Contents lists available at ScienceDirect
Agricultural Systems
journal homepage: www.elsevier.com/locate/agsy
Maize yield and profitability tradeoffs with social, human and
environmental performance: Is sustainable intensification feasible?
Sieglinde S. Snappa,b,⁎, Philip Grabowskia,b, Regis Chikowoa,b, Alex Smitha, Erin Andersa,
Dorothy Sirrinea, Vimbayi Chimonyoa, Mateete Bekundac
a Plant Soil and Microbial Sciences Department, Michigan State University, 1066 Bogue St A286, East Lansing, MI 48824, USA
b Center for Global Change Science and Earth Observations, Michigan State University, 1066 Bogue St A286, East Lansing, MI 48824, USA
c International Institute of Tropical Agriculture (IITA), c/o AVRDC, The World Vegetable Center, P.O. Box 10, Duluti, Arusha, Tanzania
A R T I C L E I N F O
Keywords:
Indicators
Sustainable intensification
Causal loop
Tradeoffs
Legume
A B S T R A C T
Sustainable intensification (SI) has been regarded as the basis for environmentally sound and equitable agri-
cultural development. Field based assessment of technologies needs to move beyond production and economic
performance to include environment, social and human condition. In this study we systematically consider all
five domains of SI based on participatory action research (PAR) initiated in 2012 at three Central Malawi sites
that varied in agroecology from low to high potential. Fifteen SI indicators were assessed for four technologies:
sole maize (Zea mays L.) with 0 and recommended fertilization (69 kg N ha−1 and 9 kg P ha−1), pigeonpea
(Cajanus Cajun (L.) Millsp.)-maize intercrop (half rate fertilizer), and doubled up legume rotation (DLR, a pi-
geonpea-groundnut intercrop) sequenced with maize at half rate fertilizer in that phase. Through radar charts SI
performance and tradeoffs were visualized, and causal loop analysis allowed identification of research gaps. SI
indicator assessments included crop performance from on-farm trials, profitability, modeled probability of food
sufficiency, risk of crop failure and ratings of technologies by women farmers who were engaged in evaluation of
technologies through participatory research. The PAR included six mother trials, 236 baby trial farmers and a
survey that was carried out with 324 farmers (baby trial farmers plus control farmers) to document socio-
economic factors and management practices on focal fields. Replicated mother trials further provided the basis
for simulation modeling (APSIM) of weather-associated crop failure risk and slow processes such as soil carbon
(C) accrual. Radar charts were used to visualize SI performance of the technologies. Environmental performance
of the two pigeonpea-diversified technologies was variable, but generally high compared with sole maize sys-
tems, due to gains in vegetative biomass, duration of cover and biological nitrogen (N) fixation. Maize pro-
duction and economic assessment varied by site, and with steeper tradeoffs for legume diversification in the
mesic site, less so in the marginal site. The domains of social and human capacity building were superior for
legume integration, notably in terms of diverse diet, food security and farmer preferences (notably, female
farmers generally favored legume crops). Performance varied by site with legume systems most beneficial at the
most marginal site, including less risk of crop failure than unfertilized maize. Causal loop analyses identified
regulators of SI that require further attention, notably: crop-livestock conflicts and opportunities, male-female
control of legume crop production, and residue management. Overall, the SI indicators framework provided a
systematic means to consider tradeoffs and opportunities associated with novel crop combinations and man-
agement practices.
1. Introduction
Sustainable intensification (SI) has been put forward as the pathway
to address global requirements for food, fuel and fiber production,
while simultaneously protecting environmental services and conserving
resources. The nature and scope of SI, however, is highly contested
(Gunton et al., 2016). Initial definitions put forward for SI focused
primarily on agronomic production, profitability and protecting the
environment (Baulcombe et al., 2009; Tilman et al., 2002). Recently
there have been important efforts to broaden the definition of SI to
include social aspects such as gender equity and human capacity
(Godfray, 2015; Pretty et al., 2011; Loos et al., 2014). The scope
https://doi.org/10.1016/j.agsy.2018.01.012
Received 22 April 2017; Received in revised form 12 December 2017; Accepted 15 January 2018
⁎ Corresponding author.
E-mail address: snapp@msu.edu (S.S. Snapp).
Agricultural Systems 162 (2018) 77–88
0308-521X/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
T
encompassed by SI is becoming broader, as the science of sustainability
is maturing and multidisciplinary approaches are coming to the fore
(Gunton et al., 2016; Petersen and Snapp, 2015).
1.1. Framework for SI indicators
We apply a SI framework here that considers indicators for five
domains, namely: production, economics, environment, social and
human condition ((http://www.k-state.edu/siil/resources/index.html).
This framework was developed through a systematic process led by a
steering committee with participants from eight institutions and multi-
disciplinary perspectives from human and natural sciences, including
economists, soil scientists, modelers, agronomists, nutritionists, gender
specialists and livestock scientists (Smith et al., 2017). The domains
were identified so as to explicitly consider multiple and sometimes
contradictory SI objectives such as increasing yields, increasing income,
improving the natural resource base, reducing inequity and improving
food security, with the aim of supporting research in sustainable de-
velopment. Identification of SI indicators for the framework took into
account cost-effective, and practical, means for research in develop-
ment teams to support cyclic, co-learning processes and consideration
of tradeoffs over time. Through this framework we aim to move beyond
linear, project defined considerations, to iteratively engage with
farmers, educators, communities, policymakers, and other stakeholders
(Bell and Morse, 2004; Falconnier et al., 2017; Snapp et al., 2002a).
The literature on SI assessments provides few examples of practical
applications using field based action research data to evaluate tech-
nologies in multiple domains. A climate-smart agricultural assessment
framework has been applied to livestock systems innovation, with
consideration given to the domains of productivity, adaptation, and
mitigation, and indicators used in a participatory, iterative process
(Notenbaert et al., 2017). Interestingly, simple educational initiatives,
rather than complex climate smart agricultural technologies, were the
best performers in all three domains in this earlier study. We found few
such examples and seek to fill this gap. Policy makers, scientists, and
extension educators urgently need examples of how to identify tech-
nologies and visualize relative SI performance across multiple domains,
while taking into consideration relevance to farmers preferences.
1.2. Malawi context
The application of a SI indicators framework is explored here by
assessing Malawian agriculture through an Africa Research in
Sustainable Intensification for the Next Generation (Africa RISING)
participatory action research (PAR) project (Mungai et al., 2016). Ma-
lawi is dominated by smallholder maize (Zea mays)-based mixed
farming systems, with population density on the rise and intensification
processes underway (Jayne et al., 2014). This context provides an op-
portunity to explore SI processes within a maize-based rain-fed system
that is important across Africa to over 200 million smallholders (Blackie
and Dixon, 2016). Soil fertility is a key productivity-limiting factor on
smallholder farms throughout the country, and fertilizer subsidies have
been promoted widely as a means to address this challenge (Chirwa and
Dorward, 2013). Fertilizer use has been promoted through subsidy
programs and policies that have increased access from about 30% of
farmers (as documented in a 1999 survey of sites near the study area;
Snapp et al., 2002b) to 76–92% of farmers in the current study area
(Mungai et al., 2016). However, the sustainability of this approach is
contested, both in terms of the economic implications of reliance on
externally-funded access to fertilizers, and the extent to which the soil
resource base can sustain continuous production of maize.
Malawi farming systems are representative of high population
density locations in Southern and Eastern Africa, with limited livestock
population densities, a narrow range of crops grown and farms less than
1.0 ha in size (Jayne et al., 2014). Almost all farmers grow maize, and
generally a small portion of farmland is dedicated to a food legume and
or cash crop (Snapp et al., 2002b). Degradation of soil resources may be
the basis for remote sensing observations of negative trends in net
primary productivity on agricultural lands (Messina et al., 2017). There
is urgent need for sustainable agricultural practices to reverse this
trend.
1.3. SI technology options
Africa RISING Malawi is testing legume ‘best bet’ species and re-
commended varieties, grown in mixtures or sequenced with maize
combined with an integrated soil fertility management (ISFM) strategy,
to sustainably improve crop yields, while paying attention to local
priorities (Snapp et al., 2002a; Bezner-Kerr et al., 2007; Mhango et al.,
2013). Farmer-engagement and co-learning on SI technologies has been
promoted in Malawi since 1996, with an initial focus on agroforestry
that has evolved to emphasize multipurpose legumes combined with
targeted fertilizer use. Technologies were identified as best bet options
based on criteria that included yield, profitability and potential to
protect the environment, tested in country-wide trials and site-specific
PAR (Snapp et al., 2010). The SI framework provides an opportunity to
take this research to the next level and assess performance across a
broader, more holistic set of sustainability goals.
The objective of this study is to assess the value of applying a
practical SI indicator framework to evaluate maize-legume diversifi-
cation and ISFM technologies. The assessment of technologies includes:
1) evaluation of performance in rain-fed maize systems, in terms of
agronomic, economic, environmental, social and human dimensions; 2)
analysis of trade-offs using radar charts to visualize performance along
multiple dimensions and 3) identification of research gaps and reg-
ulators of SI through causal loop diagrams.
2. Methods
2.1. Context
2.1.1. Bio-physical environment
Central Malawi is characterized by tremendous variation in topo-
graphy. A steep escarpment bisects Central Malawi, supporting inter-
mixed agricultural areas of sub-humid, uni-modal rainfed upland
maize-based cropping. The seasonal precipitation is 800 to 1100 mm
with high inter-annual variation in quantity and in distribution
(Table 1). The three research locations were chosen in a stratified
randomized manner, with initial characterization of Central Malawi
administrative areas (designated as extension planning areas, EPAs)
into three levels of agricultural potential, low, medium and high, then
random selection of an EPA to represent each level (Mungai et al.,
Table 1
Environmental and soil characteristics of three sites where Africa RISING has conducted
participatory action research with farmers and Malawi extension since 2012, annual
average rainfall and temperature based on weather stations (Mungai et al., 2016) and a
soil characterization survey carried out on 220 farmer fields located in the project area.
Golomoti Kandeu Linthipe
Annual average rainfall (mm) 884 NAa 875
Elevation (meters above sea
level)
555 904 1238
Mean minimum temperature (°C) 13.3 10.8 7.8
Mean maximum temperature
(°C)
32.4 29.9 27.7
Top soil pH (mean, min-max) 6.4 (5.6–7.5) 6.3 (5.3–7.8) 6.2 (5.6–6.9)
Soil P (ppm, mean, min-max) 67.9 (9.4–139) 19 (1–64) 7.3 (1.8–38)
Sandy clay soils (%) 41.1 37.4 74.7
Sandy soils (%) 58.1 61.9 7.0
Clay or loamy soils (%) 0.8 0.7 18.3
a Not available = NA. For comparison purposes, Annual Average Rainfall for this lo-
cation based on a downscaled estimate = 866 mm (TRIMM, 2016).
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
78
2016). Low potential Golomoti has high evapotranspiration and poorly
distributed rainfall, and is located at low altitude along the Lake Malawi
lakeshore plain. Linthipe is a high potential site with generally well
distributed rainfall and is located in the Lilongwe mid-altitude plain.
Relative to the other two sites, medium potential Kandeu is inter-
mediate in rainfall and growth potential and is located in the highly
dissected landscape of the escarpment.
Agriculture in Central Malawi is primarily maize-based rainfed
farming with cash crops such as tobacco (Nicotiana tabacum L.) or
cotton (Gossypium hirsutum L.), and in some areas, a grain legume. Two
soil types, Luvisols and Lixisols, are widely present in the project area,
and across Malawi (22 and 26%, respectively) with good drainage
characteristics, low to moderate fertility, and generally moderate
acidity (Table 1; Dijkshoorn et al., 2016). Soil P levels are often below
the critical level of 15 mg P kg soil−1, soil organic matter status is
generally low, which contributes to the presence of degraded agri-
cultural lands (Li et al., 2017; Snapp, 1998).
2.1.2. Socio-economic environment
We developed farm typologies by clustering farms on key farm re-
sources (farm size, income sources, labor availability, livestock and
other assets) and variables related to production objectives and market
orientation (the percent of income from crop sales, and the percent of
household expenditure on food). Initial analysis showed two functional
groups for technology targeting: 1) a small number (16%) of resource
endowed large farms– that derive most of the income on-farm, and 2) a
wide band of resource poor farms with (74%) that rely on both on-farm
and off farm income (Chikowo et al., 2018). Larger farms are better
positioned to invest in mineral fertilizers and growing maize sequenced
with grain legumes, either as sole crops or intercropped with another
legume as a doubled up legume rotation (DLR). In contrast, small farms
could use low input combinations of maize-legume intercrops and DLR.
The third group of farmers (10%) focuses on off-farm activities and are
not likely to relate well with any of the SI technologies.
2.2. Participatory action research for SI farming systems
The Africa RISING program started in 2012 in Central Malawi, with
site selection, training in PAR approaches and SI technology options
(Mungai et al., 2016). Participatory research on best bet SI options was
based on a mother and baby trial design (Snapp et al., 2002a). This
involved centrally located, on-farm ‘mother trial’ located in two com-
munities at each of the three project locations for a total of six mother
trials (Fig. 1). Full suites of SI options were tested at mother trials,
encompassing ten to twelve technologies, replicated three times. About
sixty farmers in the surrounding community with each mother trial
carried out baby trials to experiment with farmer-chosen subsets of
technologies. This approach supports engagement of farmers in trying
out and adapting SI technologies on baby trial plots, including spacing,
seeding rates and combinations of crops and soil amendments (Johnson
et al., 2003; Snapp et al., 2002a).
The full suite of technologies evaluated in the mother trials included
maize grown at different fertilizer rates, manure and fertilizer applied
to maize, intensified soybean (Glycine max L.) production (double rows,
inoculum and micro-dosing of fertilizer), and maize intercropped with
pigeonpea (Cajanus cajan (L.) Millsp.), and sequenced with a DLR, pi-
geonpea grown with an understory grain legume species (Snapp et al.,
2010; Van Vugt et al., 2016). Two SI technologies that included
Fig. 1. Central Malawi Africa RISING sites in Golomoti, Kandeu and Linthipe. Baby trial locations shown as orange dots and mother trial locations as blue dots (http://
globalchangescience.org/eastafricanode/) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
79
pigeonpea were chosen for a detailed SI assessment in this study, and
systematically compared with fertilized and unfertilized maize for a
total of four technologies (Table 2). Pigeonpea is grown in Southern
Malawi, but it is a novel crop in the Central and Northern region of
Malawi (Peter et al., 2017). It was chosen based on its unique properties
that support multiple services, such as longevity of growth, high bio-
logical nitrogen fixation capacity, superior fodder material, and pro-
duction of peas that can be consumed or sold (Mhango et al., 2013; Orr
et al., 2015). Preliminary evaluations conducted in the late 1990s in-
dicated the potential of pigeonpea to address multiple environmental
and food security objectives; however, farmer acceptance varies across
Malawi, and lags in many areas (Snapp et al., 2002a, 2003; Waldman
et al., 2016).
Pigeonpea integration within Malawi cropping systems has been
focused on intercrops with maize. A novel approach evaluated here is
the DLR which involves the complementary growth habit of a slow
growing pigeonpea crop (5 to 8 month maturity) and a fast growing
understory crop such as groundnut (Arachis hypogaea L.) that matures in
about 4 months, or other annual pulses such as soybean, see http://
globalchangescience.org/eastafricanode/index.php/innovations/). The
attributes of the technologies and farmer ratings from earlier PAR in
Malawi are shown in Table 2.
The full rate fertilizer application used for maize is based on the
Malawi government recommendation and provides 9 kg P ha−1 and
69 kg N ha−1, with P applied at planting and N applied in a split
manner, at planting and about one month after crop emergence (MAFS,
Malawi Guide to Agriculture, 2012). All crops are grown following
recommended planting arrangements, with ridges prepared by hand-
hoe, 0.75 cm between ridges, and in-row spacing appropriate to the
crop species planted, as indicated below:
● Maize: seeding at 25 cm within row spacing for a plant population of
53,000 plants ha−1.
● PP-Mz intercrop: Seeding 3 plants per station of pigeonpea at 90 cm
in-row located in the same row half way between maize planting
stations at the same spacing and seeding rate; Both crops have po-
pulations of 44,000 plants ha−1 for a total of 88,000 plant ha−1
● DLR rotated with maize: Groundnut planted at a spacing of 9 cm in-
row in the space between the pigeon pea plants at
79,000 plants ha−1 (90% additive design). Pigeonpea population at
44,000 ha−1. Maize rotation phase in year two equals sole maize
plant population density.
2.3. SI indicators assessment
The analysis of SI indicators was carried out to assess how two of the
most promising technologies (PP-Mz intercrop and DLR) compared with
two conventional maize systems (continuous sole maize with and
without fertilizer) across the five SI domains. The introduction of a new
semi-perennial crop, pigeonpea, posed unique challenges due to un-
controlled grazing that limited our ability to measure yield potential,
and this required using a combination of model and field-experi-
mentation based indicators. This is a less than ideal approach as these
involve different types of metrics. However, many SI technologies rely
on introducing long duration vegetative cover species in order to sus-
tain resources, and we present this example of a mixed methods ap-
proach as an imperfect but practical means to address the challenges
posed in on-farm research where there are issues such as changing
norms in livestock control.
We selected indicators relevant to this objective and constructed
radar charts and a causal loop diagram to identify the most important
direct and indirect consequences of the legume systems across the five
domains, which is presented in the discussion. Data gaps and the as-
sumptions used in the analysis are also discussed. The following sec-
tions outline how the various SI indicators were assessed.
2.3.1. Agronomic performance and environmental impacts
Crop yields were evaluated on the six ‘mother trials’, two trials lo-
cated at each of the project sites, implemented over three growing
seasons, 2012–13, 2013–14, and 2014–15, for a total of
2 × 3 × 3= 18 site-years. Mother trials included three replicates per
site, with plots sized 6 m× 4 m. Maize and groundnut yields were
measured in May of each year by harvesting net plots (5 m × 2.25 m),
and two smaller randomly situated net plots per plot were used to
measure pigeonpea biomass and grain yield in July. Maize grain
moisture from the net plot was determined using a moisture meter, and
all grain yields are reported at 12.5% moisture content (Smith et al.,
2016). Residue biomass was collected from the net plot, and weighed in
the field to determine fresh weight of samples and subsamples. Pi-
geonpea grain yields were almost always damaged by livestock grazing
so these were modeled using APSIM as described below (and see
Ollenburger and Snapp, 2014), after calibration of APSIM with pi-
geonpea data from the sites and years where it was possible to measure
performance. Calibration involved the model being run for the three
sites using meteorological data from 1980 to 2005 to estimate the long-
term consequences of continuous use of each of the technologies, e.g.,
change over time for organic C and total N expressed as percent change
relative to initial conditions (APSIM Initiative, 2013; Holzworth et al.
2014). We followed mother trial management practices in APSIM si-
mulations, such as incorporation of three-quarters of crop residues,
removal of remainder, and moderate N fertilizer doses as shown in
Table 2; these were designed to be representative of farming in the
study area, based on survey data (Mungai et al., 2016).
Further calibration was carried out for this study based on pi-
geonpea grain from mother trial field experimentation 2015–2017,
with good agreement between observed and predicted (RMSE = 159).
Hence, model outputs for 2013–14 and 2014–15 were deemed sa-
tisfactory for use in this study (Fig. S1). We note that this model has
been extensively calibrated for Malawi soil types and crop varieties
(Robertson et al., 2005).
The diversity added to the cropping system through the legumes has
several potential environmental benefits, including months of living
cover and associated living roots that provide habitat for microorgan-
isms and enhance soil health, as well as prevent erosion. Fossil-fuel
conservation is indicated by efficient maize response to N fertilizer
(through rotational diversity and biological N fixation) (Table 2).
Table 2
‘Best bet’ maize-legume ISFM technology options for SI experimentation at Africa RISING
sites. Characteristics shown in terms of agro-ecological traits, plant life form diversity,
number of food crops present and farmer assessment.
Adapted from Snapp et al., 2010). Fertilizer rate per recommended practice, on an an-
nualized basis, Malawi Guide to Agriculture, MAFS, 2012.
Technology Option Life
forms
Months
covera
Food
crops
Fertilizer
kg/ha
Farmer
Ratingb
Sole maize
without
fertilizer
Mz0 Annual
grass
4 1 0 N
0 P
1.0
Sole maize full
fertilizer
MzNP Annual
grass
4 1 69 N
9 P
2.5
Pigeonpea-maize
intercrop
½ fertilizer
PP-Mz Shrub
legume
2.7
Annual
grass
9 2 34.5 N
4.5 P
Doubled up
legume/
maize
rotation
¼ fertilizer
DLR Shrub
legume
3.2
Annual
legume
6.5 3 17.3 N
Annual
grass
2.3 P
a Months of cover shown an annualized basis, divided by two for 2-year rotation.
b Farmer rating based on farmer survey conducted in 1998, where n = 30, scale
1 = poor, 2 = moderately poor, 3 = moderately good, 4 = good (Snapp et al., 2010).
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
80
2.3.2. Economics
We calculated net income to assess the profitability of each tech-
nology at the three research sites. We used the actual input and yields
from the mother trials in the calculations. Labor costs were not included
due to lack of data on total labor requirements for each system and due
to the complexity of appropriately valuing the opportunity cost of
household labor. Associated input/output prices for maize grain, pi-
geonpea and groundnut were based on a combination of prices received
by farmers as reported in the 2013 baseline survey (unpublished data,
see survey description in Mungai et al., 2016), and from the project's
monitoring of markets in these locations. Prices were kept constant by
location, and a sensitivity analysis was carried out by including two
price scenarios: one with average maize price ($0.21/kg) and average
legume prices to estimate the most likely profitability of these systems
and one with a high maize price ($0.45/kg, which a farmer may obtain
by storing the harvest for six months or more) and a low price for the
legumes (which could be possible if supply increased dramatically).
Values for all prices are listed in Supplementary Table S1.
We also calculated returns to fertilizer in terms of maize production
per elemental N applied as synthetic fertilizer. This is referred to as
partial factor productivity from applied N fertilizer (PFPN) in the lit-
erature (Cassman et al., 2002). This is an important consideration for
farmers who have limited resources to purchase fertilizer. Due to sea-
sonal fluctuation in prices we presented the returns in terms of grain
production per fertilizer N; PFPN which is also a metric of efficient use
of applied fertilizer. Conservation of fertilizer N is key to preservation
of scarce fossil fuel resources and limiting greenhouse gas emissions
from agriculture (Robertson and Vitousek, 2009).
2.3.3. Food security and gender equity
Increased availability of legume grain produced could contribute to
improved family nutrition, and although testing this was beyond the
scope of the study, a seminal study in Asia points in this direction
(Darmadi-Blackberry et al., 2004). As described in Smith et al. (2016),
we used modeled yield results based on an average farm size per site,
and 100% adoption of a technology to estimate the probability of a
household producing all the calories and protein that it needed, as a
metric related to food security. Mean household demographics and
farm size was based on survey data from the project's baseline (Mungai
et al., 2016), and used to calculate calorie and protein requirements.
We also used weather series and simulation modeling to predict failed
maize harvest (less than 33% of the mean harvest) for each system over
the simulation period to demonstrate the riskiness of the technology.
These are indicators that are practical to estimate, based on data gen-
erated through PAR with farmers, linked to crop modeling, in a manner
typical of many agronomic research in development projects.
As an indicator to evaluate potential gender equity implications of
the technologies, we report the percent of female farmers who tested
the legume technologies and preferred them to sole maize systems. This
indicator is straight forward to assess and can be used as a means to
take into consideration the views of a group that is often marginalized.
The data was collected in a July 2014 survey of 236 baby trial farmers
that documented pairwise rankings of technologies (not all farmers
tested all technologies). The results were gender disaggregated, and we
report the percent of farmers who preferred PP-Mz and DLR over sole
maize, and maize to PP-Mz.
3. Results
The data for each of the SI indicators are presented separately by
site, due to important interactions with agro-ecology (Tables 3–5). The
results are presented visually as radar charts to facilitate comparison of
the technologies across the five domains (Figs. 2–4), to identify tech-
nology candidates to implement on farms, and promote through policy.
3.1. Productivity of the land
At all sites, unfertilized sole maize yields were very low and ferti-
lized maize were high. Sole maize was responsive to environment,
where grain yield was highest at the mesic Linthipe site (Table 3 and
Fig. 2), and lowest at the marginal Golomoti site (Table 5 and Fig. 4).
Maize yields from the PP-Mz intercrop were maintained at high levels,
Table 3
Indicators for sustainable intensification for four technologies at Linthipe, Malawi.
Indicators by domain Units Maize0 MaizeNP PP-Mz DLR Data sources
Productivity
Maize yielda kg/ha 1087 5908 4536 2567 Mother trial – 3 years
Maize residues kg/ha 3845 10,051 6935 4653 Model average - 25 yrs.
Legume residues kg/ha 0 0 4279 4949 Mother trials and APSIM model
Legume yield (total) kg/ha 0 0 487 792 Combined
Pigeonpea yield kg/ha 0 0 487 282 Modeled avg. - 25 years
Groundnut yieldb kg/ha 0 0 0 511 Mother trials – 3 years
Economic
Profitability – avg. prices $/ha $205 $1118 $1028 $841 Prices from surveys
Profitability – high maize price $/ha $470 $2560 $2031 $1280 Maize price at seasonal high
Partial factor productivity of nitrogen fertilizer kg maize/kg N n.a. 86 146 195 Mother trial – 3 years
Environment
Months of soil cover Month 5 5 9 9 Pigeonpea duration
Soil carbon - relative change % change −10.8% −7.5% −0.3% −3.0% Modeled 25-year simulation (APSIM)
Annual change in soil carbon change in % C −0.00964 −0.00665 −0.00031 −0.00266 Modeled % of soil dry mass (top 15 cm)- APSIM
Soil N - relative change % change −10.6% −6.7% 1.3% −2.3% Modeled 25-year simulation (APSIM)
Soil N – end value of 25-yr simulation Total soil N - % mass 0.1158 0.1208 0.1314 0.1277 Modeled % of soil dry mass (top 15 cm)- APSIM
Human condition
Probability of food sufficiency % 12% 96% 92% 100% Modeled based on survey data
% years without crop failure % 92% 96% 100% 100% Modeled avg. - 25 years
Social
Gender (women preferring system)c % farmers 33% 33% 67% 50% Pairwise ranking (n = 46)
Farmers preferring system % farmers 33% 33% 67% 60% Pairwise ranking (n = 69)
a n = 18 for all but DLR, whose n = 6; Least significant difference across sites, systems and seasons = 875.7.
b n = 6; Least significant difference across sites and seasons = 735.7.
c For all pairwise rankings maize compared with PP-Mz, legumes systems compared with Mz.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
81
although not quite as high as fertilized sole maize. In addition to maize
grain, the PP-Mz intercrop has the potential to produce an additional
crop, of a nutrient-dense pigeonpea grain that may fetch a higher price
than maize (but not in all cases).
The DLR is sequenced as follows: pigeonpea-groundnut intercrop is
produced before maize, so only one year in two has a maize crop. Thus,
on an annualized basis maize grain yields were modest (where an-
nualized maize is a metric closely related to food security on farms with
small landholdings; see Droppelmann et al., 2017). At the same time,
DLR produced the largest amount of combined yields (from maize plus
legume crops), and this grain production required modest fertilizer
inputs (one-quarter rate, on an annualized basis). The ~800 kg of
Table 4
Indicators for sustainable intensification for four technologies at Kandeu, Malawi.
Indicators by domain Units Maize0 MaizeNP PP-Mz DLR Data sources
Productivity
Maize yielda kg/ha 1003 5018 4253 1742 Mother trial – 3 years
Maize residues kg/ha 3583 9975 6943 4854 Model average - 25 yrs.
Legume residues kg/ha 0 0 6573 5496 Mother trials and APSIM model
Legume yield (total) kg/ha 0 0 751 858 Combined
Pigeonpea yield kg/ha 0 0 751 492 Modeled avg. - 25 years
Groundnut yieldb kg/ha 0 0 0 367 Mother trials – 3 years
Economic
Profitability – avg. prices $/ha $188 $935 $1054 $637 Prices from surveys
Profitability – high maize price $/ha $432 $2159 $1932 $866 Maize price at seasonal high
Partial factor productivity of N fertilizer kg maize/kg N n.a. 72.7 145.1 150.7 Mother trial – 3 years
Environment
Months of soil cover Month 5 5 9 9 Pigeonpea duration
Soil carbon - relative change % change −5.9% 1.7% 11.2% 10.0% Modeled 25-year simulation (APSIM)
Annual change in soil carbon change in % C −0.00238 0.00067 0.00453 0.00405 Modeled % of soil dry mass (top 15 cm)- APSIM
Soil N - relative change % change −5.3% 3.5% 14.3% 12.6% Modeled 25-year simulation (APSIM)
Soil N – end value of 25-yr simulation Total soil N - % mass 0.0712 0.0778 0.0860 0.0852 Modeled % of soil dry mass (top 15 cm)- APSIM
Human condition
Probability of food sufficiency % 35% 100% 85% 100% Modeled based on survey data
% years without crop failure % 92% 96% 96% 96% Modeled avg. - 25 years
Social
Gender (women preferring system)c % farmers 67% 67% 33% 43% Pairwise ranking (n = 55)
Farmers preferring system % farmers 57% 57% 43% 42% Pairwise ranking (n = 82)
a n = 18 for all but DLR, whose n = 6; Least significant difference across sites, systems and seasons = 875.7.
b n = 6; Least significant difference across sites and seasons = 735.7.
c For all pairwise rankings maize compared with PP-Mz, legumes systems compared with Mz.
Table 5
Indicators for sustainable intensification for four technologies at Golomoti, Malawi.
Indicators by domain Units Maize0 MaizeNP PP-Mz DLR Data sources
Productivity
Maize yielda kg/ha 363 3543 3197 1993 Mother trial – 3 years
Maize residues kg/ha 2127 8217 8218 4015 Model average - 25 yrs.
Legume residues kg/ha 0 0 2855 3145 Mother trials and APSIM model
Legume yield (total) kg/ha 0 0 326 796 Combined
Pigeonpea yield kg/ha 0 0 326 172 Modeled avg. - 25 years
Groundnut yieldb kg/ha 0 0 0 624 Mother trials – 3 years
Economic
Profitability – avg. prices $/ha $56 $631 $701 $767 Prices from surveys
Profitability – high maize price $/ha $145 $1496 $1411 $1061 Maize price at seasonal high
Partial factor productivity of N fertilizer kg maize/kg N n.a. 51.3 102.1 161.7 Mother trial – 3 years
Environment
Months of soil cover Month 5 5 9 9 Pigeonpea duration
Soil carbon - relative change % change −9.8% −1.4% 2.7% 2.4% Modeled 25-year simulation (APSIM)
Annual change in soil carbon change in % C −0.00322 −0.00046 0.00089 0.00078 Modeled % of soil dry mass (top 15 cm)- APSIM
Soil N - relative change % change −9.3% 0.3% 5.0% 4.6% Modeled 25-year simulation (APSIM)
Soil N – end value of 25-yr simulation Total soil N - % mass 0.0558 0.0618 0.0647 0.0651 Modeled % of soil dry mass (top 15 cm)- APSIM
Human Condition
Probability of food sufficiency % 0% 92% 85% 92% Modeled based on survey data
% years without crop failure % 62% 88% 88% 88% Modeled avg. - 25 years
Social
Gender (women preferring system)c % farmers 20% 20% 80% 50% Pairwise ranking (n = 43)
Farmers preferring system % farmers 17% 17% 83% 25% Pairwise ranking (n = 90)
a n = 9 for all but DLR, whose n = 3; Least significant difference across sites, systems and seasons = 875.7.
b n = 6; Least significant difference across sites and seasons = 735.7.
c For all pairwise rankings maize compared with PP-Mz, legumes systems compared with Mz.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
82
legume grain ha−1 that was produced annually by DLR was consistent
across all sites, including the marginal as well as the high yield po-
tential sites.
3.2. Economics
Net incomes were very low for unfertilized maize at all sites. For the
other systems, there was a strong interaction with site: the three tech-
nologies had similar levels of net income at high potential Linthipe,
whereas in moderate potential Kandeu the PP-Mz technology had the
highest net income (Table 4 and Fig. 3) and in marginal Golomoti the
DLR system had the highest. Returns to fertilizer were greater with
systems that had a larger legume presence: PFPN varied from 50 to
85 kg grain kg N−1 for continuous maize, from 100 to 145 for PP-Mz
and was the highest for maize sequenced with DLR, from 150 to 194.
Linthipe was associated with the highest PFPN, for each technology.
3.3. Environment
The technologies with long-lived legumes produced large amounts
of legume biomass, and had substantially greater living coverage (~¾
of the year). In comparison, a sole maize system can produce large
amounts of maize residues, but only has a living plant present for
~4 months (Tables 3–5). Environmental benefits through conservation
of fertilizer include energy saved to produce the fertilizer, and pre-
servation of water and air quality (Robertson and Vitousek, 2009). The
amount of fertilizer required to support farming system productivity
was halved through the introduction of a PP-Mz intercrop (fertilized at
Fig. 2. Indicators of sustainable intensification for legume
systems (pigeonpea-maize intercrop half-rate NP
fertilizer = PP-Mz and doubled up legume rotation of pi-
geonpea-groundnut rotated with maize = DLR) and sole
maize systems (unfertilized = Mz0 and full fertilizer
MzNP), in Linthipe, Malawi. Crop and residue yields are
reported on an annualized basis, i.e., halved for 2 year
rotation DLR. Note: For specific values and details on data
sources see Table 3.
Fig. 3. Indicators of sustainable intensification for legume
systems (pigeonpea-maize intercrop half-rate NP
fertilizer = PP-Mz and doubled up legume rotation of pi-
geonpea-groundnut rotated with maize = DLR) and sole
maize systems (unfertilized = Mz0 and full fertilizer
MzNP), in Kandeu, Malawi. Crop and residue yields are
reported on an annualized basis, i.e., halved for 2 year
rotation DLR. Note: For specific values and details on data
sources see Table 4.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
83
half rate) and halved again in a DLR rotation (unfertilized in the DLR
phase and fertilized at half rate in the maize phase of the two year
system) (Table 2).
Soil accrual of N and C varied with site, as assessed from APSIM
simulations. Soil organic matter gain was often modest, but some ac-
crual was observed in Kandeu (Table 4 and Fig. 3). The PP-Mz and DLR
systems generally were associated with gains in soil organic matter (or
maintenance in the case of Linthipe), whereas sole maize was asso-
ciated with losses in all cases.
3.4. Human condition
Apart from unfertilized sole maize, all technologies assessed here
allowed the average household to meet 100% of calorie requirements in
almost all seasons. The legume systems had the same or fewer years of
crop failure than fertilized maize. Unfertilized sole maize was asso-
ciated with high crop failure, particularly in the marginal Golomoti site.
3.5. Social
Farmer ratings from our PAR research indicated that the technolo-
gies were perceived as having similar labor requirements, with an
average rating (based on a scale of 1 to 4, where 1 = low and 4 = high
labor demand) of 3.33 for sole maize, 2.99 for PP-Mz intercrop and 3.20
for the DLR maize system. Female farmers at two sites, Linthipe and
Golomoti, preferred PP-Mz over all other technologies. Notably, a large
majority of women at Golomoti prefer the PP-Mz intercrop, presumably
for direct access to the nutritious pigeonpea grain along with maize
grain at this marginal site with moderate access to markets (Table 5 and
Fig. 4).
4. Discussion
The SI indicators in five domains showed high performance overall
of pigeonpea-diversified systems, compared with monoculture maize,
and some important tradeoffs. Benefits included modest fertilizer re-
quirements, soil building properties, production of nutritionally-bene-
ficial legume grain and women farmers rated these technologies highly
(with the exception of one site). In the pigeonpea diversified systems
there were no major reductions in total grain yield relative to sole
maize systems, nor was food security notably altered, however maize
grain was sometimes reduced. Income goals were met in most cases; yet
we note that this was strongly dependent on the three-way price ratio of
maize grain-legume grain-fertilizer. There are few studies that take into
account this dynamic economic context: the moderate grain yields of
legume crops (which are biologically constrained) are often offset by
high market value relative to maize grain, but not always.
Generally, the PP-Mz intercrop produced the highest net value per
area, with the exception of the high maize grain price scenario where
continuous, fertilized maize performed well. Production of grain legume
may provide a nutritional boost and an extra source of income (Foyer
et al., 2016); however the necessity to produce sufficient maize to meet
family requirements is also recognized in our analysis by including an-
nualized maize production as one of the production indicators
(Droppelmann et al., 2017). At the Linthipe site fertilized sole maize was
highly productive, with associated high economic returns. Similar re-
sults were observed in Kenya, where an on-farm study evaluated eco-
nomic performance of legume-diversified maize systems where legu-
me:maize price ratios were varied from 1:1 to 4:1 (Rao and Mathuva,
2000). As the price of legume grain increased relative to maize, net-
present-value returns associated with a PP-Mz intercrop increased by
~20 to 80%, compared with continuous maize. A maize-DLR rotation
system with a pigeonpea-cowpea intercrop, had modest returns relative
to a PP-Mz intercrop, with the exception of the highest (4:1) legume:-
maize price ratio where DLR performed well (Rao and Mathuva, 2000).
Similarly, in our study the DLR system performed as well as the PP-Mz
intercrop but only at high legume grain prices (Figs. 2–4).
Many assessments of SI technologies focus on tradeoffs, and com-
monly find inverse relationships between productivity and environ-
mental services (Droppelmann et al., 2017). Our assessment used the SI
framework to move beyond productivity vs. environment, to consider
aspects of SI such as women's ratings of technologies, and modeled food
security risks. Our results are consistent with others in that diversified
SI technologies did not produce the highest amount of grain yield or
income (Snapp et al., 2010), but were highly associated with con-
servation of resources. In addition, pigeonpea diversified technologies
were highly ranked by women farmers at two out of three sites, and had
no out-sized risks of food insecurity (Figs. 2–4). These results are con-
sistent with an earlier study from Northern Malawi (Bezner-Kerr et al.,
2007). Altogether, we recognize that are many other aspects of human
condition that were beyond the scope of this study, where we used
survey data and models to explore food security risk; we recognize that
Fig. 4. Indicators of sustainable intensification for legume
systems (pigeonpea-maize intercrop half-rate NP
fertilizer = PP-Mz and doubled up legume rotation of pi-
geonpea-groundnut rotated with maize = DLR) and sole
maize systems (unfertilized = Mz0 and full fertilizer
MzNP), in Golomoti, Malawi. Crop and residue yields are
reported on an annualized basis, i.e., halved for 2 year
rotation DLR. Note: For specific values and details on data
sources see Table 5.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
84
much remains to be done.
Food security is challenging to assess, and we focused on risk of crop
failure as a key aspect. We found few instances of pigeonpea diversified
system failure in low rainfall seasons or in the dry environment of
Golomoti (Fig. 4). The long duration of growth, indeterminant yield
production, combined with soil cover from senesce of pigeonpea leaves
(which drops leaves throughout the growing season), together these
traits may contribute to environmental services and buffering of risk
economically. A study of smallholder farmers located on Mt. Kenya
found income was higher with crop diversification (McCord et al.,
2015). By growing drought tolerant legumes, farmers may be able help
ensure food security through direct consumption of protein and oil-rich
harvest, as well as through sales of the legumes that support food
purchases, even if maize fails (Foyer et al., 2016). In Northern Malawi
where communities were supported to grow DLR crops, combined with
participatory training and nutrition education, many were found to
have improved child height by weight scores (Bezner Kerr et al., 2011).
Taken together with our results, these are indications that pigeonpea
diversification can help support sustainable trajectories of intensifica-
tion. At the same time, indepth tradeoff analysis is needed to elucidate
barriers, and potential costs, associated with uptake of these technol-
ogies.
4.1. Tradeoff analysis
The radar charts presented here illustrate that environmental con-
ditions provide an important context for tradeoff analysis. For example,
diversified systems raise concerns due to potential competition with
maize, illustrated here by the high potential site – Linthipe - where
maize yield was suppressed in a PP-Mz intercrop (Fig. 2). There is a
maize yield penalty, although overall yield (combined maize and le-
gume grain) is often maintained. In the marginal Golomoti site, by
contrast, the tradeoff was little to nil in a PP-Mz intercrop (Fig. 4). This
finding is similar to a Kenya study of paired row maize-grain legume
systems: this legume diversified system was more productive and
profitable compared with conventional sole maize, and this was most
pronounced at a marginal site (Mucheru-Muna et al., 2010).
Our study illustrates that the large amount of maize grain produced
and net income from fertilized maize may be a barrier to adoption of
alternative systems. For farmers with small land holdings, this tradeoff
can be steep, and fertilized maize is the most appropriate technology
according to many of the metrics shown here (Fig. 2). However, there
are sustainability issues with this technology, as indicated by the poor
land cover and (predicted) declines in soil C (Fig. 2). More attention is
needed to the adoption barrier faced by poorly-resourced households
who may not be able to diversify beyond maize production without
government investments, in food safety nets, and in agronomic educa-
tion. For example, knowledge is key to fine-tuning SI technologies,
through crop population densities and use of ratooning practices,
combined with judicious use of inputs (Letourneau et al., 2011; Rogé
et al., 2016).
In systems theory, a causal loop diagram is a way of representing the
positive and negative linkages among complex system components,
which is a critical first step in understanding system behavior and
tradeoffs. In Fig. 5 we present a causal loop diagram of legume di-
versified systems, including direct and indirect effects across the five SI
domains. Starting with increased production of groundnut and pi-
geonpea (through the intercrop and DLR systems), predicted improve-
ments in soil N and C in turn could result in more stable maize yields
and reduced need for fertilizer. The legumes provide more stable profits
as well as nutritious food and increased security of staples via maize
production. One clear tradeoff mentioned above is that of less land
available for maize grain yield with the DLR system (double-lined
arrow at top of Fig. 5). In developing the diagram, we were also able to
identify linkages that are not currently analyzed through our research.
Those data gaps are summarized next, as a focal point for future in-
vestigations.
4.2. Social conflict regarding residue management
A key assumption for the analysis presented in the results is that
farmers will be able to utilize the legume residues to enhance soil re-
sources (dashed box in Fig. 5). Currently many farmers burn residues,
while others transport legume residues to the home with the harvest,
where they may be consumed by livestock or composted (Mungai et al.,
2016). More research is required to document farmer perceptions and
practices around residue management but other studies in Malawi have
indicated little knowledge regarding long-term negative consequences
Fig. 5. Causal loop diagram of the direct and indirect effects driven by increased production of groundnut and pigeonpea. Notes: Black arrows show positive effects, bold arrows show
negative effects and dashed arrows may be positive or negative, where tentative linkages indicate further research needs. The strength of the effect is shown by the number of + or – signs
near the head of the arrow.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
85
of burning residues, and widespread concerns about pests associated
with residue incorporation practices (Snapp et al., 2002b).
Limited production of residue biomass, and complex, competing
demands for this resource, is emerging as a key SI issue (Valbuena et al.,
2012). Integration of crops with longer growth cycles into production
systems supports intensification of crop-livestock systems in a sustain-
able manner, through generation of nutrient-enriched feed, extension of
photosynthetic activity, and soil-root biology for enhanced nutrient
access (Drinkwater and Snapp, 2008). However, there is also risk that
introduction of long-lived legume crops could acerbate community
conflicts among crop producers and livestock owners (double-lined
arrow to the left in Fig. 5, and see Baudron et al., 2014). Studies of
community and farm level resource flows in Kenya indicate a status quo
that favors livestock owners who control the manure from their ani-
mals, though the residues they feed on come from the entire community
(Tittonell et al., 2005). Long term soil benefits can accrue from legume
production if the residues enhance animal production and the increased
manure is returned to the fields (Vesteeg and Koudokpon, 1993); edu-
cation on this topic has been identified as a simple and cost-effective
approach to supporting climate-smart mitigation and adaptation out-
comes in Tanzania livestock systems (Notenbaert et al., 2017).
The potential for crop-livestock conflict in resource use, as well as
potential complementarity, has emerged within the Africa RISING
project. In 2012 on-farm research trials were located close to villages
and were subjected to high livestock pressure, which led to a complete
loss of many pigeonpea plants, and grain as well. Many of pigeonpea's
benefits are derived from its long growth period, which leaves it vul-
nerable to grazing once other crops are harvested (Rogé et al., 2016).
Farmers who have expanded their experimentation with pigeonpea
have located these new ‘baby trial expansion plots' at a distance from
the village. How to support innovations in community management of
livestock is emerging as a key research gap, one that needs to be ad-
dressed concurrent with crop diversification to support en-
vironmentally-sound practices. Supporting this finding, crop-livestock
integration with improved forages was recently identified as one of the
only smallholder farming systems predicted to generate successful win-
wins for mitigation and adaptation to climate change (Hammond et al.,
2017).
4.3. Complex gender effects
There are complex gender effects related to the legume systems
(dashed arrow to the left in Fig. 5). Through application of a SI in-
dicator framework we explicitly considered women's preferences for
technologies, which provided novel insights at two locations (Figs. 2
and 4). Women are primarily responsible for growing crops for nutri-
tion in Malawi and this should be taken into consideration if human
nutrition is a policy goal. Indeed, we found that women ranked PP-Mz
intercrops and DLRs substantially higher than men at two of the Africa
RISING sites. When a legume becomes a cash crop, literature provides
precedents whereby there is a shift towards male production of the crop
and control of generated cash (Njuki et al., 2011). This could have
negative impacts, if women and children are worse off due to legume
crop sales and reduced consumption of legume products. Malawi's
groundnut exports have increased by 18% per year from 2004 to 2014
(Edelman and Aberman, 2015).
If legumes are grown for soil fertility improvement, then they may
be valued by both genders. This is illustrated by a recent study in
Central Malawi utilizing choice experiment methodology to explore
farmer perceptions and preferences for legumes (Waldman et al., 2016).
Both men and women were found to be interested in soil fertility at-
tributes of legumes, although there were some surprising gaps in
knowledge regarding which legume types and management practices
build soil fertility. A survey of Central Malawi farmers found that
women were twice as likely as men to be involved in experimenting
with legume varieties and combinations with other SI technologies
(Hockett and Richardson, 2016). This is consistent with a seminal study
on the gendered nature of bean knowledge in Malawi (Ferguson, 1994).
4.4. Uncertain labor requirements
Another data gap involves the challenges to obtain reliable in-
formation on labor requirements for SI technologies on smallholder
farms. Work is under way to have farmers rate each technology that
they tested in their baby trials according to their perception of labor
requirements. Initial feedback from farmers is that harvesting pi-
geonpea requires an extra trip to the field during a time of processing
the maize harvest (dashed double-lined arrow pointing down in Fig. 5),
yet overall farmer ratings of labor requirements were similar across
technologies (ranged from 3 to 3.3, on scale of 1 = low to 4 = high).
The challenges associated with understanding labor ratings point out
that an iterative assessment of SI technologies is needed, as farmers
progress from concept to testing, and from adaptation to wide scale
adoption. This is illustrated by participatory research in Mali, which
provides an example of how stakeholders could be engaged with to
further the process of adaptation to improve labor ratings and address
other locally identified priorities (Falconnier et al., 2017).
5. Conclusion
This analysis of options for sustainable intensification in Malawi has
demonstrated the importance of assessing sustainability in a holistic
manner through application of a SI indicator framework. The mixed
methods approach using data from participatory on-farm trials linked to
modeling showed the potential for achieving improved soil fertility
while still enabling farmers to meet their immediate consumption needs
and address less easily quantified traits as ascertained by farmer ratings.
Although social dimensions were not studied indepth, the importance
of considering female farmer ratings was illustrated as these did not
always line up with profitability or productivity traits. Another im-
portant lesson was that agro-ecozone markedly affected tradeoffs
around production, profits and soil impacts. Extension education may
be particularly necessary in marginal environments where commercial
agriculture has modest returns and improving soil quality is complex,
requiring greater knowledge and a longer time horizon. Also, risk mi-
tigation is crucial in these marginal areas, which may require govern-
ment support to motivate farmers to invest in SI technologies.
Overall, the analysis of indicators for SI supported elucidation of
tradeoffs and synergies as they relate to the diverse objectives of
farmers, and of society more broadly. It also facilitated the identifica-
tion of important research gaps. For promising SI options to be widely
adopted it will be important to ascertain how to avoid or minimize
conflicts with livestock owners and to better understand the complex
gender effects related to legume production. It will also be important to
document labor requirements of each system and adjust the technolo-
gies where possible. Taken together, this study found evidence that SI is
achievable to varying degrees, if attention is paid to agroecosystem
potential, and to developing policy interventions that consider the
implications of subsidies that influence tradeoffs among investments in
cereals, legumes and nutrient management.
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.agsy.2018.01.012.
Funding
This work was supported by the United States Agency for
International Development (USAID) under Cooperative Agreement No.
AID-OAA-L-14-00006, Feed the Future Innovation Lab for Sustainable
Intensification and USAID Africa RISING-Malawi (USAID No. AID-BFS-
G-11-00002), led by IITA with Michigan State University (MSU),
Lilongwe University of Agriculture and Natural Resources (LUANAR),
International Center for Tropical Agriculture (CIAT) and Malawi DAES.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
86
The contents are the sole responsibility of the authors and do not ne-
cessarily reflect the views of USAID or the United States Government.
Acknowledgments
The Malawi Government Ministry of Agriculture Department of
Agricultural Extension Services (DAES) and smallholder farmers in
Central Malawi are gratefully acknowledged as our inspiring partners.
References
APSIM Initiative. 2013. APSoil. [http://www.apsim.info/Products/APSoil.aspx].
Baudron, F., Jaleta, M., Okitoi, O., Tegegn, A., 2014. Conservation agriculture in African
mixed crop-livestock systems: expanding the niche. Agric. Ecosyst. Environ. 187,
171–182.
Baulcombe, D., Crute, I., Davies, B., Dunwell, J., Gale, M., Green, N., Jones, J., Mee, S.,
Pretty, J., Simpson, A., Stilgoe, J., Sutherland, W., Toulmin, C., 2009. Reaping the
Benefits: Science and the Sustainable Intensification of Global Agriculture. The Royal
Society, London.
Bell, S., Morse, S., 2004. Delivering sustainability therapy in a projectified world. In:
Proceedings International Sustainable Development Research Conference, 2004,
Manchester, UK, . http://oro.open.ac.uk/114/.
Bezner Kerr, R.B., Berti, P.R., Shumba, L., 2011. Effects of a participatory agriculture and
nutrition education project on child growth in northern Malawi. Public Health Nutr.
14 (08), 1466–1472.
Bezner-Kerr, R., Snapp, S.S., Chirwa, M., Shumba, L., Msachi, R., 2007. Participatory
research on legume diversification with Malawian smallholder farmers for improved
human nutrition and soil fertility. Exp. Agric. 43, 437–453.
Blackie, M., Dixon, J., 2016. Maize mixed farming systems: an engine for rural growth. In:
Dixon, J., Garrity, D., Boffa, J.-M., Williams, T., Amede, T., Auricht, C., Lott, R.,
Mburathi, G. (Eds.), Farming Systems and Food Security in Africa: Priorities for sci-
ence and policy under global change. Routledge, London and New York.
Cassman, K.G., Dobermann, A., Walters, D.T., 2002. Agroecosystems, nitrogen-use effi-
ciency, and nitrogen management. AMBIO 31 (2), 132–140.
Chikowo, R., Snapp, S.S., Grabowski, P., Odhong, J., Hoeschle-Zeledon, I., Bekunda, M.,
2018. Farm typologies use in targeting sustainable intensification. Agron. Sustain.
Dev (Ms. in press).
Chirwa, E., Dorward, A., 2013. Agricultural Input Subsidies: The Recent Malawi
Experience. Oxford University Press, Oxford, UK.
Darmadi-Blackberry, I., et al., 2004. Legumes: the most important dietary predictor of
survival in older people of different ethnicities. Asia Pac. J. Clin. Nutr. 13, 217–220.
Dijkshoorn, J.A., Leenaars, J.G.B., Huting, J.H.M., Kempen, B., 2016. Soil and terrain
database of the Republic of Malawi. In: Report 2016/01. ISRIC―World Soil
Information, Wageningen, Netherlands, http://dx.doi.org/10.17027/isric-wdcsoils.
20160001.
Drinkwater, L.E., Snapp, S.S., 2008. Nutrients in agroecosystems: rethinking the man-
agement paradigm. Adv. Agron. 92, 163–186.
Droppelmann, K.J., Snapp, S.S., Waddington, S., 2017. Sustainable intensification tech-
nologies for smallholder maize-based farming systems in sub-Saharan Africa. Food
Secur. 9, 133. http://dx.doi.org/10.1007/s12571-016-0636-0.
Edelman, B., Aberman, N.L., 2015. Promoting exports of low-aflatoxin groundnut from
Malawi. In: MaSSP Policy Note 21. Washington D.C, International Food Policy
Research Institute (IFPRI). http://ebrary.ifpri.org/cdm/ref/collection/p15738coll2/
id/129180.
Falconnier, G.N., Descheemaeker, K., Van Mourik, T.A., Adam, M., Sogoba, B., Giller,
K.E., 2017. Co-learning cycles to support the design of innovative farm systems in
southern Mali. Eur. J. Agron. 89, 61–74.
Ferguson, A., 1994. Gendered science: a critique of agricultural development. Am.
Anthropol. 96, 540–552.
Foyer, C.H., Lam, H.M., Nguyen, H.T., Siddique, K.H., Varshney, R.K., Colmer, T.D.,
Cowling, W., Bramley, H., Mori, T.A., Hodgson, J.M., Cooper, J.W., 2016. Neglecting
legumes has compromised human health and sustainable food production. Nature
Plants 2, 16112.
Godfray, H.C.J., 2015. The debate over sustainable intensification. Food Secur. 7,
199–208. http://dx.doi.org/10.1007/s12571-015-0424-2.
Gunton, R.M., Firbank, L.G., Inman, A., Winter, D.M., 2016. How scalable is sustainable
intensification? Nat. Plants. 2, 1–4. http://dx.doi.org/10.1038/NPLANTS.2016.65.
Hammond, J., Fraval, S., van Etten, J., Suchini, J.G., Mercado, L., Pagella, T., Frelat, R.,
Lannerstad, M., Douxchamps, S., Teufel, N., Valbuena, D., van Wijk, M.T., 2017. The
rural household multi-indicator survey (RHoMIS) for rapid characterisation of
households to inform climate smart agriculture interventions: description and ap-
plications in East Africa and central America. Agric. Syst. 151, 225–233.
Hockett, M., Richardson, R.B., 2016. Examining the drivers of agricultural experi-
mentation among smallholder farmers in Malawi. Exp. Agric. 1–21. http://dx.doi.
org/10.1017/S0014479716000673.
Holzworth, D.P., Huth, N.I., Zurcher, E.J., Herrmann, N.I., McLean, G., Chenu, K., van
Oosterom, E.J., Snow, V., Murphy, C., Moore, A.D., Brown, H., 2014.
APSIM–evolution towards a new generation of agricultural systems simulation.
Environ. Model Softw. 62, 327–350.
Jayne, T.S., Chamberlin, J., Headey, D.D., 2014. Land pressures, the evolution of farming
systems and development strategies in Africa: a synthesis. Food Policy 48, 1–17.
Johnson, N., Lilja, N., Ashby, J.A., 2003. Measuring the impact of user participation in
agricultural and natural resource management research. Agric. Syst. 78, 287–306.
Letourneau, D.K., Armbrecht, I., Rivera, B.S., Lerma, J.M., Carmona, E.J., Daza, M.C.,
Escobar, S., Galindo, V., Gutiérrez, C., López, S.D., Mejía, J.L., 2011. Does plant di-
versity benefit agroecosystems? A synthetic review. Ecol. Appl. 21 (1), 9–21.
Li, G., Messina, J.P., Peter, B.G., Snapp, S., 2017. Mapping land suitability for agriculture
in Malawi. Land Degrad. Dev. http://dx.doi.org/10.1002/ldr.2723.
Loos, J., Abson, D.J., Chappell, M.J., Hanspach, J., Mikulcak, F., Tichit, M., Fischer, J.,
2014. Putting meaning back into “sustainable intensification”. Front. Ecol. Environ.
12, 356–361. http://dx.doi.org/10.1890/130157.
McCord, P.F., Cox, M., Schmitt-Harsh, M., Evans, T., 2015. Crop diversification as a
smallholder livelihood strategy within semi-arid agricultural systems near Mount
Kenya. Land Use Policy 42, 738–750.
Messina, J.P., Peter, B.G., Snapp, S.S., 2017. The myth of the Malawian farm input sub-
sidy program. Nat. Plants 3. http://dx.doi.org/10.1038/nplants.2017.13.
Mhango, W., Snapp, S.S., Kanyama-Phiri, G.Y., 2013. Opportunities and constraints to
legume diversification for sustainable cereal production on African smallholder
farms. Renew. Agric. Food Syst. 28, 234–244.
Ministry of Agriculture and Food Security (MAFS), 2012. Malawi Guide to Agriculture.
Published by Agricultural Communication Branch. Government of Malawi, Lilongwe,
Malawi.
Mucheru-Muna, M., Pypers, P., Mugendi, D., Kung'u, J., Mugwe, J., Merckx, R.,
Vanlauwe, B., 2010. A staggered maize–legume intercrop arrangement robustly in-
creases crop yields and economic returns in the highlands of Central Kenya. Field
Crop Res. 115 (2), 132–139.
Mungai, L.M., Snapp, S., Messina, J.P., Chikowo, R., Smith, A., Anders, E., Richardson,
R.B., Li, G., 2016. Smallholder farms and the potential for sustainable intensification.
Front. Plant Sci. 7. http://dx.doi.org/10.3389/fpls.2016.01720.
Njuki, J., Kaaria, S., Chamunorwa, A., Chiuri, W., 2011. Linking smallholder farmers to
markets, gender and intra-household dynamics: does the choice of commodity
matter? Eur. J. Dev. Res. 23, 426–443.
Notenbaert, A., Pfeifer, C., Silvestri, S., Herrero, M., 2017. Targeting, out-scaling and
prioritising climate-smart interventions in agricultural systems: lessons from applying
a generic framework to the livestock sector in sub-Saharan Africa. Agric. Syst. 151,
153–162.
Ollenburger, M., Snapp, S.S., 2014. Model applications for sustainable intensification of
African water-limited cropping systems in a changing world. Adv. in Model 5,
375–398.
Orr, A., Kambombo, B., Roth, C., Harris, D., Doyle, V., 2015. Adoption of integrated food-
energy systems: improved cookstoves and pigeonpea in southern Malawi. Exp. Agric.
51, 191–209.
Peter, B., Messina, J.P., Frake, A.N., Snapp, S.S., 2017. Scaling Agricultural Innovations:
Pigeonpea in Malawi. The Professional Geographer, Ms (in press).
Petersen, B., Snapp, S.S., 2015. What is sustainable intensification: views from experts.
Land Use Policy 46, 1–10. http://dx.doi.org/10.1016/j.landusepol.2015.02.002.
Pretty, J., Toulmin, C., Williams, S., S., 2011. Sustainable intensification in Africa agri-
culture. Int. J. Agric. Sustain. 9, 5–24.
Rao, M.R., Mathuva, M.N., 2000. Legumes for improving maize yields and income in
semi-arid Kenya. Agric. Ecosyst. Environ. 78 (2), 123–137.
Robertson, G.P., Vitousek, P.M., 2009. Nitrogen in agriculture: balancing the cost of an
essential resource. Annu. Rev. Environ. Resour. 34, 97–125.
Robertson, M.J., Sakala, W., Benson, T., Shamudzarira, Z., 2005. Simulating response of
maize to previous velvet bean (Mucuna pruriens) crop and nitrogen fertilizer in
Malawi. Field Crop Res. 91, 91–105.
Rogé, P., Snapp, S.S., Kakwera, M.K., Mungai, L., Jambo, I., Peter, B., 2016. Ratooning
and perennial staple crops in Malawi: a review. Agron. for. Sustain. Dev. 36, 50.
http://dx.doi.org/10.1007/s13593-016-0384-8.
Smith, A., Snapp, S., Dimes, J., Gwenambira, C., Chikowo, R., 2016. Doubled-up legume
rotations improve soil fertility and maintain productivity under variable conditions in
maize-based cropping systems in Malawi. Agric. Syst. 145, 139–149.
Smith, A., Snapp, S.S., Chikowo, R., Thorne, P., Bekunda, M., Glover, J., 2017. Measuring
sustainable intensification in smallholder agroecosystems: a review. Glob. Food
Secur. 12, 127–138.
Snapp, S.S., 1998. Soil nutrient status of smallholder farms in Malawi. Commun. Soil Sci.
Plant Anal. 29, 2571–2588.
Snapp, S.S., Kanyama-Phiri, G.Y., Kamanga, B., Gilbert, R., Wellard, K., 2002a. Farmer
and researcher partnerships in Malawi: developing soil fertility technologies for the
near-term and far-term. Exp. Agric. 38, 411–431.
Snapp, S.S., Rohrbach, D.D., Simtowe, F., Freeman, H.A., 2002b. Sustainable soil man-
agement options for Malawi: can smallholder farmers grow more legumes? Agric.
Ecosyst. Environ. 91, 159–174.
Snapp, S.S., Jones, R.B., Minja, E.M., Rusike, J., Silim, S.N., 2003. Pigeon pea for Africa: a
versatile vegetable - and more. Hortscience 38, 1073–1078.
Snapp, S.S., Blackie, M.J., Gilbert, R.A., Bezner-Kerr, R., Kanyama-Phiri, G.Y., 2010.
Biodiversity can support a greener revolution in Africa. Proc. Natl. Acad. Sci. 107,
20840–20845.
Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R., Polasky, S., 2002. Agricultural sus-
tainability and intensive production practices. Nature 418, 671–677.
Tittonell, P., Vanlauwe, B., Leffelaar, P.A., Rowe, E.C., Giller, K.E., 2005. Exploring di-
versity in soil fertility management of smallholder farms in western Kenya: I.
Heterogeneity at region and farm scale. Agric. Ecosyst. Environ. 110, 149–165.
Tropical Rainfall Measuring Mission Project (TRMM), 2016. TRMM Product 3b43, NASA/
JAXA. Goddard Space Flight Center Distributed Active Archive Center (GSFC DAAC).
http://disc.sci.gsfc.nasa.gov/precipitation/documentation/TRMM_README/TRMM_
3B43_readme.shtml, Accessed date: 23 November 2016.
Valbuena, D., Erenstein, O., Tui, S.H.K., Abdoulaye, T., Claessens, L., Duncan, A.J.,
Gerard, B., Rufino, M.C., Teufel, N., van Rooyen, A., van Wijk, M.T., 2012.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
87
Conservation agriculture in mixed crop-livestock systems: scoping crop residues
trade-offs in Sub-Saharan Africa and South Asia. Field Crop Res. 132, 175–184.
Van Vugt, D., Franke, A.C., Giller, K.E., 2016. Participatory research to close the soybean
yield gap on smallholder farms in Malawi. Exp. Agric. 1–20. http://dx.doi.org/10.
1017/S0014479716000430.
Vesteeg, M.N., Koudokpon, V., 1993. Participative farmer testing of four low external
input technologies to address soil fertility decline in mono province (Benin). Agric.
Syst. 42, 266–276.
Waldman, K.B., Ortega, D.L., Richardson, R.B., Clay, D., Snapp, S.S., 2016. Preferences for
legume attributes in maize-legume system cropping systems in Malawi. Food Secur.
8, 1087. http://dx.doi.org/10.1007/s12571-016-0616-4.
S.S. Snapp et al. Agricultural Systems 162 (2018) 77–88
88