PLOS ONE
RESEARCH ARTICLE
Farm production diversity, household dietary
diversity, and nutrition: Evidence from
Uganda’s national panel survey
Haruna Sekabira 1
I *, Zainab Nansubuga2, Stanley Peter Ddungu2, Lydia Nazziwa2
D
1 Department of Natural Resources Management, International Institute of Tropical Agriculture (IITA),
Kacyiru, Rwanda, 2 Department of Agricultural Research, LADS Consult, Kampala, Uganda
* H.Sekabira@cgiar.org
a1111111111
a1111111111
a1111111111
a1111111111 Abstract
a1111111111
Improved food security and nutrition remain a notable global challenge. Yet, food security
and nutrition are areas of strategic importance regarding the United Nations’ Sustainable
Development Goals. The increasingly weakening global food production systems pose a
threat to sustainable improved food security and nutrition. Consequently, a significant popu-
OPEN ACCESS
lation remains chronically hungry and severely malnourished. As a remedy, farm production
Citation: Sekabira H, Nansubuga Z, Ddungu SP, diversity (FPD) remains a viable pathway through which household nutrition can be
Nazziwa L (2022) Farm production diversity,
improved. However, evidence is mixed, or unavailable on how FPD is associated with key
household dietary diversity, and nutrition: Evidence
from Uganda’s national panel survey. PLoS ONE nutrition indicators like household dietary diversity, energy, iron, zinc, and vitamin A (micro-
17(12): e0279358. https://doi.org/10.1371/journal. nutrients). We use the Uganda National Panel Survey (UNPS) data for rural households to
pone.0279358
analyze differential associations of sub-components of FPD on dietary diversity, energy,
Editor: Francesco Caracciolo, University of Naples and micronutrient intake. Panel data models reveal that indeed crop species count, and ani-
Federico II: Universita degli Studi di Napoli Federico
mal species count (sub-components of FPD) are differently associated with household die-
II, ITALY
tary diversity score (HDDS), energy, and vitamin A sourced from markets. Moreover, when
Received: March 5, 2022
volumes of these nutrition outcomes were disaggregated by source (own farm vs. markets),
Accepted: December 6, 2022 the animal species count was only positively significantly associated with nutrition outcomes
Published: December 16, 2022 sourced from consumption of produce from own farm. Associations were insignificant for
nutrition indicators sourced from markets except vitamin A. The crop species count, how-
Peer Review History: PLOS recognizes the
benefits of transparency in the peer review ever, consistently showed a strong positive and significant association with energy, and all
process; therefore, we enable the publication of studied micronutrients sourced from own farm produce consumption, as well as those
all of the content of peer review and author
sourced from markets except Vitamin A, which was negative but insignificant. Therefore,
responses alongside final, published articles. The
editorial history of this article is available here: inclusive, pro-poor, and pro-nutrition rural policy initiatives in the context of rural Uganda
https://doi.org/10.1371/journal.pone.0279358 and similar ones, could more widely improve household nutrition through prioritizing crop
Copyright: © 2022 Sekabira et al. This is an open species diversification on own farms because crops fetch wider nutrition gains.
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
Introduction
author and source are credited.
Hunger and malnutrition remain a strong challenge in much of the developing world, despite
Data Availability Statement: The data is freely
publicly available at the World Bank website on the food and nutrition security being a strategic aspect of central importance as regards the United
following link: https://www.worldbank.org/en/ Nations’ Sustainable Development Goals [1]. The efforts of the United Nations (UN) have
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 1 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
programs/lsms/initiatives/lsms-ISA#8. We also consistently focused on a holistic food systems approach to effectively operationalize these
provide the specific data for this particular paper in SDGs [1–3]. Under this approach, the full array of food systems actors that add value to food
supporting information S1 File.
from production, collection, processing, delivery, consumption, and waste management are
Funding: HS 2104060035 International Fund for considered [1–5]. This broader set of activities principally aims at achieving sustainable food
Agricultural Development https://www.ifad.org/en/ systems that must bring food security and nutrition to all actors [3–7] The envisioned sustain-
The funders had no role in study design, data
able food system must principally be sustainable economically, socially, and environmentally)
collection, and analysis, decision to publish, or
[4–7]. To achieve these goals, therefore, production, distribution, processing, consumption,
preparation of the manuscript.
and other processes must be optimally prioritized [4, 5, 7]. Optimal prioritization can be
Competing interests: The authors have declared
achieved by optimally prioritizing activities at the different sub-systems of the food system, for
that no competing interests exist.
instance, the farming system, and others [4, 6]. Subsequently, the needed structural changes in
a food system that might ensure a food system’s sustainability can be initiated from optimal
changes in another sub-system like farming systems’ choices of crops or animal species grown
and consumed [1, 4–7]. Hence, guidance regarding a particular food system is essential to
guide public policy and private investments on optimal choices across [4, 5, 7]. Generally, the
UN aims to achieve a sustainable food system via the United Nations’ Sustainable Develop-
ment Goals (SDGs) for instance 1) ending hunger, 2) achieving food security and 3) improving
nutrition for all by 2030 [1]. The UN adds that to achieve these goals, food systems must be: 1)
more productive, 2) more inclusive, 3) resilient environmentally, and 4) capable of delivering
healthy and nutritious foods to all [1–3]. However, since food systems are context-specific
(depending on global location, culture, environment, preferences, etc.), country-specific
(local) guidance is indispensable to achieving sustainable food systems [4–7]. Because of lack
of such local guidance, and other inefficiencies nearly a billion people globally are chronically
hungry (lack access to calories), and nearly 2 billion are exposed to micronutrient malnutri-
tion, a matter that is largely attributed to inaccessibility to available food [1, 2, 6, 7]. However,
access to food is largely determined by the availability of food and having sufficient financial
resources to purchase food [8, 9]. Regrettably, access to financial resources is not guaranteed
to most of the global population, hence cementing chronic poverty, hunger, and malnutrition
[9, 10]. Moreover, some of the common infrastructure from which households can access
food–the markets–may be rendered ineffective in availing food to the majority of the world’s
poorest, especially the rural poor in countries where the market infrastructure is even inade-
quately available [11, 12]. Therefore, farm production diversity–a key component of farming
systems, remains a viable alternative to avail food to millions of the world’s poorest, especially
the rural majority excluded from the market infrastructure by remoteness [2, 8, 13–15].
Unfortunately, there are knowledge gaps in understanding comprehensively the nexus
interconnections and linkages around associations between farm production diversity (FPD)
and key nutrition outcomes [8, 16]. Such improper understanding of this nexus hinders proper
policy formulation to optimally guide context-specific farming systems, thus hindering the
scaling of appropriate and optimal innovations and investments against food insecurity and
malnutrition, more so among vulnerable rural smallholder farmers [16]. Subsequently, such a
lack of guidance and the failure to have appropriate optimal activities, render the general food
systems non-inclusive [8–12]. Moreover, in most instances, evidence on how best smallholder
households can access diverse diets is mixed. Some evidence points to market access being
more important than diversifying farm production [17–20]. However, in the context of the
least-developed countries like Uganda [21, 22], where market infrastructure is poor and rural
smallholder households are trapped in poverty and remoteness, such evidence may be inappli-
cable. Certain evidence has documented FPD to be more important for diversity in household
diets and nutrition gains [8, 11, 12, 15, 17–19, 23–29]. Hence, empirical evidence linking FPD,
and nutrition outcomes is often mixed, disjointed, and incomprehensive [16, 30, 31]. For
instance, none of the studies above explored the differential impacts of FPD sub-components
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 2 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
(animal and crop species count) on daily energy, iron, zinc, vitamin A intake, or other micro-
nutrients. Closer efforts have been done by Muthini et al. [26], however, these covered only as
far as dietary diversity without further consideration of the specific micronutrients. We con-
tribute to this body of literature by answering the following research questions:
1. Are sub-components (crop and animal species counts) of FPD differentially associated with
household dietary diversity score (HDDS), daily energy, iron, zinc, and vitamin A?
2. Which of the two FPD sub-components is associated with better nutrition gains?
Our contributions provide country-specific guidance–on potential optimal farming options
in the rural Ugandan context and similar ones to reshape food production and consumption
patterns [4, 5, 7]. Over the past two decades, Uganda witnessed improvements in the produc-
tion of staple food crops, even though cereals like rice and wheat have also been increasingly
imported but with low reliance on their consumption as staples [32–35]. The regular Ugandan
diet is majorly made up of starchy roots and tubers (cassava, sweet potatoes), plantains (mostly
cooking bananas), and cereals (maize, millet, sorghum). Pulses, nuts, and leafy green vegeta-
bles are usual accompaniments [32–34]. Sometimes, depending on household financial needs,
food crops like bananas, sweet potatoes, pulses, and cereals can be sold for cash, alongside reg-
ular cash crops like coffee [32–35]. However, in urban areas, a nutrition transition is evolving
towards imported and processed foods like rice and fast foods largely composed of meat prod-
ucts like chicken [32, 33, 35]. Nevertheless, the general diet especially for the poor and rural
households remains needy, especially regarding foods that are rich and sufficient in micronu-
trients [32–35]. Therefore, food and nutrition insecurity remain persistent, due to poverty, low
agricultural production, civil unrest, and climate change [32, 33]. Fortunately, in much of
civil-secure rural Uganda (southern, eastern, central, western, and near northern region) the
dietary energy supply from available foods, meets household energy needs, but the share of key
macronutrients like lipids and proteins and micronutrients like iron, zinc, and vitamin A that
are crucial for human development is grossly at the lower side of thresholds recommended by
the United Nations [32–36]. For instance, globally, 2 billion persons are exposed to micronu-
trient malnutrition, because of low intake of vitamins and essential minerals like vitamin A,
iron, and zinc [37]. These nutritional insufficiencies (micronutrient malnutrition) are caused
by both the low quantities of food consumed (calories) and poor dietary diversity and quality,
leading to diminished mental (learning), and physiological human growth and development,
vulnerability to diseases, early deaths, and eroded economic productivity [17, 37]. About 45%
(3.1 million) of deaths in children under 5 years annually are connected to micronutrient mal-
nutrition, and in some cases lead to severe stunting and wasting [24, 37]. In 2020, 22% (149.2
million) of children below 5 years were stunted, and about 7% (45.4 million) suffered from
wasting [37]. In Uganda, a third of children below five years are stunted, more so in rural areas
(30%) than urban ones (24%) [38].
The human body requires calories to live, and if deprived of energy the body dies because
vital organs like the heart will collapse [37, 38]. Energy is absorbed from foods consumed, and
energy requirements are variant based on one’s sex, age, size, and activeness [37]. In 2018
FAO, ranked Uganda 168th among 171 countries based on the volume of daily energy intake
per capita, with 1,981 kilocalories, which is quite less than the African average of 2,331 (least
globally) [39]. Regarding iron, globally, about 50% of anemia stands attributed to iron defi-
ciency, ranking 9th amongst the top 26 risk components of the world’s disease burdens,
accounting for nearly a million deaths annually [37, 40]. The least developed world, Africa and
Asia share 71% of this mortality burden, and about 50 billion USD is lost annually globally in
GDP due to iron-deficiency anemia (IDA) [40]. Iron is necessary for the formation of
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 3 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
hemoglobin (stores and transports oxygen) [40]. Therefore, iron is so essential for life, and its
deficiency severely affects children and lactating women [37, 40]. Children without sufficient
iron in their life’s initial thousand days experience reduced; understanding capacity, social
adaptability, and language grasping skills, while foetuses would have higher probabilities of
premature births, low weight, and suffocation at birth [37, 40]. For pregnant women, IDA
presents higher risks of infections, and bleeding–accounting for 20% of global maternal deaths
[37–40]. In Uganda, the prevalence of anemia amongst children stands at 53% [41]. Regarding
zinc, about 31% of the global population suffers from zinc deficiency with rates of prevalence
reaching 73% in some least developed countries, contributing to an immense burden of disease,
accounting for about 58% of child deaths in Africa [37, 42]. Zinc is essential for proper human
growth, reproduction, and building strong immunity against diseases, and a key anti-inflamma-
tory mineral that is central in nearly 300 enzymes [37, 42]. The prevalence of zinc deficiency in
Uganda is around 66% [28], yet zinc, iron, vitamin A, and energy have been found essential in
school children’s learning [43, 44]. Lastly, Vitamin A deficiency (VAD) is one of the most preva-
lent deficiencies globally, mostly affecting children (30% of children below 5 years) and
accounting for nearly 2% of all children’s deaths annually [37, 45]. VAD causes avoidable blind-
ness in children, and VA ingestion by children via breast milk is dependent on the mother’s VA
status, hence VAD usually manifests early in children’s lives especially in communities consum-
ing diets that are short in vitamin A [46]. In Uganda, 28% of infant children are exposed to
VAD [45], and in some regions, VAD has been found as high as 85% [28].
Therefore, understanding how food systems can substantially provide energy and these
essential micronutrients is noble to attain sustainable human and economic development.
With this study, we contribute to closing the knowledge gap on how farm production diversity
can be tilted to ensure sufficient supplies of energy and key micronutrients, to rural Uganda
where 39% of the employed population is engaged in subsistence agriculture (eat what they
farm of crops and animals) [32, 35]. Regarding animal species, according to the Uganda
Bureau of Statistics [35], Uganda’s smallholder farmers are also engaged in livestock farming,
bringing livestock numbers to about 16 million goats, 14 million cattle, 5 million sheep, 48 mil-
lion poultry, and 4 million pigs. In fact, Uganda is a net exporter of live livestock especially cat-
tle, eggs, and dairy products, most of which are produced in rural Uganda [35]. Hence, it was
informative that farm production diversity is studied with disaggregated data.
The rest of the paper is organized as follows: next, we present the conceptual framework
and then elaborate on the materials and methods used. We then present and discuss results
while highlighting policy implications, and finally draw conclusions.
Conceptual framework
Generally, we hypothesized that FPD bears a positive influence on food security and thus
nutrition outcomes, and we diagrammatically illustrate this in Fig 1. Conceptually, following
Sekabira & Nalunga [12], policies (agriculture, nutrition, or investment) influence the diversity
of crops and animals species produced by farmers, thus influencing which crops or livestock
species are prioritized for either direct consumption within households (own farm produce
consumption pathway) or for sale to earn income and then buy food items from markets (mar-
ket consumption pathway), that in the end dictate nutrition outcomes. Based on the conceptu-
alization in Fig 1, and the empirical methodology highlighted above but elaborated later in this
paper, we hypothesize that crop species count and animal species count, which are the key
sub-components of FPD, associate differently with HDDS, energy, and micronutrients, and
that crop species count attract stronger nutrition gains for rural households especially via the
own farm consumption pathway.
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 4 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Fig 1. Conceptual frame for farm production diversity (FPD) and nutrition nexus, adapted from Sekabira &
Nalunga [12].
https://doi.org/10.1371/journal.pone.0279358.g001
Materials and methods
Data
Data structure. Generally, the data used for this study is motivated by the 50x2030 Smart
Agriculture Data Initiative of the International Fund for Agricultural Development (IFAD).
Specifically, we use the rural component of the Uganda National Panel Survey (UNPS) data
collected by the Uganda Bureau of Statistics (UBOS), with technical support from the Living
Standards Measurement Study–Integrated Surveys on Agriculture (LSMS–ISA) section of the
World Bank. The sample size of the UNPS is about 3,200 households, that were previously
selected and interviewed during the 2005/2006 Uganda National Household Survey (UNHS).
Furthermore, the UNPS sample contains a randomly selected segment of split-off households
that came into existence after the 2005/2006 UNHS. Moreover, the UNPS is both regionally
and nationally representative. Each UNPS household is interviewed two times every year in an
interval of six months to enhance respondent recall abilities. Data are collected and entered
concurrently using computer-assisted interview applications (CAPI), installed on mobile per-
sonal computers that are operated by trained graduate enumerators. Subsequently, when data
are fully cleaned and documented, they are made available to the public in a period of twelve
months [47]. The UNPS has seven waves including 2009/10, 2010/11, 2011/12, 2013/14, 2015/
16, 2018/19, and 2019/20. This data is collected once every year, even though the exercise is
planned to be completed in two calendar years given other data management needs. We only
use the latest 3 waves because these were possible to merge and analyze together given a com-
mon structure of households’ identification. We had also separately worked with earlier waves
on studying similar nutrition outcomes. Essentially, we built on analyses of Sekabira &
Nalunga [12] and Sekabira et al. [28] who used 2009/10, 2010/2011, and 2011/2012 waves.
Data are sourced from across Uganda covering the different farming systems under which dif-
ferent crops and livestock species are produced. Most commonly are the intensive banana and
coffee farming systems around the lake shores in southern Uganda, the annual cropping and
cattle farming systems in northern Uganda, the banana-coffee-cattle farming systems in
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 5 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Fig 2. Major farming systems of Uganda, as sourced from Global Yield Gap Atlas [48].
https://doi.org/10.1371/journal.pone.0279358.g002
western Uganda, and the pastoral/ annual crops farming systems in the northeast [32, 35, 48].
Details of these farming systems and geographic areas of rural Uganda where data was col-
lected are illustrated in Fig 2, sourced from Global Yield Gap Atlas [48].
Measurement of key variables. Farm production diversity (FPD) was measured using
the biodiversity index, which is a simple count of all crops and livestock produced on farm, as
previously used by [11, 12, 15, 17, 49]. Therefore, before generating the biodiversity index for
FPD, we generated its sub-components, the species count based on crops, and the animal spe-
cies count based on livestock. Commercial but edible crops and animal species have also been
considered in the calculation of FPD–since the markets consumption pathway which is insti-
gated by market sales based on different crop and animal species grown by households is con-
sidered in this study. Moreover, some crops and animal species are used for both food and
cash making purposes, depending on prevalent household needs. Examples of some of the
crops used in calculation of the FDP include, rice, maize, millet, sorghum, beans, cowpeas,
chickpeas, groundnuts, soya beans, sunflower, Simsim, cabbage, tomatoes, carrots, onions,
pumpkins, amaranth, eggplants, cucumber, green pepper, sugarcane, potatoes, sweet potatoes,
cassava, yams, oranges, pawpaw, pineapples, bananas, mangoes, avocado, passion fruit, water-
melon, coffee, and ginger. Some of the animal species considered included cattle, goats, sheep,
pigs, chicken, and rabbits. Dietary diversity was measured using the aggregated food index
which measures the sum of food groups (12 in total) consumed in the household, including
cereals, white roots and tubers, vegetables, fruits, meat and its products, eggs, fish, legumes
nuts and seeds, dairy and its products, oils and fats, sweets and sugars, and spices condiments
and beverages. The index has been elaborated by Swindale & Bilinsky [50], and recently widely
used to study dietary diversity [12, 17, 26, 51]. Energy, iron, zinc, and vitamin A (micronutri-
ents) available per adult per household have been measured by computing quantities of food
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 6 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
items consumed by households in kilograms and then computing edible proportions for each
food item available. From the edible quantities, we computed quantities of energy in kilocalories
and respective micronutrients, following Uganda food consumption tables documented by
Hotz et al. [52]. For comparability of nutrition outcomes across households with different
demographic compositions, we standardized household size into adult equivalents (AE) follow-
ing (FAO,WHO & UNU [53], using an adult male as the person category with the highest nutri-
tional requirements to survive. Therefore using this threshold, other persons’ nutritional
requirements based on their sex (female or male) and age, are computed, and then standardized
to their adult male equivalent. Edible quantities of energy and micronutrients were then divided
by respective adult equivalents to produce comparable nutrition indicators available per adult
across households. Following FAO,WHO & UNU [53], we also computed deficiencies for these
micronutrients using as critical levels; 2400 kilocalories, 18 milligrams, 15 milligrams, and 625
retinal activity equivalent micrograms, for energy, iron, zinc, and vitamin A respectively.
Data description. From Table 1, the sample was on average aged 48 years, with a house-
hold size of 6 persons and was barely educated (1 year of formal education). This being a
purely rural sample, low education levels may not be a surprise. The value of annual household
assets averaged at seven (7) million UGX (2,000 USD). However, land size averaged at 0.7
acres. Concerning discrete variables, most of the sample (76%) had experienced shocks
(weather issues like drought, famine, storms etc., health issues like death of the head, chronical
illnesses etc.). Furthermore, most of the sample (63%) used mobile phones, and heavily relied
on agriculture (59%) as their main income source. On the other hand, males (66%) dominated
household headship. Concerning production diversity, on average, households farmed nearly
5 species of both crops and livestock of which majority (69%) were crop species. However, as
seen in Fig 3, the average count of crop species farmed across the years, slightly declined
between 2015 and 2019, while that for animals slightly increased over the same period. On the
other hand, household dietary diversity score (HDDS) slightly increased from 2015 to 2019,
and this variation is illustrated in Fig 4.
Furthermore, average consumption of energy and all considered micronutrients (2,430
kilocalories, 20 milligrams, 13 milligrams, and 701 rae-micrograms respectively) was slightly
above FAO recommended thresholds per adult, except for zinc. Energy, iron, and zinc were
mostly (64%, 56%, and 66% respectively) sourced from markets, whereas vitamin A was mostly
(59%) sourced from own farm produce. From Fig 3, FPD was dominated by crop species per-
haps because our sample is totally rural and dominantly composed of smallholder farmers
who mostly grow on crops [7, 12, 21, 20, 32]. Overall, the 9 most grown crops species were
beans (16.3%), cassava (16.1%), maize (14.9%), cooking bananas (matooke) (13.9%), coffee
(8.5%), sweet potatoes (7.8%), groundnuts (3.6%), sorghum (2.7%) finger millet (2.4%),
whereas the 5 most farmed animal species were cattle (30.2%), goats (29.5%), chicken (23.3%),
pig (8%), and sheep (5.3%). See S1 Table for more details on the composition of the FPD
index. We also note that there was slightly more food produced in 2015 than in 2018 or 2019, a
fact that is also collaborated by the official data, for instance indicating cereal (maize, rice, sor-
ghum, and millet) production at 4.1 metric tonnes in 2018, and 4.3 in 2019, but 4.5 in 2015
[54–56]. This is also collaborated by FOASTAT data presented in the supplementary data file,
the S2 File. We do not expect these differences by time to affect our analysis since we used
panel data models to control for time or year specific effects.
Empirical model for data analysis
We implemented the specification of the panel regression model in Eqs (1) and (2), to study
the nature of association between FPD, the two FPD sub-components and various nutrition
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 7 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Table 1. Sample descriptive statistics (means or percentages) (N = 6,992).
Variables 2015 (N = 2,381) 2018 (N = 2,348) 2019 (N = 2,263) All sample (N = 6,992)
Male head (dummy) 0.655 0.662 0.669 0.662
Age of head (years) 47.91 (16.14) 47.69 (15.90) 47.78 (15.94) 47.80 (15.99)
Household size (persons) 5.746 (3.240) 5.662 (3.118) 5.785 (3.189) 5.730 (3.183)
Education of head (years) 1.440 (3.387) 1.333 (3.396) 1.309 (3.263) 1.362 (3.351)
Mobile phone use (dummy) 0.631 0.628 0.627 0.629
Total assets (million UGX) 6.35 (2.09) 6.93 (2.79) 6.64 (2.73) 6.64 (2.53)
Experienced shocks (dummy) 0.733 0.778 0.783 0.764
Land Size (Acres by GPS) 0.686 (3.609) 0.682 (1.858) 0.624 (1.742) 0.664 (2.565)
Farming (main income source) 0.585 0.585 0.595 0.589
FPD (bio index) 4.911 (3.319) 4.790 (3.282) 4.931 (3.282) 4.877 (3.295)
Crops FPD (bio index) 3.483 (2.761) 3.345 (2.748) 3.431 (2.777) 3.420 (2.762)
Animals’ FPD (bio index) 1.429 (1.307) 1.445 (1.338) 1.499 (1.340) 1.457 (1.328)
HDDS (food groups) 9.972 (2.594) 10.09 (2.558) 10.18 (2.371) 10.08 (2.513)
Energy (kilocalories/AE) 3,286 (2,838) 2,048 (1,643) 1,925 (2,139) 2,430 (2,347)
Iron (milligrams/AE) 25.75 (23.55) 17.65 (13.37) 17.60 (18.76) 20.39 (19.43)
Zinc (milligrams/AE) 17.36 (16.70) 11.20 (9.260) 11.42 (10.33) 13.37 (12.90)
Vitamin A (rae_micrograms/AE) 850.3 (1,217) 549.6 (789.0) 712.2 (3,504) 704.6 (2,168)
From markets
Energy (kilocalories/AE) 2,154 (2,291) 1,293 (1,396) 1,209 (1,280) 1,559 (1,776)
Iron (milligrams/AE) 14.42 (17.97) 9.760 (10.41) 10.22 (10.69) 11.50 (13.70)
Zinc (milligrams/AE) 11.53 (13.75) 7.203 (7.863) 7.747 (7.976) 8.851 (10.46)
Vitamin A (rae_micrograms/AE) 378.7 (920.5) 223.7 (507.8) 272.7 (705.9) 292.4 (735.2)
From own production
Energy (kilocalories/AE) 1,132 (1,369) 755.1 (879.8) 716.2 (1,741) 870.9 (1,384)
Iron (milligrams/AE) 11.33 (13.88) 7.887 (9.030) 7.382 (15.83) 8.895 (13.31)
Zinc (milligrams/AE) 5.834 (7.503) 3.994 (4.931) 3.676 (6.555) 4.518 (6.492)
Vitamin A (rae_micrograms/AE) 471.6 (827.4) 325.9 (592.4) 439.4 (3,440) 412.3 (2,045)
Deficiencies
Energy 0.457 0.664 0.699 0.605
Iron 0.451 0.602 0.622 0.557
Zinc 0.567 0.751 0.722 0.679
Vitamin A 0.572 0.727 0.696 0.664
FPD is farm production diversity, HDDS is household dietary diversity score, UGX is Uganda shillings (1USD = 3,557 UGX over considered years), in parentheses are
standard deviations. Values without standard deviations are percentages, GPS is global positioning system, AE is adult equivalent, rae_mg is retinal activity equivalents
micrograms
https://doi.org/10.1371/journal.pone.0279358.t001
outcomes or indicators respectively.
HNOit ¼ a0 þ b1FPDit þ yXit þ gTt þ εit ð1Þ
Where β1 in this case is the effects of FPD on the nutrition outcome that we estimated.
HNOit ¼ a0 þ b1Animal SpeciesCountit þ b2Crop SpeciesCountit þ yXit þ gTt þ εit ð2Þ
Where HNOit is household nutrition outcome of interest (dietary diversity, energy, zinc,
iron, or vitamin A available per adult) of household i in year t. α0 is the constant. β1 and β2 are
respectively the effects of the animal, and crop species components of FPD that we aim to
establish. θ is a vector of coefficients for observed household, and contextual characteristics,
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 8 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Fig 3. Farm production diversity (FPD) as generated from different sources (crops or livestock).
https://doi.org/10.1371/journal.pone.0279358.g003
while γ is a time fixed effects parameter. εit is the normally distributed error term, and Xit is
the vector of observed household characteristics (education, gender, and age of head, house-
hold size, assets, use of mobile phones, major source of income, and exposure to shocks, and
farm (land size). These characteristics could alongside the considered FPD or FPD compo-
nents, influence household nutrition outcomes. t is the year identifier variable capturing yearly
fixed effects. We use Eq (2) to empirically study the associations elaborated above for which
we do not claim causality.
Although in Eq (1) we controlled for FPD itself, in Eq (2), we controlled for the two sub-
components of FPD to examine magnitudes of their coefficients to see which FPD sub-compo-
nent is associated with better nutrition gains for households. We estimated both Eqs (1) and
Fig 4. Variation of household dietary diversity score (HDDS) across the panel waves.
https://doi.org/10.1371/journal.pone.0279358.g004
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 9 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
(2) with random effects (RE) to control for heterogeneity within observed time variant and
time invariant household characteristics, and fixed effects (FE) to control for unobserved het-
erogeneity [57, 58]. Moreover, because the UNPS data is collected randomly, and are a panel,
this also helped to reduce potential biases. But, because farmers self-select which crops and
livestock species to farm based on own characteristics, and supposedly time-invariant covari-
ates like gender of household heads become variant when headship changes for example due
to death or divorce, this may yield systemic bias in results generated by the FE estimator [57].
Moreover, even the RE estimator’s strong assumption that FPD cannot correlate with unob-
served factors that may influence HDDS, energy, or micronutrients intake is also violated by
self-selection [57, 58]. Therefore, to control for potential violations of these assumptions, we
use the Mundlak (MK) estimator, a pseudo-fixed effects estimator, that also controls biases
caused by time-invariant unobserved heterogeneity, as would do a FE estimator, [59]. Essen-
tially, the MK estimator helps bridge the FE and RE estimations by controlling for means of
variables, such that the FE assumption (there is a correlation between specific effects of studied
individuals and the independent variables), and the RE assumption (there is no correlation
between unobserved heterogeneity of studied individuals and the independent variables) are
not violated–which if violated would yield biased estimates [57, 59]. Therefore, we interpret
the MK estimator results. Nevertheless, in the first regression results involving HDDS, we pres-
ent results from both FE and MK estimators for comparisons, and to minimize space we leave
out RE results but include the Hausman test value for all FE specifications that consistently
showed that the FE model was better suited to the data than the RE specification. However, we
only present the MK estimator results concerning energy and micronutrients, to avoid bulki-
ness. To correctly interpret the MK model, we generate elasticities following Bellemare &
Wichman [60], after inverse hypergolic sine (IHS) transformations of the FPD indices. The
IHS transformations are preferred for the explanatory variables of interest because they allow
retention of preferred properties of the log transformations while retaining the negative and
zero-valued observations in the data instead of merely dropping such observations [61].
Therefore, the stated empirical model was appropriately used to find answers to our
research questions, using the panel survey data from Uganda covering a representative sample
of nearly 3,000 households consisting of the of 2009/10, 2010/2011, and 2011/2012 waves of
the Uganda National Panel Survey (UNPS). The data was collected by the Uganda Bureau of
Statistics annually and is freely available from the World Bank’s Living Standards Measure-
ment Study–Integrated Surveys on Agriculture (LSMS-ISA) section. Because the data is a
panel, we used panel data models specifying fixed effects and random effects to analyze the
data. These panel data models enable variation in parameters of the model across studied
households, which improves efficiency, hence in the quality of estimated results generated
from combining households across the different data waves [57, 58]. However, because both
the fixed effects and random effects estimators have assumptions that could easily be violated,
we finally estimate the Mundlak which sufficiently connects the fixed and random effects esti-
mations [59]. The Mundlak concept is premised on assumptions of the FE and RE estimators
[57, 59]. Answers to the above questions have enhanced the understanding of the linkages
between FPD and nutrition. Such an interlinked understanding is indispensably important in
designing appropriate food systems interventions [16, 29–31]. Since we were focused on study-
ing farm production diversity, we only analyzed rural households of the UNPS. Our results
(that crop species count is strongly associated with better nutrition outcomes via the two main
consumption pathways–own farm production, and markets) have also generated evidence to
inform pro-nutrition and food security policies in Uganda, and those of a similar context, on
how inclusiveness in nutrition gains especially among poor rural smallholder farmers can be
achieved. More specifically, for instance, our results have among others informed the nutrition
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 10 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
following policy initiatives: 1) The Uganda Nutrition Action Plan II (UNAP II) that spans
between 2020–2025, and aims to leave behind none among Ugandans in scaling up nutrition
outcomes. 2) The Uganda Food and Nutrition Policy (UFNP). 3) The Uganda National Agri-
culture Policy (UNAP). 4) Uganda’s Multisectoral Food Security and Nutrition Project
(UMFSNP) funded by the World Bank aiming mostly to eradicating malnutrition in children
and rural dwellers.
Results and discussions
From Table 2, model 1 we see that the combined (animal and livestock species count) FPD
index is significantly and positively associated with HDDS. However, the Hausman test value
is significant and for the reasons explained earlier we interpret model 3 the MK specification,
which shows that diversification is negatively associated with HDDS. However, this blanket
interpretation could be misleading since FPD contains 2 major subcomponents that may influ-
ence HDDS differently given the rural context of our sample. We thus re-run the FE and MK
estimators each with the two main sub-components of the FPD index to further investigate if
each of the two FPD sub-components will differentially be associated with HDDS. On disag-
gregating the FPD index (crop and livestock) in models 4 and 5 (without controls), and models
6 and 7 (with controls), the two sub-components are associated significantly with HDDS dif-
ferently. Moreover, in all these models the animal species count shows a positive and signifi-
cant association (models 4 and 6), while the crop species count shows a negative and
significant association. The negative association with crop species could be linked to domi-
nance of crop-farming by most rural households which is also less capital intensive but nearly
sufficiently done and usually provides the key needed primary energy sources for households
to survive, that more diversification of the same may yield diminishing returns. For quantita-
tive empirical elaborations, we interpret the elasticities in Table 3 for MK estimators particu-
larly model 7.
From model 7, Table 3, each additional animal species grown on farm is associated with
increases of 2.4 percentage points in food groups consumed on farm, implying and increment
of 0.3 food groups. On the other hand, the association of HDDS with crop species count is neg-
ative but with decreases of only 0.8 percentage points on food groups consumed, an implica-
tion of 0.1 food groups. Thus, the two sub-components of FPD are differently associated with
HDDS. Since our rural sample largely and abundantly grew staple food crops (see S1 Table for
details) that are largely cereals or roots and tubers thus contributing mainly to household
energy and general micronutrients needs [26, 33–36], it is surprising that a crops species count
was significantly and negatively associated with HDDS. However, since largely grown crops
species were yielding common dietary quality, for instance the dominant cereals and starchy
crops, it may also imply that increments in their species, only yielded diminishing retards on a
dietary quality (diversity of diets) indicator, HDDS [17, 18, 62]. On the other hand, animal spe-
cies count showing a positive and significant association with HDDS is not surprizing.
Muthini et al. [26], found that producing animals enabled households’ access to a diversity of
nutrition benefits in energy, proteins, fats, and micronutrients. However, since HDDS is an
aggregated indicator of dietary quality [18, 19, 62], we re-run model 7 with specific nutrition
outcomes to establish the exact nature of association between FPD sub-components and cer-
tain micronutrients. As expected, these unearthed positive and significant associations of both
the animal species count and the crop species count with various micronutrients and energy
especially those sourced from own-farm produce, that was difficult to detect under a qualita-
tive indicator, HDDS. We present only the elasticities of these results later in respective Tables
4–7, and details are presented in respective S1–S6 Tables.
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 11 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Table 2. Association of farm production diversity (FPD) on dietary diversity score (HDDS).
Models RE (1) FE (2) MK (3) FE (4) MK (5) FE (6) MK (7)
Variables HDDS HDDS HDDS HDDS HDDS HDDS HDDS
IHS of FPD (bio index) 0.183��� -0.210��� -0.206���
(0.032) (0.062) (0.056)
IHS of Animal FPD (bio index) 0.484��� 0.483��� 0.485��� 0.259���
(0.078) (0.071) (0.079) (0.043)
IHS of Crop FPD (bio index) -0.561��� -0.557��� -0.643��� -0.070��
(0.064) (0.059) (0.066) (0.034)
Male head (dummy) 0.030 -0.004
(0.262) (0.226)
Mobile phone use (dummy) 0.002 -0.007
(0.107) (0.093)
Age of head (years) -0.021� -0.007
(0.013) (0.011)
Household size (adult equivalents) 0.013 -0.005
(0.039) (0.032)
Education of head (years) -0.039 -0.054��
(0.027) (0.022)
Total assets (million UGX) -0.038 -0.125���
(0.036) (0.027)
Experienced shocks (dummy) 0.104 0.075
(0.102) (0.088)
Land Size (Acres by GPS) -0.009 -0.025
(0.025) (0.022)
Farming is the main income source (dummy) -0.003 -0.083
(0.111) (0.096)
Year is 2018 0.138�� -0.085 0.127� -0.084 0.127� -0.085 0.102
(0.070) (0.082) (0.070) (0.081) (0.069) (0.084) (0.070)
Year is 2019 0.196��� -0.107 0.186��� -0.113 0.180�� -0.092 0.077
(0.071) (0.083) (0.071) (0.082) (0.070) (0.083) (0.070)
Means of covariates YES YES YES
Constant 7.028��� 7.986��� 6.672��� 8.007��� 6.751��� 9.220��� 5.218���
(0.081) (0.136) (0.091) (0.123) (0.082) (0.639) (0.219)
Observations 6,992 6,992 6,992 6,992 6,992 6,828 6,828
No. of households 2,838 2,838 2,838 2,838 2,838 2,804 2,804
F value 4.46��� 23.22��� 8.72���
Hausman test value 105.61��� 155.98��� 158.75���
Wald Chi2 value 41.28��� 112.39��� 217.64��� 329.68���
Standard errors in parentheses
��� p<0.01
�� p<0.05
� p<0.1
UGX is Uganda shillings (1USD = 3,557 USD); GPS is Global positioning system; RE is Random effects, FE is Fixed effects, MK is Mundlak, IHS is Inverse hyperbolic
sine. Full table with means of variables is presented in S2 Table.
https://doi.org/10.1371/journal.pone.0279358.t002
However, Table 2, model 7 results do also highlight other factors that are significantly asso-
ciated with HDDS. For instance, our rural sample was barely educated with an average of one
(1) year of formal education. Moreover, each additional year of education for such a grossly
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 12 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Table 3. Elasticities for the associations of farm production diversity (FPD) on dietary diversity score (HDDS).
Nutrition indicator Household dietary diversity score (HDDS)
Models MK (3) MK (5) MK (7)
IHS of FPD (bio index) -0.025���
(0.006)
IHS of animals’ FPD (bio index) 0.045��� 0.024���
(0.007) (0.004)
IHS of crops FPD (bio index) -0.063��� -0.008��
(0.007) (0.004)
Other model attributes
Other covariates are included NO NO YES
Means of covariates YES YES YES
Observations 6,992 6,992 6,828
No. of households 2,838 2,838 2,804
Wald Chi2 value 112.39��� 217.64��� 329.68���
Standard errors in parentheses
��� p<0.01
�� p<0.05
� p<0.1
IHS is Inverse hyperbolic sine
https://doi.org/10.1371/journal.pone.0279358.t003
uneducated sample was negatively associated with reductions in food groups consumed. In
fact, all our sample was rural (100%) which is characteristic of strong traditions that are heavily
aligned towards consumption of staples and limited education [63, 64]. Therefore, it may not
be surprising that associated effects of education towards HDDS were negative–contrary to
Table 4. Elasticities of associations of farm production diversity (FPD) and daily energy intake per adult equivalent (AE).
Nutrition indicator Daily energy intake (kilocalories/AE)
Models MK (1) MK (2) MK (3) MK (4)
Variables Total Total Own farm-sourced Markets source
IHS of FPD (bio index) 0.034���
(0.004)
IHS of animals’ FPD (bio index) -0.003 0.031��� -0.003
(0.004) (0.009) (0.006)
IHS of crops FPD (bio index) 0.039��� 0.165��� 0.024���
(0.004) (0.009) (0.006)
Other model attributes
Other covariates YES YES YES YES
Means of covariates YES YES YES YES
Observations 6,828 6,828 6,828 6,828
No. of households 2,804 2,804 2,804 2,804
Wald Chi2 value 986.33��� 1024.91��� 1409.65��� 883.99���
Standard errors in parentheses
��� p<0.01
�� p<0.05
� p<0.1
MK is Mundlak, IHS is Inverse hyperbolic sine. Full model table is in S3 Table.
https://doi.org/10.1371/journal.pone.0279358.t004
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 13 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
our expectations. However, education in substantially more years (higher education of domi-
nantly rural samples), has been found to positively associate with nutrition outcomes [28, 51].
On the other hand, further surprisingly, assets also showed a significant negative association
with HDDS. Usually, most assets among smallholder households including productive assets
(communication and transport equipment like mobile phones, motorcycles, or bicycles etc.)
and non-productive ones are controlled by males who may work largely in non-farm activities
[51, 65, 66]. However, considerable household financial resources, that would be used to pur-
chase food or invest in food production–are diverted daily to service costs related to the use of
these assets for instance, buying airtime and fuel, repairs etc. Hence, the variable costs burden
presented to the household by availability of these assets may render assets to be negatively
associated to HDDS. Moreover, even when such assets are liquidated by households, generated
incomes are turned to strategic household investments like housing, off-farm business invest-
ments, and medication, but not food consumption [51]. Nevertheless, some evidence has
found assets to contribute importantly to household welfare [66].
Regarding micronutrients and energy, results are presented in separate tables for each
nutrition indicator. From Tables 4–7 model (1), as was expected, FPD in general is positively
and significantly associated with total daily energy, iron, zinc, and vitamin A intake per adult
equivalent at 1% level. Each additional species of crops and livestock grown on farm increases
energy, iron, zinc, and vitamin A available to the household by 3.4 (83 kilocalories per AE), 4.9
(0.99 milligrams/AE), 6.0 (0.8 milligrams/AE), and 2.9 (20.4 rae_micrograms/AE) respec-
tively.4.9 However, in model (2) of each of Tables 4–7, disaggregation of FPD into subcompo-
nents (animal and crop species), shows differential associations with micronutrients and
energy, some of which is insignificant. Nevertheless, the insignificance of the associations
between micronutrients and some FPD subcomponents could be due to aggregation of micro-
nutrients irrespective of the source (own farm and markets), yet we consider only rural house-
holds that mostly rely on one source (own farm produce) for their nutrition needs. Therefore,
in models (3) and (4) of Tables 4–7, we analyze FPD subcomponents and their associations
with energy and micronutrients by source.
More specifically, from Table 4, when FPD was disaggregated into the two subcomponents
in model (2) while energy was still aggregated, the association with energy is negative with the
animal species count but insignificant but positive and significant with crops species count
(3.9 percentage points). However, in models 3 and 4 when total energy is disaggregated by
source–own farm and markets source, the animal species count then shows a strong significant
and positive association with energy sourced from own farm produce consumption. Specifi-
cally, each additional animal species kept on farm is associated with increases in daily energy
intake per adult equivalent of 3.1 (27 kilocalories/AE) percentage point. The association of the
animal species count with energy sourced from markets is not significant.
Similarly, from Table 4, models (3) and (4) the association of the crops’ species count with
energy sourced from both own farm produce and markets consumption was strongly signifi-
cant and positive, with each additional crops’ species being associated with an additional
16.5and 2.4 percentage points via each source respectively. Associations with energy sourced
from own farm produce yielded starker increments.
From Table 5, disaggregating FPD while iron was aggregated in model 2, the animal species
count shows negative but insignificant associations with iron intake. However, the crop species
count shows a significant and positive association with increments of 5.3 percentage points in
food groups consumed for every additional crop species. When iron is disaggregated by
source, again, both the animal and crops species count show a significant and positive associa-
tion with daily iron intake sourced from own farm. Each additional animal species is associ-
ated with an additional 10.9 percentage points, while that of crops is associated with an
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 14 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Table 5. Elasticities of associations of farm production diversity (FPD) and daily iron intake per adult equivalent (AE).
Nutrition indicator Daily iron intake (milligrams/AE)
Models MK (1) MK (2) MK (3) MK (4)
Variables Total Total Own farm-sourced Markets source
IHS of FPD (bio index) 0.049���
(0.007)
IHS of animals’ FPD (bio index) -0.0004 0.109�� 0.004
(0.007) (0.043) (0.013)
IHS of crops FPD (bio index) 0.053��� 0.752��� 0.025��
(0.007) (0.042) (0.012)
Other model attributes
Other covariates YES YES YES YES
Means of covariates YES YES YES YES
Observations 6,828 6,828 6,828 6,828
No. of households 2,804 2,804 2,804 2,804
Wald Chi2 value 581.57��� 602.69��� 1298.19��� 284.01���
Standard errors in parentheses
��� p<0.01
�� p<0.05
� p<0.1
MK is Mundlak, IHS is Inverse hyperbolic sine. Full table is in S4 Table.
https://doi.org/10.1371/journal.pone.0279358.t005
additional 75.2 percentage points on the daily iron intake sourced from own farm. Associa-
tions with iron sourced from markets were insignificant although positive with animal species
count, but significant and positive with crop species count, where each additional species was
associated with increases 2.5 percentage points in iron intake sourced from markets.
Table 6. Elasticities of associations of farm production diversity (FPD) and daily zinc intake per adult equivalent (AE).
Nutrition indicator Daily zinc intake (milligrams/AE)
Models MK (1) MK (2) MK (3) MK (4)
Variables Total Total Own farm-sourced Markets source
IHS of FPD (bio index) 0.060���
(0.009)
IHS of animals’ FPD (bio index) 0.002 0.753��� 0.005
(0.009) (0.222) (0.015)
IHS of crops FPD (bio index) 0.065��� 0.812��� 0.050���
(0.009) (0.216) (0.015)
Other model attributes
Other covariates YES YES YES YES
Means of covariates YES YES YES YES
Observations 6,828 6,828 6,828 6,828
No. of households 2,804 2,804 2,804 2,804
Wald Chi2 value 529.53��� 548.02��� 1353.46��� 396.52���
Standard errors in parentheses
��� p<0.01
�� p<0.05
� p<0.1
MK is Mundlak; IHS is Inverse hyperbolic sine. Full table is in S5 Table.
https://doi.org/10.1371/journal.pone.0279358.t006
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 15 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
Table 7. Elasticities of associations of farm production diversity (FPD) and daily vitamin-A intake per adult equivalent (AE).
Nutrition indicator Daily vitamin-A intake (rae_micrograms/AE)
Models MK (1) MK (2) MK (3) MK (4)
Variables Total Total Own farm-sourced Markets source
IHS of FPD (bio index) 0.029���
(0.006)
IHS of animals’ FPD (bio index) 0.021��� 0.051��� 0.038���
(0.006) (0.017) (0.012)
IHS of crops FPD (bio index) 0.019��� 0.260��� -0.011
(0.006) (0.017) (0.011)
Other model attributes
Other covariates YES YES YES YES
Means of covariates YES YES YES YES
Observations 6,828 6,828 6,828 6,828
No. of households 2,804 2,804 2,804 2,804
Wald Chi2 value 240.75��� 247.29��� 1062.00��� 212.63���
Standard errors in parentheses
��� p<0.01
�� p<0.05
� p<0.1
MK is Mundlak; IHS is Inverse hyperbolic sine. rae_mg is retinal activity equivalents micrograms. Full table is in S6 Table.
https://doi.org/10.1371/journal.pone.0279358.t007
From Table 6, when FPD was disaggregated while zinc was still aggregated, again, the ani-
mal species count was insignificantly associated with zinc, although positively. On the other
hand, the crop species count is significant and positively associated with daily zinc intake with
each additional crop species being associated with increments of 6.5 percentage points in zinc
intake. However, when zinc is disaggregated by source, both the animal and crops species
count show a strongly significant and positive association with zinc sourced from own farm.
Each additional species to animals and crops grown is associated respectively with an addi-
tional 75.3 and 81.2 percentage points on daily zinc intake sourced from own farms. The large
size of these incremental margins could be explained by the high (highest of all studied nutri-
tion indicators) levels of zinc deficiency in the sample, as well as Ugandan population at large
[28, 33, 35]. Associations of animal species count with zinc sourced from markets were insig-
nificant but positive. However, such association was positive and significant for the crop spe-
cies count, where each additional crop species was associated with increments of 5 percentage
points in daily zinc intake.
Taking a slightly different pattern, from Table 7, when FPD was disaggregated while vita-
min A was still aggregated, both the animal and crop species count yields a positive and
strongly significant association with daily vitamin A intake. Each additional animal and crop
species grown yields respectively an associated additional 2.1 and 1.9 percentage points of total
daily vitamin A intake per adult. Furthermore, if total daily vitamin intake is disaggregated by
source, like with other micronutrients and energy, both animal and crops species count inde-
pendently yield a positive and strongly significant association with vitamin A sourced from
own farm foods consumption. Each additional species for animals and crops grown is associ-
ated respectively with an additional 5.1 and 26 percentage points to daily vitamin intake
sourced from own farm. Unlike with earlier nutrition indicators studied here, the associated
effect of the animal species count with vitamin A sourced from markets is positive and strongly
significant. Each additional animal species on the farm, is associated with 3.8 percentage points
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 16 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
increments in daily vitamin A intake sourced from markets. The association with crops species
count was negative but insignificant, unlike other micronutrients.
Generally, from Tables 4–7, aggregated FPD showed significant and positive associations
with total daily energy, iron, zinc, and vitamin A intake. When FPD was disaggregated in the
two components–the animal species count, showed insignificant associations with total energy
and totals of other micronutrients except vitamin A. At this level, all associations between the
crop species count and the aggregated totals of energy and all micronutrients were all positive
and strongly significant. However, when total energy and totals of micronutrients were also
disaggregated by source (own farm or markets), and individually analyzed alongside disaggre-
gated FPD, both the animal and the crops species count, consistently exhibited a positive and
strongly significant association with household daily energy and micronutrients intake
sourced from own farm consumption. In all cases, incremental percentage points yielded by
each additional crop species were larger than those yielded by each additional animal species
grown on farm. Moreover, FPD in general has been previously found to be positively associ-
ated with nutrition outcomes [8, 15, 17, 24–29, 62]. In disaggregated terms, small animals kept
on farm like goats, sheep, and rabbits and poultry species like chicken and ducks that formed
most of the animal species count can easily be consumed for food within households anytime
of the year without their availability being dependent on farming seasons. Moreover, larger
animals like cattle and even small animals can regularly provide products like milk, and eggs
that are good sources of energy and micronutrients. Therefore, regular consumption of ani-
mals and their products, makes it possible for households to enhance their available energy
and micronutrients. Our findings agree with Muthini et al. [26] who found the animal species
count to be important to household dietary diversity. Gaillard et al. [8] also found that women
consumed more dairy products if these were produced on farm. However, although the associ-
ation is as expected negative for energy, and positive for iron and zinc, the animal species
count is not significantly associated with daily energy, iron, and zinc intake sourced from mar-
kets. Surprisingly, the association was positive and strongly significant with vitamin A sourced
from markets.
On the other hand, the crop species count showed a consistent, strongly significant, and
positive association with daily energy and micronutrients intake regardless of the source–
except for vitamin A sourced from markets. The association was strongly significant with
energy and all micronutrients sourced from own farm consumption. In fact, the associations
of FPD components (animal and crop species count) and nutrition indicators, were strongest
via those components of nutrition indicators that had been sourced from own farm consump-
tion. Considering the two FPD components, the crop species count showed the strongest asso-
ciations with nutrition indicators within a particular source (except for vitamin A sourced
from markets) and across sources (own farm vs. markets). For example, each additional crop
species count yielded 16.5 incremental percentage points of daily energy intake via own farm
sources compared to 2.4 via markets. The strong positive association of the crop species count
with energy and micronutrients sourced from own farm produce is not surprising since most
smallholder farmers are engaged in subsistence agriculture, and our sample was totally rural.
Hence one would expect that since our sample mostly consume what they grow, then a crop
species count should bear a strong positive association with nutrition outcomes, as has been
established previously [11, 12, 15, 17–19, 24, 26, 28]. Moreover, the crop species count has also
showed positive and strongly significant associations with energy, iron, and zinc intake
sourced from markets, despite this being a dominantly rural sample with poor market infra-
structure Such consistent, positive and significant associations, may further confirm the
importance of the markets consumption pathway which is only possible to farmers after gain-
ing income from selling their produce, in this case crops as has been asserted in literature [12,
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 17 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
17–19, 24, 28]. In good seasons, farmers sell their surplus crops or sell cash crops in all seasons
to accumulate income that is used in purchasing foods from markets [33–36]. Moreover, some
strategically valuable crops like coffee, and vegetables are farmed within households but sold to
earn money and smooth other consumption and non-consumption needs [25, 36, 67, 68]. Yet,
households usually never regularly consume such valuable crops like vegetables within house-
holds but spare them for sale, and usually consider it a luxury to consume these crops [33, 34,
36]. Such may also explain why associations of the crop species and vitamin A derived from
market sources was negative and insignificant. Our findings partly concur with two important
meta-analysis reviews around farm production diversity and food consumption–that indeed
there are strong (significant) associations between farm production diversity and certain nutri-
tion outcomes as Jones [29] established, which are mostly realized via the diversification of
crops species contributing towards own farm sources of energy and micronutrients intake.
However, the association is also positive and significant, but the proportional magnitudes of
these associations are small, as found Sibhatu and Qaim [18], especially through the animal
species count component of the FPD via the own farm and market sources, and the crop spe-
cies count component contributing to energy and micronutrients intake via the market
sources.
Policy implications
We cautiously point out to policy that investments (scientific, technical, physical, and finan-
cial) in farmers’ own farm production of both livestock and crop species in rural Uganda, do
positively influence nutrition outcomes of households. Therefore, wherever such investment
opportunities are available–whether from national, regional, or international governments,
then these must be harnessed and carefully scaled-out. However, for the contextual structure
of rural Uganda (remote with poor roads and poor market infrastructure), investments in pro-
duction of crops species at farm household level yield better nutritional outcomes than invest-
ments in livestock species–especially via the own farm produce consumption pathway.
Therefore, for the government of Uganda, and those partners who are flexible with what they
support, the emphasis could be put on diversification in production of crop species. However,
our analysis did not consider what crop or animal species were more feasible than others
regarding the studied nutrition outcomes, hence we are cautious not to make specific crops or
animal species’ recommendations. Moreover, this can be an interesting area of research to be
exploited in the future. Nevertheless, on-going food security and nutrition initiatives can be
effectively guided by this research on the general areas of prioritization for instance, prioritiz-
ing investments in crops species diversification. Strategic examples of these policy initiatives
among others include: 1) The Uganda Nutrition Action Plan II (UNAP II) that spans between
2020–2025, and aims to leave behind none among Ugandans in scaling up nutrition outcomes.
2) The Uganda Food and Nutrition Policy (UFNP). 3) The Uganda National Agriculture Pol-
icy (UNAP). 4) and the Uganda’s Multisectoral Food Security and Nutrition Project
(UMFSNP).
Study limitations
From literature some of which was cited in this paper; indeed, the population consumes fewer
protein foods as would be expected of rural populations, and we agree that studying protein
consumption patterns would be noble. Unfortunately, we did not study proteins for this study.
This is because, we build on the work of Sekabira et al. [28], where in the original data, did not
compute for proteins. However, we make a stark recommendation to expand consideration of
these strategic macro and micronutrients in empirical studies on nutrition. Even though such
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 18 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
wider coverage of many macro and micronutrients may be difficult in one paper, but done
individually for each macro or micronutrient would suffice, to avoid getting lost into so much
detail.
Conclusions
Using nationally representative rural households’ panel data from Uganda, we establish that
indeed, FPD in general is positively and significantly associated with HDDS. However, the two
sub-components of FPD (animal species count and crops species count) are differentially asso-
ciated with HDDS. For instance, the animal species count is significantly and positively associ-
ated with HDDS, while the crop species count showed a negative association. Differential and
insignificant associations could stem from the high categorization (aggregation) embedded in
HDDS. On analyzing nutrition indicators (energy, iron, zinc, and vitamin A) embedded in
HDDS, singly, and disaggregating these by source (own farm vs. markets), the crops species
count is significantly and positively associated with all studied micronutrients and energy irre-
spective of their source–except for vitamin A sourced from markets where the association is
negative but insignificant. On the other hand, the animal species count is also significantly and
positively associated with energy and all micronutrients sourced from consumption of own
farm produce, and vitamin A sourced from markets. However, such association is insignificant
but positive for zinc and iron sourced from markets, while it is negative for energy. The wider
significance of the crops’ species count clearly highlights the strategic importance of crops
towards better smallholder households’ nutrition–especially in a rural sample like ours, where
remoteness and poor market infrastructure are persistently prevalent. Crops can easily be con-
sumed directly or sold to markets for income to buy other food items. Generally, concerning
individual micronutrients and energy intake, the crop species count shows a stronger associa-
tion via the own farm produce consumption pathway. Therefore, in a smallholder farmer con-
text, diversification in crop species could be more important than animal species
diversification towards availing more energy, and micronutrients per adult. Hence, compara-
tive efforts (household or policy level) targeted towards crop species diversification in farm
production could still yield better household nutrition outcomes. However, notice should be
taken that our sample is fully rural, and traditionally dominantly reliant on crops than animals
to satisfy their food needs and general livelihoods. Hence, our results may not be binding in a
context of countries that are predominantly dependent on animals (pastoralists) hence, must
be interpreted cautiously in dominantly pastoral and urban contexts.
Supporting information
S1 Table. Animal and crops species used in the calculation of the FPD bio index.
(DOCX)
S2 Table. Association of farm production diversity (FPD) on Household dietary diversity
score (HDDS).
(DOCX)
S3 Table. Association of farm production diversity (FPD) and daily energy intake per
adult equivalent (AE).
(DOCX)
S4 Table. Association of farm production diversity (FPD) and daily iron intake per adult
equivalent (AE).
(DOCX)
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 19 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
S5 Table. Association of farm production diversity (FPD) and daily zinc intake per adult
equivalent (AE).
(DOCX)
S6 Table. Association of farm production diversity (FPD) and daily vitamin-A intake per
adult equivalent (AE).
(DOCX)
S1 File. Data used for the generation of these model results.
(DTA)
S2 File. Data from FAOSTAT used to further show that food production in 2015 was gen-
erally more than that in 2018 or 2019.
(CSV)
Acknowledgments
We are grateful to the World Bank for making this data available, and UBOS for collecting the
data. We also gratefully acknowledge Heath Henderson, and our anonymous reviewers for
their immensely constructive comments.
Author Contributions
Conceptualization: Haruna Sekabira.
Data curation: Haruna Sekabira, Zainab Nansubuga.
Formal analysis: Haruna Sekabira.
Funding acquisition: Haruna Sekabira.
Investigation: Haruna Sekabira.
Methodology: Haruna Sekabira.
Project administration: Haruna Sekabira, Zainab Nansubuga, Stanley Peter Ddungu, Lydia
Nazziwa.
Resources: Haruna Sekabira, Zainab Nansubuga.
Software: Haruna Sekabira.
Supervision: Haruna Sekabira, Zainab Nansubuga, Stanley Peter Ddungu.
Validation: Haruna Sekabira, Stanley Peter Ddungu, Lydia Nazziwa.
Visualization: Haruna Sekabira, Zainab Nansubuga, Stanley Peter Ddungu, Lydia Nazziwa.
Writing – original draft: Haruna Sekabira.
Writing – review & editing: Haruna Sekabira, Zainab Nansubuga, Stanley Peter Ddungu,
Lydia Nazziwa.
References
1. FAO. FAO and the 17 Sustainable Development Goals. Food and Agriculture Organization of the
United Nations, Rome–Italy; 2015.
2. FAO, IFAD, UNICEF, WFP, WHO. The State of Food Security and Nutrition in the World 2019. Safe-
guarding against economic slowdowns and downturns. FAO, Rome, Italy. Accessed from http://www.
fao.org/3/ca5162en/ca5162en.pdf on 26th August 2021.
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 20 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
3. FAO. Sustainable food value chain development–Guiding principles. Rome, Italy, 2014.
4. High-Level Panel of Experts on Food Security and Nutrition (HLPE). Food losses and waste in the con-
text of sustainable food systems. A report by the High-Level Panel of Experts on Food Security and
Nutrition of the Committee on World. Food Security. Rome, Italy, 2014.
5. Ingram J. A food systems approach to researching food security and its interactions with global environ-
mental change. Food Security, 2011, 3 (4): 417–431. https://doi.org/10.1007/s12571-011-0149-9
6. United Nations Environment Programme (UNEP). Food Systems and Natural Resources. A Report of
the Working Group on Food Systems of the International Resource Panel. Nairobi, Kenya, 2016.
7. International Panel of Experts on Sustainable Food Systems (IPES). The new science of sustainable
food systems: Overcoming barriers to food system reform, 2015. Accessed on 26th May 2022 from:
https://www.ipes-food.org/_img/upload/files/NewScienceofSusFood.pdf
8. Gaillard C, Verger EO, Dury S, Dop MC, El Ati J. Farm production diversity and women’s dietary diver-
sity: Evidence from central Tunisia. PLoS ONE,2022; 17(2): e0263276. https://doi.org/10.1371/journal.
pone.0263276 PMID: 35130282
9. Van Campenhout B, Sekabira H, Aduayom HD. Poverty and its Dynamics in Uganda–Explorations
Using a New Set of Poverty Lines. In Arndt C., McKay A., & Tarp F., (Eds.), Growth and Poverty in Sub-
Saharan Africa. Oxford, UK: Oxford University Press, 2016; pp.138 – 160.
10. Reardon T, Taylor JE, Stamoulis K, Lanjouw P, Balisacan A. Effects of Non-Farm Employment on
Rural Income Inequality in Developing Countries: An Investment Perspective. Journal of Agricultural
Economics, 2000; 51 (2): 266–288. https://doi.org/10.1111/j.1477-9552.2000.tb01228.x
11. Islam AHMS, Von Braun J, Thorne-Lyman LA, Ahmed UA. Farm diversification and food and nutrition
security in Bangladesh: Empirical evidence from nationally representative household panel data. Food
Security, 2018; 10, 701–720, https://doi.org/10.1007/s12571-018-0806-3
12. Sekabira H, Nalunga S. Farm Production Diversity: Is it Important for Dietary Diversity? Panel Data Evi-
dence from Uganda. Sustainability, 2020; 12(3), 1028. https://doi.org/10.3390/su12031028
13. Minten B, Barrett CB. Agricultural technology, productivity, and poverty in Madagascar. World Develop-
ment, 2008; 36,797–822. https://doi.org/10.1016/j.worlddev.2007.05.004
14. Godfray HCJ, Beddington JR, Crute IR, Haddad L, Lawrence D, Muir JF, et al. Food Security: The Chal-
lenge of Feeding 9 billion People. Science, 2010; 327, 812–818. https://doi.org/10.1126/science.
1185383 PMID: 20110467
15. Jones AD, Shrinivas A, Bezner-Kerr R. Farm production diversity is associated with greater household
dietary diversity in Malawi: Findings from nationally representative data. Food Policy, 2014; 46, 1–12,
https://doi.org/10.1016/j.foodpol.2014.02.001
16. Webb P, Kennedy E. Impacts of Agriculture on Nutrition: Nature of the evidence and research gaps.
Food and Nutrition Bulletin, 2014; 35 (1), 126–132. https://doi.org/10.1177/156482651403500113
PMID: 24791585
17. Sibhatu KT, Krishna VV, Qaim M. Production diversity and dietary diversity in smallholder farm house-
holds. Proceedings of the National Academy of Sciences of USA, 2015; 112, 10657–10662. https://doi.
org/10.1073/pnas.1510982112 PMID: 26261342
18. Sibhatu KT, Qaim M. Review: Meta-analysis of the association between production diversity, diets and
nutrition in smallholder farm households. Food Policy, 2018; 77, 1–18. https://doi.org/10.1016/j.
foodpol.2018.04.013
19. Sibhatu KT, Qaim M. Farm production diversity and dietary quality: linkages and measurement issues.
Food Security, 2018; 10(1), 47–59. https://doi.org/10.1007/s12571-017-0762-3
20. Olabisi M, Obekpa OH, Liverpool-Tasie OSL. Is growing your own food necessary for dietary diversity?
Evidence from Nigeria. Food Policy, 2021; 104, https://doi.org/10.1016/j.foodpol.2021.102144
21. Wikipedia. The economy of Uganda. Available from: https:/en.wikipedia.org/wiki/Economy_of_Uganda
accessed on, 26th August 2021.
22. United Nations. Least Developed Country Category: Uganda Profile. Available from: https://www.un.
org/development/desa/dpad/least-developed-country-category-uganda.html accessed on 26th August
2021.
23. Haddinott J., Agriculture health, and nutrition: towards conceptualizing the linkages. In: Fan S., Pandya-
Lorch R., Eds. Reshaping Agriculture for Nutrition and Health. Washington DC–United States: Interna-
tional Food Policy Research Institute, 2012.
24. Koppmair S, Kassie M, Qaim M. Farm production, market access and dietary diversity in Malawi. Public
Health Nutrition, 2017; 20, 325–335, https://doi.org/10.1017/S1368980016002135 PMID: 27609557
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 21 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
25. Whitney CW, Luedeling E, Hensel O, Tabuti JRS, Krawinkel M, Gebauer J, et al. The role of home gar-
dens for food and nutrition security in Uganda. Human Ecology, 2018; 46, 497–514. https://doi.org/10.
1007/s10745-018-0008-9
26. Muthini D, Nzuma J, Nyikal R. Farm production diversity and its association with dietary diversity in
Kenya. Food Security, 2020; (12), 1107–1120. https://doi.org/10.1007/s12571-020-01030-1
27. Chegere JM, Stage J. Agricultural production diversity, dietary diversity, and nutritional status: Panel
data evidence from Tanzania. World Development, 2020; 129. https://doi.org/10.1016/j.worlddev.
2019.104856
28. Sekabira H, Nalunga S, Umwungerimwiza YD, Nazziwa L, Ddungu S. Household Farm Production
Diversity and Micronutrient Intake: where are the linkages? Panel Data Evidence from Uganda. Sustain-
ability, 2021; 13, 4041. https://doi.org/103390/su13074041
29. Jones DA. Critical review of the emerging research evidence on agricultural biodiversity, diet diversity,
and nutritional status in low- and middle-income countries, Nutrition Reviews, 2017; 75 (10), 769–782,
https://doi.org/10.1093/nutrit/nux040 PMID: 29028270
30. Shariff ZM, Khor GL. Obesity and Household Food Insecurity: Evidence from a Sample of Rural House-
holds in Malaysia. European Journal of Clinical Nutrition, 2005; 59, 1049–1058. https://doi.org/10.
1038/sj.ejcn.1602210 PMID: 16015271
31. Mello JA, Gans KM, Risica PM, Kirtania U, Strolla LO, Fournier L. How Is Food Insecurity Associated
with Dietary Behaviors? An Analysis with Low-Income, Ethnically Diverse Participants in a Nutrition
Intervention Study. Journal of the Academy of Nutrition and Dietetics, 2010; 110 (12), 1906–1911.
https://doi.org/10.1016/j.jada.2010.09.011 PMID: 21111099
32. FAO. Uganda Nutrition Profile. Nutrition and Consumer Protection Division, FAO, Rome, Italy, 2010.
33. Food and Nutrition Technical Assistance II Project (FANTA-2). The Analysis of the Nutrition Situation in
Uganda. FHI 360, Washington DC, USA, 2010.
34. Benson T. An assessment of the likely impact of rising global food prices on Ugandan households. Inter-
national Food Policy Research Institute (IFPRI). Kampala, Uganda, 2008.
35. Uganda Bureau of Statistics, Statistical Abstract, 2018. Accessed on 25th May 2022 from: https://www.
ubos.org/wp-content/uploads/publications/05_2019STATISTICAL_ABSTRACT_2018.pdf
36. Kabunga N, Ghosh S, Griffiths J. Can smallholder fruit and vegetable production systems improve
household food security and nutritional status of women? Evidence from rural Uganda International
Food Policy Research Institute (IFPRI); Washington, DC, USA, 2014; Discussion Paper No. 01346
37. UNICEF. UNICEF/WHO/World Bank Joint Child Malnutrition Estimates, 2021 Edition. Accessed from:
https://data.unicef.org/topic/nutrition/malnutrition/#:~:text=In%202020*%2C%2022%20per%
20cent,203.6%20million%20to%20149.2%20million. on 25th July 2022
38. USAID. Uganda: Nutrition Profile. Accessed from: https://www.usaid.gov/sites/default/files/documents/
tagged_Uganda-Nutrition-Profile.pdf. on 25th July 2022
39. Wikipedia. List of countries by food energy intake. Accessed from: https://en.wikipedia.org/wiki/List_of_
countries_by_food_energy_intake. on 25th July 2022
40. Stoltzfus RJ. Iron deficiency: global prevalence and consequences. Food Nutrition Bulletin, 2003; 24
(4), 99–103. https://doi.org/10.1177/15648265030244S206 PMID: 17016951.
41. Nankinga O, Aguta D, Kabahuma C (2019). Trends and Determinants of Anemia in Uganda: Further
Analysis of the Demographic and Health Surveys. DHS Working Paper No. 149. Rockville, Maryland,
USA.
42. Kumera G, Awoke T, Melese T et al. Prevalence of zinc deficiency and its association with dietary,
serum albumin and intestinal parasitic infection among pregnant women attending antenatal care at the
University of Gondar Hospital, Gondar, Northwest Ethiopia. BMC Nutrition, 2015, 1 (31). https://doi.org/
10.1186/s40795-015-0026-6
43. Kikafunda JK, Walker AF, Allan EF, Tumwine JK. Effect of zinc supplementation on growth and body
composition of Ugandan preschool children: a randomized, controlled, intervention trial. The American
journal of clinical nutrition, 1998, 68(6), 1261–1266. https://doi.org/10.1093/ajcn/68.6.1261 PMID:
9846856
44. Kajjura RB (2007). Assessment of dietary zinc intake, serum zinc concentration and nutritional status of
school children aged 60–120 months in Wakiso District in Central Uganda. MSc. Thesis. Makerere
University Kampala, Uganda.
45. Fiedler JL, Afidra R. Vitamin A fortification in Uganda: comparing the feasibility, coverage, costs, and
cost-effectiveness of fortifying vegetable oil and sugar. Food and Nutrition Bulletin, 2010; 31(2),193–
205. https://doi.org/10.1177/156482651003100202 PMID: 20707225
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 22 / 23
PLOS ONE Farm production diversity, household dietary diversity and nutrition
46. Wirth JP, Petry N, Tanumihardjo SA, et al. Vitamin A supplementation programs and country-level evi-
dence of vitamin A deficiency. Nutrients, 2017; 9(3), 190. https://doi.org/10.3390/nu9030190 PMID:
28245571
47. World Bank. Country Data–Uganda. Accessed from: https://www.worldbank.org/en/programs/lsms/
initiatives/lsms-ISA#8. on 29th May 2021
48. Global Yield Gap Atlas (2022). Description of cropping systems, climate, and soils in Uganda. Accessed
from: https://www.yieldgap.org/uganda# on 2nd September 2022.
49. Di Falco S, Chavas JP. On crop biodiversity, risk exposure, and food security in the highlands of Ethio-
pia. American Journal of Agricultural Economics, 2009; 91, 599–611. https://doi.org/10.1111/j.1467-
8276.2009.01265.x
50. Swindale A, Bilinsky P. Household Dietary Diversity Score for Measurement of Household Food
Access: Indicator Guide. Washington DC, USA: Food and Nutrition Technical Assistance Project
(FANTA); 2006
51. Sekabira H, Qaim M. Can mobile phones improve gender equality and nutrition? Panel data evidence
from farm households in Uganda, Food Policy, 2017; 73, 95–103. https://doi.org/10.1016/j.foodpol.
2017.10.004
52. Hotz C, Lubowa A, Sison C, Moursi M, Loechl CA. Food Composition Table for Central and Eastern
Uganda. Harvest Plus Technical Monograph 9. Washington DC, USA: FHarvest Plus; 2012.
53. FAO, WHO, UNU. Human Energy Requirements: Report of a Joint FAO/WHO/UNU (Food and Agricul-
ture Organization, World Health Organization and United Nations University). Rome, Italy, FAO; 2001.
54. UBOS. Statistical Abstract 2016. Uganda Bureau of Statistics, Kampala, Uganda. 2016.
55. UBOS. Statistical Abstract 2019. Uganda Bureau of Statistics, Kampala, Uganda. 2019.
56. UBOS. Statistical Abstract 2020. Uganda Bureau of Statistics, Kampala, Uganda. 2020.
57. Wooldridge JM (Ed). Econometric Analysis of Cross section, and Panel Data. Cambridge, London:
MIT, UK; 2010.
58. Cameron AC, Trivedi PK. Microeconometrics: Methods and Applications. Cambridge, UK: Cambridge
University Press; 2005.
59. Mundlak Y. On the pooling of time series and cross section data. Econometrica, 1978; 46, 69–85,
https://doi.org/10.2307/1913646
60. Bellemare M. F., & Wichman C. J. (2020). Elasticities and the inverse hyperbolic sine transformation.
Oxford Bulletin of Economics and Statistics, 82(1), 50–61. https://doi.org/10.1111/obes.12325
61. Bellemare M. F., Barrett C. B., & Just D. R. (2013). The welfare impacts of commodity price volatility:
evidence from rural Ethiopia. American Journal of Agricultural Economics, 95(4), 877–899. https://doi.
org/10.1093/ajae/aat018
62. Fongar A, Gödecke T, Aseta A, Qaim M. How well do different dietary and nutrition assessment tools
match? Insights from rural Kenya. Public Health Nutrition, 2019; 22, 391–403. https://doi.org/10.1017/
S1368980018002756 PMID: 30376906
63. Kyomuhendo GB. Low Use of Rural Maternity Services in Uganda: Impact of Women’s Status, Tradi-
tional Beliefs and Limited Resources. Reproductive Health Matters, 2003; 11(21), 16–26. https://doi.
org/10.1016/s0968-8080(03)02176-1 PMID: 12800700
64. Appiah-Opoku S. Indigenous Beliefs and Environmental Stewardship: A Rural Ghana Experience. Jour-
nal of Cultural Geography, 2007; 24(2), 79–98. https://doi.org/10.1080\08873630709478212
65. Sekabira H, Qaim M. Mobile Money, Agricultural Marketing, and Off-Farm Income in Uganda. Agricul-
tural Economics, 2017; 48 (5), 597–611. https://doi.org/10.1111/agec.12360
66. Barrett CB, Reardon T, Webb P. Nonfarm income diversification and household livelihood strategies in
rural Africa: concepts, dynamics, and policy implications. Food Policy, 2001; 26 (4), 315–331. https://
doi.org/10.1016/S0306-9192(01)00014-8
67. Rahman S, Rahman M. Impact of land fragmentation and resource ownership on productivity and effi-
ciency: The case of rice producers in Bangladesh. Land Use Policy, 2009; 26(1), 95–103. https://doi.
org/10.1016/j.landusepol.2008.01.003
68. Ntakyo PR, van den Berg M. Effect of market production on rural household food consumption: evi-
dence from Uganda. Food Security, 2019; 11, 1051–1070. https://doi.org/10.1007/s12571-019-00959-
2
PLOS ONE | https://doi.org/10.1371/journal.pone.0279358 December 16, 2022 23 / 23