting by the experts, due to the relatively large total
number of ex situ germplasm accessions held for the species, whereas the gap analysis assessed
the species as high priority due to large geographic and ecological gaps in ex situ collections in
comparison to the total potential distribution. The taxa with relatively few and dispersed
occurrence records and resulting CA50 potential distribution models (C. albicans, C.
cajanifolius, C. mollis and C. platycarpus) were also among those species receiving the lowest
index scores (Supplementary Figure 3D). The evaluations thus served to highlight those species
with greatest need of further investigation in regard to their distributions, and at the same time
confirmed the robustness of the Maxent models in creating valid depictions of the general range
of those CWR with sufficient data availability. Furthermore, the evaluations were useful in
identifying erroneous occurrence records for the species, and in highlighting factors
contributing to sampling bias in existing collections (e.g., proximity to roads or to research
sites), which may affect species distribution modeling.
Identification of CWR with potential traits of use in breeding for abiotic stress tolerance
Strong linear relationships were found between ecogeographic variables within the study area,
justifying the application of the PCA, with 70.3% of variance explained through the first three
principal components. The first component (37.9% of variation) was generally positively
correlated with high and variable temperatures, soil bulk density and pH, and negatively with
precipitation and soil organic carbon. The second component (20.8% of variation) was
determined by extreme temperature variables. The third component (11.6% of variation), was
positively related with precipitation and finer soils (Supplementary Figure 4A; Supplementary
Table 2).
Occurrence data were segregated into three ecogeographic clusters. The first cluster,
corresponding to lowland areas of Southeast Asia and southern India, was characterized
generally by highly variable temperatures and finer soils. The second cluster corresponded to
more temperate and/or highland regions in South Asia as well as the eastern coast of Australia,
202

Crop wild relatives of pigeonpea
Table 2. Agronomic traits prioritized
ecogeographic niches of CWR.
Taxon
Gene Gap
pool analysis
priority
C. acutifolius
2
HPS
C. albicans
2
MPS
C. cajanifolius
2
HPS
C. cinereus
2
HPS
C. confertiflorus 2
HPS
C. crassus
3
HPS

in pigeonpea breeding objectives potentially associated with
Mean
expert
priority
HPS
MPS
HPS
MPS
HPS
MPS

Eco
geographic
cluster
3
1,3,2
3,2
3
2,3
1,3

C. lanceolatus
C. latisepalus
C. lineatus

2
2
2

HPS
HPS
HPS

HPS
HPS
MPS

3
3
1,3

C. mollis

3

LPS

MPS

2

C. platycarpus

3

LPS

MPS

3,2

C. reticulatus
C. scarabaeoides

2
2

HPS
HPS

MPS
MPS

3,2
1,3,2

C. sericeus

2

HPS

HPS

3,1,2

C. trinervius

2

HPS

HPS

2,1

Potential traits of value for major
breeding objectives for abiotic stress
tolerance in pigeonpea
Heat, drought
Heat
Heat, drought
Cold
Temperature variation/ seasonality,
high precipitation, waterlogging
Heat, drought
Heat, drought
High precipitation, waterlogging,
drought
Cold
Heat, temperature
variation/seasonality, cold
Heat, temperature variation/
seasonality
High precipitation, waterlogging,
drought
Cold, high precipitation

Genepool 2 refers to the secondary genepool, and 3 to the tertiary. Priority categories for the CWR of pigeonpea
included high (HPS), medium (MPS), and low priority species (LPS) for further collecting for ex situ conservation.

and was characterized by dry conditions and colder temperatures. The third cluster,
corresponding more generally to India, the Mekong region, and northern Australia, was
characterized by low precipitation and low soil carbon (Supplementary Figure 4B). The
exploration of germplasm through the lens of ecogeographic clusters may facilitate the
identification of populations of species with valuable traits, in this case for temperature stress
resistance and waterlogging tolerance; cold tolerance; and tolerance to drought and low soil
fertility, respectively. The great majority of presence records of Australian species C.
acutifolius, C. cinereus, C. latisepalus, and C. lanceolatus, fell within one cluster, while the
South and Southeast Asian species generally contained populations falling into two or three
clusters (Table 2, Supplementary Figure 4C).
The assessment of climatic and edaphic niches of CWR species based upon occurrence data
points revealed large differences in adaptation to temperature, precipitation, and soil
characteristics variables (Supplementary Figure 5), including populations of species tolerant of
203

Chapter 6
low and high temperatures, as well as very low and very high rainfall. Such variables for many
species fell well outside the ecogeographic niche of the cultivated taxon, particularly for low
temperatures and high precipitation (Figure 5A).
For adaptation to high temperatures, populations of northern Australian species such as C.
latisepalus, C. cinereus, C. acutifolius, and C. lanceolatus stood out, as did C. platycarpus and
C. cajanifolius (Table 2, Supplementary Figure 5B,F,K). Taxa with large spreads in their
adaptation to temperature generally were those more relatively widespread species such as C.
scarabaeoides, C. crassus, and potentially C. platycarpus. Species with populations of notable
adaptation to low temperatures included C. mollis, C. trinervius, C. confertiflorus, and again C.
platycarpus (Supplementary Figure 5B,G,L). Scant occurrence information was available for a
number of these species and further exploration is needed to determine the full range of the
taxa, which may result in the identification of populations with even greater tolerance to
extreme temperatures, e.g., from populations at higher elevations.
Pigeonpea CWR occurring in areas of notably high annual and/or seasonal precipitation
included the central and southern Indian species C. lineatus, C. sericeus, and C. trinervius, as
well as C. crassus (Table 2, Supplementary Figure 5M,N,Q). As populations of most of these
species are adapted to soils with relatively high clay content, these may also represent
candidates for traits for waterlogging tolerance (Figure 5B, Supplementary Figure 5W).
Populations of CWR encountered in regions of very low annual and/or seasonal precipitation
included those from Australian species C. latisepalus, C. cinereus, C. acutifolius, and C.
lanceolatus, as well as C. sericeus and C. lineatus, among others (Table 2, Supplementary
Figure 5M,O,R). The pigeonpea crop was also modeled as being tolerant to very low rainfall
regions. Despite such tolerance in pigeonpea, we identified eleven CWR species with
distributions occurring in regions with less annual precipitation than the driest areas modeled
within the middle 90% of occurrence data inputs in regard to the area of cultivation of the crop
(i.e., <500 mm). Maps of potential distributions for a selected number of these CWR that are
not currently represented in ex situ genebanks are displayed in Figure 6, and the differentiation
of the occurrence data of these species in the PCA is shown in Supplementary Figure 4D.

204

Crop wild relatives of pigeonpea

Figure 5. Ecogeographic niches of pigeonpea CWR for (A) annual mean temperature and precipitation
and (B) percent clay and annual precipitation. Niches per species represent the middle 90% of
occurrence points, i.e., 10% outliers were excluded. For niches per ecogeographic variable per species,
see Supplementary Figure 5.

Figure 6. Potential distributions of selected CWR in (A) south Asia and (B) Australia that are not
currently represented in germplasm collections and which occur in regions with <500 mm annual
precipitation.
205

Chapter 6
DISCUSSION
With 80% of the CWR of pigeonpea in this study assessed as high priority for further collecting
for ex situ conservation, agreement from expert evaluators of medium to high importance for
all species, and with significant geographic gaps in ex situ collections for virtually all species,
it is clear that further conservation action is needed to safeguard the wild genetic resources of
this important grain legume. Included in this list of priorities are species with very few
germplasm accessions accessible to the global community in genebank information systems,
including C. confertiflorus, C. trinervius, C. latisepalus, and the putative progenitor C.
cajanifolius. Such taxa represent the highest level of priority for further collecting to fill gaps
in germplasm collections at the species level. As the species diversity gaps in genebank
collections largely align with the geographic distribution of species richness of pigeonpea
CWR, hotspots in India and in northern Australia represent particularly high value regions for
efficient collecting of the taxa (Figure 4). These areas may also be of interest for encountering
genetic variation created through hybridization between CWR species.
Additional unrepresented distributions of high priority species such as C. crassus and C.
scarabaeoides occur outside these regions, thus targeted collecting throughout the geographic
distributions of the species is necessary in order to form germplasm collections that are
comprehensive at the population level. Non-native distributions of widespread species,
particularly C. scarabaeoides, may also be considered for further collecting in the search for
useful traits for crop improvement. As techniques for the efficient utilization of wide diversity
of plant genetic resources improve, the collection, conservation, and availability of more distant
relatives of the crop may also become more worthwhile. We assessed the representation of the
other 17-19 species within genus Cajanus in ex situ conservation as currently minimal, with
only 41 accessions from seven species listed in germplasm repositories.
Cajanus scarabaeoides and C. platycarpus were identified as exhibiting potential adaptation to
climatic extremes, and are represented by some diversity of accessions conserved ex situ.
Existing collections should therefore be further screened for adaptive traits. Other species
identified as having potentially useful adaptations in contribution to major abiotic stress
tolerance breeding objectives for pigeonpea are represented by very few germplasm accessions,
especially C. confertiflorus, C. trinervius, C. cajanifolius, C. latisepalus, C. lanceolatus, and C.
cinereus. The climatic extremes of potential distribution models, such as those displayed in
206

Crop wild relatives of pigeonpea
Figure 6, may represent particular areas of interest for exploration. As Maxent models are based
upon known presence points for species and are thus subject to sampling bias, they may not
fully capture the possibility of occurrence of populations of CWR species in unique climates
(Araújo & Guisan 2006, Loiselle et al. 2008, Kramer-Schadt et al. 2013). Further field
exploration of climatic extremes beyond the edges of the distributions created through these
methods may therefore lead to the discovery of new populations with particularly valuable
adaptations to abiotic stress (Williams et al. 2009).
Despite sizable existing germplasm collections, species such as C. scarabaeoides and C.
albicans were categorized as medium or high priority for further collecting due to geographic
and ecological gaps in the collections. As the cost of conserving and investigating germplasm
ex situ is significant, particularly for CWR, a further assessment informed by genotypic
diversity analyses of what constitutes sufficient germplasm collections for pigeonpea CWR is
warranted. Given adequate resources, further collecting should be considered for these species,
as extremely valuable traits sourced from CWR of native Southeast Asian crops such as rice
have been found in only a few populations despite screening of a large number of accessions
(Brar & Khush 1997), and accessions of individual CWR species of pigeonpea have been shown
to possess notable variation in traits such as resistance to insect pests (Sharma et al. 2003), seed
protein content, and days to maturity (Upadhyaya et al. 2013a).
The regions of distribution of pigeonpea CWR species occur in areas undergoing habitat change
due to conversion to agriculture, logging, urbanization, mining, invasive species, and climate
change, among other factors (Sodhi et al. 2004, Sodhi et al. 2009, Upadhyaya et al. 2013a,
Sahai et al. 2014). It is clear that the window of opportunity for comprehensive collecting for
ex situ conservation of pigeonpea CWR will not exist indefinitely.
CONCLUSION
Pigeonpea cultivation is still largely limited to its origins in South Asia and in East Africa. Due
to its high nutritive value, agronomic versatility, stress-tolerance, and multiple uses, increasing
yield in existing production lands as well as expanding the crop into other areas of Asia and
Africa, as well as the Americas, can contribute toward greater agricultural sustainability and
improved human nutrition in sub-tropical and tropical regions. While pigeonpea already
possesses very favorable agronomic characteristics compared to other major grain legumes, its
207

Chapter 6
productivity can be improved via breeding, and its wild relatives show promise in providing
critical adaptive traits for major breeding objectives, including heat and cold tolerance, high
precipitation, waterlogging, and drought tolerance. Further collecting for ex situ conservation
of this diversity, securing long-term funding for this conservation and associated research,
ensuring safety-duplication of unique germplasm, and sharing of this diversity with the global
research community are critical to this process (FAO 2002, Esquinas-Alcázar 2005). Greater
investment in genotypic and phenotypic characterization and evaluation for traits of interest
(Mallikarjuna et al. 2011, Varshney et al. 2011, Upadhyaya et al. 2013a) and in breeding
programs using CWR, represent equally urgent steps (Tester and Langridge 2010, Guarino &
Lobell 2011, Henry 2014). Through such actions the crop research community will contribute
to ensuring the long term viability of this important crop, and be better prepared to adapt to the
challenges facing present and future grain legume production.

208

Supplementary Figure 1. Distribution of gap analysis values per CWR species for (A) sampling representativeness score (SRS), (B) geographic
representativeness score (GRS), and (C) ecological representativeness score (ERS).

Crop wild relatives of pigeonpea

SUPPLEMENTARY INFORMATION

209

Chapter 6

Supplementary Figure 2. Number of CWR species prioritized for further collecting per country. HPS =
high priority species, MPS = medium priority species, and LPS = low priority species for further
collecting.

210

Crop wild relatives of pigeonpea

Supplementary Figure 3. Expert evaluation accordance with gap analysis results. (A) correlation
between gap analysis results and comparable expert evaluation scores. (B) correlation between gap
analysis results and contextual expert evaluation scores. (C) correlation circle of all evaluation variables
[comparable expert priority score (Comparable); contextual expert priority score (Contextual);
evaluation of gap analysis results score (Evaluation); evaluation of occurrence data (Occ_data);
evaluation of potential species distribution models (SDM_map); and evaluation of collecting priorities
map (Gap_map)]. (D) combined expert evaluation index score per species.

211

Supplementary Figure 4. Principal component analysis of ecogeographic variables associated with occurrence data for crop wild relative species.
(A) contribution of variables to the first three components.

Chapter 6

212

Crop wild relatives of pigeonpea

Supplementary Figure 4. Principal component analysis of ecogeographic variables associated with
occurrence data for crop wild relative species. (B) geographic origin of clustered occurrence data points.
(C) contribution of cluster identified occurrence data points per species. (D) species’ occurrence data
points mapped over three principal components, with all points from the selected species highlighted in
Figure 6 colored.

213

Chapter 6
Supplementary Figure 5. Ecogeographic niches of crop wild relative (CWR) species based upon their
occurrence data presence locations, and the pigeonpea crop, per bioclimatic and edaphic variable. The
dark line represents median values, boxplots between 25-75% variation, and open circles outliers.

Supplementary Figure 5A. Niches of CWR species and the pigeonpea crop for altitude.

Supplementary Figure 5B. Climatic niches of CWR species and the pigeonpea crop for annual mean
temperature.
214

Crop wild relatives of pigeonpea

Supplementary Figure 5C. Climatic niches of CWR species and the pigeonpea crop for mean diurnal
range.

Supplementary Figure 5D. Climatic niches of CWR species and the pigeonpea crop for isothermality.
215

Chapter 6

Supplementary Figure 5E. Climatic niches of CWR species and the pigeonpea crop for temperature
seasonality.

Supplementary Figure 5F. Climatic niches of CWR species and the pigeonpea crop for maximum
temperature of the warmest month of the year.
216

Crop wild relatives of pigeonpea

Supplementary Figure 5G. Climatic niches of CWR species and the pigeonpea crop for minimum
temperature of the coldest month of the year.

Supplementary Figure 5H. Climatic niches of CWR species and the pigeonpea crop for temperature
annual range.
217

Chapter 6

Supplementary Figure 5I. Climatic niches of CWR species and the pigeonpea crop for mean temperature
of the wettest quarter of the year.

Supplementary Figure 5J. Climatic niches of CWR species and the pigeonpea crop for mean temperature
of the driest quarter of the year.
218

Crop wild relatives of pigeonpea

Supplementary Figure 5K. Climatic niches of CWR species and the pigeonpea crop for mean
temperature of the warmest quarter of the year.

Supplementary Figure 5L. Climatic niches of CWR species and the pigeonpea crop for mean
temperature of the coldest quarter of the year.
219

Chapter 6

Supplementary Figure 5M. Climatic niches of CWR species and the pigeonpea crop for annual
precipitation.

Supplementary Figure 5N. Climatic niches of CWR species and the pigeonpea crop for precipitation of
the wettest month of the year.
220

Crop wild relatives of pigeonpea

Supplementary Figure 5O. Climatic niches of CWR species and the pigeonpea crop for precipitation of
the driest month of the year.

Supplementary Figure 5P. Climatic niches of CWR species and the pigeonpea crop for precipitation
seasonality.
221

Chapter 6

Supplementary Figure 5Q. Climatic niches of CWR species and the pigeonpea crop for precipitation of
the wettest quarter of the year.

Supplementary Figure 5R. Climatic niches of CWR species and the pigeonpea crop for precipitation of
the driest quarter of the year.
222

Crop wild relatives of pigeonpea

Supplementary Figure 5S. Climatic niches of CWR species and the pigeonpea crop for precipitation of
the warmest quarter of the year.

Supplementary Figure 5T. Climatic niches of CWR species and the pigeonpea crop for precipitation of
the coldest quarter of the year.
223

Chapter 6

Supplementary Figure 5U. Edaphic niches of CWR species and the pigeonpea crop for bulk density.

Supplementary Figure 5V. Edaphic niches of CWR species and the pigeonpea crop for cation exchange
capacity.
224

Crop wild relatives of pigeonpea

Supplementary Figure 5W. Edaphic niches of CWR species and the pigeonpea crop for percent clay.

Supplementary Figure 5X. Edaphic niches of CWR species and the pigeonpea crop for organic carbon.
225

Chapter 6

Supplementary Figure 5Y. Edaphic niches of CWR species and the pigeonpea crop for pH.

Supplementary Figure 5Z. Edaphic niches of CWR species and the pigeonpea crop for percent silt.
226

Crop wild relatives of pigeonpea

Supplementary Figure 5AA. Edaphic niches of CWR species and the pigeonpea crop for percent sand.

227

Chapter 6
Supplementary Table 1. Sources of occurrence data for assessed CWR species.
Data provider
Record type
Australian Grains Genebank (AGG)
International Center for Tropical Agriculture (CIAT)
International Crops Research Institute for the Semi-Arid
Tropics (ICRISAT)
International Livestock Research Institute (ILRI)
Millennium Seed Bank Partnership, Royal Botanic
Gardens, Kew
World Vegetable Center (AVRDC)
Australia's Virtual Herbarium (AD, BRI, CANB, CBG,
DNA, HO, MEL, NSW, PERTH)
Bioversity International
California Academy of Sciences Herbarium (CAS)
Consortium of Pacific Northwest Herbaria (CPNWH)
Harvard University Herbarium (HUH)
Missouri Botanical Garden Herbarium (MO)
Nationaal Herbarium Nederland (NHN)
National Herbarium of New South Wales (NSW)
Natural History Museum UK Herbarium (BM)
New York Botanical Garden Herbarium (NY)
Plants of Taiwan
Royal Botanic Gardens, Edinburgh (E)
Royal Botanic Gardens, Kew (K)
Smithsonian Institution, National Herbarium (US)
USDA, National Plant Germplasm System, Germplasm
Resources Information Network (USDA NPGS GRIN)
van der Maesen (1986)
Western Australia Herbarium (PERTH) WAHerb

228

Germplasm accession
Germplasm accession
Germplasm accession

Number
of
records
201
75
89

Germplasm accession
Germplasm accession

2
9

Germplasm accession
Herbarium or other record

1
1459

Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record
Herbarium or other record

60
7
1
70
9
114
74
13
49
125
40
165
61
4

Herbarium or other record
Herbarium or other record

518
25

Crop wild relatives of pigeonpea
Supplementary Table 2. Ecogeographic variables utilized in the species potential distribution modeling
and niche analyses.
Variable Variable name
Units
number
0
Altitude
m
1
Annual mean temperature
ºC
2
Mean diurnal temperature range
ºC
3
Isothermality
N/A
4
Temperature seasonality (standard deviation)
ºC
5
Maximum temperature of warmest month
ºC
6
Minimum temperature of coldest month
ºC
7
Temperature annual range
ºC
8
Mean temperature of wettest quarter
ºC
9
Mean temperature of driest quarter
ºC
10
Mean temperature of warmest quarter
ºC
11
Mean temperature of coldest quarter
ºC
12
Annual precipitation
mm
13
Precipitation of wettest month
mm
14
Precipitation of driest month
mm
15
Precipitation seasonality (coefficient of variation)
%
16
Precipitation of wettest quarter
mm
17
Precipitation of driest quarter
mm
18
Precipitation of warmest quarter
mm
19
Precipitation of coldest quarter
mm
20
Bulk density
kg/m3
21
Cation exchange capacity
cmol/kg
22
Percent clay
%
23
Organic carbon
g/kg
24
pH in H2 O
pH
25
Percent silt
%
26
Percent sand
%

229

2

2

2

2

2

3

2

2

2

3

3

2

2

2

2

Taxon

C. albicans

C. cajanifolius

C. cinereus

C. confertiflorus

C. crassus

C. lanceolatus

C. latisepalus

C. lineatus

C. mollis

C. platycarpus

C. reticulatus

C. scarabaeoides

C. sericeus

C. trinervius

Total Records

134

96

577

483

38

15

180

233

140

34

124

273

52

175

594

Total Records- Distinct

25

29

162

228

12

8

34

86

41

14

47

132

13

45

192

H Records- Total

131

85

397

467

9

3

170

228

134

21

122

267

47

145

562

H Records- Distinct
23

20

113

228

1

1

26

86

39

11

46

132

8

27

185

G Records- Total
3

11

180

16

29

12

10

5

6

13

2

6

5

30

32

G Records- Distinct
3

9

57

0

11

7

8

0

2

3

1

0

5

18

7

SRS
0.22

1.15

3.12

0.33

7.63

8.00

0.56

0.21

0.43

3.82

0.16

0.22

0.96

1.71

0.54

Ecogeographic
Variables
2, 4, 8, 9, 15,
18, 19, 24
0, 5, 11, 12, 14,
15, 19, 20
4, 10, 13, 14,
20, 21, 22, 23,
24
3, 7, 9, 12, 14,
19, 20, 26

3, 7, 14, 18, 23

1, 3, 7, 12, 14,
15, 17, 19, 24
2, 8, 12, 15, 18,
20
11, 20, 22, 23,
24, 25
2, 3, 14, 15, 19,
20, 24, 25, 26
3, 6, 12, 15, 17,
19
2, 3, 8, 9, 12,
17, 23, 24
2, 3, 17, 26

1, 2, 9, 17, 18,
20, 23, 24
0, 7, 18, 20, 23,
24
8, 11, 14, 18, 24

ATAUC
0.96

0.94

0.95

0.95

0.75

0.67

0.99

0.96

0.98

0.89

0.97

0.93

0.78

0.93

0.93

STAUC
0.05

0.02

0.02

0.01

0.10

0.05

0.01

0.02

0.02

0.02

0.01

0.01

0.11

0.10

0.01

ASD15
4.71

1.68

0.42

0.48

0.19

0.00

1.63

0.00

0.70

0.00

2.55

1.74

0.07

10.90

0.28

cAUC
0.46

0.48

0.50

0.45

0.30

0.18

0.49

0.47

0.49

0.41

0.48

0.44

0.32

0.43

0.44

GRS
0.67

0.43

0.61

0.00

1.27

3.57

1.18

0.00

0.38

0.09

0.09

0.00

0.63

1.01

0.56

2.40

3.82

2.21

0.00

10.00

8.18

4.67

0.00

1.33

0.61

3.00

0.00

5.45

7.65

3.33

1.10

1.80

1.98

0.11

6.30

6.58

2.13

0.07

0.71

1.51

1.08

0.07

2.35

3.46

1.48

FPCAT
HPS

HPS

HPS

HPS

LPS

LPS

HPS

HPS

HPS

HPS

HPS

HPS

HPS

MPS

HPS

EPS 1
0

0

0

1

0

0

0

0

0

1

1

0

0

0

1

EPS 2
3

6

8

6

9

6

9

5

6

8

5

8

3

8

3

EPS 3
1

2

6

2

2

5

1

1

2

1

1

2

2

4

2

EPS 4
1

2

4

3

4

3

1

1

0

2

0

1

3

3

1

1.25

2.5

4.5

3

3.75

3.5

2.75

1.75

2

3

1.75

2.75

2

3.75

1.75

EPS mean

Genepool 2 is secondary; 3 is tertiary. Distinct records denote occurrence records with unique coordinates. H records denote herbarium and other locality data used to inform species distribution models.G
records denote germplasm accessions accessible to the global community. T he final priority score (FPS) for the species is the mean of the sampling representativenessscore(SRS),geographicrepresentativeness
score (GRS), and ecological representativeness score (ERS). Ecogeographic variables lists those species-specific bioclimatic and edaphic variables used in Maxent modeling per taxon. Maxent validation
techniques include the 5-fold average area under the ROC curve of test data (AT AUC), the standard deviation of the test AUC of the 5 different folds (ST AUC), the proportion of the potentialdistribution
coverage with standard deviation above 0.15 (ASD15), and the calibrated AUC ( cAUC). Final priority categories (FPCAT ) for the CWR of pigeonpea included high (HPS), medium (MPS),andlowpriority
species (LPS) for further collecting for ex situ conservation. EPS denotes comparable expert priority scores for 4 experts, along with their mean value.

Genepool

C. acutifolius

ERS

230
FPS

Supplementary Table 3. Gap analysis and comparable expert evaluation priorities results per crop wild relative species.

Chapter 6

Crop wild relatives of pigeonpea
Supplementary Table 4. Countries identified for potential further collecting per crop wild relative
species.
Taxon
Priority Countries identified for further collecting
C. acutifolius
HPS
Australia
C. albicans
MPS
India, Sri Lanka
C. cajanifolius
HPS
India
C. cinereus
HPS
Australia
C. confertiflorus HPS
Australia
C. crassus
HPS
China, India, Indonesia, Lao PDR, Malaysia, Papua New Guinea,
Philippines, Thailand, Timor-Leste, Vietnam
C. lanceolatus
HPS
Australia
C. latisepalus
HPS
Australia
C. lineatus
HPS
India, Sri Lanka
C. mollis
LPS
Nepal
C. platycarpus

LPS

C. reticulatus
HPS
C. scarabaeoides HPS
C. sericeus
C. trinervius

HPS
HPS

None modeled, but expected in India
Australia
Australia, Bangladesh, Bhutan, China, Fiji, India, Indonesia, Japan,
Madagascar, Malaysia, Myanmar, Nepal, Papua New Guinea,
Philippines, Sri Lanka, Taiwan, Thailand, Timor-Leste, Vietnam
India
India, Sri Lanka

231

Chapter 6
REFERENCES
Araújo, M.B., Guisan, A. (2006). Five (or so) challenges for species distribution modelling. J. Biogeogr.
33 (10), 1677-1688.
Ariyanayagam, R.P., Rao, A.N., Zavery, P.P. (1995). Cytoplasmic-genic male-sterility in interspecific
matings of Cajanus. Crop Sci. 35, 981-985.
Bivand, R., Keitt, T., Rowlingson, B. (2014). Rgdal: Bindings for the geospatial data abstraction
library. R package version 0.8-16. http://CRAN.R-project.org/package=rgdal.
Bivand, R., Lewin-Koh, N. (2014). Maptools: Tools for reading and handling spatial objects. R package
version 0.8-30. http://CRAN.R-project.org/package=maptools.
Bivand, R., Pebesma, E., Gomez-Rubio, V. (2013). Applied spatial data analysis with R, Second edition.
Springer, NY. http://www.asdar-book.org/.
Bogdan, A.V. (1977). Tropical Pasture and Fodder Plants (New York: Longman).
Brar, D.S., Khush, G.S. (1997). Alien introgression in rice. Plant Mol. Biol. 35 (1-2), 35-47.
Choudhary, A.K., Kumar, S., Patil, B.S., Bhat, J.S., Sharma, M., Kemal, S., et al. (2013). Narrowing
yield gaps through genetic improvement for Fusarium wild resistance in three pulse crops of the
semi-arid tropics. SABRAO J. Breed. Genet. 45 (3), 341-370.
Choudhary, A.K., Sultana, R., Pratap, A., Nadarajan, N., Jha, U.C. (2011). Breeding for abiotic stresses
in pigeonpea. J. Food Legumes 24 (3), 165-174.
Cordell, D., Drangert, J-O., White, S. (2009). The story of phosphorus: global food security and food
for thought. Global Environ. Chang. 19 (2), 292-305.
Costa, G.C., Nogueira, C., Machado, R.B., Colli, G.R. (2010). Sampling bias and the use of ecological
niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodivers.
Conserv. 19, 883-899.
Dempewolf, H., Eastwood, R.J., Guarino, L., Khoury, C.K., Müller, J.V., Toll, J. (2014). Adapting
agriculture to climate change: A global initiative to collect, conserve, and use crop wild relatives.
Agroecol. Sust. Food Syst. 38 (4), 369-77.
Elith, J., Graham, C.H., Anderson, R.P., Dudik, M., Ferrier, S., Guisan, A., et al. (2006). Novel methods
improve prediction of species’ distributions from occurrence data. Ecography 29, 129-151.
Esquinas-Alcázar, J. (2005). Protecting crop genetic diversity for food security: political, ethical and
technical challenges. Nat. Rev. Genet. 6 (12), 946-953.
ESRI (2011). ArcGIS Desktop: Release 10.1.
FAO (Food and Agriculture Organization of the United Nations). (2002). International Treaty on Plant
Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2010a). Crop Calendar. Online
Resource. http://www.fao.org/agriculture/seed/cropcalendar/welcome.do. (Rome: FAO). Accessed
December 2013.
FAO (Food and Agriculture Organization of the United Nations). (2010b). The Second Report on the
State of the World’s Plant Genetic Resources for Food and Agriculture (Rome: FAO).
Flower, D.J., Ludlow, M.M. (1987). Variation among accessions of pigeonpea (Cajanus cajan) in
osmotic adjustment and dehydration tolerance of leaves. Field Crop. Res. 17 (3-4), 229-243.
Foley, J.A., Ramankutty, N., Brauman, K.A., Cassidy, E.S., Gerber, J.S., Johnston, M., et al. (2011).
Solutions for a cultivated planet. Nature 478 (7369), 337-342.
Google Maps Geocoder (2013). Online resource. https://developers.google.com/maps/documentation/
geocoding/?hl=pl (accessed June 2013).
GRIN (Germplasm Resources Information Network), USDA, ARS, National Genetic Resources
Program (2012). National Germplasm Resources Laboratory, Beltsville, Maryland. Online resource.
232

Crop wild relatives of pigeonpea
http://www.ars-grin.gov/~sbmljw/cgi-bin/index.pl?language=en. Accessed March 2013.
Guarino, L., Lobell, D.B. (2011). A walk on the wild side. Nat. Clim. Change 1 (8), 374-375.
Harlan, J.R., de Wet, J.M.J. (1971). Toward a rational classification of cultivated plants. Taxon 20, 509517.
Hengl, T., de Jesus, J.M., MacMillan, R.A., Batjes, N.H., Heuvelink, G.B.M., Ribeiro, E., et al. (2014).
SoilGrids1km — Global soil information based on automated mapping. PLoS One 9 (8), e105992.
Henry, R.J. (2014). Genomics strategies for germplasm characterization and the development of climate
resilient crops. Front. Plant Sci. 5, 68.
Hijmans, R.J. (2012). Cross-validation of species distribution models: removing spatial sorting bias and
calibration with a null model. Ecology 93, 679-688.
Hijmans, R.J. (2014). Raster: geographic data analysis and modeling. R package version 2.2-31.
http://CRAN.R-project.org/package=raster.
Hijmans, R.J., Cameron, S., Parra, J., Jones, P.G., Jarvis, A. (2005). WorldClim, Version 1.3 (Berkeley:
University of California).
Hijmans, R.J., Guarino, L., Cruz, M., Rojas, E. (2001). Computer tools for spatial analysis of plant
genetic resources data: 1. DIVA-GIS. Plant Genetic Resources Newsletter 127, 15-19.
Hijmans, R.J., Phillips, S., Leathwick, J., Elith, J. (2013). Dismo: Species distribution modeling. R
package version 0.9-3. http://CRAN.R-project.org/package=dismo.
Hijmans, R.J., Schreuder, M., De La Cruz, J., Guarino, L. (1999). Using GIS to check co-ordinates of
genebank accessions. Genet. Resour. Crop Ev. 46, 291-296.
Husson, F., Josse, J., Le, S., Mazet, J. (2009). FactoMineR: multivariate exploratory data analysis and
data mining with R. v.1.26. http://factominer.free.fr.
Jarvis, A., Lane, A., Hijmans, R.J. (2008). The effect of climate change on crop wild relatives. Agr.
Ecosyst. Environ. 126 (1-2), 13-23.
Jarvis, A., Williams, K., Williams, D., Guarino, L., Caballero, P.J., Mottram, G. (2005). Use of GIS for
optimizing a collecting mission for a rare wild pepper (Capsicum flexuosum Sendtn.) in Paraguay.
Genet. Resour. Crop Ev. 52 (6), 671-682.
Jones, R.B., Freeman, H.A., Lo Monaco, G. (2002). Improving the access of small farmers in Eastern
and Southern Africa to global pigeonpea markets. Agricultural Research and Extension Network
120, 1-11.
Kassa, M.T., Varma Penmetsa, R., Carrasquilla-Garcia, N., Sarma, B.K., Datta, S., Upadhyaya, H.D.,
et al. (2012). Genetic patterns of domestication in pigeonpea (Cajanus cajan (L) Millsp) and wild
Cajanus relatives. PloS One 7 (6), e39563.
Khoury, C., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic resources:
a review of global crop and regional conservation strategies. Genet. Resour. Crop Ev. 57 (4), 625639.
Kramer-Schadt, S., Niedballa J., Pilgrim J.D., Schroder B., Lindenborn J., Reinfelder, V., et al. (2013).
The importance of correcting for sampling bias in MaxEnt species distribution models. Divers.
Distrib. 19 (11), 1366-1379.
Liu, C., Berry, P.M., Dawson, T.P., Pearson, R.G. (2005). Selecting thresholds of occurrence in the
prediction of species distributions. Ecography 28 (3), 385-393.
Lobell, D.B., Burke, M.B., Tebaldi, C., Mastrandrea, M.D., Falcon, W.P., Naylor, R.L. (2008).
Prioritizing climate change adaptation needs for food security in 2030. Science 319 (5863), 607-610.
Loiselle, B., Jørgensen, P.M., Consiglio, T., Jiménez, I., Blake, J.G., Lohmann, L.G., et al. (2008).
Predicting species distributions from herbarium collections: does climate bias in collection sampling
influence model outcomes? J. Biogeogr. 35, 105-116.
Lopez, F.B., Setter, T.L., McDavid, C.R. (1987). Carbon dioxide and light responses of photosynthesis
233

Chapter 6
in cowpea and pigeonpea during water deficit and recovery. Plant Physiology 85 (4), 990-995.
Mallikarjuna, N., Jadhav, D., Reddy, M.V., Usharani, D.-T. (2005). Introgression of Phytophthora
blight disease resistance from Cajanus platycarpus into short duration pigeonpea [Cajanus cajan
(L.) Millsp.]. Indian J. Genet. Pl. Br. 65 (4), 261-263.
Mallikarjuna, N., Saxena, K.B. (2005). A new cytoplasmic nuclear male-sterility system derived from
cultivated pigeonpea cytoplasm. Euphytica 142 (1-2), 143-48.
Mallikarjuna, N., Saxena, K.B., Jadhav, D.R. (2011). “Cajanus,” in: Wild Crop Relatives: Genomic and
Breeding Resources, Legume Crops and Forages, ed. C. Kole (Berlin, Heidelberg: Springer-Verlag),
21-33.
Mallikarjuna, N., Sharma, H.C., Upadhyaya, H.D. (2007). Exploitation of wild relatives of pigeonpea
and chickpea for resistance to Helicoverpa armigera. Jounal of SAT Agricultural Research 3 (1), 4.
McCouch, S., Baute, G., Bradeen J., Bramel, P., Bretting, P.K., Buckler, E., et al. (2013). Agriculture:
feeding the future. Nature 499, 23-24.
Mligo, J.K., Craufurd, P.Q. (2005). Adaptation and yield of pigeonpea in different environments in
Tanzania. Field Crop. Res. 94 (1), 43-53.
Monfreda, C., Ramankutty, N., Foley, J.A. (2008). Farming the planet: 2. Geographic distribution of
crop areas, yields, physiological types, and net primary production in the year 2000. Global
Biogeochem. Cy. 22, GB1022. Data: http://www.geog.mcgill.ca/landuse/pub/Data/175crops2000/.
Mudaraddi, B., Saxena, K.B., Saxena, R.K., Varshney, R.K. (2013). Molecular diversity among wild
relatives of Cajanus cajan (L.) Millsp. Afr. J. Biotechnol. 12 (24), 3797-3801.
Odeny, D.A. (2007). The potential of pigeonpea (Cajanus cajan (L.) Millsp.) in Africa. Nat. Resour.
Forum 31 (4), 297-305.
Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V.N., Underwood, E.C.,
et al. (2001). Terrestrial ecoregions of the world: a new map of life on earth. BioScience 51 (11),
933-938.
Pebesma, E.J., Bivand, R.S. (2005). Classes and methods for spatial data in R. R News 5 (2), http://cran.rproject.org/doc/Rnews/.
Phillips, S.J., Anderson, R.P., Schapire, R.E. (2006). Maximum entropy modeling of species geographic
distributions. Ecol. Model. 190 (3-4), 231-259.
Phillips, S.J., Dudik, M. (2008). Modeling of species distributions with maxent: new extensions and
comprehensive evaluation. Ecography 31, 161-175.
R Core Team (2013). R: A language and environment for statistical computing. http://www.rproject.org/.
Ramírez-Villegas, J., Khoury, C., Jarvis, A., Debouck, D.G., Guarino, L. (2010). A gap analysis
methodology for collecting crop genepools: a case study with Phaseolus beans. PloS One 5 (10),
e13497.
Ray, D.K., Mueller, N.D., West, P.C., Foley, J.A. (2013). Yield trends are insufficient to double global
crop production by 2050. PLoS One 8 (6), e66428.
Reddy, L.J. (1990). “Pigeonpea: Morphology,” in: The Pigeonpea, eds. Y.L. Nene, S.D. Hall, V.K.
Sheila, (Wallingford: CAB International). 47-87.
Reddy, L.J., Green, J.M., Singh, U., Bisen, S.S., Jambunathan, R. (1979). “Seed protein studies on
Cajanus cajan, Atylosia spp. and some hybrid derivatives,” in: Seed protein improvement in cereals
and grain legumes. Proceedings of an international symposium jointly organized by IAEA and FAO
in cooperation with GSF Neuherberg, 4-8 September 1978, 105-117.
Rosenzweig, C., Elliot, J., Deryng, D., Ruane, A.C., Muller, C., Arneth, A., et al. (2013). Assessing
agricultural risks of climate change in the 21st century in a global gridded crop model
intercomparison. Proc. Natl Acad. Sci. 111 (9), 3268-3273.
234

Crop wild relatives of pigeonpea
Sahai, K., Rawat, K.K. (2014). A survey of the degrading population of Cajanus lineatus (Wight &
Arn.) Maesen, in parts of the Western Ghats, India. Genet. Resour. Crop Ev., 1-9. doi:
10.1007/s10722-014-0176-4.
Sanchez, G. (2012). Package ‘plsdepot’. Partial Least Squares (PLS) Data Analysis Methods, v. 0.1.17.
Sardana, V., Sharma, P., Sheoran, P. (2010). “Growth and production of pulses,” in: Soils, Plant Growth
and Crop Production- Vol III- Growth and Production of Pulses. Encyclopedia of Life Support
Systems. http://www.eolss.net/sample-chapters/c10/e1-05a-44-00.pdf.
Saxena, K.B. (2005). Pigeonpea [Cajanus cajan (L.) Millsp.]. in: Genetic Resources, Chromosome
Engineering, and Crop Improvement: Volume 1, Grain Legumes, eds. R.J. Singh, P.P. Jauhar (Boca
Raton: CRC Press), 86-110.
Saxena, K.B., Kumar, R.V. (2003). Development of a cytoplasmic nuclear male-sterility system in
pigeonpea using C. scarabaeoides (L.) Thouars. Indian J. Genetics 63 (3), 225-229.
Saxena, K.B., Kumar, R.V., Gowda, C.L.L. (2010). Vegetable pigeonpea- a review. J. Food Legumes
23 (2), 91-98.
Saxena, K.B., Kumar, R.V., Srivastava, N. Shiying, B. (2005). A cytoplasmic-nuclear male-sterility
system derived from a cross between Cajanus cajanifolius and Cajanus cajan. Euphytica 145 (3),
289-294.
Saxena, K.B., Kumar, R.V., Tikle, A.N., Saxena, M.K., Gautam, V.S., Rao, S.K., et al. (2013). ICPH
2671 – the world’s first commercial food legume hybrid. Plant Breeding 132, 479-485.
Saxena, K.B., Sawargaonkar, S. L. (2014). First information on heterotic groups in pigeonpea [Cajanus
cajan (L.) Millsp.]. Euphytica 200 (2), 187-196.
Saxena, K.B., Srivastava, D.P, Tikkla, S.B.S. (1998). “Breaking yield barrier in pigeonpea through
hybrid breeding”. in: Proceedings of the National Symposium on Biotic and Abiotic Stress of Pulse
Crops (Kanpur: Indian Institute of Pulses Research), 55-64.
Sharma, H.C., Pampapathy, G., Reddy, L.J. (2003). Wild relatives of pigeonpea as a source of resistance
to the pod fly (Melanagromyza obtusa Malloch) and pod wasp (Tanaostigmodes cajaninae La Salle).
Genet. Resour. Crop Ev. 50, 817-824.
Sodhi, N.S., Koh, L.P., Brook, B.W., Ng, P.K.L. (2004). Southeast Asian biodiversity: an impending
disaster. Trends Ecol. Evol. 19 (12), 654-660.
Sodhi, N.S., Posa, M.R.C., Lee, T.M., Bickford, D., Koh, L.P., Brook, B.W. (2009). The state and
conservation of Southeast Asian biodiversity. Biodivers. Conserv. 19 (2), 317-328.
Subbarao, G.V., Chauhan, Y.S. Johansen, C. (2000). Patterns of osmotic adjustment in pigeonpea - its
importance as a mechanism of drought resistance. Eur. J. Agron. 12 (3-4), 239-249.
Subbarao, G.V., Johansen, C., Jana, M.K., Kumar Rao J.V.D.K. (1990). Comparative salinity responses
among pigeonpea wild relatives. Crop Sci. 31, 415-418.
Tapia, C., Torres, E., Parra-Quijano, M. (2014). Searching for adaptation to abiotic stress:
ecogeographical analysis of highland Ecuadorian maize. Crop Sci. 55 (1), 262-274.
Tester, M, Langridge, P. (2010). Breeding technologies to increase crop production in a changing world.
Science 327 (5967), 818-822.
The Plant List. (2010). Version 1. http://www.theplantlist.org/.
Tilman, D., Clark, M. (2014). Global diets link environmental sustainability and human health. Nature
515 (7528), 518-522.
Upadhyaya, H.D. (2006). Improving pigeonpea with the wild. SATrends January 2006.
Upadhyaya, H.D., Reddy, K.N., Pundir, R.P.S., Singh, S., Gowda, C.L.L., Irshad Ahmed, M. (2013a).
Diversity and geographical gaps in Cajanus scarabaeoides (L.) Thou. germplasm conserved at the
ICRISAT genebank. Plant Genetic Resources 11 (1), 3-14.
Upadhyaya, H.D., Reddy, K.N., Sastry, D.V.S.S.R., Gowda, C.L.L. (2007). Identification of
235

Chapter 6
photoperiod insensitive sources in the world collection of pigeonpea at ICRISAT. J. SAT. Agri. Res.
J. 3 (1), 46-49.
Upadhyaya, H.D., Reddy, K.N., Singh, S., Gowda, C.L.L. (2013b). Phenotypic diversity in Cajanus
species and identification of promising sources for agronomic traits and seed protein content. Genet.
Resour. Crop Ev. 60 (2), 639-659.
Valenzuela, H., Smith J. (2002). “Pigeonpea,” in: Sustainable Agriculture (Honolulu: University of
Hawaii).
van der Maesen, L.J.G. (1986). Cajanus DC. and Atylosia W. & A. (Leguminosae). Agricultural
University Wageningen Papers 85-4 (Wageningen: Wageningen University Press).
van der Maesen, L.J.G. (1989). “Cajanus cajan (L.) Millsp,” in: Plant Resources of South-East Asia No
1. Pulses, eds. L.J.G. van der Maesen, S. Somaatmadja (Wageningen: Pudoc/Prosea), 39-42.
van der Wal, J., Falconi, L., Januchowski, S., Shoo, L., Storlie, C. (2014). SDMTools: species
distribution modelling tools: tools for processing data associated with species distribution modelling
exercises. R package version 1.1-20. http://CRAN.R-project.org/package=SDMTools.
Varshney, R.K., Chen, W., Li, Y., Bharti, A.K., Saxena, R.K., Schlueter, J.A., et al. (2011). Draft
genome sequence of pigeonpea (Cajanus cajan), an orphan legume crop of resource-poor farmers.
Nat. Biotechnol. 30 (1), 83-89.
Vincent, H., Wiersema, J., Kell, S., Fielder, H., Dobbie, S., Castañeda-Álvarez, N.P., et al. (2013). A
prioritized crop wild relative inventory to help underpin global food security. Biol. Conserv. 167,
265-275.
Wanjari, K.B., Patil, A.N., Manapure, P., Manjayya, J.G., Manish, P. (1999). Cytoplasmic male sterility
in pigeonpea with cytoplasm from Cajanus volubilis. Annals of Plant Physiology 13, 170-174.
West, P.C., Gerber, J.S., Engstrom, P.M., Mueller, N.D. Brauman, K.A., Carlson, K.M., et al. (2014).
Leverage points for improving global food security and the environment. Science 345 (6194), 325328.
Wickham, H. (2009). Ggplot2: Elegant Graphics for Data Analysis (New York: Springer).
Williams, J.N., Seo, C., Thorne, J., Nelson, J.K., Erwin, S., O’Brian, J.M., et al. (2009). Using species
distribution models to predict new occurrences for rare plants. Divers. Distrib. 15 (4), 565-576.
Xiao, J., Grandillo, S., Ahn, S.N., McCouch, S.R., Tanksley, S.D., Li, J., et al. (1996). Genes from wild
rice improve yield. Nature 384, 223-224.

236

CHAPTER 7
Interdependence among countries on crop genetic diversity
Colin K. Khoury1,2, Harold A. Achicanoy1, Anne D. Bjorkman3, Carlos Navarro-Racines1,4,
Luigi Guarino5, Ximena Flores-Palacios6, Johannes M. M. Engels7, John H. Wiersema8,
Hannes Dempewolf5, Julian Ramírez-Villegas1,4,9, Nora P. Castañeda-Álvarez1,10,
Cary Fowler5, Andy Jarvis1,4, Loren H. Rieseberg11,12,13, and Paul C. Struik2
International Center for Tropical Agriculture (CIAT), Km 17, Recta Cali-Palmira, Apartado Aéreo
6713, 763537 Cali, Colombia.
2
Centre for Crop Systems Analysis, Wageningen University, Droevendaalsesteeg 1, 6708 PB
Wageningen, The Netherlands.
3
German Centre for Integrative Biodiversity Research, Leipzig, Germany.
4
CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Km 17,
Recta Cali-Palmira, Apartado Aéreo 6713, 763537 Cali, Colombia.
5
Global Crop Diversity Trust, Platz der Vereinten Nationen 7, 53115 Bonn, Germany. 6Auckland
University of Technology, 55 Wellesley St E, Auckland 1010, New Zealand. 7Bioversity
International, Via dei Tre Denari, 472/a, 00057 Maccarese, Italy.
8
United States Department of Agriculture, Agricultural Research Service, National Germplasm
Research Laboratory, Building 003, BARC-West, Beltsville, Maryland 20705-2350, USA.
9
School of Earth and Environment, University of Leeds, Leeds, UK.
10
School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. 11The
Biodiversity Research Centre, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada.
12
Department of Botany, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada.
13
Department of Biology, Indiana University, Bloomington, Indiana, 47405, USA.
1

Submitted for publication in an international scientific journal
Published as Research Study 8, International Treaty on Plant Genetic Resources for Food and
Agriculture (Rome: FAO). A modified version of this chapter with the title “Estimation of
countries’ interdependence in plant genetic resources provisioning national food supplies and
production systems” was published by the Secretariat of the International Treaty on Plant
Genetic Resources for Food and Agriculture following a call by the Secretariat and its Ad Hoc
Open-Ended Working Group to Enhance the Functioning of the Multilateral System of Access
and Benefit-Sharing, to encourage international organizations to make available information
relevant to an analysis of the factors that influence the willingness of stakeholder groups to make
contributions to the Benefit-sharing Fund and to access plant genetic resources for food and
agriculture from the Multilateral System.

Chapter 7
Abstract
The crops that feed the world originated in specific geographic regions across the planet. Genetic
diversity within these crops and their wild relatives is considered to be historically particularly rich
within these regions. These genetic resources are regularly employed in crop improvement: thus,
preventing erosion of remnant genetic diversity occurring in situ, and ensuring the long-term access to
this diversity conserved in genebanks ex situ, are critical to continued increases in agricultural
productivity. The geopolitical significance of the geography of crop genetic diversity has not been
quantified. Here we assess the degree to which the food supplies and production systems of countries
worldwide are comprised of crops from each of these regions of diversity. We then examine dependence
of countries upon crops from regions of diversity other than their own (“foreign crops”), and determine
change in this dependence over the past 50 years. National food systems are thoroughly interconnected
worldwide in regard to the geographic origins of crop diversity. Countries are highly dependent on
foreign crops in their food supplies (68.7% as a global mean across food variables) and in their national
production systems (69.3%). This reliance is evident even in countries located in regions of high
indigenous crop diversity and has increased significantly over the past half century, stressing the need
for effective national and international policies to promote genetic resource conservation and exchange.
Keywords: Food security, Crop diversity, Crop origins, Interdependence, Plant genetic resources

238

Interdependence among countries on crop genetic diversity
INTRODUCTION
As the source of traits employed for adaptation to biotic and abiotic stresses and for yield
increases through breeding, as well as the palette from which food systems may be further
diversified, crop genetic diversity represents a cornerstone of food security (Xiao et al. 1996,
Hoisington et al. 1999, Zhu et al. 2000, Gepts 2006, Guarino & Lobell 2011). The need for
utilization of this diversity to maintain or enhance crop productivity is likely only to grow given
rising food demand and increasing constraints on the use of non-renewable agricultural inputs,
limitations in further expansion of arable lands, soil degradation, and global climatic change
(Lobell et al. 2008, Cordell et al. 2009, Kearney 2010, Kastner et al. 2012).
Crop genetic diversity is generated through genetic mutation and recombination, and further
transformed through natural and artificial selection. Therefore, high levels of crop genetic
diversity are associated with the persistence of crops and their wild relatives in specific
geographic regions worldwide. A century ago, N. I. Vavilov described these “centers of origin”,
characterized by notably high levels of inter- and intra-specific variability in food crops. These
included Central America and Mexico; the Andes, Chile and Brazil-Paraguay; the
Mediterranean; the Near East; Ethiopia; Central Asia; India; China; and Indo- Malaysia
(Vavilov 1926, 1951, 1992). Since then, the number and boundaries of these centers have been
investigated and refined (Harlan 1951, Zhukovsky 1968, Harlan 1971, Zeven & Zhukovsky
1975, Zeven & de Wet 1982). Here we use the term “primary regions of diversity” to describe
these areas, which typically include the general geographic locations of the initial domestication
of crops, encompass major geographic zones of crop genetic diversity generated since that time,
and generally also include high species richness in related wild taxa.
New forms and combinations of crop genetic diversity may arise wherever farmers plant,
harvest and re-sow their seed (Harlan 1975, Nuijten et al. 2009). The spread of crops outside
their centers of origin and their increasing contribution to the diets of diverse cultures in
different regions (Khoury et al. 2014) have therefore led to the development of “secondary
centers of diversity”, as well as novel genotypes arising outside of any such defined centers.
While all crop genetic diversity, regardless of geographic distribution, is of potential value to
crop improvement, the diversity generated over time in the primary regions is considered to be
the centerpiece of historical, current, and future crop improvement efforts due to its
comparatively high values at allelic, genotypic, and species levels (Vavilov 1926, Harlan 1971,
239

Chapter 7
Hoisington et al. 1999, Gepts 2006).
Severe erosion of crop genetic diversity is considered to have occurred over the past half
century, particularly through the adoption of improved crop varieties or substitute crop species
and subsequent neglect of traditional varieties and crops, economic development and associated
shifts in consumer demand, land use change and habitat destruction, and urbanization and the
displacement of cultures associated with particular crops and varieties (Hoisington et al. 1999,
Gepts 2006, Wilkes 2007, van de Wouw et al. 2009, 2010). In some crops, a small fraction of
the diversity once present is thought to still be found today in farmers’ fields, e.g., in wheat
varieties in parts of the Fertile Crescent (Harlan 1971, Gepts 2006). Due to the loss of variation
in regions of diversity, the world’s genebanks originally established to make plant genetic
resources readily available to breeders for crop improvement, have become essential
repositories for crop diversity conservation. A substantial portion of the world’s remaining
heritage of food crop genetic resources is likely now conserved exclusively in genebanks
(Hoisington et al. 1999, Gepts 2006, FAO 2010, Thormann et al. 2015).
Thus, the long-term productivity and resilience of food systems depend upon the conservation
and use of crop genetic diversity, accessed either directly from regions of diversity or via
genebanks (Hoisington et al. 1999, McCouch et al. 2013). Unfortunately, significant gaps
remain in the conservation of crop genetic resources globally (Hoisington et al. 1999, Gepts
2006, FAO 2010) and international access to this diversity requires improvement (Fowler &
Hodgkin 2004, Bjørnstad et al. 2013). These deficiencies may be due, in part, to lack of
information on the significance to modern food systems of the historical geographic
distributions of crop genetic diversity. Here we determine the degree to which the national food
supplies (measured in calories, protein, fat, and food weight) and national production systems
(measured in production quantity, harvested area, and production value) of countries worldwide
are comprised of crops from all the different primary regions of genetic diversity. In order to
assess the level to which international collaboration is required in order to achieve access to crop genetic
resources, we estimate the degree of dependence of countries upon crops from primary regions

of diversity other than their own (i.e., “foreign” crops), and determine change in this
dependence over the past 50 years.

240

Interdependence among countries on crop genetic diversity
MATERIALS AND METHODS
We used food supplies and production systems data provided by FAO (2015) [for food supplies:
calories (kcal/capita/day), protein (g/capita/day), fat (g/capita/day), and food weight
(g/capita/day); for production systems: production quantity (tonnes), harvested area (ha), and
gross production value (million US$)]. National food supply from plants represents national
production plus imports plus or minus stock changes over the survey period; minus exports,
quantities used for seed, animal feed, and in the manufacture of non-food products, and losses
during storage and transport (Khoury et al. 2014). We utilized the full set of food crop
commodities included in food supply and production data. While food supplies data accounts
for direct human consumption, production data for crops such as maize and soybean is
potentially inclusive of livestock and industrial uses as well as human food. In the production
analysis we also included agricultural crops indirectly contributing to human food supplies via
livestock production (i.e., alfalfa, clover, and vetch). Non-food (e.g., industrial and fiber) crops
as well as animal product commodities were not included in the analysis. Plant commodities
comprised of the same crop species were aggregated into single commodities representing the
crop, e.g., sesame seed oil and sesame seed. After aggregation, 53 crop commodities remained
in food supplies data, and 132 crop commodities in production data (Supplementary Table 1).
See Table S1 of Khoury et al. (2014) for a comprehensive listing of the crop species included
in the commodities treated in food supplies data.
For current food supplies and production systems, we analyzed data for each crop commodity
per country per measurement over the most recent three years for which sufficient data were
available (2009-2011). All (177) countries consistently reported during the time period were
included for food supplies variables, as well as for production quantity and harvested area
(Supplementary Table 2), covering 98.5% of the world’s population. All (141) countries
reported for (current million US$) production value were included, covering 94.1% of the
world’s population (FAO 2015).
For the analysis of change in dependence over time, food supplies data were assessed for each
year from 1961-2009, and production systems from 1961-2011, utilizing the full set of
commodity and country listings, standardized across all years. In order to align all time periods
and include as much of the world’s population as possible, the current countries formerly
comprising the USSR, Yugoslav SFR, Ethiopia PDR, and Czechoslovakia were aggregated into
241

Chapter 7
their former countries, with national data summed per year for production measurements, and
merged by weighted average based upon the population of the respective states during the
respective reporting year for per capita food supplies measurements. Belgium and Luxembourg
were reported together during 1961-1999 and therefore recent years listing the countries
separately were merged as above. Countries that did not have estimates in every year between
1961 and 2009/2011 were removed from the analysis. The resulting 152 comparable countries
treated in food supplies data comprised 98% of the world’s population across the study period
(Khoury et al. 2014). The 182 comparable countries covered in production quantity and
harvested area data comprised 99.7% of the global population, and the 115 countries covered
in (constant 2004-2006 million US$) production value data covered 88.5% (Supplementary
Table 2).
Primary regions of diversity were assigned based upon primary and secondary literature
regarding centres of crop diversity, origins of crop domestication, and high species richness of
closely related wild plants (Harlan 1951, Zhukovsky 1968, Harlan 1971, Zeven & Zhukovsky
1975, Zeven & de Wet 1982, Flores-Palacios 1998, Engels et al. 2001, Vincent et al. 2013,
GRIN 2014, Prota 2014). Regional classifications followed those listed in Annex 2 of the FAO
State of the World’s Plant Genetic Resources for Food and Agriculture (2010), modified to
more accurately represent eco-geographic parameters driving plant species distributions.
Specifically, both western and eastern Europe were split into north and south regions to account
for temperate versus Mediterranean ecologies; Australia and New Zealand were segregated
from remaining (tropical) islands of the Pacific region; and South America was split into
Andean, temperate, and tropical regions. A total of 23 eco-geographic regions were delineated
(Supplementary Figure 1). Countries whose boundaries included more than one eco-geographic
region were included in all appropriate regions (e.g., Colombia was assigned both to Andean
and to tropical South American regions) (Supplementary Table 2).
Crops whose primary areas of diversity encompassed more than one eco-geographic region
were listed in all appropriate regions (e.g., wheat was listed in Central Asia, West Asia, and the
South and East Mediterranean). Forty-two of the 53 crop commodities treated in food supplies
data, and 116 of the 132 crops in production data, were capable of being attributed to primary
regions of diversity, with the remaining general commodities which were not clearly
attributable to specific crop species listed as “not specified” (Supplementary Table 1).
242

Interdependence among countries on crop genetic diversity
We constructed circular plots displaying the relative importance of primary regions of diversity
as sources of crops comprising current (2009-2011 average) national food supplies and
production systems, using methods and code adapted from Abel and Sander (2014). For
recipient data, regional food supply values (kcal or g, /capita/day) were formed per variable by
deriving a weighted average across countries comprising each region, with national values
weighted by population. Regional production values were calculated by summing values across
countries for each variable.
We estimated the degree to which a country’s food supplies and production systems are
dependent upon crops of “foreign” primary regions of diversity by determining the extent to
which such supplies/systems are composed of crops whose primary regions of diversity do not
coincide with the regions within which that country is located. The method was initiated with
the assumption that the primary diversity of crops within a given country’s food
supplies/production systems was completely foreign (100% dependence). The percent
contribution of all crops whose primary diversity was identified as in the same region as the
country was then subtracted to estimate a “maximum dependence” metric per country. In this
metric, those general crop commodities whose regions could not be specified were assumed to
be of foreign primary regions of crop diversity. The sum of the percent contribution of these
non-specified general crop commodities was then subtracted, resulting in a “minimum
dependence” metric which assumes that all non-specified crop commodities possess primary
diversity within the same region as the country (modified from Flores-Palacios 1998).
Mean dependence in food supplies and production systems per country was estimated using an
interval censoring method, where the response variable (the calculated dependence value in
each country in each year) was bounded between the minimum and maximum dependence
estimates for each observation. A model of this type allows the uncertainty around an
observation to be incorporated into the parameter estimates for the parameter of interest. For
estimates of current dependence, we modeled the mean of the most recent three years (20092011). For estimates of change in dependence from 1961-2009/2011, intercepts and slopes per
country were modeled as random effects, where the mean hyper-parameter for the random
slopes represented the estimated slope (change in dependence over time) across all countries.
We allowed a correlation between country-level intercepts and slopes to account for the fact
that countries with high dependence have weaker dependence-time relationships than countries
243

Chapter 7
with low dependence (Gelman & Hill 2007). The interval-censored models were implemented
using a Bayesian framework in JAGS (v. 3.4.0) called from R (v.3.1.1), using the packages
rjags and R2jags. Non-informative (“flat”) priors were used for all coefficients. Convergence
was assessed using the Gelman-Rubin diagnostic (Gelman & Rubin 1992) and by visual
inspection of trace plots. Dependency values reported in the text represent the model-estimated
coefficient, ± the standard deviation. Credible intervals for each parameter are reported in
Supplementary Table 3-5.
We used Simpson’s diversity index (2009-2011 mean) to correlate the degree of contributing
crop diversity in current national food supplies/production systems with dependence on foreign
primary regions of diversity. The diversity-dependence relationship was modeled using a
simple linear model with both linear and quadratic terms. Diversity analyses were performed
using the vegan package in R (v. 3.1.1). We also correlated dependence with national Gross
Domestic Product (GDP) per capita purchasing power parity, using a mean GDP value across
2009-2011 for 169 available countries (World Bank 2014).
Crops were assigned importance individually for each food supplies and production systems
variable into 10% quantiles, from 1 (low importance) to 10 (high importance), based upon their
global aggregate (food supplies) and total global production values. A combined assessment
was performed on (136) unique crop commodities covered in food supplies and production
systems data (Supplementary Table 6). Thirty-seven of these commodities possessed both food
supplies and production systems data and were directly compared. An additional 92 crop
commodities with production systems values were embedded within 12 general commodities
in food supplies data (i.e., cereals, other; fruits, other; oilcrops, other; oranges & mandarines;
pulses, other; rape & mustard; roots, other; spices, other; sugar; tea; treenuts; and vegetables,
other). Food supplies values for most of the individual commodities were estimated by dividing
their total general commodity values equally across listed crops. For the sugar commodity,
sugarcane was assigned 70% and sugar beet 30% of the total value; for the tea commodity, tea
[Camellia sinensis (L.) Kuntze] was assigned 80%, mate 10%, and “not elsewhere specified”
(nes) tea 10%. Three additional production systems crop commodities (alfalfa, clover, and
vetches), which are livestock feed/forage crops and therefore are not recorded in food supplies
data, were assessed through quantile values derived solely from production systems variables.
Four general food supplies commodities (beverages, alcoholic; beverages, fermented;
244

Interdependence among countries on crop genetic diversity
miscellaneous; and sweeteners, other) were not recorded in production systems variables, thus
these commodities were assessed through quantile values derived solely from food supplies
variables. Coverage of each crop in the Multilateral System (MLS; i.e., Annex 1) of the
International Treaty on Plant Genetic Resources for Food and Agriculture (FAO 2002) was
assessed, listing crops as covered, partially covered (often in the case of general crop
commodities, in which some portion of the crops within the commodity are covered in the MLS
and others not), or not covered (Supplementary Table 6). The extent of geographic importance
of crops was additionally documented by counting the number of countries listing each
commodity (>0) for each variable, as well as listing the plant commodities by decreasing
importance until the total contribution equaled 90% of each country’s food supply/ production
for each variable, a threshold which is inclusive of major contributors to supply/ production
systems and exclusive of commodities contributing very small quantities (Prescott-Allen &
Prescott-Allen 1990, Khoury et al. 2014). The total count of countries including each crop
commodity as important was then derived per crop commodity.
RESULTS
Primary regions of genetic diversity of agricultural crops were identified across the tropics and
subtropics, extending into temperate regions in both hemispheres (Figure 1, Supplementary
Figure 2). The food supplies and production systems of countries worldwide were found to be
comprised of a wide range of crops from diverse geographic backgrounds, indicating a
thoroughly interconnected global food system in regard to the geographic origins of crop
genetic diversity (Figure 2, Supplementary Figure 3). The evident widespread importance in
global food supplies particularly of major crops such as wheat, rice, sugarcane, maize, soybean,
potatoes, barley, oil palm, beans, tomatoes, bananas & plantains, and sugarbeet, among others,
lead to particular significance of key primary regions of diversity, including West, Central,
South, Southeast, and East Asia, the South and East Mediterranean, West and Central Africa,
Central America and Mexico, Andean and tropical South America, and southern Europe (Figure
2, Supplementary Figure 3-4). Cassava, rape & mustard, groundnut, grapes, apples, alfalfa,
sorghum, and millets were among other crops of particular international importance for one or
more food supply and/or production variables.
Dependence upon crops from foreign primary regions of genetic diversity in national food
supplies and production systems was highest (i.e., up to 100%) in those countries
245

Chapter 7

Figure 1. Primary regions of diversity of major agricultural crops worldwide. See Supplementary Table
1 for a list of primary regions for all assessed crop commodities.

geographically isolated from and/or located at great distance from the primary regions of
diversity of major staple crops (Figure 3, Supplementary Figure 5, Supplementary Table 3).
This includes Australia and New Zealand, the Indian Ocean Islands, the Caribbean, southern
South America, North America, southern Africa, and northern Europe. These countries are
generally in temperate climates, although tropical islands and some continental tropical regions,
such as Central Africa, also demonstrated very high levels of dependence for most variables.
Conversely, dependence upon foreign crops was lowest in countries located within the primary
regions of diversity of major crops, and where traditional staples are still cultivated and
consumed, such as Southeast Asia, the South and East Mediterranean, South Asia, Central
Asia, West Asia, and West Africa. The lowest levels of dependence were found in countries
with food systems dominated by a limited number of traditional staples such as rice, wheat,
yams, sorghum, and millets (see Supplementary Figure 6). Island nations predominantly
dependent upon native crops for fat, such as coconut in the tropical Pacific Region, and
countries with extreme agroecological conditions limiting national production to the cultivation
of a select number of native crops (e.g., dates in the United Arab Emirates and other arid nations
of West Asia) also exhibited very low levels of dependence for relevant food supply
246

Interdependence among countries on crop genetic diversity

Figure 2. Circular plots indicating the importance of primary regions of diversity as sources of crops
comprising (A-B) calories in national food supplies, and (C-D) production quantity in national
production systems, averaged over 2009-2011. For recipient data, regional caloric food supply values
(kcal/capita/day) were formed by deriving a weighted average across countries comprising each region,
with national values weighted by population. Regional production quantity values were formed by
summing values across countries. For countries within regions, see Supplementary Table 2. Region
names are shortened in the figures; IOI = Indian Ocean Islands, ANZ = Australia and New Zealand, and
C. America = Central America and Mexico; and in production quantity only, Car = Caribbean, and Pac
= Tropical Pacific Region. The direction of the importance contribution is indicated by both the origin
region’s color and a gap between the connecting line and the destination region’s segment. Arrows
indicating direction of contribution are also included as examples. The magnitude of contribution is
indicated by the width of the connecting line. Because the line width is nonlinearly adapted to the
curvature, it corresponds to the contribution size only at the start and end points. Figures on the left (A,
C) display only the most significant contributions (i.e., 95th percentile) for visibility, which correspond
to an approximate contribution of at least 1180 kcal/cap/day (A), and 212 million tonnes (C). In these
figures, the importance of rice, wheat, maize, sugarcane, and oil palm are evident. Figures on the right
(B, D) display all contributions. See Supplementary Figure 3 for circular plots for all measured food
supply and production variables.
247

Chapter 7

Figure 3. Dependence on foreign primary regions of diversity of crops per country in regard to (A)
calories in national food supplies, and (B) production quantity in national production systems, as a
modeled mean between minimum and maximum dependence per country, 2009-2011. Dependence scale
is degree of dependence (1 = completely dependent). See Supplementary Figure 5 for world maps
displaying dependence per country for all measured food supply and production variables.

and/or production metrics. In such extreme cases, though, low dependence was generally
evident in only one or a few food supplies or production metrics, while other variables exhibited
much higher dependence.
Although food supplies and production systems variables were highly correlated in degree of
dependence (Supplementary Figure 7-8), variation was also visible across variables, with
highest overall dependence evident in fat, production value, production quantity, and food
weight (Supplementary Table 3). Considerable variation in dependence was found within
geographic areas, e.g., ranging from (mean ± SD) 48.2% ± 1.6 for calories in Mexico, to 86.5%
± 1.4 in Panama, within the Central America and Mexico region. Large variation in dependence
in production systems was also found within regions, e.g., ranging from 25.7% ± 3.5 for
production value in the Philippines, to 94.1% ± 1.1 for Malaysia, within the Southeast Asia
region. Countries with very high dependence for such production variables were exemplified
by the presence of extensive production systems dedicated to a limited number of high value,
foreign commodity crops, such as oil palm in Malaysia.
Dependence upon crops from foreign primary regions of genetic diversity was positively
correlated with diverse food supplies/production systems, although high dependence also
occurred in numerous countries with exhibiting low diversity (Supplementary Figure 6). Very
few countries, on the other hand, showed high diversity in their food supplies and/or production
systems and at the same time low dependence on crops of foreign primary regions of diversity.
National Gross Domestic Product was also associated with dependence, although with
considerable variation worldwide (Supplementary Figure 9).
248

Interdependence among countries on crop genetic diversity

Figure 4. Change in dependence on foreign primary regions of diversity of crops in (A) national food
supplies (1961-2009) and (B) production systems (1961-2011). Lines represent change over time in the
mean between minimum and maximum dependence for each country in each year for each variable as
predicted by a quadratic regression. Transparent ribbons represent modeled mean change across all
countries (± 95% credible interval) in dependence for each variable, estimated using a Bayesian model
with an interval censored response variable bounded between minimum and maximum dependence.

Mean dependence across all countries on crops of foreign primary regions of diversity in food
supplies was 65.8% ± 1.8 for calories, 66.6% ± 2.1 for protein, 73.7% ± 1.6 for fat, and 68.6%
± 1.4 for food weight. Mean dependence in production systems was 71.0% ± 1.8 for production
quantity, 64.0% ± 2.2 for harvested area, and 72.9% ± 1.9 for production value. The combined
mean dependence across food supply variables was estimated at 68.7%, across production
systems at 69.3%, and across food systems worldwide (i.e., both food supplies and production
systems, across all countries and all variables) at 68.9% (Supplementary Table 3).
National dependence upon crops of foreign primary regions of diversity increased significantly
as a global mean for all food supplies and production systems variables over the past half
century (Figure 4, Supplementary Table 4-5). Dependence in regard to calories increased from
62.7 to 67.4%, protein from 63.1 to 68.1%, fat from 63.4 to 73.2%, and food weight from 65.2
to 69.7% as measured in change in dependence from the mean of the first three years (19611963) to the last three years (2007-2009) per country, averaged across countries worldwide.
Likewise, dependence in regard to production quantity increased from 64.1 to 69.0%, harvested
249

Chapter 7
area from 59.0 to 62.1%, and production value from 64.4 to 70.5% between 1961 and 2011.
Countries with the greatest increases in dependence over the period were located in Africa,
West, South, Southeast, and East Asia, Central America and Mexico, and Andean and tropical
South America (Supplementary Figure 10). A number of countries with the largest changes in
dependence upon foreign crops in contribution to their food supplies were also those with major
transitions in their production systems during the past 50 years (e.g., the growth of oil palm
cultivation in Malaysia and Indonesia, a crop whose primary regions of diversity are located in
West and Central Africa and the Neotropics; and soybean in Brazil, a crop of East Asian origin).
Most regions also contained countries with decreases in dependence over the period. Growing
consumption of major staples within the native regions of these crops, such as soybean in China,
or wheat in West Asia, may be a factor in this decrease. Dependence in regard to fat in food
supplies increased the greatest degree over the past 50 years among all variables, a trend that is
concordant with significant changes in the contributing crop species composition of national
food supplies globally over this period (Khoury et al. 2014).
DISCUSSION
The food supplies and production systems of countries worldwide are primarily composed of
crops that were initially domesticated and largely diversified elsewhere around the world. While
geography and climate constrained for some period following the agricultural revolutions ca.
10,000 BP the availability of crops to their regions, growth in migration, colonialism, and trade,
among other historical forces (Diamond 2004), increased the availability of crops beyond their
primary regions, and current economic and agricultural development, urbanization, and
globalization trends are making important food crops comprehensively available worldwide
(Khoury et al. 2014). Even countries located within the most ancient and richest primary regions
of diversity, e.g., West Asia, now exhibit considerable dependence on foreign crops in their
food supplies and production systems.
The range of crops covered in this analysis is not fully inclusive of all foodstuffs produced and
consumed in national food systems. Therefore, an underestimation and/or overgeneralization
of diversity is assumed, particularly in regard to plants primarily encountered in home gardens
and local markets, seasonally important foods, and culinary herbs, spices and other crops
consumed in relatively small quantities (Prescott-Allen & Prescott-Allen 1990). Some of these
250

Interdependence among countries on crop genetic diversity
crops may be important to diets, particularly in their contribution of micronutrients (Doughty
1979). The aggregation of some crop commodities, the generality of the defined eco-geographic
regions, uncertainty for some crops as to their primary regions of diversity, and the subjective
nature of the boundaries of such regions, also lead to a degree of uncertainty in dependence
metrics. Acknowledging these limitations, the results are a very strong indication of the extent
of globalization of food systems and the resulting interdependence among nations on crop
genetic resources.
Complementary approaches to the current study have been taken to assess the degree of
interdependence among countries in provisioning the global food system. Analyses of
production versus consumption in nations globally have revealed increasing interdependence
in regard to trade in food (Fader et al. 2013, Porkka et al. 2013, D’Odorico et al. 2014,
MacDonald et al. 2015). Investigations of exchange of crop genetic resources in the form of
accession requests from international genebanks have also shown expanding transfers of
germplasm (Kloppenburg & Kleinman 1987, Dudnik et al. 2001, Smale & Day Rubenstein
2002, Day Rubenstein & Smale 2004, Fowler & Hodgkin 2004). The increasingly global
contribution of diverse breeding materials to the development of modern crop cultivars has also
been documented for major crops (Smale 1996, Gollin 1998, Brennan et al. 1999, Zhou et al.
2000, Cassaday et al. 2001, Fowler et al. 2001, Smale et al. 2002, Johnston et al. 2003). Such
studies bolster evidence of increased interdependence among nations in concert with greater
globalization of the crops and crop varieties providing our global food supplies.
The importance of continued access to diverse crop genetic resources through international
exchange in support of national production, and its corollary impact on national economies, is
unequivocal. Yet access to genetic diversity in important crops by major producers, wherever
their location, is equally critical for the reliable provisioning of global food supplies via
international trade, especially as countries have transitioned from food insecurity to trade
dependence (Fader et al. 2013, Porkka et al. 2013, D’Odorico et al. 2014, MacDonald et al.
2015). Production of the major crops is unevenly distributed across countries and for many
crops now generally occurs outside of the primary regions of diversity of those crops, e.g.,
China, India, the USA, the Russian Federation, France and Canada for wheat; the USA, China,
Germany, France, Brazil, and Argentina for maize; the USA, Brazil, Argentina and India for
soybean; and China, India, the Russian Federation, Ukraine, and the USA for potatoes.
251

Chapter 7
The evidently very high levels of interdependence among countries in regard to crop genetic
diversity supports the rationale for internationally coordinated mechanisms to facilitate access
to these resources worldwide, such as the Multilateral System (MLS) created within the
International Treaty on Plant Genetic Resources for Food and Agriculture (Plant Treaty) (FAO
2002). This interdependence also reinforces the importance of the international genebank
collections safeguarded as global public goods by the CGIAR international agricultural research
centres (Hoisington et al. 1999), which are covered under Article 15 of the Plant Treaty. While
the long-term sustainability of funding for these collections has partially been achieved (Global
Crop Diversity Trust 2013), an increased level of international support will be needed to secure
their role in conserving and distributing the genetic resources of their mandate crops. Moreover,
large gaps remain in the conservation of crop diversity not covered by CGIAR collections (FAO
2010, Khoury et al. 2010). The window of opportunity for securing the world’s agricultural
diversity threatened in situ and in under-funded genebanks will not remain open indefinitely
(Wilkes et al. 2007, FAO 2010).
A comprehensive MLS should engender facilitated access to the genetic resources of all crops
of present and future international importance, but current access is suboptimal (Bjørnstad et
al. 2013). As food systems undergo further transition due both to dietary change (Kearney 2010,
Khoury et al. 2014) and to novel production challenges (Lobell et al. 2008, Cordell et al. 2009,
Jarvis et al. 2009), a broadly inclusive and adaptable effort to conserve and provide access to
agricultural diversity internationally is prudent.
The Plant Treaty MLS has focused mostly on cereal, pulse, starchy root, and forage crops (listed
in Annex 1 of the Treaty) (FAO 2002), thus oil crops, vegetables and fruits are not well covered.
As much as 28.7% of global aggregate calories in food supplies, 19.0% of protein, 61.0% of
fat, 43.4% of food weight; and 41.0% of total global production quantity, 27.0% of harvested
area, and 41.2% of global production value are comprised of crops not covered by the MLS.
These include soybean, oil palm, sugarcane, groundnut, tomatoes, onions, grapes, coffee, cocoa
beans, and a variety of other vegetable, nut and fruit crops (Supplementary Figure 11,
Supplementary Table 6). As limited genetic resources research on these crops has been carried
out by CGIAR (Khoury & Jarvis 2014), their diversity is likewise underrepresented in the
international collections. These crops therefore constitute major gaps in coordinated efforts
among nations to conserve and make accessible valuable genetic resources. Furthermore, as
252

Interdependence among countries on crop genetic diversity
greater variation in genetic materials is increasingly needed for future breeding of major crops
(Hoisington et al. 1999, Gepts 2006, Burke et al. 2009, Jarvis et al. 2009, McCouch et al. 2013),
the expansion of MLS coverage to include wild relatives of staples such as maize and cassava
will be important to the strengthening of a global system for the conservation and exchange of
crop genetic diversity.

253

Chapter 7
SUPPLEMENTARY INFORMATION

Supplementary Figure 1. Eco-geographic regions utilized in the analysis. See Supplementary Table 2
for a list of countries per region.

254

Supplementary Figure 2. Richness maps of primary regions of diversity of crop commodities. (A) Primary regions of diversity of 42 crop commodities
assessed in food supplies; and (B) Primary regions of diversity of 116 crop commodities assessed in production systems. Scale is number of crops;
darker colors represent regions where numerous primary regions of diversity of crops overlap.

Interdependence among countries on crop genetic diversity

255

Supplementary Figure 3A. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (A) calories in national
food supplies, averaged over 2009-2011. For recipient data, regional food supply values (kcal/capita/day) were formed per variable by deriving a
weighted average across countries comprising each region, with national values weighted by population. For countries within regions, see Supplementary
Table 2. Region names are shortened in the figures; IOI = Indian Ocean Islands, ANZ = Australia and New Zealand, and C. America = Central America
and Mexico. The direction of the importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the
destination region’s segment. The magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly
adapted to the curvature, it corresponds to the contribution size only at the start and end points. Figures on the left (1) display only the most significant
contributions (i.e., 95th percentile) for visibility; figures on the right (2) display all contributions.

Chapter 7

256

Supplementary Figure 3B. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (B) protein in national
food supplies, averaged over 2009-2011. For recipient data, regional food supply values (g/capita/day) were formed per variable by deriving a weighted
average across countries comprising each region, with national values weighted by population. For countries within regions, see Supplementary Table
2. Region names are shortened in the figures; IOI = Indian Ocean Islands, ANZ = Australia and New Zealand, and C. America = Central America and
Mexico. The direction of the importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the
destination region’s segment. The magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly
adapted to the curvature, it corresponds to the contribution size only at the start and end points. Figures on the left (1) display only the most significant
contributions (i.e., 95th percentile) for visibility; figures on the right (2) display all contributions.

Interdependence among countries on crop genetic diversity

257

Supplementary Figure 3C. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (C) fat in national
food supplies, averaged over 2009-2011. For recipient data, regional food supply values (g/capita/day) were formed per variable by deriving a weighted
average across countries comprising each region, with national values weighted by population. For countries within regions, see Supplementary Table
2. Region names are shortened in the figures; IOI = Indian Ocean Islands, ANZ = Australia and New Zealand, and C. America = Central America and
Mexico. The direction of the importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the
destination region’s segment. The magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly
adapted to the curvature, it corresponds to the contribution size only at the start and end points. Figures on the left (1) display only the most significant
contributions (i.e., 95th percentile) for visibility; figures on the right (2) display all contributions.

Chapter 7

258

Supplementary Figure 3D. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (D) food weight in
national food supplies, averaged over 2009-2011. For recipient data, regional food supply values (g/capita/day) were formed per variable by deriving a
weighted average across countries comprising each region, with national values weighted by population. For countries within regions, see Supplementary
Table 2. Region names are shortened in the figures; IOI = Indian Ocean Islands, ANZ = Australia and New Zealand, and C. America = Central America
and Mexico. The direction of the importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the
destination region’s segment. The magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly
adapted to the curvature, it corresponds to the contribution size only at the start and end points. Figures on the left (1) display only the most significant
contributions (i.e., 95th percentile) for visibility; figures on the right (2) display all contributions.

Interdependence among countries on crop genetic diversity

259

Supplementary Figure 3E. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (E) production quantity
in national production systems, averaged over 2009-2011. For recipient data, regional production values were formed by summing values across
countries for each variable. For countries within regions, see Supplementary Table 2. Region names are shortened in the figures; IOI = Indian Ocean
Islands, ANZ = Australia and New Zealand, C. America = Central America and Mexico, Car = Caribbean, and Pac = Tropical Pacific Region. The
direction of the importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the destination region’s
segment. The magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly adapted to the curvature,
it corresponds to the contribution size only at the start and end points. Figures on the left (1) display only the most significant contributions (i.e., 95th
percentile) for visibility; figures on the right (2) display all contributions.

Chapter 7

260

Supplementary Figure 3F. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (F) harvested area in
national production systems, averaged over 2009-2011. For recipient data, regional production values were formed by summing values across countries
for each variable. For countries within regions, see Supplementary Table 2. Region names are shortened in the figures; IOI = Indian Ocean Islands, ANZ
= Australia and New Zealand, C. America = Central America and Mexico, Car = Caribbean, and Pac = Tropical Pacific Region. The direction of the
importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the destination region’s segment. The
magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly adapted to the curvature, it corresponds
to the contribution size only at the start and end points. Figures on the left (1) display only the most significant contributions (i.e., 95th percentile) for
visibility; figures on the right (2) display all contributions.

Interdependence among countries on crop genetic diversity

261

Supplementary Figure 3G. Circular plots indicating the importance of primary regions of diversity as sources of crops comprising (G) production value
in national production systems, averaged over 2009-2011. For recipient data, regional production values were formed by summing values across
countries for each variable. For countries within regions, see Supplementary Table 2. Region names are shortened in the figures; IOI = Indian Ocean
Islands, ANZ = Australia and New Zealand, C. America = Central America and Mexico, Car = Caribbean, and Pac = Tropical Pacific Region. The
direction of the importance contribution is indicated by both the origin region’s color and a gap between the connecting line and the destination region’s
segment. The magnitude of contribution is indicated by the width of the connecting line. Because the line width is nonlinearly adapted to the curvature,
it corresponds to the contribution size only at the start and end points. Figures on the left (1) display only the most significant contributions (i.e., 95th
percentile) for visibility; figures on the right (2) display all contributions.

Chapter 7

262

Interdependence among countries on crop genetic diversity

Supplementary Figure 4. Importance of primary regions of diversity of crops in contribution to global
aggregate food supplies [(A) calories, (B) protein, (C) fat, and (D) food weight] and total global
production systems [(E) production quantity, (F) harvested area, and (G) production value], averaged
over 2009-2011.

263

Chapter 7

Supplementary Figure 5. Dependence on foreign primary regions of diversity of crops per country in
national food supplies [(A) calories, (B) protein, (C) fat, and (D) food weight] and production systems
[(E) production quantity, (F) harvested area, and (G) production value] as a modeled mean between
minimum and maximum dependence per country, 2009-2011. Dependence scale is degree of
dependence (1 = completely dependent).

264

Supplementary Figure 6. Correlation within countries between national (A) food supply or (B) production system diversity, and dependence on foreign
primary regions of crop diversity as a modeled mean between minimum and maximum dependence per country, 2009-2011. Dependence scale is degree
of dependence (1 = completely dependent). Simpson’s Index scale is degree of diversity in food supplies/production systems (1 = highly diverse). Shaded
areas around each line represent the 95% confidence interval of the predicted relationship between mean diversity and mean dependence.

Interdependence among countries on crop genetic diversity

265

Supplementary Figure 7. Correlation within countries in regard to dependence on foreign primary regions of diversity in (A) food supplies variables,
and (B) production systems variables as a modeled mean between minimum and maximum dependence per country, 2009-2011. Dependence scale is
degree of dependence (1 = completely dependent).

Chapter 7

266

Supplementary Figure 8. Correlation within countries in regard to dependence on foreign primary regions of diversity between food supplies and
production systems [(A) production quantity, (B) harvested area, and (C) production value] variables, as a modeled mean between minimum and
maximum dependence per country, 2009-2011. Dependence scale is degree of dependence (1 = completely dependent).

Interdependence among countries on crop genetic diversity

267

Supplementary Figure 9. Correlation within countries between per capita gross domestic product (GDP) and national (A) food supply or (B) production
systems dependence on foreign primary regions of crop diversity as a modeled mean between minimum and maximum dependence per country, 20092011. Dependence scale is degree of dependence (1 = completely dependent). Shaded areas around each line represent the 95% confidence interval of
the predicted relationship between mean GDP and mean dependence.

Chapter 7

268

Interdependence among countries on crop genetic diversity

Supplementary Figure 10. Slope of change in dependence from 1961-2009 on foreign primary regions
of diversity in regard to national food supplies [(A) calories, (B) protein, (C) fat, and (D) food weight]
and from 1961-2011 for national production systems [(E) production quantity, (F) harvested area, and
(G) production value], measured as change in the modeled mean value between minimum and maximum
dependence for each country in each year for each variable.

269

Chapter 7

Supplementary Figure 11. Relevance of crop commodities in global food supplies and production
systems worldwide, and their coverage in Annex 1 of the International Treaty on Plant Genetic
Resources for Food and Agriculture (FAO 2002). Crops were assigned importance individually for each
variable into 10% quantiles, from 1 (low importance) to 10 (high), based upon their global aggregate
(food supplies) and total global production values. The center of the web is equivalent to 1; the outside
of the web to 10 in importance. All specific crops are displayed; 20 general commodities are not shown
here (see Supplementary Table 6 for all crops). Ani-Fen-Cor denotes Anise, Fennel and Coriander
treated together; Mango Guav denotes Mango and Guava; and Nut Card denotes Nutmeg and
Cardamoms. Blue outlines identify crop commodities covered in Annex 1, while red are not covered.

270

Interdependence among countries on crop genetic diversity
The following supplementary tables contain extensive information on dependence metrics per
country and importance metrics per crop. These multi-sheet tables are too large to be
reproduced in the supplementary information of this chapter. They are available in the
“Interdependence among countries on crop genetic resources” project on Figshare (permanent
links below).
Supplementary Table 1. Crop commodities and eco-geographic regions assessed in food supplies and
production systems analyses.
permanent link: figshare.com/s/667b2840f33111e4a89b06ec4bbcf141
Supplementary Table 2. Countries and eco-geographic regions assessed in food supplies and production
systems analyses.
permanent link: figshare.com/s/87d4197cf4d311e4971b06ec4bbcf141
Supplementary Table 3. Estimated percent dependence on foreign primary regions of diversity in
national food supplies and production systems. Data includes the raw mean minimum and maximum
dependence values across 2009-2011 per country, and the mean value between minimum and maximum
per country across these years, as well as modeled mean values and variation metrics as estimated in a
Bayesian framework using an interval-censored response variable bounded between minimum and
maximum dependence estimates.
permanent link: figshare.com/s/f78dad72f33011e4b9f606ec4b8d1f61
Supplementary Table 4. Change in dependence on foreign primary regions of diversity in national food
supplies, 1961-2009. Data includes minimum and maximum dependence values for each variable for
each country in each year, as well as slopes of change and variation metrics over the time period as
estimated in a Bayesian framework using an interval-censored response variable bounded between
minimum and maximum dependence estimates. Year was centered at 1985 for modeling purposes, thus
model intercepts represent mean dependence in this year.
permanent link: figshare.com/s/2e0455fef33111e4898306ec4bbcf141
Supplementary Table 5. Change in dependence on foreign primary regions of diversity in national
production systems, 1961-2011. Data includes minimum and maximum dependence values for each
variable for each country in each year, as well as slopes of change and variation metrics over the time
period as estimated in a Bayesian framework using an interval-censored response variable bounded
between minimum and maximum dependence estimates. Year was centered at 1985 for modeling
purposes, thus model intercepts represent mean dependence in this year.
permanent link: figshare.com/s/4134f534f33111e4a89b06ec4bbcf141
Supplementary Table 6. Relevance of crop commodities in national food supplies and production
systems, 2009-2011. Importance by quantiles: crops were assigned importance individually for each
variable into 10% quantiles, from 1 (low importance) to 10 (high), based upon their global aggregate
(food supplies) and total global production values. Data also includes the number of countries counting
the crop within food supplies/production systems (>0) for each variable, as well as the number of these
countries in which the crop is important (i.e. within the top 90% of food supply/production system
variable). Within these importance counts, the number of foreign (i.e., countries whose location do not
overlap with the primary regions of diversity of the crop) countries and native countries are listed. In
addition, details on coverage in Annex 1 of the International Treaty on Plant Genetic Resources for Food
and Agriculture (FAO 2002) are provided for each crop.
permanent link: figshare.com/s/5029166af33111e48d2906ec4bbcf141

271

Chapter 7
REFERENCES
Abel, G.J., Sander, N. (2014). Quantifying global international migration flows. Science 343, 15201522.
Bjørnstad, Å., Tekle, S., Göransson, M. (2013). ‘Facilitated access’ to plant genetic resources: does it
work? Gen. Res. Crop Evol. 60, 1959-1965.
Brennan, J.P., Godden, D., Smale, M., Meng, E. (1999). Breeder demand for and utilization of wheat
genetic resources in Australia. Plant Varieties and Seeds 12, 113-127.
Burke, M.B., Lobell, D.B., Guarino, L. (2009). Shifts in African crop climates by 2050, and the
implications for crop improvement and genetic resources conservation. Global Environ. Chang. 19
(3), 317-325.
Cassaday, K., Smale, M., Fowler, C., Heisey, P.W. (2001). Benefits from giving and receiving genetic
resources: the case of wheat. Plant Gen. Res. News. 127, 1-10.
Cordell, D., Drangert, J-O., White, S. (2009). The story of phosphorus: global food security and food
for thought. Global Environ. Chang. 19 (2), 292-305.
D’Odorico, P., Carr, J.A., Laio, F., Ridolfi, L., Vandoni, S. (2014). Feeding humanity through global
food trade. Earth’s Future 2, 458-469.
Day Rubenstein, K., Smale, M. (2004). International Exchange of Genetic Resources, The Role of
Information and Implications for Ownership: The Case of the U.S. National Plant Germplasm
System. EPTD Discussion Paper no. 119 [Washington D.C.: Environment and Production
Technology Division, International Food Policy Research Institute (IFPRI)].
Diamond, J. (2004). The wealth of nations. Nature 429, 616-617.
Doughty, J. (1979). Dangers of reducing the range of food choice in developing countries. Ecol. Food
Nutr. 8, 275-283.
Dudnik, N.S., Thormann, I., Hodgkin, T. (2001). The extent of use of plant genetic resources in research
- a literature survey. Crop Sci. 41, 6-10.
Engels, J., Hodgkin, T., Thormann, I., Robinson, J., Fowler, C. (2001). Crops Proposed for the
Multilateral System: Centres of Diversity, Locations of Ex Situ Collections, and Major Producing
Countries. Background Study Paper No. 12 (CGRFA/CG-4/00/Inf.4) of the Fourth Inter-Sessional
Meeting of the Contact Group of the CGRFA Secretariat (Rome: FAO).
Fader, M., Gerten, D., Krause, M., Lucht, W., Cramer, W. (2013). Spatial decoupling of agricultural
production and consumption: quantifying dependences of countries on food imports due to domestic
land and water constraints. Environ. Res. Lett. 8 (1), 014046.
FAO (Food and Agriculture Organization of the United Nations). (2015). FAOSTAT (Rome: FAO).
http://faostat3.fao.org/.
FAO (Food and Agriculture Organization of the United Nations). (2002). The International Treaty on
Plant Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2010). Second Report on the State of
the World’s Plant Genetic Resources for Food and Agriculture (Rome: FAO).
Flores-Palacios, X. (1998). Contribution to the Estimation of Countries’ Interdependence in the Area of
Plant Genetic Resources. Commission on Genetic Resources for Food and Agriculture, Background
Study Paper No. 7, Rev. 1 (Rome: FAO).
Fowler, C., Hodgkin,T, (2004). Plant genetic resources for food and agriculture: assessing global
availability. Annu. Rev. Env. Resour. 29 (1), 143-179.
Fowler, C., Smale, M., Gaiji, S. (2001). Unequal exchange? Recent transfers of agricultural resources
and their implications for developing countries. Devel. Pol. Rev. 19, 181-204.
Gelman, A., Hill, J. (2007). Data Analysis using Regression and Multilevel/Hierarchical Models (New
272

Interdependence among countries on crop genetic diversity
York: Cambridge University Press).
Gelman, A., Rubin, D.B. (1992). Inference from iterative simulation using multiple sequences.
Statistical Sci. 7, 457-472.
Gepts, P. (2006). Plant genetic resources conservation and utilization: the accomplishments and future
of a societal insurance policy. Crop Sci. 46 (5), 2278-2292.
Global Crop Diversity Trust. (2013). Fundraising Strategy 2014-2018 (Bonn: Global Crop Diversity
Trust).
Gollin, D. (1998). “Valuing farmers’ rights”, in Agricultural Values of Plant Genetic Resources, eds.
R.E. Evenson, D. Gollin, V. Santaniello (CAB International, Wallingford, UK), pp. 233–245.
GRIN (Germplasm Resources Information Network). (2014). USDA, ARS, National Genetic Resources
Program, Beltsville. http://www.ars-grin.gov/cgi-bin/npgs/html/tax_search.pl.
Guarino, L., Lobell, D.B.A. (2011). A walk on the wild side. Nat. Climate Chang. 1. 374-375.
Harlan, J.R. (1951). Anatomy of Gene Centers. American Naturalist 8.
Harlan, J.R. (1971). Agricultural origins: centres and noncentres. Science 174, 468-474.
Harlan, J.R. (1975). Crops and Man (Madison: American Society of Agronomy).
Hoisington, D., Khairallah, M., Reeves, T., Ribout, J.-M., Skovmand, B., Taba, S., et al. (1999). Plant
genetic resources: what can they contribute toward increased crop productivity? Proc. Natl. Acad.
Sci. 96, 5937-5943.
Jarvis, A., Ramirez-Villegas, J., Hanson, J., Leibing, C. (2009). “Crop and forage genetic resources:
international interdependence in the face of climate change”, in: The Impact of Climate Change on
Countries’ Interdependence on Genetic Resources for Food and Agriculture. Background Study
Paper No.48, eds. S. Fujisaka, D. Williams, M. Halewood (Rome: FAO, Commission on Genetic
Resources for Food and Agriculture).
Johnson, N.L., Pachico, D., Voysest, O. (2003). The distribution of benefits from public international
germplasm banks: the case of beans in Latin America. Agric. Econ. 29, 277-286.
Kastner, T., Rivas, M.J.I., Koch, W., Nonhebel, S. (2012). Global changes in diets and the consequences
for land requirements for food. Proc. Natl. Acad. Sci. 109, 6868-6872.
Kearney, J. (2010). Food consumption trends and drivers. Philos. T. R. Soc. B. 365 (1554), 2793-2807.
Khoury, C.K., Bjorkman, A.D., Dempewolf, H., Ramírez-Villegas, J., Guarino, L., Jarvis, A., et al.
(2014). Increasing homogeneity in global food supplies and the implications for food security. Proc.
Natl. Acad. Sci. 111 (11), 4001-4006.
Khoury, C.K., Jarvis, A. (2014). The Changing Composition of the Global Diet: Implications for CGIAR
Research. CIAT Policy Brief No. 18. (Cali: Centro Internacional de Agricultura Tropical), 6 pp.
Khoury, C., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic resources:
a review of global crop and regional conservation strategies. Gen. Res. Crop Evol., 57, 625-639.
Kloppenburg, J.Jr., Kleinman, D.L. (1987). Analyzing empirically the distribution of the world’s plant
genetic resources: the plant germplasm controversy. Bioscience 37, 190-198.
Lobell, D.B., Burke, M.B., Tebaldi, C., Mastrandrea, M.D., Falcon, W.P., Naylor, R.L. (2008).
Prioritizing climate change adaptation needs for food security in 2030. Science 319 (5863), 607-610.
MacDonald, G.K., Brauman, K.A., Sun, S., Carlson, K.M., Cassidy, E.S., Gerber, J.S., et al. (2015).
Rethinking agricultural trade relationships in an era of globalization. Bioscience 65, 275-289.
McCouch, S., Baute, G.J., Bradeen, J., Bramel, P., Bretting, P.K., Buckler, E., et al. (2013). Agriculture:
feeding the future. Nature 499 (7456), 23-24.
Nuijten, E., van Treuren, R., Struik, P.C., Mokuwa, A., Okry, F., Teeken, B., et al. (2009). Evidence for
the emergence of new rice types of interspecific hybrid origin in West African farmers’ fields. PLoS
One 4, e7335.
Porkka, M., Kummu, M., Siebert, S., Varis, O. (2013). From food insufficiency towards trade
273

Chapter 7
dependency: a historical analysis of global food availability. PLoS One 8, e82714.
Prescott-Allen, R., Prescott-Allen, C. (1990). How many plants feed the world? Conserv. Biol. 4 (4),
365-374.
Prota (Plant Resources of Tropical Africa). (2014). Prota4u online database. http://www.prota4u.info/.
Smale, M. (1996). Understanding Global Trends in the Use of Wheat Diversity and International Flows
of Wheat Genetic Resources. Economics Working Paper, 96 (2). [International Maize and Wheat
Improvement Center (CIMMYT)].
Smale, M., Day Rubenstein, K. (2002). The demand for crop genetic resources: international use of the
U.S. National Plant Germplasm System. World Dev. 30, 1639-1655.
Smale, M., Reynolds, M.P., Warburton, M., Skovmand, B., Trethowan, R.M., Singh, R.P., et al. (2002).
Dimensions of diversity in modern spring bread wheat in developing countries from 1965. Crop Sci.
42, 1766–1779.
Thormann, I., Fiorino, E., Halewood, M., Engels, J.M.M. (2015). Plant genetic resources collections
and associated information as a baseline resource for genetic diversity studies: an assessment of the
IBPGR-supported collections. Genet. Resour. Crop Evol. doi 10.1007/s10722-015-0231-9.
van de Wouw, M., Kik, C., van Hintum, T., van Treuren, R., Visser, B. (2009). Genetic erosion in crops:
concept, research results and challenges. Plant Genetic Resources 8, 1-15.
van de Wouw, M., van Hintum, T., Kik, C., van Treuren, R., Visser, B. (2010). Genetic diversity trends
in twentieth century crop cultivars: a meta analysis. Theor. Appl. Genet. 120 (6), 1241-1252.
Vavilov, N.I. (1926). Tzentry proiskhozhdeniya kulturnykh rastenii (The centers of origin of cultivated
plants). Works of Applied Botany and Plant Breeding 16, 248pp.
Vavilov, N.I. (1951). The origin, variation, immunity and breeding of cultivated plants (Translated by
K. Start). Cron. Bot. 13, 1-366.
Vavilov, N.I. (1992). Origin and Geography of Cultivated Plants (Translated by Doris Löve)
(Cambridge: Cambridge University Press).
Vincent, H., Wiersema, J., Kell, S., Fielder, H., Dobbie, S., Castañeda-Álvarez, N.P., et al. (2013). A
prioritized crop wild relative inventory to help underpin global food security. Biol. Conserv. 167,
265-275.
Wilkes, G. (2007). Urgent notice to all maize researchers: disappearance and extinction of the last wild
teosinte population is more than half completed. A modest proposal for teosinte evolution and
conservation in situ: the Balsas, Guerrero, Mexico. Maydica 52, 49-58.
World Bank. (2014). International Comparison Program database. http://data.worldbank.org/.
Xiao, J., Grandillo, S., Ahn, S.N., McCouch, S.R., Tanksley, S.D. (1996). Genes from wild rice improve
yield. Nature 384, 223-224.
Zeven, A.C., De Wet, J.M.J. (1982). Dictionary of Cultivated Plants and Their Regions of Diversity:
Excluding Most Ornamentals, Forest Trees and Lower Plants (Wageningen: CAPD).
Zeven, A.C., Zhukovsky, P. (1975). Dictionary of Cultivated Plants and Their Centres of Diversity:
Excluding Most Ornamentals, Forest Trees and Lower Plants (Wageningen: CAPD).
Zhou, X., Carter, T.E.Jr., Cui, Z., Miyazaki, S., Burton, J.W. (2000). Genetic base of Japanese soybean
cultivars released during 1950 to 1988. Crop Sci. 40, 1794-1802.
Zhu, Y., Chen, H., Fan, J., Wang, Y., Li, Y., Chen, J., et al. (2000). Genetic diversity and disease control
in rice. Nature 406 (6797), 718-722.
Zhukovsky, P.M. (1965). Main gene centres of cultivated plants and their wild relatives within the
territory of the U.S.S.R. Euphytica 14, 177-188.
Zhukovsky, P.M. (1968). New centres of origin and new gene centres of cultivated plants including
specifically endemic microcentres of species closely allied to cultivated species. Bot. J. (Russian Bot
Z.) 53, 430-460.
274

CHAPTER 8
General discussion
DISCUSSION OF RESULTS AND IMPLICATIONS
This thesis aimed to contribute to the knowledge required to answer a set of key questions
regarding the need for, potential of, challenges and constraints regarding, and necessary steps
to enhance the conservation and use of crop genetic diversity. A discussion of the results and
their implications is presented in the following subsections.
The status of diversity and associated vulnerabilities in the global food system
In the past half century, very substantial changes have occurred in human diets worldwide and
in the production systems that sustain them. National food supplies around the world have
become increasingly similar (Figure 1), gaining in calories, protein, and fat, as animal-derived
foods and high-calorie plant foods (oils and sugars) have risen in importance. The proportion
of diets consisting of major cereals, sugar crops and oil crops has increased, while regionally
and locally important cereals, root crops, and oil crops have generally become further
marginalized [Khoury et al. 2014 (Chapter 2)].
These changes have been driven by globalization, urbanization, and economic development,
including agricultural research (Khoury & Jarvis 2014). While this nutrition transition has
enhanced food security by making macronutrients more readily available worldwide, it has had
mixed effects on micronutrient sufficiency and the over-consumption of macronutrients has
contributed to a global surge in diet-related non-communicable diseases (Popkin 2006, Pingali
2007, Kearney 2010). Dietary change is also linked with greater homogeneity in farmers’ fields
and the associated commodity trading systems, thus heightening concerns about genetic
vulnerability to biotic and abiotic stresses as well as food system vulnerability to climatic and
political instability (Chapter 2).
What can be done to mitigate the vulnerabilities created by placing our proverbial eggs in one
basket - an increasingly interconnected global food system, highly contingent upon trade in a
handful of selected food crop commodities, highly dependent on the extensive application of

Chapter 8

Figure 1. Increase in similarity (homogeneity) in national food supplies, 1961-2009. The figure depicts
a multivariate ordination of crop commodity composition in contribution to calories in national food
supplies in 1961, 1985 and 2009. Blue points represent the multivariate commodity composition of each
country in 1961, yellow points in 1985, and red points in 2009. Circles represent 95% confidence
intervals around the centroid in each year. Both x- and y-axis have been deliberately removed for
visualization purposes. See Khoury et al. (2014) (Chapter 2) for further details.

renewable and non-renewable resources, and creating major environmental impacts?
A comprehensive view of food and nutrition security encompasses consistent availability and
access to adequate, culturally acceptable and nutritious food as well as empowerment of
consumers to use this food for improved health. Over the long term, food security also requires
actions to mitigate the negative ecological effects of food systems and adapt agriculture to
276

General discussion
climatic change as well as to natural resource limitations. Achieving long-term food security
thus requires a holistic approach to agricultural development. From a crop diversity perspective,
such an approach aims to build food systems in which diverse varieties of a broad range of
crops can flourish in terms of production, markets, and consumption.
Key steps that can help to mitigate global food system vulnerabilities, increase agricultural
productivity and sustainability, and enhance human nutrition are proposed below. These actions
are divided into sections devoted to major staple crops, increasingly dominant oil crops, and
neglected or marginalized crops, with particular focus on the role of crop genetic resources in
enhancing food and nutrition security.
As wheat, rice, maize, and other critical staples gain importance in the global food system
(Figure 2), their production stability, nutritional quality, and environmental impacts become
ever more critical issues. Improving these crops in the face of land, water, and other resource
limitations, climate change, and agriculture’s increasing pressure on ecosystems, will
necessitate a range of research and political actions, including (a) conserving and describing
the genetic variation within these crops and their wild relatives and making diverse genetic
material and associated information available to researchers and breeders, (b) facilitating
sharing of diversity through the International Treaty on Plant Genetic Resources for Food and
Agriculture (Plant Treaty) and aligned political efforts, (c) breeding more productive and
resilient varieties, with emphasis on enhancing nutritional quality, stress tolerance, and resource
use efficiency, (d) promoting widespread adoption of genetically diverse, locally adapted
varieties to minimize vulnerability associated with genetic uniformity, and (e) developing more
ecologically efficient agronomic, storage, processing and distribution practices in order to
reduce their negative environmental impacts (Khoury & Jarvis 2014).
Challenges related to the production of oil crops such as soybean and oil palm that have become
an increasingly important part of global food supplies over the past 50 years (Figures 2 and 3)
include major impacts on natural ecosystems through deforestation, greenhouse gas emissions,
and vulnerabilities due to genetic uniformity in crop varieties. Key objectives for production
systems focused on oil production should be on reducing these ecological impacts, improving
species diversity at the system level (i.e., diversify crops) and enhancing genetic diversity at
the varietal level. Steps that can be taken to address these challenges include:
277

Chapter 8

Figure 2. Contribution of crop commodities to mean food supplies in developing countries for calories
(kcal/capita/day), 1969 and 2009 [Khoury & Jarvis 2014, based on Khoury et al. 2014 (Chapter 2)].

(a) developing and promoting ecologically efficient production and processing methods, (b)
broadening the genetic diversity within major varieties, and (c) diversifying oil crop production,
processing, and markets by promoting less globally dominant oil crops such as coconut,
cottonseed, groundnut, olive, rape and mustardseed, sesame, shea nut, and sunflower.
278

General discussion

Figure 3. Median percent change in the relative contribution to calories (kcal/capita/day), from assorted
crops in national food supplies in developing countries, 1969-2009 [Khoury & Jarvis 2014, based on
Khoury et al. 2014 (Chapter 2)].

A number of cereal, root, and oil crop species have generally declined in terms of their relative
contribution to national and global food supplies in the past half century (Figure 3), although
they may remain significant particularly for rural communities in some developing regions.
These include sorghum, cassava, millets, sweetpotato, coconut, yams, and grain legumes,
among others. Agricultural research funding for these crops is minimal compared with funding
279

Chapter 8
for the major cereals (Khoury & Jarvis 2014). Since many of these plants are both stress tolerant
and nutritionally rich, investment in their conservation, improvement, and promotion offers a
wise long-term option for diversifying global food supplies, particularly as the environmental
challenges that agriculture faces intensify and as more people suffer from the negative health
effects of the nutrition transition (Popkin 2006, Pingali 2007, Kearney 2010).
The recent rise of quinoa (Zurita-Silva et al. 2014) demonstrates how global food supplies can
be diversified, particularly against a background of growing consumer interest in diverse and
healthy food alternatives. A potential benefit of globalization and urbanization trends may be
an increased tolerance for healthy non-traditional foods. Increasing the contribution of
marginalized crops in global food supplies requires focus both on establishment or
improvement of incentives for farmers and seed systems at the production level as well as
promotion at the consumer level. Specific actions may include: (a) identifying regionally and
locally important crop species that show potential for improved productivity, enhanced
nutritional quality, and greater competitiveness under challenging conditions, (b) conserving as
well as fostering the use of their genetic diversity, (c) breeding productive and resilient varieties
of these crops, with particular emphasis on traits such as increased micronutrient levels, storage
life, and versatility in processing and consumer use, (d) making these materials widely available
to breeders and other researchers, facilitating their uptake through agricultural extension and
training, and developing robust seed systems for their multiplication and distribution, and (e)
stimulating policy measures that strengthen market demand for these crops.
Potential, geographic distributions, and conservation priorities for crop wild relatives
A number of key impressions emerge from the present research devoted to identifying,
investigating the potential of, and determining the conservation status of the genetic diversity
represented in crop wild relatives. First, a great deal of wild diversity exists and is therefore
potentially available for use in breeding. The national assessment for the USA, a country
generally considered to be depauperate in regard to crop diversity (Vavilov 1926), identified a
great range of native and introduced species associated with a long list of important crops
(Chapter 3). Case studies on the wild relatives of beans, sweetpotato and pigeonpea revealed
diverse crop genetic resources widely distributed over regions and continents (Chapters 4-6).
Second, wild genetic resources have considerable potential to contribute traits of value to crop
280

General discussion
improvement. Populations of wild species related to sweetpotato (Chapter 5) and pigeonpea
(Chapter 6) are distributed in areas modeled to be well outside of the temperature, precipitation,
and soil characteristic comfort zone of these crops (see Figure 4 of Chapter 5, and Figures 5
and 6 of Chapter 6). Similar results have been reported for wild relatives and/or landraces of
other crops (e.g., Endresen et al. 2011, Tapia et al. 2014). Given the continual advances in
information on wild relatives, as well as improving breeding technologies facilitating the
utilization even of distantly related species (Zamir 2001, Cao et al. 2009, McCouch et al. 2013),
there is good reason to consider crop wild relatives as genetic resources of increasing potential
for major contributions to crop improvement.
Third, these wild genetic resources are drastically under-represented as a whole in ex situ
conservation systems. With an estimated 57% of wild relatives of bean (Chapter 4), 79% of
sweetpotato (Chapter 5), and 80% of pigeonpea (Chapter 6) assessed as high priority for further
collecting, and virtually all crop wild relative species determined to be insufficiently
represented in genebanks in regard to the full range of geographic and ecological variation in
their distributions, it is evident that much remains to be done to conserve their genetic diversity.
Such results are concordant with conservation assessments for other crop genepools (Khoury
et al. 2010).
Given such large gaps remaining in conservation needs for crop wild relatives, significant
resources needed for their maintenance ex situ, and seemingly chronic funding deficits for
genetic resources conservation (Khoury et al. 2010), a further assessment ideally informed by
genotypic diversity analyses of what constitutes sufficient germplasm collections for wild
relatives is warranted. As extremely valuable traits sourced from crop wild relatives have been
found in only a few populations despite screening of a large number of accessions (Brar &
Khush 1997), and given modeling of large variation within species in regard to traits such as
resistance to insect pests (Sharma et al. 2003), seed protein content and days to maturity
(Upadhyaya et al. 2013), and tolerance to high precipitation (Martin & Jones 1973,
Nimmakayala et al. 2011), the case can be made for increasing funding for collecting and
conservation so as to comprehensively conserve wild genetic resources. The window of
opportunity to fully collect and conserve this genetic diversity will not remain open indefinitely
(Wilkes 2007, FAO 2010).

281

Chapter 8
Along with ensuring their maintenance ex situ, further investment in genebank information
systems, ex situ conservation technologies (i.e., storage, testing, and regeneration), safety
duplication of unique germplasm, characterization and evaluation for traits of interest,
taxonomic and systematic studies, and breeding research are needed to mobilize wild genetic
resources for use in crop improvement (FAO 2010, Khoury et al. 2010, Tester and Langridge
2010, Guarino & Lobell 2011, Henry 2014).
Access priorities for crop genetic diversity
Given the global geography of crop genetic diversity, it is evident that international
collaboration is critical to achieving access to these resources. The vast majority of countries,
even those located within the most ancient and richest primary regions of crop genetic diversity,
e.g., West Asia, are significant consumers and producers of crops whose genetic diversity
largely sources from outside their borders (Chapter 7). The importance of ‘foreign’ crops in
national food supplies and production systems is growing over time as national food systems
become more diverse and at the same time more homogeneous worldwide. These trends bolster
the rationale for considering genetic resources of internationally important crops as global
public goods, the common heritage of humankind, the conservation and facilitated access of
which should be proactively supported by all countries worldwide (FAO 2002, EsquinasAlcázar 2005). They also reaffirm the critical importance of the germplasm collections
safeguarded by CGIAR and key national genebanks, which have born most of the burden of
germplasm distribution to the global agricultural research community over the past few
decades.
LIMITATIONS AND UNCERTAINTIES
The main limitations and uncertainties inherent in the thesis research relate to the reliability and
sufficiency of input information, and the robustness of modeling methods. Food supply and
production data managed by the Food and Agriculture Organization of the United Nations
(FAO) lack the specificity needed for an exhaustive analysis of changes in crop diversity
[Khoury et al. 2014 (Chapter 2)], and although significant efforts are being taken to improve
data variably reported by national institutions (e.g., FAO 2015), it is disappointing that annual
data was probably of greater scope and quality 35 years ago than it is now (R. Prescott-Allen,
personal communication 2014). Broad support for and increased investment in agricultural
statistics by all countries are critically needed in order to resolve these deficiencies.
282

General discussion
Significant further research is still needed to identify wild plant taxa, assess species boundaries
within taxonomic groups, and determine degrees of relatedness with associated crops, as
evidenced by the study on sweetpotato (Chapter 5). Incorrect or outdated species identifications,
synonymy, and erroneous or low resolution occurrence information are recognized challenges
in the utilization of the current openly available biodiversity resources (Gaiji et al. 2013).
The rapidly evolving field of species distribution modeling is continuously improving in its
methodologies, but deficiencies and uncertainty are acknowledged (Dormann 2006, Hijmans &
Graham 2006, Graham et al. 2008, Jimenez-Valverde et al. 2008, Lobo et al. 2008, Loiselle et
al. 2008, Wisz et al. 2008, Fitzpatrick & Hargrove 2009), particularly in regard to sampling
bias due to the lack of absence information and imperfect sampling techniques (Araújo &
Guisan 2006, Costa et al. 2010, Kramer-Schadt et al. 2013). The methods utilized in the thesis
drew upon the scientific literature to refine data inputs, capitalized on a large set of climatic and
edaphic variables to determine distributions, performed modeling with different variations of
this data in order to test robustness, and employed a set of statistical analyses to determine
confidence in the results. Further studies that validate modeling results (Jarvis et al. 2005,
Cobben et al. 2014) will be useful in the application of species distribution modeling to genetic
resources conservation planning. As the potential distributions maps created in this thesis and
elsewhere (e.g., Castañeda-Alvarez et al., 2015) are informing ongoing collecting missions
(Dempewolf et al. 2014), feedback from the fieldwork can result in improvements in the
methods used here.
While the analyses presented in Chapters 5 and 6 of this thesis revealed large eco-geographic
variation in wild populations compared with the cultivated species, further studies are needed
to correlate such variation with useful adaptive traits that can be bred into commercial
genotypes. Such analyses would improve the knowledge base on potential uses of crop wild
relatives.
An important tool employed in the gap analysis methodology to mitigate deficiencies and
uncertainty in input data and modeling techniques is the expert evaluation process, which can
involve a number of researchers and utilizes both quantitative and qualitative metrics (Chapters
4-6). Agricultural research in parallel with technology appears to be transitioning from high
quality restricted scope studies performed by a limited number of experts, to ‘big data’
283

Chapter 8
approaches drawing overall trends from large amounts of information. The expert evaluation
and the data processing and modeling steps bring together these classical and emerging methods
in the gap analysis, attempting to draw upon the strengths of both worlds. While the expert
assessments were considered to be extremely valuable in verifying the modeled results for the
crop genepools assessed here, the unfortunate reality is that there are fewer and fewer experts
left that have a full career of experience in taxonomy, genetic resources, and/or breeding. The
loss of these human encyclopedias on crop diversity has some regrettable parallels to the folk
tale of John Henry (Bradford 1931). Henry's ability as a (human) steel-driver was measured in
a race against the emerging technology of a steam powered hammer. Henry won, only to die in
victory with his hammer in his hand as his heart gave out from stress. As research transitions
from dependence on knowledge stored in human brains to data stored in online platforms,
improvements in data quality, modeling methods, and statistical analyses of significance of
trends will become ever more critical to the validity of results.
An additional uncertainty in regard to conceptualizing the future potential of crop genetic
diversity is that advancing molecular and breeding tools are expanding the capacity to
successfully introgress traits from distant relatives (Zamir 2001, McCouch et al. 2013, Henry
2014), making the boundaries of what might be considered potentially valuable related species
less obvious. Developments in cisgenic, transgenic, and gene editing techniques (Sander &
Joung 2014) will only widen the scope of possibility, with implications for genetic resource
conservation, exploration, and use.
This thesis represents concrete steps taken to conceptualize global change in crop diversity at
the species level, advance information on crop wild relatives for one country and for three crop
genepools, and contribute to policy arguments in support of facilitated access to crop genetic
resources of global importance. A comprehensive understanding of the diversity, conservation,
and utilization needs for genetic resources associated with the world’s crops requires parallel
and extended research efforts across other levels of diversity, regions and crop genepools. As
the present study proposes methods that rely upon openly available data, tools and software, it
is hoped that the effort will facilitate this further needed research.

284

General discussion
IMPACT
The results presented in this thesis have been received by the scientific community and the
public with remarkable interest.
The research on homogeneity in global food supplies (Khoury et al. 2014) was covered
extensively by the media, with over 300 articles, blogs, and podcasts published in the first year
following its release, including in top tier newspapers and journals such as Scientific American,
Time Magazine, and National Geographic (see the Publications of author section of this thesis
for references for a selection of these articles). This attention led to an altmetric score
considering the research among the top 100 articles ever published in PNAS, and within the 99th
percentile of all scientific articles ever tracked. The study was remarked upon by leaders of the
World Bank and CGIAR, inspired significant discussion on research and funding priorities in
CGIAR, generated a series of invited articles, blogs, and presentations given internationally,
and was further reproduced in academic textbooks, radio and television shows.
Research on crop wild relatives has also received considerable attention. The highly cited gap
analysis methodology (Ramírez-Villegas et al. 2010) has become a central method paper in the
field of exploration of diversity of wild relatives, and subsequent articles (e.g., Khoury et al.
2015) have a high number of viewings by scientists and the public. The national inventory of
the United States (Khoury et al. 2013) was also covered in American media and received the
2014 Crop Science Society of America C8 Division “Outstanding Papers in Plant Genetic
Resources” award.
A number of concrete actions associated with the present research are ongoing. Inspired by the
assessment of increasing homogeneity in global food supplies and subsequent research focused
on its implications in regard to agricultural development (Khoury & Jarvis 2014), CGIAR
international research centers are engaging in wider discussions regarding funding and research
priorities. Numerous crop genepool based gap analyses, bringing together a variety of
associated researcher experts, are in planning or have been initiated. A cumulative global
analysis communicating conservation concerns for over 1100 wild species related to 80 crops
is also nearing completion at the time of writing. Building upon the national inventory of the
United States, agencies are collaborating on refining knowledge on conservation concerns,
performing targeted collecting, and establishing management plans for the conservation of key
285

Chapter 8
iconic wild relatives in situ. Finally, the research on interdependence among countries on crop
genetic resources is being used as a formal submission to the Plant Treaty in contribution to
current negotiations for the potential expansion of scope and membership.
CLOSING REMARKS
How humanity will feed itself into the future remains an open question. Whether our ingenuity
will prove successful in finding new ways to innovate with natural resources to overcome
nutrition and production challenges as we surpass what appear to be safe planetary boundaries
(Rockström et al. 2009) remains to be seen. More than six decades ago, Aldo Leopold, a
preeminent American ecologist and conservationist, warned of the importance of natural
resource conservation to human wellbeing:
“To keep every cog and wheel is the first precaution of intelligent tinkering”
(Leopold 1949).
Sixty years later, it is clear that much work remains to be done to accomplish a comprehensive
global system for the conservation, exploration, access to, and use of crop genetic resources.
This work is more pressing than ever, given increasing homogeneity in the global food system,
the myriad of challenges facing agriculture, and threats to the survival of this diversity in the
wild, in farmers’ fields, and in under-funded genebanks. N. I. Vavilov’s oft cited statement on
the urgency of these efforts is still poignant:
“Time is short, time is short, there is so much to do” [N. I. Vavilov, as recorded in
Cohen (1991)].

286

General discussion
REFERENCES
Araújo, M.B., Guisan, A. (2006). Five (or so) challenges for species distribution modelling. J. Biogeogr.
33 (10), 1677-1688.
Bradford, R. (1931). John Henry (Harper & Brothers).
Brar, D.S., Khush, G.S. (1997). Alien introgression in rice. Plant Mol. Biol. 35 (1-2), 35-47.
Cao, Q., Zhang, A., Ma, D., Li, H., Li, Q., Li, P. (2009). Novel interspecific hybridization between
sweetpotato (Ipomoea batatas (L.) Lam.) and its two diploid wild relatives. Euphytica 169, 345-352.
Castañeda-Álvarez, N.P., de Haan, S., Juarez, H., Khoury, C.K., Achicanoy, H., Sosa, C.C., et al. (2015).
Ex situ conservation priorities for the wild relatives of potato (Solanum L. section Petota). PLoS One
10 (4), e0122599.
Cobben, M.M.P., van Treuren, R., Castañeda-Álvarez, N.P., Khoury, C.K., Kik, C., van Hintum, T.J.L.
(2014). Robustness and accuracy of Maxent niche modelling for Lactuca species distributions in
light of collecting expeditions. Plant Genetic Resources: Characterization and Utilization, 1-9. doi:
10.1017/S1479262114000847.
Cohen, B.M. (1991). Nikolai Ivanovich Vavilov: the explorer and plant collector. Econ. Bot. 45 (1), 3846.
Costa, G.C., Nogueira, C., Machado, R.B., Colli, G.R. (2010). Sampling bias and the use of ecological
niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodiv.
Conserv. 19, 883-899.
Dempewolf, H., Eastwood, R.J., Guarino, L., Khoury, C.K., Müller, J.V., Toll, J. (2014). Adapting
agriculture to climate change: A global initiative to collect, conserve, and use crop wild relatives.
Agroecol. Sust. Food Syst. 38 (4), 369-377.
Dormann, C.F. (2006). Promising the future? Global change projections of species distributions. Basic
Appl. Ecol. 8, 387-397.
Endresen, F., Terje, D., Street, K., Mackay, M., Bari, A., De Pauw, E. (2011). Predictive association
between biotic stress traits and eco-geographic data for wheat and barley landraces. Crop Sci. 51 (5),
2036-2055.
Esquinas-Alcázar, J. (2005). Protecting crop genetic diversity for food security: political, ethical and
technical challenges. Nat. Rev. Genet. 6 (12), 946-953.
FAO (Food and Agriculture Organization of the United Nations). (2010). The Second Report on the
State of the World’s Plant Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2002). International Treaty on Plant
Genetic Resources for Food and Agriculture (Rome: FAO).
FAO (Food and Agriculture Organization of the United Nations). (2015). CountrySTAT.
http://www.fao.org/economic/ess/ess-capacity/countrystathome/en/.
Fitzpatrick, M.C., Hargrove, W.W. (2009). The projection of species distribution models and the
problem of non-analog climate. Biodiv. Conserv. 18, 2255-2261.
Gaiji, S., Chavan, V., Ariño, A.H., Otegui, J., Hobern, D., Sood, R., et al. (2013). Content assessment
of the primary biodiversity data published through GBIF network: status, challenges and potentials.
Biodiversity Informatics 8, 94-172.
Graham, C., Elith, J., Hijmans, R.J., Guisan, A., Peterson, A.T., Loiselle, B.A., et al. (2008). The
influence of spatial errors in species occurrence data used in distribution models. J. Appl. Ecol. 45,
239-247.
Guarino, L., Lobell, D.B. (2011). A walk on the wild side. Nat. Clim. Change 1 (8), 374-375.
Henry, R.J. (2014). Genomics strategies for germplasm characterization and the development of climate
resilient crops. Front. Plant Sci. 5, 68.
287

Chapter 8
Hijmans, R.J., Graham, C. (2006). The ability of climate envelope models to predict the effect of climate
change on species distributions. Glob. Change Biol. 12, 2272-2281.
Jarvis, A., Williams, K., Williams, D.E., Guarino, L., Caballero, P.J., Mottram, G. (2005). Use of GIS
for optimizing a collecting mission for a rare wild pepper (Capsicum flexuosum Sendtn.) in Paraguay.
Genet. Resour. Crop Ev. 52, 671-682.
Jimenez-Valverde, A., Lobo, J.M., Hortal, J. (2008). Not as good as they seem: the importance of
concepts in species distribution modelling. Divers. Distrib. 14, 885-890.
Kearney, J. (2010). Food consumption trends and drivers. Philos. T. R. Soc. B. 365 (1554), 2793-2807.
Khoury, C., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic resources:
a review of global crop and regional conservation strategies. Genet. Resour. Crop Ev. 57 (4), 625639.
Khoury, C.K., Bjorkman, A.D., Dempewolf, H., Ramírez-Villegas, J., Guarino, L., Jarvis, A., et al.
(2014). Increasing homogeneity in global food supplies and the implications for food security. Proc.
Natl. Acad. Sci. 111 (11), 4001-4006.
Khoury, C.K., Greene, S., Wiersema, J., Maxted, N., Jarvis, A., Struik, P.C. (2013). An inventory of
crop wild relatives of the United States. Crop Sci. 53 (4), 1496.
Khoury, C.K., Heider, B., Castañeda-Álvarez, N.P., Achicanoy, H.A., Sosa, C.C., Miller, R.E., et al.
(2015). Distributions, ex situ conservation priorities, and genetic resource potential of crop wild
relatives of sweetpotato [Ipomoea batatas (L.) Lam., I. series Batatas]. Front. Plant Sci. 6, 251.
Khoury, C.K., Jarvis, A. (2014). The Changing Composition of the Global Diet: Implications for CGIAR
Research. CIAT Policy Brief No. 18 (Cali: Centro Internacional de Agricultura Tropical).
Krame-Schadt, S., Niedballa J., Pilgrim J.D., Schroder B., Lindenborn J., Reinfelder, V., et al. (2013).
The importance of correcting for sampling bias in MaxEnt species distribution models. Divers.
Distrib. 19 (11), 1366-1379.
Leopold, A. 1949. A Sand County Almanac and Sketches Here and There (New York: Oxford University
Press).
Lobo, J.M., Jiménez-Valverde, A., Real, R. (2008). AUC: a misleading measure of the performance of
predictive distribution models. Global Ecol. Biogeogr. 17 (2), 145-151.
Loiselle, B.A., Jorgensen, P.M., Consiglio, T., Jimenez, I., Blake, J.G., Lohmann, L.G., et al. (2008).
Predicting species distributions from herbarium collections, does climate bias in collection sampling
influence model outcomes? J. Biogeogr. 35(1), 105-116.
Martin, F.W., Jones, A. (1973). The species of Ipomoea closely related to the sweetpotato. Econ. Bot.
26, 201-215.
McCouch, S., Baute, G.J., Bradeen, J., Bramel, P., Bretting, P.K., Buckler, E., et al. (2013). Agriculture:
feeding the future. Nature 499 (7456), 23-24.
Nimmakayala, P., Vajja, G., Reddy, U.K. (2011). “Ipomoea,” in: Wild Crop Relatives: Genomic and
Breeding Resources, ed. C. Kole (Berlin: Springer), 123-132.
Pingali, P. (2007). Westernization of Asian diets and the transformation of food systems: implications
for research and policy. Food Pol. 32 (3), 281-298.
Popkin, B.M. (2006). Technology, transport, globalization and the nutrition transition food policy. Food
Pol. 31 (6), 554-569.
Ramírez-Villegas, J., Khoury, C., Jarvis, A., Debouck, D.G., Guarino, L. (2010). A gap analysis
methodology for collecting crop genepools: a case study with Phaseolus beans. PloS One 5, e13497.
Rockström, J., Steffen, W., Noone, K., Persson, Å., Chapin, F.S., Lambin, E.F., et al. (2009). A safe
operating space for humanity. Nature 461 (7263), 472-475.
Sander, J.D., Joung, J.K. (2014). CRISPR-Cas systems for editing, regulating and targeting genomes.
Nat. Biotechnol. 32, 347-355.
288

General discussion
Sharma, H.C., Pampapathy, G., Reddy, L.J. (2003). Wild relatives of pigeonpea as a source of resistance
to the pod fly (Melanagromyza obtusa Malloch) and pod wasp (Tanaostigmodes cajaninae La Salle).
Genet. Resour. Crop Ev. 50, 817-824.
Tapia, C., Torres, E., Parra-Quijano, M. (2014). Searching for adaptation to abiotic stress:
ecogeographical analysis of highland Ecuadorian maize. Crop Sci. 55 (1), 262-274.
Tester, M, Langridge, P. (2010). Breeding technologies to increase crop production in a changing world.
Science 327 (5967), 818-822.
Upadhyaya, H.D., Reddy, K.N., Pundir, R.P.S., Singh, S., Gowda, C.L.L., Irshad Ahmed, M. (2013).
Diversity and geographical gaps in Cajanus scarabaeoides (L.) Thou. germplasm conserved at the
ICRISAT genebank. Plant Genetic Resources 11 (1), 3-14.
Vavilov, N.I. (1926). Studies on the origin of cultivated plants. Bulletin of Applied Botany, Genetics and
Plant Breeding 16, 1-248.
Wilkes, G. (2007). Urgent notice to all maize researchers: disappearance and extinction of the last wild
teosinte population is more than half completed. A modest proposal for teosinte evolution and
conservation in situ: The Balsas, Guerrero, Mexico. Maydica 52, 49-58.
Wisz, M.S., Hijmans, R.J., Li, J., Peterson, A.T., Graham, C.H., Guisan, A., and NCEAS Predicting
Species Distributions Working Group. (2008). Effects of sample size on the performance of species
distribution models. Divers. Distrib. 14, 763-777.
Zamir, D. (2001). Improving plant breeding with exotic genetic libraries. Nat. Genet. 2, 983-989.
Zurita-Silva, A., Fuentes, F., Zamora, P., Jacobsen, S.-E., Schwember, A.R. (2014). Breeding quinoa
(Chenopodium quinoa Willd.): potential and perspectives. Mol. Breed. 34 (1), 13-30.

289

Summary
The challenges to long-term global food and nutrition security are complex and compounding.
The increasing availability of energy-dense foods worldwide is reducing stunting and other
measures of undernutrition, but this nutrition transition demonstrates mixed success in
resolving micronutrient deficiencies, and the over-consumption of macronutrients is
contributing to the global surge in diet-related non-communicable diseases. Our growing
population and dietary expectations are projected to increase demand on food systems for at
least the next four decades, outpacing current yield trends. Limitations in land, water, and
natural resource inputs, competition for arable soils with non-food crops and other land uses,
the need to minimize harmful impacts on biodiversity and other ecosystem services, and greater
climatic variability further constrain production potential. Although future gains in food
availability may partially be obtained through dietary change and food waste reduction, an
increase both in productivity and sustainability on current agricultural lands is necessary. This
increase will be achieved through improved agronomic practices combined with the use of
varieties of crops with reliable yields under more adverse conditions.
As the source of traits employed for adaptation to biotic and abiotic stresses and for yield
increases through breeding, and the palette from which food systems may be further diversified,
crop genetic diversity is critical to increased productivity and sustainability. This biological
cornerstone of food security is generated through genetic mutation and recombination, and
further transformed through natural and artificial selection, and is therefore the product of the
evolution over time of crops and closely related wild plants in their agricultural and natural
habitats.
Expectations of increases in utilization of crop genetic diversity in order to address production
challenges are made under the assumption that adequate variation will be available for
exploration. Unfortunately, considerable erosion of crop genetic diversity is occurring through
economic development, demographic change, and habitat destruction. Due to the disappearance
of this diversity in situ, i.e., in farmers’ fields and natural habitats, the world’s ex situ genebanks
originally established to make plant genetic diversity readily available to breeders for crop
improvement, have become essential repositories for crop diversity conservation, but these too
are vulnerable due to insufficient support and resources.

Summary
Two points regarding the current state of the world’s heritage of crop genetic diversity are thus
clear - much remains to be done to secure the diversity threatened in situ and in under-funded
genebanks, and the window of opportunity to accomplish a comprehensive global system for
genetic resources conservation will not remain open indefinitely.
Among the factors hindering the conservation of crop genetic resources is a lack of essential
information regarding this diversity. A number of primary questions have not been fully
answered, including: (a) what is the status of diversity in our food systems, and where are the
greatest vulnerabilities?, (b) where can genetic diversity be found that can be useful in
increasing productivity and mitigating these vulnerabilities?, (c) is this genetic diversity
available in the present and in the long term?, and (d) if not available, what steps are needed to
improve the ability for researchers to access genetic resources critical for present and future
crop improvement?
This thesis aims to contribute to the knowledge required to answer these questions through an
exploration of the need for, potential of, challenges and constraints regarding, and necessary
steps to enhance the conservation and use of crop genetic resources. The research starts with an
investigation of the current state of diversity in global food supplies (Chapter 2). This also
represents an exploration into a longstanding two-part assumption in the field of genetic
resources – that humanity relies on relatively few crops for its survival, and that this list of
important crops is growing smaller over time.
In order to understand the state of diversity in global food supplies, trends over the past 50 years
in the richness, abundance, and composition of crop species in the national food supplies of
countries worldwide were assessed. Over this period national per capita food supplies expanded
in total quantities of food calories, protein, fat, and weight, with increased proportions of those
quantities sourcing from energy-dense foods. At the same time, the number of measured crop
commodities contributing to national food supplies increased, the relative contribution of these
commodities within these supplies became more even, and the dominance of the most
significant commodities decreased. As a consequence, national food supplies worldwide
became more similar in composition, correlated particularly with an increased supply of a
number of globally important cereal and oil crops, and a decline of other cereals, oil crops, and
starchy root and tuber species. The increase in homogeneity worldwide portends the
292

Summary
establishment of a global standard food supply, which is relatively species rich in regard to
measured crops at the national level, but species poor globally. This research contextualizes the
need for the use of diverse genetic resources in crop breeding in order to mitigate vulnerability
created by greater homogeneity in global food supplies.
The thesis then delves into the potential for utilization of a particular set of genetic resources of
increasing interest globally – crop wild relatives. These weedy and wild cousins of cultivated
species have been used for many decades in order to introduce traits of value to crops through
breeding, particularly for pest and disease resistance. Their use in crop improvement is likely
only to increase because (a) information is improving regarding the identity, potential, and
conservation status of these plants, and digital data platforms are making this information more
readily available, (b) advancements in geographic information systems, due to increasingly high
resolution ecological data as well as evolving models and methods, are enabling a more
comprehensive conceptualization of the geographic distribution and conservation status of wild
species, (c) rapidly progressing classical and genomic tools, technologies, and methods are
facilitating their use in crop breeding, and (d) there is a growing interest in the use of exotic
genetic diversity in order to address compounding agronomic challenges.
Crop wild relatives are genetic resources at the nexus of a number of critical global challenges.
They have the potential to contribute significantly to crop improvement, helping to address food
security and development goals, while improving the adaptation of crops to climate change. At
the same time, as wild plants they are subject to a myriad of human caused threats to natural
ecosystems, including habitat modification, urbanization, mining, logging, changing fire
regimes, pollution, invasive species, overharvesting, and climate change. A focus on wild
genetic resources is thus timely both for conservation and food security objectives.
Research on crop wild relatives first concentrates on the identification of potentially important
wild genetic resources at the national level (Chapter 3). Focusing on the United States of
America, a large and eco-geographically diverse country with relatively advanced conservation
policy and active national genetic resources conservation efforts, the chapter presents an
inventory of crop wild relatives and other wild species of potential use in agricultural research.
The chapter also develops a method for prioritizing these wild species based upon their potential
to contribute to food security. The resulting National Inventory listed 4,600 taxa from 985
293

Summary
genera and 194 plant families, including wild relatives of potential value via breeding as well
as wild species of direct use for food, forage, medicine, herb, industrial, ornamental, and
environmental restoration purposes. Crop wild relatives were found to be related to a broad
range of important crops. Some potentially valuable species are threatened in the wild, and few
accessions of such taxa are currently conserved ex situ. The prioritization identified 821 taxa
from 69 genera primarily related to major food crops, with emphasis on approximately 285
native taxa from 30 genera that are most closely related to such crops.
Once potentially valuable crop genetic resources are identified, subsequent information is
needed regarding where they occur, what diversity they may possess, and how well conserved
and therefore available to crop breeders they are. Chapter 4 offers a ‘gap analysis’ methodology
to answer these questions at the crop genepool level, i.e., for the wild relatives associated with
any particular crop, with a case study on the wild relatives of bean (Phaseolus L.). This chapter
capitalizes on developments in the generation of and access to digital occurrence and ecogeographic data as well as improvements in modeling wild plant species distributions,
intentionally utilizing freely available software and data. The method also includes a novel
expert evaluation methodology, using researchers knowledgeable in the distributions and
conservation concerns of crop wild relatives to assess the results. Of 85 assessed taxa, over half
were found to be highly under-represented in genebanks. Priority areas for collecting were
identified, particularly in central Mexico.
Chapters 5 and 6 represent advances on the basic methodology outlined in Chapter 4, taking
advantage of improvements in species targeting, occurrence data, modeling, and expert
feedback methods. The chapters also take an additional step by utilizing eco-geographic
information to indicate the potential for species and specific populations to possess traits of
value to crop improvement, particularly for abiotic stress tolerance. The chapters focus on crops
differing substantially in the state of existing information regarding associated wild relatives,
as well as historical use of the resources in crop improvement.
The potential for use of the crop wild relatives of sweetpotato [Ipomoea batatas (L.) Lam., I.
series Batatas] is constrained by uncertainty in regard to species boundaries and their
phylogenetic relationships, the limited availability of germplasm with which to perform crosses,
and the difficulty of introgressing genes from wild species. Chapter 5 modeled the distributions
294

Summary
of the fourteen species that are considered the closest wild relatives of sweetpotato, and found
that currently designated species differed among themselves and in comparison with the crop
in their adaptations to temperature, precipitation, and edaphic characteristics, and most species
also showed considerable intraspecific variation. With 79% of species identified as high priority
for further collecting, these crop genetic resources were found to be highly under-represented
in ex situ conservation systems and thus inadequately available to breeders and researchers.
Taxa and specific geographic locations were prioritized for further collecting, particularly in
diversity hotspots in Mesoamerica and in the extreme southeastern United States, in order to
improve the completeness of germplasm collections.
The use of crop wild relatives in pigeonpea [Cajanus cajan (L.) Millsp.] breeding has been
successful in providing important resistance, quality, and breeding efficiency traits. Current
breeding objectives for pigeonpea include increasing its tolerance to abiotic stresses, including
heat, cold, drought, and waterlogging. Like the sweetpotato study, Chapter 6 found considerable
variation among the fifteen wild relatives of the crop in regard to adaptations to climatic and
soil conditions. Likewise, the research assessed that these wild genetic resources are broadly
under-represented in ex situ conservation systems, with 80% of species identified as high
priority for further collecting. Species and geographic locations particularly in southern India
and northern Australia were highlighted for further collecting in order to improve the
completeness of germplasm collections, with particular emphasis on collecting and conserving
populations possessing tolerance to abiotic stresses.
While conservation of crop genetic resources is fundamental to the availability of this diversity
for breeding, it is not the only major constraint to utilization. National and international policies
on crop genetic resources determine the real capacity for researchers to acquire diversity of
potential interest. The thesis culminates in an exploration of the implications of the global
geographic distribution of crop genetic diversity for food security, in particular the level to
which international collaboration is required in order to achieve access to genetic resources
where they are needed.
The research presented in Chapter 7 demonstrated that rich historical areas of crop genetic
diversity occur across the tropics and subtropics, extending into temperate regions in both
hemispheres. National food systems are thoroughly interconnected worldwide in regard to the
295

Summary
geographic origins of crop diversity. Countries are highly dependent on crops whose genetic
diversity largely sources from outside their borders in their food supplies (68.7% as a global
mean across food variables) and in their national production systems (69.3%). This reliance is
evident even in countries located in regions of high indigenous crop diversity and has increased
significantly over the past half century, bolstering evidence for the need for effective national
and international policies to promote genetic resource conservation and exchange.
Chapter 8 provides a general discussion of the research and its main results. The chapter offers
a summary of the current status of conservation of the crop genetic resources targeted in the
thesis, identifying limitations in the study and suggesting future research in order to make
further progress on key questions in the field. The chapter also discusses the impact of the
research to date, and highlights ongoing activities that are building upon the efforts documented
here. Increasing awareness and information about the critical role of crop genetic diversity in
overcoming nutrition and production challenges offer a renewed opportunity to establish a more
comprehensive global system for its conservation and availability for use. Collecting,
improving conservation, performing further taxonomic, breeding, and associated research, and
resolving the politics of access to genetic resources are urgently needed in order to maximize
the potential for crop genetic diversity to contribute to improving food security.

296

Publications of author
Scientific journals
Castañeda-Álvarez, N.P., de Haan, S., Juarez, H., Khoury, C.K., Achicanoy, H., Sosa, C.C., et
al. (2015). Ex situ conservation priorities for the wild relatives of potato (Solanum L. section
Petota). PLoS One 10 (4), e0122599.
Cobben, M.M.P., van Treuren, R., Castañeda-Álvarez, N.P., Khoury, C.K., Kik, C., van
Hintum, T.J.L. (2014). Robustness and accuracy of Maxent niche modelling for Lactuca
species distributions in light of collecting expeditions. Plant Genetic Resources:
Characterization and Utilization, 1-9. doi: 10.1017/S1479262114000847.
Dempewolf, H., Eastwood, R.J., Guarino, L., Khoury, C.K., Müller, J.V., Toll, J. (2014).
Adapting agriculture to climate change: a global initiative to collect, conserve and use crop
wild relatives. Agroecol. Sust. Food Syst. 38 (4), 369-377.
Khoury, C.K., Bjorkman, A.D., Dempewolf, H., Ramírez-Villegas, J., Guarino, L., Jarvis, A.,
et al. (2014). Increasing homogeneity in global food supplies and the implications for food
security. Proc. Natl. Acad. Sci. 111 (11), 4001-4006.
Khoury, C.K., Castañeda-Álvarez, N.P., Achicanoy, H., Sosa, C.C., Bernau, V., Kassa, M.T.,
et al. (2015). Crop wild relatives of pigeonpea [Cajanus cajan (L.) Millsp.]: Distributions,
ex situ conservation status, and potential genetic resources for abiotic stress tolerance. Biol.
Conserv. 184, 259-270.
Khoury, C.K., Greene, S., Wiersema, J., Maxted, N., Jarvis, A., Struik, P.C. (2013). An
inventory of crop wild relatives of the United States. Crop Sci. 53 (4), 1496.
Khoury, C.K., Heider, B., Castañeda-Álvarez, N.P., Achicanoy, H.A., Sosa, C.C., Miller, R.E.,
et al. (2015). Distributions, ex situ conservation priorities, and genetic resource potential of
crop wild relatives of sweetpotato [Ipomoea batatas (L.) Lam., I. series Batatas]. Front.
Plant Sci. 6, 251.
Khoury, C.K., Laliberté, B., Guarino, L. (2010). Trends in ex situ conservation of plant genetic
resources: a review of global crop and regional conservation strategies. Genet. Resour. Crop
Ev. 57 (4), 625-639.
Ramírez-Villegas, J., Khoury, C.K. (2013). Reconciling approaches to climate change
adaptation for Colombian agriculture. Clim. Chang. 119, 575-583.
Ramírez-Villegas, J., Khoury, C., Jarvis, A., Debouck, D.G., Guarino, L. (2010). A gap analysis
methodology for collecting crop genepools: a case study with Phaseolus beans. PLoS One 5
(10), e13497.
Book chapters, policy briefs and other published articles
Arnaud, E., Castañeda-Álvarez, N., Dulloo, E., Drucker, A., de Haan, S., Jarvis, D., et al.
(2014). Global Information System for In Situ Conservation and On-farm Management of
Plant Genetic Resources for Food and Agriculture. Input Paper for the ITPGRFA
Consultation on the Global Information System on Plant Genetic Resources for Food and
Agriculture (COGIS-PGRFA) (Rome: FAO).
Guarino, L., Khoury, C., Fowler, C. (2011). “Adapting to climate change: the importance of ex

Publications of author
situ conservation of crop genetic diversity,” in: Food security and Climate Change in Dry
Areas: Proceedings of an International Conference, eds. M. Solh, M.C. Saxena, 1-4
February 2010, Amman, Jordan [Aleppo: International Center for Agricultural Research in
the Dry Areas (ICARDA)].
Hunter, D., Guarino, L., Khoury, C., Dempewolf, H. (2011). “A community divided: lessons
from the conservation of crop wild relatives around the world,” in: Agrobiodiversity
Conservation: Securing the diversity of Crop Wild Relatives and Landraces, eds. N. Maxted,
E. Dulloo, B. Ford-Lloyd, L. Frese, J Iriondo (Wallingford: CAB International).
Khoury, C.K., Achicanoy, H.A., Bjorkman, A.D., Navarro-Racines, C., Guarino, L., FloresPalacios, X., et al. (2015). Estimation of Countries’ Interdependence in Plant Genetic
Resources Provisioning National Food Supplies and Production Systems. International
Treaty on Plant Genetic Resources for Food and Agriculture, Research Study 8 (Rome:
FAO).
Khoury, C.K., Dempewolf, H., Castañeda-Álvarez, N.P. (2014). Wild Relatives. The Scientist.
June 2014.
Khoury, C., Guarino, L. (2010). Back to the roots: wild genes for food security. Samara 18/19,
January-December 2010 (London: Kew Publications), 1-2.
Khoury, C.K., Jarvis, A. (2014). The Changing Composition of the Global Diet: Implications
for CGIAR Research. CIAT Policy Brief No. 18 (Cali: Centro Internacional de Agricultura
Tropical), 1-6.
Khoury, C., Laliberté, B., Guarino, L. (2009). Trends and Constraints in Ex Situ Conservation
of Plant Genetic Resources: A Review of Global Crop and Regional Conservation Strategies
(Rome: Global Crop Diversity Trust), 1-56.
Thesis
Khoury, C. (2007.) An Ecogeographic Survey and Conservation Analysis for Arid North
American Phaseolus L. (Fabaceae), Sections Acutifolii, Coriacei, Minkelersia and Rugosi
Species (Birmingham: University of Birmingham), UK MSc. thesis.
Publications coordinated
Global Strategy for the Ex Situ Conservation of Chickpea (Cicer L.) (2008). (Rome: Global
Crop Diversity Trust).
Global Strategy for the Ex Situ Conservation of Faba Bean (Vicia faba L.) (2009). (Rome:
Global Crop Diversity Trust).
Global Strategy for the Ex Situ Conservation of Lentil (Lens Miller) (2008). (Rome: Global
Crop Diversity Trust).
Media articles featuring research (selected works)
Media articles regarding Khoury et al. (2014) Proc. Natl. Acad. Sci.
Baragona, S. (2014). Global diets growing similar. Voice of America. 3 March 2014.
Becker, M. (2014). Food security declines with global crop diversity. Conservation Magazine.
17 March 2014.
Charles, D. (2014). In the new globalized diet, wheat, soy and palm oil rule. National Public
298

Publications of author
Radio. 4 March 2014.
Cherfas, J. (2014). The global standard diet. Eat this podcast. 31 March 2014.
Doyle, A. (2014). Global diets get more similar in threat to food security: study. Reuters. 3
March 2014.
Doyle, A. (2014). Global diets get more similar in threat to food security: study. Chicago
Tribune. 3 March 2014.
Doyle, K. (2014). Decreasing diversity in global food supplies threatens crop production and
nutrition. Crop Science Society of America. 5 March 2014.
Graber, C. (2014). Lower diet diversity threatens crops and us. Scientific American. 10 March
2014.
Kinver, M. (2014). Crop diversity decline 'threatens food security'. BBC News. 3 March 2014.
MacVean, M. (2014). Global food supply grows increasingly homogeneous, study says. Los
Angeles Times. 3 March 2014.
Mintz, S. (2014). World food security at risk as variety of crops shrinks. Thomson Reuters
Foundation. 4 March 2014.
Muller, G. (2014). ‘Global average diet’ threatens food security. Australian Broadcasting
Corporation, Bush Telegraph. 14 March 2014.
Roach, J. (2014). Increasingly similar global food supply poses risks, Study Says. NBC News.
3 March 2014.
Stuart, H. (2014). A narrowing global food supply will make us all fat, sick and less safe, study
says. Huffington Post. 6 March 2014.
Tran, M. (2014). Growing reliance on fewer crops will increase risk of drought and disease.
The Guardian. 4 March 2014.
Treat, J., Scalamogna, A. (2014). Diet similarity- we’ll have what they’re having. National
Geographic Magazine. September 2014, p. 41.
Vence, T. (2014). Crop coalescence. The Scientist. 5 March 2014.
Walsh, B. (2014). Our global diet is becoming increasingly homogenized - and that’s risky.
Time Magazine. 4 March 2014.
Media articles regarding Khoury et al. (2013) Crop Sci.
Fisher, M., Fisk, S. (2013). U.S. a surprisingly large reservoir of crop plant diversity. Press
release. 29 April 2013.
Gewin, V. (2013). US takes stock of valuable “weeds”. Frontiers in Ecology and the
Environment. Dispatches. The Ecological Society of America.
Suszkiw, J. (2014). National inventory takes stock of crops’ wild relatives. USDA Agricultural
Research. January 2014.
Media articles regarding crop wild relative initiatives
Braw, E. (2013). World food day: the search for sustainable crops. The Guardian. 16 October
2013.
Fisher, M. (2012). Crop wild relatives and their potential for crop improvement. CSA News.
May 2012, p. 4-10.
Fisher, M., Fisk, S. (2013). Protecting the weedy and wild kin of globally important crops. Press
299

Publications of author
release. American Society of Agronomy (ASA), Crop Science Society of America (CSSA),
and Soil Science Society of America (SSSA) International Annual Meetings, Nov. 3-6 in
Tampa, Florida.
Gewin, V. (2013). Weeds warrant urgent conservation. Nature News. 22 July 2013.
Gilbert, N. (2010). Searching for wild relatives. Nature News Blog. 10 December 2010.
Kew. (2013). Major global analysis offers hope for saving the wild side of staple food crops.
Press release. 22 July 2013.

300

PE&RC Training and Education Statement
With the training and education activities listed below the PhD
candidate has complied with the requirements set by the C.T. de Wit
Graduate School for Production Ecology and Resource Conservation
(PE&RC) which comprises of a minimum total of 32 ECTS (= 22 weeks
of activities)
Review of literature (6 ECTS)
Adapting agriculture to climate change: collecting,
protecting and preparing crop wild relatives
Writing of project proposal (4.5 ECTS)
Adapting agriculture to climate change: collecting, protecting and preparing crop wild
relatives
Post-graduate courses (4.5 ECTS)
Contemporary approaches in genetic resources conservation and use: integrated and
participatory approaches in agro-biodiversity management; WUR (2010)
Multivariate analysis; WUR (2011)
Invited review of (unpublished) journal manuscript (2 ECTS)
Evolutionary Applications: The red queen and seed bank: pathogen resistance of ex situ
and in situ conserved barley (2011)
Journal of Plant Breeding and Crop Science: Selection of yam seeds from open pollination
for adoption in yam (Dioscorea rotundata Poir) production (2012)
Plant Genetic Resources: Characterization and utilization: Identification of gaps in the
world collection of pigeonpea germplasm from East and Southern Africa conserved at the
ICRISAT genebank (2013)
Economic Botany: Factors influencing diversity of farmers’ varieties of sweetpotato in
Uganda: implications for conservation (2014)
Genetic Resources and Crop Evolution: Evaluation and improvement of the
ecogeographical representativeness of a collection of the genus Trichloris in Argentina
(2014)
Journal for Nature Conservation: Private reserves in Brazil: distribution patterns, logistical
challenges and conservation contributions (2015)
Deficiency, refresh, brush-up courses (0.6 ECTS)
Introduction to R (2011)
Competence strengthening / skills courses (1.5 ECTS)
Ethics and Philosophy in Life Sciences (E&P LS); WUR (2012)
PE&RC Annual meetings, seminars and the PE&RC weekend (1 ECTS)
PE&RC Weekend (2010)
PE&RC Day (2011)
Discussion groups / local seminars / other scientific meetings (4.5 ECTS)
Modelling and statistics network; WUR (2011-2013)
GIS Modelling group – CIAT (DAPA SDM group) (2011-2015)
International symposia, workshops and conferences (9 ECTS)
USDA National Plant Germplasm System (NPGS) Plant Germplasm Operations Committee
(PGOC); oral presentation; Washington DC (2011)
Botany conference (Society for Economic Botany); poster presentations (2011)
USDA NPGS Curators and PGOC; oral presentation; Spokane WA (2012)
USDA NPGS Curators and PGOC; oral presentation; Washington DC (2013)
ASA, CSSA and SSSA International Water, Food, Energy and Innovation for Sustainable
World; oral presentation (2013)
Center for Development Research (ZEF); oral presentation (2014)
University of Arizona WEES Food Systems Scholars & Native Seeds/SEARCH; oral

PE&RC Training and Education Statement

-

302

presentation; University of Arizona (2014)
USDA NPGS Curators and PGOC; oral presentation; Davis CA (2014)
USDA National Program Leaders; oral presentation; Washington DC (2014)
ASA, CSSA and SSSA International Grand Challenges, Great Solutions; oral and poster
presentations (2014)
International Food Policy Research Institute (IFPRI); oral presentation; Washington DC
(2014)

Curriculum vitae
Colin Kahlil Khoury was born on the 14th of March, 1975 in La Jolla, California, USA. Raised
at the interface of wild lands, subtropical fruit orchards, and urban areas, his mother regularly
phoned the California Poison Control System to inquire as to the toxicity of the wild plants that
he often sampled. Satsuma mandarins, pineapple guavas, and avocados remain his cultivated
fruits of preference. His feeling of foreboding as natural habitats and agricultural soils are
urbanized continues as well. Colin completed two years of undergraduate studies in Biology at
Oberlin College before leaving university to pursue an agricultural career full time. He
cultivated vegetables, fruits, and grains in Argentina, Mexico, and various parts of the USA
before returning to complete a Bachelor of Arts in Environmental Studies with an emphasis in
Agricultural Ecology from Prescott College, Arizona, USA in 2000. He then served as a crop
curator at Native Seeds/SEARCH, a regional seedbank dedicated to the conservation and
celebration of agricultural seeds and their associated traditions adapted to the arid southwestern
USA and northwestern Mexico. Following a further year exploring seed diversity movements
in Latin America, he returned to California, where he surveyed rare wild plants in the coastal
scrub, mountains, and deserts of the region. In 2007, Colin completed a Master of Science in
Conservation and Utilisation of Plant Genetic Resources at the University of Birmingham, UK.
For his MSc. thesis, he analyzed the distributions and conservation status of crop wild relatives
of bean (Phaseolus spp.) native to Mexico and the USA. He then served as a scientific associate
at the Global Crop Diversity Trust, located in Rome, Italy, where he informed strategies for
more comprehensive conservation of crop diversity worldwide. In 2011, he transferred to the
International Center for Tropical Agriculture (CIAT), Colombia, to work as a visiting research
scientist on a Crop Trust managed project while pursuing his doctoral studies at the Centre for
Crop Systems Analysis, C.T. de Wit Graduate School of Production Ecology and Resource
Conservation, Wageningen University, The Netherlands. He is a member of the Species
Survival Commission, Crop Wild Relative Specialist Group of the International Union for
Conservation of Nature (IUCN); the Standing Committee on Science Communications of the
Crop Science Society of America; the Crop Science Society of America, Soil Science Society
of America, and American Society of Agronomy; the Botanical Society of America; and the
California Native Plant Society.

Funding
Funding for this thesis came from the initiative "Adapting Agriculture to Climate Change:
Collecting, Protecting and Preparing Crop Wild Relatives" which is supported by the
Government of Norway. The project is managed by the Global Crop Diversity Trust with the
Millennium Seed Bank of the Royal Botanic Gardens, Kew, UK, and implemented in
partnership with national and international genebanks and plant breeding institutes around the
world (Dempewolf et al. 2014). Support also came from the International Center for Tropical
Agriculture (CIAT), the CGIAR Research Program on Climate Change, Agriculture, and Food
Security (CCAFS), and the CGIAR Research Program on Roots, Tubers, and Bananas (RTB).
The Grains Research and Development Corporation (GRDC); the Global Public Goods Phase
2 project of the System-wide Genetic Resources Program (SGRP), Bioversity International; and
the World Bank supported the early stages of development of the gap analysis methodology
and occurrence data compilation.
Dempewolf, H., Eastwood, R.J., Guarino, L., Khoury, C.K., Müller, J.V., Toll, J. (2014). Adapting
agriculture to climate change: a global initiative to collect, conserve and use crop wild relatives.
Agroecol. Sust. Food Syst. 38 (4), 369-377.