Asiimwe et al.

Pastoralism: Research, Policy and Practice (2020)

10:5 Pastoralism: Research,
https://doi.org/10.1186/s13570-020-0160-x
Policy
and Practice

RESEARC Open
H Access
The role of camel production on household
resilience to droughts in pastoral and
agro-pastoral households in Uganda
Robert Asiimwe1*, John Herbert Ainembabazi1, Anthony Egeru2,3, Rosemary Isoto1, Daniel Knox
Aleper4, Justine Namaalwa2 and Gracious M. Diiro1

Abstract
Recurrent and prolonged droughts have exacerbated the problems of pasture and water scarcity in arid a
Keywords: Resilience, Pastoralism, Karamoja, Camels, Droughts

Introduction markets (Opiyo et al. 2015). These reinforcing effects

Pastoral and dry land livelihoods are negatively affected constrain pastoralists’ income and family nutritional se-
by climate variation-induced shocks (United Nations curity in the affected areas.
Development Program 2009) through recurrent and It is expected that increasing variability in climates will
prolonged droughts which constrain access to water and continue affecting livestock production systems in all
pasture for livestock. This has led to serious ecological parts of the world, including the rural poor who have
and economic consequences to rangelands and range- livestock as their main source of livelihood (Kima et al.
land users, especially in Africa’s dry lands (Vetter 2009). 2015). Increase in evapotranspiration due to increased
Recurrent droughts lead to increases in incidences of warming of the Earth may outpace increases in precipi-
certain livestock diseases and livestock deaths; deterior- tation, thus increasing terrestrial aridity1 (Brookshire
ation of livestock conditions at times force pastoralists and Weaver 2015). These effects will alter landscapes,
to alter their herd structure; and a collapse of livestock ecosystems, and conditions where people live. According

* Correspondence: robertasiimwe4@gmail.com 1
Aridity describes the water deficiency of a given climate and is
1
School of Agricultural Sciences, Makerere University, Kampala, commonly characterized by the ratio of precipitation to potential
Uganda Full list of author information is available at the end evapotranspiration (Lickley and Solomon 2018).
of the article

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0
International License, which permits use, sharing, adaptation, distribution and reproduction in any medium
or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article's Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 2 of

to Lickley and Solomon (2018), increased aridity contrib- because of their inability to withstand many days with-
utes to more frequent droughts, severe soil moisture out water during the drought periods. In the dry north,
def- icits, decreases in carbon uptake, and the preference of integrating camels into mainly cattle pro-
expansion of global drylands. As the global climate duction systems in dry lands by some communities that
continues chan- ging, arid and semi-arid lands will did not previously keep camels had been reported in
increase by 5 to 8% in Africa across different climate earlier studies (Farm-Africa 2002).
scenarios, thereby result- ing in exposure of between 75 Camels present a unique opportunity for adaptation to
and 250 million people to droughts by 2080 droughts and varying pasture quality in ASALs. They are
(Intercontinental Panel on Climate Change 2014). The able to survive well in ASALs due to their biological and
increase in arid and semi-arid lands will in part drive physiological adaptations which help them to cope with
the change in ecological systems and survival tactics of harsh environmental conditions (Kagunyu and Wanjohi
the affected communities, which are mainly pastoralist 2014; Yosef et al. 2014). They drink less water and are
communities (IPCC 2014). able to stay for many days without water due to their
The hot and dry nature with low and erratic rainfall of tolerance of dehydration compared to other livestock spe-
arid and semi-arid lands (ASALs) leaves few livelihoods cies. They are able to convert the scanty plant resources of
that are suited to this arid environment other than mo- the ASALs into milk, meat, and fibre (Ahmad et al. 2010).
bile livestock-keeping of mostly cattle, goats, and sheep They almost have no competition for feed with other ani-
which is particularly well adapted (Oxfam 2008). How- mals, as they are hardy animals and comparatively eat less
ever, climate variability manifested in the form of recur- food (Khan et al. 2003). Additionally, camels are kept as
ring drought is making it increasingly hard for cattle security against calamities and natural disasters like
— the most important livestock species in African dry droughts and diseases that may be devastating to other
lands to survive in the ASALs (Kagunyu and Wanjohi livestock species kept (Mochabo et al. 2005), hence can
2014). According to Kagunyu and Wanjohi (2014), the be integrated into other production systems.
drought of 2005 to 2006 in Kenya reduced cattle, goat, Camels are important livestock species in the subsist-
and sheep herds by up to 70% in the heavily affected ence economy of rural pastoral communities especially
areas, which left the affected communities poor and hence in ASALs (Aujla et al. 2013). They contribute to house-
more dependent on aid. hold food security by provision of meat and milk
Communities in ASALs have shown resilience over (Ahmad et al. 2010). They are used as pack animals for
time to such conditions as long droughts, and pasture transport and provide household income through sale of
and water stresses, through adoption of livestock- live animals, meat, milk, and other by-products like hair
dependent livelihoods that are well suited for such ter- and hides (Aujla et al. 2013; Faye et al. 2010; Mochabo
rains. Traditionally, pastoral communities survive the ef- et al. 2005). Field and Karuiki (2005) estimated that the
fects of drought through breeding locally adapted volume of milk produced by camels is six times higher
livestock species and diversifying livestock species kept than that produced by indigenous cattle reared in the
(like cattle, goats, sheep, camels, donkeys, and poultry), dry lands especially during the dry periods. Mochabo
and through resource management practices like calf et al. (2005) further identified that camels are given as
grazing paddocks which are becoming common in East bride price among some groups in Kenya.
Africa (The International Union for Conservation of Na- The contribution of camels to household subsist-
ture (IUCN) 2010). Pastoral tribes (Samburu and Tur- ence and national economy in developing countries is
kana) in northern Kenya have begun to increase the generally obscured by inaccurate estimates of camel
number of camels that they manage, substituting them populations due to the lack of regular censuses and
for cattle in order to have more drought-resilient live- the fact that camel products seldom enter a formal
stock (Kagunyu and Wanjohi 2014, 2015; IUCN 2010). marketing system (Ahmad et al. 2010). As such, the
Kenya’s camel population increased from 0.9 million in economic contribution of camel-rearing is often
1999 to three million in 2009 (Kagunyu and Wanjohi underestimated. Despite having unclear statistics,
2014) despite the increased occurrence of droughts camel production is increasingly receiving much at-
within this period. tention as a major source of food (meat and milk) in semi-
Zwaagstra et al. (2010), in their assessment of the im- arid and arid areas (Aujla et al. 2013). In some pastoral
pacts of the 2008–2009 droughts in northern Kenya, households, camel milk is an integral part of the food
found out that communities that traditionally never basket and household income and has a huge potential
adopted camels were increasingly integrating camels into for poverty reduction (Musinga et al. 2008; Elhadi et al.
their herds as, in part, an adaptation strategy to drought 2015; Aujla et al. 2013). A number of existing studies,
occurrence. This was due to high death rates of cattle however, highlight milk as the main product, leaving
and other livestock during the drought periods. Further, out other camel products such
some communities did not prefer keeping cattle alone
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 3 of

as meat and blood that could also contribute to poverty line (Browne and Glaeser 2010; FAO 2018). In
household income. addition to cross-border conflicts, the region is also
Extensive keeping of livestock is the major economic prone to increasingly frequent and severe natural disas-
activity in the Karamoja region of Uganda due to its semi- ters, especially droughts, in part as a result of climate
arid nature. Though well adapted for such a cli- mate, change (Caffrey et al. 2013; United Nations Development
mobility of livestock in the region has been limited by Program (UNDP) 2009; World Food Program of the
chronic insecurity of the area in the past, as well as low- United Nations (WFP) 2009).
quality forage; limited access to water, especially in the
dry season; high incidence of contagious livestock Methods
diseases, and limited access to veterinary services and Data used in this study consists of detailed household in-
livestock markets (USAID 2005; FAO 2018). These fac- formation obtained from a cross-sectional survey con-
tors have limited the livestock-based development initia- ducted in January to February 2016. Semi-structured
tives in the region, and as such the region still registers questionnaires were administered through local inter-
the lowest human development indices in the country preters to a total of 52 camel herding households and
(FAO 2018). While the Karamoja region holds 97% of 64 non-camel herding households. Key informant in-
the camel population in the country (Uganda Bureau of terviews were later conducted, guided by a key in-
Statistics 2010), little is known about the community’s formant interview guide to validate the information
motivation for adopting camels, camels’ impact on from the household interviews. Two districts (Moroto
income-generation, and the role they play in the en- and Amudat) were purposively selected due to the ex-
hancement of resilience2 to droughts in regions so prone istence of high camel numbers, and key informant in-
to recurrent and prolonged droughts. Incorporating terviews guided the selection of the sub-counties;
camels into mainstream livestock production presents Rupa in Moroto and Amudat TC, Amudat, and Looro
an opportunity for improved food security and income. in Amudat District.
Since there is little empirical evidence on the effects of A snowball approach was used to identify the camel
camel adoption, this study therefore sought to establish herding households, and the study respondents were se-
the effect of camels on household income and pastoral- lected depending on their availability and willingness to
ists’ resilience to droughts in the Karamoja sub-region. participate in the study. Information on household
demographic characteristics, economic activities, and
Study area physical, social, and economic connectivity of the house-
The study was conducted in the districts of Moroto and holds was collected. Further, information on numbers of
Amudat in north-eastern Uganda (Karamoja sub- camels kept, camel products obtained, proportions of
region) which lies between latitudes 1° 30′ and 4° N products consumed and sold, sale prices, other income
and longi- tudes 33° 30′ and 35° E. These are sources, amount, and available resources such as land
semi-arid lands characterized by unpredictable rainfall and other livestock species was also collected. Target re-
and high temper- atures (Egeru et al. 2014; Mubiru spondents were household head, spouse, or older chil-
2010). These areas border north-western Kenya, south- dren familiar with household routine activities and camel-
eastern Sudan, and south-western Ethiopia. The regions related activities.
have three types of livelihood options; namely crop
production, agro- pastoral and pastoralism, each of Data analysis
which are dictated by the amount of rainfall received in The study examined the effect of smallholder pastoralist
an area. Most of these areas have erratic rainfall or camel production in Karamoja on household resilience
extended dry conditions, diseases and pests, low to droughts, using the conventional random utility
infrastructure, cattle thefts, bush fires, periodic low model. In this framework, individuals were assumed to
water, and limited pasture availability (FAO 2018). make rational decisions by choosing to rear camels if
Karamoja is mostly semi-arid savanna cov- ered with doing so maximized their expected utilities. Following
seasonal grasses, thorny plants, and occasional small Hanemann (1984) and Baltas and Doyle (2001), a ran-
trees (Mubiru 2010). The area is dominated by in- dom utility function for a smallholder farmer in Kara-
digenous tropical grasses and mostly composed of Aca- moja facing a decision to rear camels was specified as in
cia species (Egeru et al. 2014). It is the poorest region Eq. (1) below:
in Uganda with over 80% of its people living below
—
the V ij ¼ V i j þ εi j ¼ Xijθ þ εi j; i ¼ 1; :::; n ð1Þ
2
Resilience can be defined as the ability of, and mechanisms used by, where Vij, utility of alternative j for consumer i, i s a
individuals and communities to cope with external shocks; the ability
of those communities to self-organize after experiencing a shock; and function of the deterministic component Vij and the
the ability to adapt to such shocks mainly through social learning
(Osbahr et al. 2010).
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 4 of

random component εij; X is a vector of observed socio- 2008; Ciani and Romano 2013; Giuseppe et al. 2016;
economic and demographic characteristics of the indi- Guyu and Muluneh 2018; Osbahr et al. 2010). The direct
vidual; and ε is the stochastic component of the utility approach is based on the construction of a Resilience
function representing the unobserved attributes affecting Capacity Index (RCI) or resilience capacity matrix
individual’s choice of the practice, heterogeneity in (RCM). The approach considers several household liveli-
tastes, and measurement errors. hood indicators such as alternative sources of income,
A rational pastoral household would choose to adopt access to productive assets, availability of and access to
camel rearing if the expected utility derived from camels basic services (healthcare, education and extension), and
(Vi1) was higher than that generated without camels social safety nets to determine household response to
(Vi0), given the constraints, such as access to resources, shocks (Ciani and Romano 2013). However, in a review
and information and knowledge about camels. Camel of RCI indicators by Giuseppe et al. (2016) , the RCI
adoption and its role in enhancing household resilience now considers access to basic services (ABS), assets
to droughts was the focus of the study. (AST), social safety nets (SSN), sensitivity (S), and adap-
Camel adoption was expected to contribute to house- tive capacity (AD) of an individual household as the fun-
hold food security through increased meat and milk damental pillars of resilience (Giuseppe et al. 2016). The
availability (Ahmad et al. 2010) and additional household mechanism of response in turn depicts the level of resili-
income through sale of live animals, meat, milk, hair, ence of such a household at any given point in time. The
and hides (Aujla et al. 2013; Faye et al. 2010; Mochabo empirical challenge with this method is that the variables
et al. 2005). Camel-adopting households obtain a higher used to determine the individual components in such an
social status and benefit by having constant milk supply index if included in a regression model would introduce
despite variations in pasture and water availability. Being potential endogeneity. Therefore, the index from the dir-
browsers, camels can thrive on the available scant vege- ect methods cannot be used in an inferential work (Giu-
tation and thus reduce the need for transhumance com- seppe et al. 2016) unless large data sets and a good
pared to the case of cattle. From the benefits of food number of relevant variables are available (Ciani and Ro-
security, increased income flows, and settlement, an mano 2013; Guyu and Muluneh 2018).
adopting household was expected to be more resilient to The indirect measure of household resilience was
the devastating effects of recurrent and prolonged adopted in the determination of household Drought Re-
droughts. Based on this background, a Drought Resili- silience Index. This method has been used in recent
ence Index was constructed for adopting and non- drought resilience (Banda 2015; Keil et al. 2008) and
adopting households. food security (Giuseppe et al. 2016) empirical work to
construct a Resilience Capacity Index that is used as a
Empirical methods dependent variable in the inferential analysis. It used a
Construction of the household Drought Resilience set of determinants of resilience (including policy-
Index Individuals, households, or communities faced relevant decision variables) to infer a given level of resili-
with cli- matic or other risks will tend to prioritize ence to droughts. Construction of this index using the
between ele- ments of the production, consumption, direct methods of resilience measuring would introduce
and ecological systems in which they are at the potential endogeneity problems since some of the vari-
moment (Osbahr et al. 2010). Drought affects a ables used in the construction of the index would be
household through its effect on crop and livestock used as independent variables in the regression (Giu-
production. For a subsistence house- hold, this implies seppe et al. 2016). To avoid the endogeneity problem
a direct effect on production and an indirect effect on that would bias the estimates and thus the inferences
consumption and/or expenditure be- haviours. Due to drawn thereafter, indirect methods were more appropri-
the moisture stress-farm output and farm output ate for this study.
consumption relationships, the effects of droughts can Arising from multiple manifestations of drought ef-
be assessed through observing changes in consumption fects, one variable alone may not have adequately repre-
or expenditure of a given household dur- ing drought sented the effect of a drought; thus, several variables
relative to normal times. Measures such as were used to observe these effects. This necessitated the
consumption or expenditure-smoothing behaviours of use of data reduction methods to adequately represent
individuals or households have thus been used to esti- multiple variables without loss of information and then
mate the effect of droughts on households (Banda proceeding to use them in regression analysis. The prin-
2015; Giuseppe et al. 2016; Keil et al. 2008). cipal component analysis (PCA) approach was adopted
Direct and indirect methods of measuring resilience for this study due to its ability to summarize individual
have been developed, with each having its own indicators while preserving the maximum possible pro-
inherent analysis challenges. Several recent studies have portion of the total variation in the original variables
used the direct approach to resilience measurement
(Alinovi et al.
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 5 of

(Nardo et al. 2008). The study computed a Drought Re- 0

Y 1 ¼ X β1 þ Y 2 γ 1 þ ε1 ε1 ¼ α1 U þ V 1 ð3Þ
silience Index (DRI) for each household using the PCA
approach. Y 2 ¼ Xβ2 þ ε2ε2 ¼ α2U þ V 2 ð4Þ
Determinants of household resilience to droughts
drawn from the literature (Banda 2015; Keil et al. 2008) where Y1 is the resilience to droughts and Y2 is the
were used to construct a Drought Resilience Index camel adoption. U denotes the individual’s unobserved
(DRI). Specifically, the variables chosen for the DRI in- motivation which could affect their decision to rear
cluded proportion of food consumed (Fc) in droughts as camels and the ability to cope with shocks; V1and V2 are
compared to normal season, months of food scarcity random errors. It is observed that some of the
(Ms), and meals per day (Md) in times of scarcity as parameters in the models cannot be identified without
measures of consumption-smoothing capacity, and num- additional information. By either imposing equality con-
ber of alternative sources of income (Ai) and proportion straints on the coefficients of X (that is using ordinary
of total income that is off-farm (Ip) as measures of in- least squares (OLS) regression), or assuming that one or
come stability in the event of droughts. As specified in more elements of β1 are equal to zero, one could identify
Eq. (2), DRI was expressed as: the parameters which would permit the estimation of
the Y1 equation using instrumental variables according
to Lewbel.
DRI ¼ wcFc þ wsMs þ wdMd þ wiAi þ wpIp However, if one assumes Z to be a vector of observed
exogenous variables (Z could be a subset of X or equal
ð2Þ
to X), Lewbel argues that if Eq. (5) moment conditions
are met, then:
where wn was the proportions explained by the given
EðXε1Þ ¼ 0; EðXε2Þ ¼ 0; CovðZ; ε1ε2Þ ¼ 0 ð5Þ
factor in the PCA analysis used as weights; n = c, s, d,
i, p, and the other variables are as described above. and there is some heteroscedasticity of εj, one can esti-
mate the above set of equations by using [Z − E(Z)]ε2 as
Estimating the effect of camel adoption on an instrument, employing either two-stage least squares
household resilience to droughts (TSLS) or generalized methods of moments (GMM).
The study estimated the effect of camel adoption on TSLS may be inefficient when heteroscedasticity is
household resilience to droughts. However, the decision present in the regression errors, which is a precondition
to adopt camels and household resilience to droughts for implementing the Lewbel approach (Mishra and
are potentially endogenous. Camel-rearing households Smyth 2015); hence, GMM method in Stata was used to
may be systematically different from non-rearing house- fit the model. Lewbel’s own empirical work and Ventura
holds with respect to observed and unobserved attri- (2018) have shown that the resulting IV estimates are
butes, resulting in inconsistent estimates of the effect of very close to those using conventional valid IVs.
camel rearing on household resilience to the effect of re-
current and prolonged droughts. For example, most mo- Results and discussion
tivated pastoralists with higher management abilities and Socio-economic and demographic characteristics
skills are more likely to adopt new resilience-enhancing of the surveyed households
technologies such as camel rearing. If so, the effect of This subsection presents selected socio-economic and
adoption would be biassed upwards due to positive demographic attributes of the farmer households in-
correlation with unobservable motivation to adopt cluded in the survey. The attributes presented in Table 1
resilience-enhancing technologies. Estimation of the ef- include main livelihood activities of the household, hu-
fect using ordinary least squares estimation would thus man capital (household size and dependence ratio,
yield biassed results. The natural candidate for circum- household education completion, and age of the house-
venting this problem would be the use of external in- hold head), access to cropping land and area of land
strumental variables for the adoption of camels. There cultivated, and financial capital (including per capita in-
were difficulties in finding valid instrumental variables come, alternative sources of income, on and off-farm in-
for the study. A method, which utilizes a heteroscedastic come, remittances, and access to credit).
covariance restriction to construct an internal instru- The summary statics in Table 1 show that the two
mental variable (IV), was used to allow parameter esti- farmer categories are generally comparable with respect
mation (Lewbel 2012). Mishra and Smyth (2015) and to most of the attributes except per capita income, in-
Ventura (2018) have recently applied this method to fa- come from the farm and non-farm sectors, and age of
cilitate the identification and estimation of parameters in the farmer. Camel adopters were older and reported
the absence of externally generated instrumental vari- higher per capita income and income from the farm
ables. Using Eqs. (3) and (4) below, the Lewbel approach
is adopted to the study as follows:
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 6 of

Table 1 Socio-economic and demographic characteristics of the surveyed households

Variable All Camel Non-camel t-statistic
households households (n = households (n = 64)
(n = 116)
52)
Mean Std. Mean Std. Mean Std. Dev.
Dev. Dev.
Per capita income (UShs ‘000) 163.06 303.76 231.67 419.31 107.32 138.063 –
7 5 2.230**
Occupation of household head (1, pastoralist; 0, 0.74 0.440 0.75 0.437 0.73 0.445 – 0.190
otherwise)
Age of household head (years) 49.62 14.975 54.19 13.776 45.91 14.982 –
3.070*
Years of experience in rearing camels 26.05 20.281
Marital status (1, married polygamous; 0, otherwise) 0.74 0.439 0.81 0.398 0.69 0.467 – 1.446
Household size (continuous) 11.05 5.138 11.27 5.010 10.88 5.272 – 0.410
Household member completed primary 0.17 0.379 0.19 0.398 0.16 0.366 – 0.507
Dependence ratio 0.67 0.818 0.71 0.823 0.64 0.819 – 0.423
On-farm income (UShs ‘000) 1085.0 1471.22 1628.8 1728.31 643.16 1044.305 –
0 7 0 7 3.791*
Off-farm income (UShs ‘000) 411.35 1172.92 318.04 526.81 487.17 1507.846 0.771
5 1
Proportion of off-farm income to total income 0.36 0.379 0.26 0.353 0.44 0.381 2.667*
Access to credit (1, yes; 0, otherwise) 0.29 0.457 0.25 0.440 0.32 0.471 0.783
Number of alternative sources of income (continuous) 1.00 0.780 0.92 0.788 1.06 0.774 0.9567
Assets owned (0 = only agric; 1 = agric and non-agric) 0.66 0.510 0.73 0.490 0.61 0.523 – 1.279
Crop area cultivated 4.40 3.691 4.86 4.037 4.02 3.371 – 1.220
Feeling of food scarcity (1 = yes; 0 = no) 0.93 0.254 0.88 0.323 0.97 0.175 1.788*
**
Months of scarcity 5.49 4.622 5.00 4.847 5.89 4.622 1.010
Meals consumed per day in times of scarcity 1.45 0.450 1.47 0.504 1.43 0.450 – 0.336
Proportion of food consumed in scarcity 0.59 0.595 0.66 0.726 0.54 0.471 – 1.063
Received extension 0.41 0.495 0.48 0.505 0.36 0.484 – 1.319
Member of social group 0.41 0.494 0.31 0.466 0.49 0.504 0.291
Received remittances 0.21 0.407 0.15 0.364 0.25 0.436 1.269
Distance to nearest input stockiest 8.31 6.756 9.19 6.473 7.63 6.942 – 1.204
Distance to nearest extension office 6.69 5.817 6.41 5.648 6.91 5.980 0.446
Distance to nearest health centre 4.08 3.520 4.17 3.555 4.01 3.520 – 0.242
Distance to nearest primary school 2.67 3.110 3.04 4.258 2.36 1.604 – 1.154
Distance to nearest secondary schools 11.09 6.663 11.44 7.124 10.83 6.355 – 0.469
***Significant at 10%, **significant at 5%, *significant at
1%

sector than farmers without camels. Higher per capita 040 (about USD 95) for the camel adopters (Table 1). It can
income may be an indicator of household resilience also be noted that the share of non-farm income is
where households with low per capita incomes may
have low savings and limited capacity to smooth
consumption in the event of shock. Furthermore, the
head of a camel- rearing household was about 8 years
older (54) than their non-camel counterparts (46). Age
of the household head is likely to be associated with
wealth accumulation in households and may lead to
increased resilience to shocks in households.
Whereas there is no significant difference in off-farm
income between the two farmer categories, non-
adopting households registered higher non-farm income
of UShs 487,170 (about USD 1145) relative to UShs 318,
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 7 of
significantly higher among the non-adopters relative to
the adopters. These results suggest that non-camel
farmers appear to compensate for the low farm income
by engaging in the non-farm sector to smooth
consumption.
Similarly, there is no difference in the level of educa-
tion between the two household categories;
completion of education is still very low, with less than
20% of the households reporting to have at least a
member who has completed primary school (Table 1).
Low education level affects the potential of household
members to involve in alternative sources of income
such as employment in the formal sector or participation
in non-farm sectors. Edu- cation attainment affects the
adoption of resilience- and income-enhancing
technologies introduced in the com- munity due to
limited ability to synthesize new informa- tion and
make an informed decision. Low education
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 8 of

level could therefore lead to generally lower resilience in scarcity decreases the opportunities of a household to
the community due to low adoption of resilience- smoothen consumption for example from saved food.
enhancing technologies introduced. Thus, as the number of months households experience a
The results in Table 1 further show that more camel- drought increase, the households are more likely to re-
rearing households felt food scarcity than the non-camel- spond to food shortages by reducing the proportions
rearing households. However, both farmer categories of con- sumed of whatever food is available. Opiyo et al.
households reported a comparable num- ber of meals (2014) and Muricho et al. (2019) stress the importance of
eaten per day (1.45) and adjusted quan- tities of food in income diversification in enhancing household
the periods of scarcity and plenty. Non-camel-rearing resilience to droughts; this study used alternative
households reduced their con- sumption by almost a half sources of in- come and proportion of off-farm income
(46%) whereas camel household reduced consumption by to total income to capture the consumption-smoothing
34% on average during times of scarcity. This means oppor- tunities in the event of a drought and, hence,
that camel- rearing households were slightly less affected income stability indicators. Most households in rural
by food scarcity as measured through the reduction in areas produce most of their food and are more likely to
food consumption during the times of scarcity—mostly re- sort to consumption adjustments in the event of
due to droughts. droughts. Months of scarcity, meals consumed per
The average numbers of years of rearing camels in the day in scarcity, and the proportion of food consumed
adopting households were 26 years. However, differences in scarcity as compared to normal times were used to
were observed between the two districts, with signifi- account for the consumption adjustments of a house-
cantly higher years of rearing in Moroto (38 years) than hold in the event of droughts and, thus, consumption
in Amudat (13 years). This shows that the rearing of stability indicators. The existing strong correlations
camels is a relatively new practice in Amudat compared justify the use of PCA as a dimension reduction
to Moroto. method on the data. Bartlett’s test of sphericity and
the Kaiser–Meyer–Olkin (KMO) measure of sampling
The level of household resilience to drought in adequacy were also done to confirm the use of PCA.
Karamoja sub-region The chi-square value for Bartlett’s test of sphericity
This study constructed an index of resilience to was 107.040 and significant at 1% level of significance,
drought for each household using PCA , which helps leading to the rejection of the null hypothesis and
to aggregate several variables measuring one latent conclusion that the variables used in the study were
variable into a single index. Following Giuseppe inter-correlated and that the correlations did not re-
et al. (2016) and Nardo et al. (2008), a correlation sult from a sampling error. This means that there
analysis was performed to determine the existence of were suitable correlations to warrant the application
a strong association among the variables used in the of PCA on the data.
analysis. The results of the correlation analysis are The Kaiser–Meyer–Olkin measure of sampling ad-
presented in Table 2. The results show a strong and equacy yields a value of 0.503, indicating a fair correl-
positive correlation between alternative sources of ation between the included variables and an adequate
income and proportion of off-farm income to total sample to carry on with PCA. The result from these
income. A strong negative correlation is noted between two tests shows confidence in the application of PCA
months of scarcity and the proportion of food consumed. as a method of dimension reduction of the data. The
The negative correlation is expected; increasing months of results from the principal component analysis are pre-
sented in Table 3.

Table 2 Correlation matrix for the variables used in construction of the index
Variable Alternative Proportion Meals Proportion of Months of food
sources of of off-farm consumed/day in food consumed in scarcity in the last
income income scarcity scarcity drought
Alternative sources of income 1.0000
Proportion of off-farm income 0.448* 1.0000
Meals consumed per day in – 0.258* – 0.147 1.000
scarcity
Proportion of food – 0.206** – 0.045 0.497* 1.000
consumed in scarcity
Months of food scarcity in 0.052 0.118 – 0.670* – 0.205* 1.00
the last drought 0
**5% and *1% level of significance
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 9 of

Table 3 Eigenvalues of the components meet their food needs from their own production. This
Component Eigenvalu Differenc Proportio Cumulativ is supported by the relatively lower factor loading for
e e n e consumption-smoothing options because droughts limit
Component 1.964 0.674 0.393 0.393 households’ consumption from own production. Using
1
the component loading as the weights for the vari-
Component 1.290 0.406 0.258 0.651
2 ables used in the construction of the DRI, equation
Component 0.885 0.301 0.177 0.829 2 earlier presented was modified to produce equation
3 6 below. Equation 6 was used to construct the drought
Component 0.584 0.307 0.117 0.945 resilience index for each household.
4
Component 0.277 0.055 1.000 DRI ¼ 0:2571 Fc−0:4216Ms þ 0:1651Mc
5 þ 0:6156Ai þ 0:5916Ip ð6Þ

Two components (1 and 2) in Table 3 had The DRI was rescaled to ease interpretation and ex-
eigenvalues above 1 thus adequately capture the planation of regression results and to allow ease of com-
variation within a set of constructs (Nardo et al. 2008). putation of treatment effects which can be interpreted as
However, compo- nent 1 yielded negative signs on the percentage increases—the rescaled index is bound be-
factor loading of the proportion of farm income and the tween 0 and 1 (Giuseppe et al. 2016). A rescaling ap-
number of meals consumed which contradict the proach used by Keil et al. (2008) and Giuseppe et al.
expected signs. Compo- nent 2 on the other hand (2016) was adopted by the study since they have both
presented all signs of factor loadings for the selected worked with resilience to climate variability-induced
variables as expected from the resilience literature and shocks. An adopted min-max scaling was used to trans-
theory. Therefore, it can be con- cluded that the second form the DRI value into a standardized index, ranging
principal component adequately captured drought between 0 and 1 (Giuseppe et al. 2016). The rescaling
resilience. The factor loading, expected signs, and formula was stated in Eq. (7) as:
summary statistics of the variables used in the
construction of the resilience index are presented in Xiω ¼ ðxi−xminÞ=ðxmax−xminÞ ð7Þ
Table 4. where Xi*and xi denote the individual households’
The results show that number of alternative sources of transformed and untransformed DRIs, respectively; xmax
income, and proportion of off-farm income to total and xmin represent the untransformed maximum and
household income are the highest factor loadings. This minimum DRIs, respectively, observed in the data set.
implies that income-smoothing opportunities had a From the results, the max and min values of untrans-
higher bearing on resilience consistent with the findings formed DRI were 2.059 and − 4.748, respectively.
of Opiyo et al. 2014 who found that complementary
The transformation was then worked using Eq. (8)
source of income as well as access to off-farm income
stated below:
was positively related with reducing household vulner-
ability to climate-induced stress—thus increasing resili- Xiω ¼ ðxi−−4:748Þ=ð2:059−−4:748Þ ð8Þ
ence among pastoral communities in Kenya. Opiyo et al.
(2014) and Keil et al. (2008) further show that the risk of Table 5 presents the summary statistics from the
exposure to droughts is almost uniform within stated obtained standardized and unstandardized DRIs for all
geographical confines. Thus, the livelihoods dependent categories of households. After performing the inde-
on natural resources are likely to be uniformly affected. pendent samples t test, the study fails to reject the
This implies that household resilience is more likely to null hypothesis that camel-rearing households and non-
be shaped by factors the households ability to diversify camel-rearing households exhibit similar levels of
its income beyond the most affected farm sectors or be-
yond the farm as a whole. Muricho et al. 2019 assert that
participation in the markets and food aid then become
the next available options when the households fail to

Table 4 Summary of variables used to construct a ‘Drought Resilience Index’ (DRI)

Variable Camel households Non-camel households
Component
Hypothesized
Mean Std. Dev. Mean Std. Dev. sign loading

Number of alternative sources of income 0.920 0.788 1.060 0.774 + 0.616

Proportion of off-farm income to total 0.260 0.353 0.440 0.381 + 0.592
income
Meals consumed per day in times of scarcity 1.470 0.504 1.430 0.450 + 0.257
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 10 of
Proportion of food consumed in scarcity 0.660 0.726 0.540 0.471 + 0.165
Months of scarcity 5.000 4.847 5.890 4.622 – 0.422
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 11 of

Table 5 Summary statistics for the unstandardized and standardized DRIs obtained from principal component analysis
DRI All camel households, mean Camel households, mean Non-camel households, mean t-
statistic
Unstandardized DRI – 1.152 (2.117) – 1.101 (2.160) – 1.190 (2.100) – 0.215
Standardized DRI 0.528 (0.311) 0.536 (0.317) 0.523 (0.309) – 0.215
Standard deviation in parentheses

resilience to droughts. This analysis is strengthened in

Econometric results and discussion
the regression analysis using independent variables to
The effect of camel rearing on household
control for differences between camel adopting and non-
resilience to droughts
adopting households.
The study used the Lewbel-IV approach to determine the
The unstandardized DRI in Table 5 ranges from
effect of camel adoption on household resilience to
negative to positive values with a mean value of − 1.152.
droughts. Table 6 presents the Lewbel-IV and OLS esti-
The findings of the study are in tandem with those of
mates. Due to the potential endogeneity between house-
Alinovi et al. (2008) and Alinovi et al. (2010) who found
hold resilience to droughts and camel adoption, OLS
that pastoral communi- ties in ASALs exhibited low
estimates could be biassed. The discussions hereafter fol-
resilience indices. The magni- tude of indices across
lows Lewbel-IV estimates. The results in Table 6 show
communities is not comparable due to the differences in
that both farm-specific (intensity of camel adoption and
other community characteristics and environments, but
location) and household-specific (main occupation of the
inferences from the signs of the indices can be compared
household head, education, and membership to associa-
(Giuseppe et al. 2016). The negative values were
tions) factors affect household resilience to droughts. The
interpreted by the authors as non-resilient and the positive
proportion of camels in the herd presents a positive and
values as resilient with reference to a given house- hold.
significant effect on household resilience to droughts. This
This study adopted the same characterization. The
value is significant at 10% level of significance.
categorization is broken into households below and above
Household resilience to droughts increased with the
the 0 mark in Table 6. While the surveyed households
increase in the proportion of camels in the herd. Pre-
were generally found to be not resilient to droughts (DRI
cisely, increasing the proportion of camels by one unit
of − 1.152), either of the categories has more than half of
increased household resilience to droughts by 0.204% at
the sur- veyed households below the 0 cut-off, indicating
10% significance level. Livestock diversity is a long-term
low drought resilience among households in general.

Table 6 OLS and Lewbel-IV estimates of the factors influencing household’s resilience to droughts
Variable (standardized resilience) Lewbel-IV Ordinary least squares (OLS)
Coef. Std. Coef. Std. error
error
Constant 0.499* 0.024 0.668* 0.123
Proportion of camels to total herd 0.204*** 0.117 0.083 0.109
Age of household head (years) 0.001 0.002 0.002 0.002
Occupation of household head (1, pastoralist; 0, otherwise) – 0.103** 0.045 – 0.092 0.059
Household size (continuous) 0.009 0.006 0.004 0.006
Dependence ratio – 0.026 0.031 – 0.022 0.030
Marital status (dummy) – 0.049 0.049 – 0.046 0.073
A household member completed primary (dummy) – 0.153* 0.054 – 0.144** 0.065
Household accessed credit (dummy) 0.078 0.048 0.072 0.056
Household received remittances (dummy) 0.017 0.048 0.021 0.062
Membership to association (dummy) 0.164* 0.055 0.163** 0.062
Asset ownership (1 = agricultural and non-agricultural assets; 0 = only agricultural – 0.016 0.044 0.003 0.048
assets)
Crop area cultivated (acres) 0.003 0.007 0.004 0.006
Received animal related extension services (dummy) 0.053 0.051 0.038 0.057
District (dummy 1 = Moroto; 0 = Amudat) – 0.494* 0.045 – 0.460* 0.064

R 2 0.522 0.541
***10%, **5%, and *1% level of significance. Standard errors in parentheses
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 12 of

adaptation to shocks such as droughts, especially where consistent with that of Alinovi et al. (2010) who found
both small and large ruminants are involved. Rushton that predominantly pastoral households in the Turkana
(2009) argues that small ruminants are important in the region were least resilient in the event of food security
immediate years after a shock to allow recovery of the hampering shocks, followed by agro-pastoral households.
stock due to their rapid proliferation capacities. In the This highlights the importance of livelihood diversifica-
long term, small ruminants (less stable) can be sold to tion by households in order to enhance resilience to
buy more stable camels and hence keep the stock stable. shocks through increased livelihood stability.
This strategy provides long-term viability of livestock Education levels of the household members were nega-
production in an uncertain environment. The large ru- tively correlated with household resilience to droughts at
minants such as camels fetch higher incomes in the bad a 1% level of significance. The implication is that house-
times which can be used to solve the households’ imme- holds with at least a member who had completed pri-
diate problems. In addition, the higher income fetched mary education were 0.153% less resilient compared to
could be used in adopting other capital-intensive those households with no primary school completion at
resilience-enhancing technologies. all. This result contradicts the findings of Keil et al.
In Table 6, the results show that the estimates for (2008) and Banda (2015) who found a positive relation-
other determinants of household resilience to droughts ship between household resilience to droughts and edu-
are generally consistent with previous studies (Alinovi cation. However, it should be noted that the studies of
et al. 2008; Banda 2015; Keil et al. 2008) except for Keil et al. and Banda had measured education as the
household levels of education. The study reveals that years of education attained by the household head. This
group membership positively affects household resilience study looked at education at the entire household level,
to droughts while household head being mainly pastoral- whether there is a member of the household who has
ist, education, and location have a statistically negative completed at least primary school. It should also be
effect on household resilience to droughts. Social capital noted that the illiteracy rate in the study area is generally
in the form of belonging to a social group is positively still high (only 17% of the households had a member
related to household resilience to droughts at a 1% level who had completed primary school). This limits drawing
of significance. A household belonging to a social group strong conclusions about the effect of education on re-
was 0.164% more resilient to droughts as compared to a silience to shocks. Education is expected to increase
household not in a social group. household’s access to off-farm income-generating activ-
This is consistent with the findings of Keil et al. (2008) ities and improve decision-making capabilities in the
who found that households that belonged to social households. However, in the scenarios where returns to
groups were more resilient to ENSO-related droughts. education are low due to limited opportunities for the
Socially connected households are able to benefit from educated, education may have no or even negative ef-
an extensive social network during times of crisis, allow- fects where it is expected to produce positive effects.
ing such households to go through a crisis with minimal This is because the investment in education turns out to
damage to their core functioning as a unit. In addition, be a loss to the investing household if the educated do
access to social safety nets can be associated with re- not have opportunities to “repay” what has been invested
duced use of asset liquidation options while coping with in them.
a shock for a given household.
The coefficient on the location of the household is Conclusions and recommendations
negative and significant (at 1% level of significance). This The main objective of the study was to determine the
implies that households in Moroto were less resilient effect of camel adoption on household resilience to
than those in Amudat. Precisely, households in Moroto droughts. The study captured the level of adoption as
were less resilient than their Amudat counterparts. Simi- the ownership of camels by a household, and intensity of
larly, the coefficient on the main occupation of the camel adoption as the proportion of TLUs of camels in
household head was negative and significant at a 5% sig- a household to the total livestock TLUs. The effect of
nificance level. This means that pastoral households camel adoption on household resilience to droughts was
were less resilient than agro-pastoral and other livelihood- estimated using Lewbel’s estimator, an innovative econo-
based houses. While households in Moroto mainly derive metric estimator that accounts for potential endogeneity
their livelihoods from livestock and live- stock products, of a regressor by generating instrumental variables in-
households from Amudat District have a mix of livestock, ternally. A household resilience index was constructed
crop production, beekeeping, and some petty cross-border using principal component analysis methods, and com-
trade with Kenya. This could imply that purely livestock- parisons were made between camel-producing and non-
based livelihoods are less resilient to droughts than camel-producing households in the communities chosen.
mixed livelihoods. This result is PCA shows that income-smoothing indicators (increased
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 13 of

off-farm income and alternative sources of income) had the design of the study and collection, analysis and interpretation
the greatest contribution to resilience. Econometric re- of the data, and writing the manuscript.
sults show that the increasing proportion of camels in a
Availability of data and materials
herd by one tropical livestock unit significantly increased The data set generated and analysed during the current study is
household resilience by 20%. Other important drivers of available from the corresponding author on reasonable
resilience in households included: social capital in the request.
form of group membership which positively affects
Ethics approval and consent to participate
household resilience to droughts; household head being The survey questionnaire solicited for consent as the first questions.
predominantly pastoralist; completion of primary educa- Our research assistants were guided not to interview anyone who
tion by any household member, and households residing was not comfortable providing information.
in Moroto District which negatively affected household
Consent for publication
resilience to droughts. Not applicable
In conclusion, the study highlights the important role
of not just livestock species diversification as a means of Competing interests
enhancing resilience to shocks but also adoption of The authors declare that they have no competing interests.
drought-resilient livestock species as well as the import-
Author details
ance of income-smoothing opportunities in enhancing 1
School of Agricultural Sciences, Makerere University, Kampala,
household resilience to droughts. The results also reveal Uganda. 2School of Forestry, Environmental and Geographical
the importance of community-based livelihood diversifi- Science, Makerere University, Kampala, Uganda. 3Regional
Universities Forum for Capacity Building in Agriculture (RUFORUM),
cation in enhancing household resilience to droughts. Kampala, Uganda. 4National Livestock Resources Research Institute
Both on-farm income-augmenting factors (location and (NaLIRRI), Tororo, Uganda.
proportion of camels) and off-farm factors (group mem-
Received: 29 August 2019 Accepted: 30 January 2020
bership and education) influence household resilience to
droughts in pastoral households. This highlights the key
complementarity relationship between the off-farm and References
on-farm sectors in enhancing household resilience to Ahmad, S., M. Yaqoob, N. Hashmi, S. Ahmad, M.A. Zaman, and M. Tariq.
2010. Economic importance of camel: Unique alternative under
droughts. The study recommends increased emphasis on crisis. Pakistan Veterinary Journal 30 (4): 191–197
income diversification both on the farm and off the farm https://pdfs.semanticscholar.org/fbf1/6
across programmes that aim to build household resili- e88d7c99b1b68b98829ad3adf23123ba44a.pdf.
Alinovi, L., M. d’Errico, E. Mane, and D. Romano. 2010. Livelihoods strategies
ence to droughts. and household resilience to food insecurity: An empirical analysis to Kenya.
Italy: DiPSA, University of Florence
Abbreviations
https://www.researchgate.net/publication/2426
ASALs: Arid and semi-arid lands; DRI: Drought Resilience Index;
93358_Livelihoods_Strategies_and_Household_Resilience_to_Foo
ENSO: El Niño–Southern Oscillation; FAO: Food and Agriculture
d_ Insecurity_An_Empirical_Analysis_to_Kenya. Accessed 27 Dec
Organization of the United Nations; IPCC: Intercontinental Panel 2019.
on Climate Change;
Alinovi, L., E. Mane, and D. Romano. 2008. Measuring household resilience to
IUCN: International Union for Conservation of Nature; IV: food insecurity. Application to Palestinian households (Working paper). Food
Instrumental variable; KMO: Kaiser–Meyer–Olkin; NEMA: National and Agriculture Organisation of the United Nations (FAO)
Environmental Management Authority; PCA: Principal component https://www.
analysis; RCI: Resilience Capacity Index; TLUs: Tropical livestock researchgate.net/publication/228609523_Towards_the_Measureme
units; UBoS: Uganda Bureau of Statistics; UNDP: United Nations nt_of_
Development Program; USAID: United States Agency for Household_Resilience_to_Food_Insecurity_An_Application_to_Palesti
International Development; USD: United States Dollar; nian_ Households. Accessed 16 Jan 2019.
UShs: Uganda shillings; WFP: The UN World Food Program
Aujla, K.M., M. Rafiq, and A. Hussain. 2013. The marketing system of live-
camels and camel products in the desert ecologies of Pakistan.
Acknowledgements
Pakistan Journal of Agricultural Research 26 (2): 130–142
Our deepest appreciation and gratitude is extended to the https://www.soas.ac.uk/ camelconference2013/file89774.pdf.
Regional Universities Forum for Capacity Building in Agriculture Baltas, G., and P. Doyle. 2001. Random utility models in marketing
(RUFORUM) for funding this research. We thank the people of
research: A survey. Journal of Business Research 51 (2001): 115–
Karamoja for willingly availing and recognizing the great work done
125.
by our research assistants in collecting the information.
Banda, T.F. 2015. Determinants of household resilience to dry spells and drought in
Malawi: A case of Chipoka. (Masters of Science in Agriculture and Applied
Authors’ contributions
Economics). University of Malawi
RA recruited, trained, and supervised the research assistant,
https://ageconsearch.umn.edu/record/24346 9/. Accessed 28 June
analysed and structured the manuscript. JHA, RI, and GMD provided
2019.
technical guidance on the econometrics methods and analysis. AE
Brookshire, E.N.J., and T. Weaver. 2015. Long-term decline in
and JN provided administrative support, technical guidance on
grassland productivity driven by increasing dryness. Nature
research methodologies, and general structuring of the
Communications 6: 1–7.
manuscript. DKA provided guidance in the field during the data
Browne, S., & Glaeser, L. 2010. Karamoja region food security assessment:
collection. All authors read, corrected, and approved the final
Uganda Famine Early Warning Systems Network (FEWS NET), Kampala,
manuscript.
Uganda.
https://fews.net/sites/default/files/documents/reports/Karamoja
Funding
%20Food%2 0Security%20Assessment%20January%202010.pdf.
The Regional Universities Forum for Capacity Building in
Accessed 28 Nov 2019.
Agriculture (RUFORUM) funded the study. RUFORUM as a funding
Caffrey, P., T. Finan, S. Trzaska, D. Miller, R. Laker-Ojok, and S. Huston.
body had no role in 2013. Uganda climate change vulnerability assessment report. Washington
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 14 of
DC: USAID
https://www.climatelinks.org/sites/default/files/asset/document/AR
CC- Uganda%2520VA-Report.pdf. Accessed 26 Jan 2020.
Ciani, F., and D. Romano. 2013. Testing for Household Resilience to Food Insecurity:
Evidence from Nicaragua. Italy: University of Florence
https://www.siecon.org/sites/siecon.org/
files/oldfiles/uploads/2013/09/Ciani-Romano.pdf. Accessed on 20 Mar
2020.
Egeru, A., O. Wasonga, J. Kyagulanyi, G.M. Majaliwa, L. MacOpiyo, and J.
Mburu. 2014. Spatio-temporal dynamics of forage and land cover
changes in Karamoja sub- region, Uganda. Pastoralism 4: 6
https://doi.org/10.1186/2041-7136-4-6.
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 15 of

T.D. Kaitho. 2005. Community perceptions of important camel

Elhadi, Y.A., M.D. Nyariki, and V.O. Wasonga. 2015. Role of camel milk in diseases in Lapur division of Turkana District, Kenya. Tropical Animal
pastoral livelihoods in Kenya: contribution to household diet and Health and Production 37 (2005): 187–204.
income. Pastoralism, Research, Policy and Practice 5 (1) Mubiru, D.N. 2010. Climate change and adaptation options in Karamoja. A report
https://doi.org/10.1186/s13570-015-0028-7. submitted to Food and Agriculture Organization and European Union, 62.
FAO, (Food and Agriculture Organisation of the United Nations). 2018. Kampala:
Resilience analysis in Karamoja, Uganda. Rome: FAO resilience analysis https://pdfs.semanticscholar.org/eae4/e9dab59ab65ebf737bccd15
report, 10. Food and Agriculture Organization of the United dbb1aba64bf33.pdf. Accessed 25 Jan 2020.
Nations http://www.fao.org/3/ i8365en/I8365EN.pdf. Accessed Muricho, D.N., D.J. Otieno, W. Oluoch-Kosura, and M. Jirström. 2019.
28 Jan 2020. Building pastoralists’ resilience to shocks for sustainable disaster
Farm-Africa. 2002. Micro-enterprise development. Best practices from risk mitigation:
FARM-Africa Nairobi.
http://www.fao.org/tempref/docrep/nonfao/LEAD/X6174e/x6174e00.
pdf. Accessed 20 Dec 2020.
Faye, B., O. Abdelhadi, A.I. Ahmed, and S.A. Bakheit. 2011. Camel in Sudan:
Future prospects. Livestock Research for Rural Development 23 (10): 11.
Field, C.R., and J. Karuiki. 2005. Where there is no development agency: A
manual for pastoralists and their promoters: (with special reference to the arid
regions of the Greater Horn of Africa). Aylesford: NR International
https://www.worldcat. org/title/where-there-is-no-development-
agency-a-manual-for-pastoralists- and-their-promoters-with-special-
reference-to-the-arid-regions-of-the-greater-
horn-of-africa/oclc/64230785. Accessed 20 Jan 2019.
Giuseppe, S.D., R. Pietrelli, F. Grazio, M. d'Erico, and L. Russo. 2016. Resilience
index measurement and analysis – II. Rome: Food and Agriculture
Organisation of the United Nations http://www.fao.org/3/a-i5665e.pdf .
Accessed 20 Nov 2019.
Guyu, F., and W. Muluneh. 2018. Food insecurity in the green
famine belt of Ethiopia: Extent and severity in Belo-jiganfoy
District, Benishangul-gumuz region. African Journal of Food
Science 12 (3): 54–62.
Hanemann, W.M. 1984. Discrete/continuous models of consumer
demand.
Econometrica 52 (3): 541–551.
IPCC (Intercontinental Panel on Climate Change). 2014. Summary for
policymakers. In Climate change 2014: Impacts, adaptation, and
vulnerability. Part A: Global and sectoral aspects
https://www.ipcc.ch/site/assets/uploads/2
018/02/ar5_wgII_spm_en.pdf. Accessed 26 Jan 2020.
IUCN (The International Union for Conservation of Nature). 2010.
Building climate change resilience for African livestock in sub-
Saharan Africa- World Initiative for Sustainable Pastoralism (WISP): A
program of IUCN. https://www.iucn.org/
sites/dev/files/content/documents/resilence2.pdf. Accessed 26 Jan
2020.
Kagunyu, A., and J. Wanjohi. 2014. Camel rearing replacing cattle
production among the Borana community in Isiolo County of
Northern Kenya, as climate variability bites. Pastoralism, Research,
Policy and Practice 4 (13): 1–5 https:// doi.org/10.1186/s13570-014-
0013-6.
Kagunyu, A., and J. Wanjohi. 2015. The emergency of Euphorbia tiruculli
as drought feeds for camels in northern Kenya. Pastoralism Research,
Policy and Practice 5 (17): 4–16 https://doi.org/10.1186/s13570-015-
0038-5.
Keil, A., M. Zeller, A. Wida, B. Sanim, and R. Birner. 2008. What determines
farmers’ resilience towards ENSO-related drought? An empirical
assessment in Central Sulawesi, Indonesia. Climatic Change 86: 291–
307 https://doi.org/10.1007/ s10584-007-9326-4.
Khan, B., A. Iqbal, and M. Riaz. 2003. Production and management of camels.
Faisalabad: T. M. Printers.
Kima, S.A., A.A. Okhimamhe, A. Kiema, N. Zampaligre, and I. Sule. 2015.
Adapting to the impacts of climate change in the sub-humid zone of
Burkina Faso, West Africa: Perceptions of agropastoralists. Pastoralism,
Research, Policy and Practice 5 (16) https://doi.org/10.1186/s13570-
015-0034-9.
Lewbel, A. 2012. Using heteroscedasticity to identify and estimate
mismeasured and endogenous regressor models. Journal of Business
& Economic Statistics 30 (1): 67–80.
Lickley, M., and S. Solomon. 2018. Drivers, timing and some impacts
of global aridity change. Environmental Research Letters 13 (2018):
104010 https://doi. org/10.1088/1748-9326/aae013.
Mishra, V., and R. Smyth. 2015. Estimating returns to schooling in
urban China using conventional and heteroskedasticity-based
instruments. Economic Modelling 47: 166–173.
Mochabo, K.O.M., P.M. Kitala, P.B. Gathura, W.O. Ogara, A. Catley, E.M.
Eregae, and
 Asiimwe et al. Pastoralism: Research, Policy and (2020) Page 16 of
ght_2010. pdf?sequence=3. Accessed 3 June 2019.
Lessons from West Pokot County, Kenya. International Journal of Disaster
Risk Reduction 34: 429–435. Publisher’s Note
Musinga, M., D. Kimenye, and P. Kivolonzi. 2008. The camel milk industry Springer Nature remains neutral with regard to jurisdictional
in Kenya Results of a study commissioned by SNV to explore the potential of claims in published maps and institutional affiliations.
Camel Milk from Isiolo District to access sustainable formal markets. Kenya:
Netherlands Development Organisation (SNV)
https://www.afraca.org/download/general_ rf_publications/Camel-
Milk-Value-Chain-The-Case-of-Isiolo-District-A-Study- Report-by-
Muli-Musinga-et-al.pdf. Accessed on 10 Mar 2020.
Nardo, M., M. Saisana, A. Saltelli, S. Tarantola, A. Hoffmann, and E.
Giovannini.
2008. Handbook on constructing composite indicators: Methodological and
user guide. https://doi.org/10.1787/533411815016 ISBN: 978-92-64-
04345-9, OECD.
Opiyo, F., O. Wasonga, and M. Nyangito. 2014. Measuring household
vulnerability to climate-induced stresses in pastoral rangelands of
Kenya: Implications for resilience programming. Pastoralism,
Research, Policy and Practice 4 (10)
http://www.pastoralismjournal.com/content/4/1/10.
Opiyo, F., O. Wasonga, M. Nyangito, J. Schilling, and R. Munang. 2015.
Drought adaptation and coping strategies among the Turkana
pastoralists of Northern Kenya. International Journal of Disaster Risk
Science 6 (3): 295–309 https://doi. org/10.1007/s13753-015-0063-
4. Accessed 27 Jan 2020.
Osbahr, H., C. Twyman, W.N. Adger, and D.S.G. Thomas. 2010.
Evaluating successful livelihood adaptation to climate variability
and change in southern Africa. Ecology and Society 15 (2): 27
http://www.ecologyandsociety.org/ vol15/iss2/art27/.
Oxfam. 2008. Survival of the fittest pastoralism and climate change
in East Africa. Oxfam briefing paper.
https://oxfamilibrary.openrepository.com/bitstream/
handle/10546/114607/bp116-survival-fittest-east-africa-
180808-en. pdf?sequence=1&isAllowed=y. Accessed 27 Jan
2020.
Rushton, J. 2009. The economics of animal health and production. United
Kingdom: MPG Books Ltd, Bodmin
https://books.google.co.ug/books?hl=
en&lr=&id=iCyABkjkRJ0C&oi=fnd&pg=PR5&dq=Rushton,+J.
+(2009).+The+ Economics+of+Animal+Health+and+Production.
+United+Kingdom:+MPG+ Books+Ltd,
+Bodmin.&ots=kUatGKZRct&sig=pBqBDd4l-oNUGyKuOk5d2
iWZfF0&redir_esc=y#v=onepage&q&f=false. Accessed 14 May
2019.
UBoS (Uganda Bureau of Statistics). 2010. The national livestock census report 2008.
Kampala:
https://www.ubos.org/wp-content/uploads/publications/05_2019THE_
NATIONAL_LIVESTOCK_CENSUS_REPORT_2008.pdf. Accessed 26 Jan 2020.
United Nations Development Program (UNDP). 2009. Climate change in
the African drylands: Options and opportunities for adaptation and
mitigation. Nairobi:
https://www.droughtmanagement.info/literature/UNCCD_climate_
change_african_drylands_2009.pdf. Accessed 3 Jan 2020.
USAID. 2005. Conflict baseline study report conducted in the Karamajong cluster
of Kenya and Uganda [online]. Washington, D.C:
https://karamojaresilience.org/ publications/item/conflict-early-
warning-and-mitigation-of-resource-based- conflicts-in-the-
greater-horn-of-africa-conflict-baseline-study-report-conducted- in-
the-karamajong-cluster-of-kenya-and-uganda. Accessed 15 Jan 2020.
Ventura, M. 2018. Testing the validity of instruments in an exactly
identified equation. International Journal of Computational Economics
and Econometrics 8 (2) https://doi.org/10.1504/IJCEE.2018.091038.
Accessed 19 Dec 2019.
Vetter, S. 2009. Drought, change and resilience in South Africa’s
arid and semi- arid rangelands. South African Journal of Science
105: 29–33.
World Food Program of the United Nations (WFP). 2009. Country
Strategy for WFP in Uganda (2009–2014).
https://docs.wfp.org/api/documents/WFP-
0000025764/download/. Accessed 22 Oct 2019.
Yosef, T., K. Kefelegn, Y.K. Mohammed, U. Mengistu, A. Solomon, D.
Tadelle, and J. Han. 2014. Morphological diversities and eco-
geographical structuring of Ethiopian camel (Camelus dromedarius)
populations. Journal of Food Agriculture and Environment 26 (4): 371–389
https://doi.org/10.9755/ejfa.v26i4.17021.
Zwaagstra, L., Sharif, Z., Wambile, A., Leeuw, J. d., Johnson, N., Njuki, J.,
Herrero, M. 2010. An assessment of the response to the 2008-
2009 drought in Kenya: A report to the European Union
Delegation to the Republic of Kenya. https://
cgspace.cgiar.org/bitstream/handle/10568/2057/assessment_drou