Livestock Research for Rural Development 24 (10) 2012 Guide for preparation of papers LRRD Newsletter

Citation of this paper

Vulnerability of rural agro-pastoral households to drought in semi-arid Botswana

K Mogotsi, M M Nyangito* and D M Nyariki*

Range and Pasture Research Program, Animal Production and Range Research Division, Department of Agricultural Research, MoA,
P. O. Box 10275, Francistown, Botswana
kbmogotsi@gov.bw
* Department of Land Resource Management and Agricultural Technology, University of Nairobi,
P. O. Box 29053, Nairobi, Kenya

Abstract

A survey was carried out during the 2009/10 season to examine the vulnerability of agro-pastoral communities to drought shocks in Bobonong and Kgalagadi North Sub-districts. The key drivers of vulnerability of households included: gender of the household head, livestock sales, advance preparations before drought, size of arable land under cultivation, the number of drought-tolerant crops planted, as well as, the overall yield of such crops. Although the main determinants of household vulnerability differed between the two study areas, Bobonong nonetheless had a higher percentage of highly vulnerable households. However, the two study areas still had a substantial number of vulnerable households – further underlining the seriousness of drought risk among agro-pastoralists in Botswana.  Thus, timing and form of intervention, including from the government, is critical and a one-size-fits-all approach to alleviating adverse impacts of drought may not always be appropriate.

Keywords: climate variability, Kalahari, livelihoods, livestock, resilience, subsistence agriculture


Introduction

If there is one environmental extreme event with the potential to disrupt livelihoods among rural small-scale agro-pastoral communities in Botswana, it is drought. With a low and highly variable rainfall regime, the country has a long history of droughts (Hitchcock 1979; Dube 1995; Setshwaelo 2001; CSO 2006; Mogotsi et al 2011a). Thus, droughts are expected phenomena in semi-arid Botswana and have ceased to be perceived as surprises. Furthermore, future regional dry spells are generally expected to increase both in frequency and severity (Hulme et al 2001) and this does not bode well for the majority of rural agro-pastoral communities who are heavily dependent on rain-fed agriculture for sustenance. While tremendous progress has been made in understanding the physical aspects of droughts and their impact on ecological systems, Vetter (2009) pointed out that there is still a need to incorporate ecological and social dimensions of rangeland use. Though subsistent communities have developed a myriad of adaptive responses to mitigate the adverse effects of droughts (Mogotsi et al 2011b), these are becoming inadequate due to continued exposure to drought shocks at increased frequencies than in the past resulting in production shortfalls and subsequent food insecurity at the household level. Failure to adequately cope and adapt could lead to gradual erosion of households’ resilience with subsequent dry spells and inevitably render households vulnerable. 

Although vulnerability is a familiar concept, there is still no consensus on its definition. IPCC (Intergovernmental Panel on Climate Change) (2001) defined vulnerability as the degree to which a system is susceptible to, or unable to cope with adverse effects of climate change, including climate variability and extremes. Furthermore, Kabat et al (2003) referred to vulnerability as the characteristic of a person, or group, or component of a natural system in terms of its capacity to resist and/or recover from and/or anticipate and/or cope with, the impacts of an adverse event. It is primarily because of this ambiguity that quantifying vulnerability is even more challenging. Downing et al (2001) argued that the defining criteria for quantifying vulnerability have proven difficult in part because of the fact that vulnerability is often not a directly observable phenomenon. Nonetheless, numerous attempts have been made to develop approaches to assess vulnerability over the years (e.g. Adger 1999; Brooks et al 2005; Vincent 2007; Heltberg et al 2009). This study draws from the notion of entitlements as proposed by Sen (1981) who contended that access to resources is a key determinant of vulnerability. According to Keijsper (1993), livestock keeping is a revered and principal activity among agro-pastoralists in Botswana. This is partly due to livestock’s ability to withstand the harsh semi-arid conditions and thus provide a more ‘secure’ source of livelihood far superior to arable farming. Subsequently, households strive to accumulate as many animals as possible, especially cattle, and can better cope with drought. Adger (1999) stated that because of their importance, income and wealth (e.g., livestock) distribution directly affects the ability of households to cope with the impacts of extreme events. In this study, cattle ownership was used as the proxy for vulnerability. This follows the livelihood approach as proposed by Ellis (2000), which enables generalization concerning the varying implications of socioeconomic and environmental shocks for households pursuing similar strategies (Eakin and Bojorquez-Tapia 2008). 

This study therefore set out to determine the main factors influencing vulnerability of households to drought among rural agro-pastoral communities in Botswana. Understanding the dynamics of vulnerability due to drought could lead to enhanced coping and adaptation not only to current climate variability, but ultimately, to future climate changes in the Southern African region. 


Materials and methods

Study areas

Kgalagadi North and Bobonong regions of Botswana were the targeted areas. Kgalagadi North is part of Kgalagadi (Kalahari) desert ecosystem, the driest district in Botswana. The Bobonong region lies within the Central district in the eastern part of Botswana. A detailed description of the study areas has been reported by Mogotsi et al (2011c). Subsistence agriculture is the main source of livelihood of the rural communities in both study areas. Livestock keeping is dominated by traditional production systems within open access communal grazing areas known as cattle posts (locally referred to as meraka). The main livestock kept are cattle, goats, sheep and donkeys. Subsistence rain-fed crop cultivation is practiced but to a lesser extent, especially in the more arid Kgalagadi district.

Data collection

Detailed standard questionnaires with open-ended, multiple response and dichotomous questions, were administered to a representative sample of the respective communities following Mogotsi et al (2011c). The two sub-districts were purposively selected. In the Bobonong sub-district, 50 randomly-selected subsistence farmers were interviewed from Lepokole and surrounding communal areas of Sekgopswe, Mmamanaka and Mmaditshwene. Kgalagadi North sub-district had 40 households interviewed in Hukuntsi, Lehututu and Tshane villages (15, 12 and 13 households respectively). The survey collected information primarily on socio-economic status (e.g. income sources), land tenure and size thereof, livestock ownership, water sources, livestock fodder production, crops planted and yields, draught power, agricultural pests and diseases, access to information, extension, technology, markets and credit as well as adaptation and coping strategies currently used to buffer drought shocks. In addition, data triangulation was done through focus group discussions and key informant interviews (community heads, farmers’ association representatives and agricultural extension agents) to get a holistic understanding of vulnerability of the communities to droughts. Relevant secondary sources augmented primary data (for example, the government’s Inter-Ministerial Drought Committee publications and Annual Poverty Monitoring reports (MFDP 2008)).

Data analysis

Data were subjected to the Statistical Package for Social Sciences (SPSS). Descriptive statistics were used to characterize respondents’ demographics, socio-economic status and other related variables. The Ordinary Least Squares (OLS) regression was used to model vulnerability of households to drought. All statistical differences were tested at a significance level of 5% (p<0.05).

Index of vulnerability

An index of vulnerability relevant to the households in the study areas was developed. Because of their unparalleled importance in rural Botswana, ownership of cattle per household was used as a proxy for vulnerability. Also, due to households’ general reluctance to divulge the exact livestock numbers they own, categories were used instead [Botswana farmers regard livestock as ‘money in the bank’ and thus would deliberately underestimate their wealth to outsiders. And because they are often blamed for overstocking and degradation (e.g. NPAD 1991), they might also view with skepticism any enquiries about herd sizes].  A participatory approach identified three categories. The first category was for households with less than 25 cattle (highly vulnerable). The second category comprised households with 26 to 50 cattle (vulnerable). The last category was for households with cattle numbers greater than 50 (less vulnerable). Old and/or unproductive animals were excluded. Given k variables and following Gujarati (1995), the following general equation for OLS was used; 

Y =α + β1X1i + β2X2i + …+ βkXki + µi

Y represents the dependent variable (in this study, livestock asset index), α is a constant, X1,…,Xk is a set of explanatory variables, i denotes ith household, µ is the error or disturbance term and β1,…,βk are coefficients representing parameter estimators of variables used. 

Regression analysis and variables used 

Vulnerability depends critically on context, and according to Brooks et al (2005), the factors that make a system vulnerable to a hazard will depend on the nature of the system and the type of hazard in question. It is thus necessary to identify specific variables likely to influence agro-pastoral communities’ vulnerability to drought in the study areas. Variables likely to influence vulnerability of households in both study areas were selected á priori in formulating the model. The variables used in the final regression model included size of arable land, gender of head of household, crop yield, use of tolerant crops, main source of income, access to information, sale of livestock, advance preparation for drought, type of draught power and whether or not households used Government aid programmes. 

The relationship of the preceding hypothesized explanatory variables and vulnerability of households to drought was determined using the OLS regression. Prior to that, a correlation regression was performed to investigate whether two or more explanatory variables could have been highly correlated - the problem of multicollinearity. To eliminate this statistical problem, one of the two highly correlated variables was dropped.  The goodness of fit was based on higher values of the coefficient of variation, R2, adjusted R2, and the number of significant explanatory variables achieved. The F statistic was also used to come up with the ‘best’/appropriate model explaining the extent of household vulnerability to drought and the proposed determinants.


Results and Discussion

The socio-economic and demographic status of the respondents in Bobonong and Kgalagadi North Sub-districts have been previously reported by Mogotsi et al (2011a).  

Socioeconomic characteristics of households
Gender, education level and main sources of income

Households in both study areas had between 5 and 7 members each. Most households were headed by members aged 41 years and older, with 72 and 89% in Bobonong and Kgalagadi North Sub-districts respectively, very few of whom had gone beyond primary education. There were 56% of heads of households with some formal education in Bobonong, while Kgalagadi had 76.3% of such respondents. There was a disparity in gender - where 64% of the households were female-headed in Bobonong while 23.7% were headed by females in Kgalagadi North. In Bobonong, of all household heads with no formal education, 68.2% were female-headed

Households also had several sources of income. In Kgalagadi North, the majority of households had sales from agriculture as their main source of income (57.9%). Other sources included salaries from working household members (23.7%) and for all elderly citizens aged 65 and above, a monthly allowance from the government’s Old Age Pension Scheme (10.5%). In contrast, most households in Bobonong earned their main income through temporary employment (72%). Sales from agriculture and allowances from the Old Age Pension Scheme were the main sources of income for some households (8% each), followed closely by salaries at 6%. (Mogotsi et al 2011a)

The results of the regression models for Bobonong and Kgalagadi North Sub-districts are presented in Table 1. 

Table 1: Factors influencing livestock numbers and thus vulnerability to drought in Bobonong and Kgalagadi North Sub-districts

Variable

Bobonong

Kgalagadi North

Bobonong

Kgalagadi North

Β

t-value

Constant

0.583

2.67

1.15*

5.94*

Gender of head of household

-0.281

-

-2.03*

-

Income

0.092

-

1.75

-

Arable land size

0.060

0.095

3.27*

2.33*

Drought tolerant crops

0.165

-

3.11*

-

Crop yield

0.010

-0.051

2.63*

-1.18

Access to climate info

0.225

-

1.64

-

Access to extension

-0.275

-

-1.77

-

Pre-drought preparation

-

0.102

-

3.14*

Government aid programmes

-

0.059

-

1.371

Livestock sales

-

-1.10

-

-4.30*

*Significant at p<0.05

(Bobonong) F = 8.618, R2 = 0.59, Adj. R2 = 0.52  (Kgalagadi North) F = 6.641, R2 = 0.56, Adj. R2 = 0.48

In Bobonong area, the significant (p<0.05) explanatory variables were gender of the head of household, size of available arable land, number of drought-tolerant crops planted as well as the yield of such crops. While all the significant variables (p<0.05) had a positive impact on household vulnerability to drought, access to extension services and the gender of the head of household had the opposite effect. The results imply that female-headed households in Bobonong’s agro-pastoral communities are likely to be vulnerable to drought. Female-headed households in the area have fewer assets compared to male-headed households and cannot adequately buffer shocks such as drought. Planting a wide variety of drought tolerant crops significantly (p<0.05) reduced household vulnerability to droughts. This is expected as drought tolerant crops would likely perform relatively better than non-adapted crops, enabling families to harvest and be food secure during droughts. On the other hand, in Kgalagadi North, the size of arable land, pre-drought preparation and sale of livestock were significant (p<0.05). Advance preparations in anticipation of upcoming droughts included storage of crop harvests for household consumption and Zea mays stover for animals. Livestock sales had a significant (p<0.05) and negative influence on household vulnerability to drought. Also, households that did not sell livestock were more likely to be vulnerable. Apart from avoiding losing animals to drought, selling livestock also enables households to have readily available cash to acquire other essentials.

The results also showed differences between the two Sub-districts in terms of the main determinants of household vulnerability to drought. This point is worth highlighting since it is a crucial consideration for government policy and subsequent aid programmes extended to such communities. ‘Blanket aid programmes’ which might be appropriate in one Sub-district may not necessarily reduce household vulnerability to drought in another area and might instead reinforce the negative effects of drought. Overall, the vulnerability of households in Bobonong and Kgalagadi North Sub-districts is shown in Figure 1.

Figure 1: Levels of vulnerability to drought of households in the study areas

The majority of households in Bobonong area were highly vulnerable (68%). This could be attributed to most households being female-headed (64%). These households are generally resource-poor (Fako and Molamu 1995), making them unable to mobilize enough resources to buffer against drought. Other contributing factors to this observed high vulnerability among Bobonong households include small area of cultivated land, failure to plant drought-tolerant crops and generally low yields. The latter is especially important because, though the cultivated land might be large enough or the adapted crops are planted, if the yield is low or there is total crop failure, the household is immediately food insecure. This first order effect of low crop yields on households is exacerbated by the damage of crops by elephants (Loxodonta africana) as well as the low and unpredictable rainfall regime in the area. In contrast, Kgalagadi North had fewer (16%) households who were highly vulnerable to drought. Even more households (37%) in Kgalagadi North fell under the less vulnerable category compared to 6% of such households in Bobonong. 

However, the two study areas still had a substantial number of households falling under the vulnerable category, 26% and 47% for Bobonong and Kgalagadi North respectively. This category is important as it shows the seriousness of drought risk among agro-pastoralists. Eakin and Bojorquez-Tapia (2008) reckon that vulnerability at the household level is likely to be more dynamic than at national level. According to Alwang et al (2001), any given household can move between different categories of vulnerability within a relatively short time. This then means that the vulnerable category of households in both Sub-districts can either remain vulnerable - or join the less vulnerable or further regress back to the highly vulnerable category. Thus, timing of intervention is critical. This scenario further demonstrates the lurking threat droughts put on rural agro-pastoral communities in Botswana.


Conclusions


Acknowledgements

The authors are grateful to the communities of Bobonong and Kgalagadi North Sub-districts for their time and cooperation. The Regional Universities Forum for Capacity Building in Agriculture (RUFORUM) provided financial support for this project while the Department of Agricultural Research (Botswana) handled logistics during field work.


References

Adger W N 1999 Social vulnerability to climate change and extremes in Coastal Vietnam. World Development. 27: 249-269

Alwang J, Siegel P B and Jorgensen S L 2001 Vulnerability: A view from different disciplines. Social Protection Discussion Paper Series. World Bank, Washington, DC. 

Brooks N, Adger W N and Kelly P M 2005 The determinants of vulnerability and adaptive capacity at the national level and the implications for adaptation. Global Environmental Change, 15: 151-163

CSO (Central Statistics Office) 2006 Ministry of Finance and Development Planning, Gaborone, Government Printers, Botswana. 

Downing T E, Butterfield R, Cohen S, Huq S, Moss R, Rahman A, Sokona Y and Stephen L 2001 Climate change vulnerability: Linking impacts and adaptation. University of Oxford, Oxford. 

Dube O P 1995 Monitoring vegetation and soil degradation in Botswana. Progress report. Australian Center for International Agricultural research. ACIAR Project No. S010001. 

Eakin H and Bojorquez-Tapia L A 2008 Insights into the composition of household vulnerability from multicriteria decision analysis. Global Environmental Change, 18: 112-127. 

Ellis F 2000. Rural livelihoods and diversity in developing countries. Oxford University Press, Oxford. 

Fako T T and Molamu L 1995 The seven-year drought, household food security and vulnerable groups in Botswana. Pula: Botswana Journal of African Studies, 9: 48-70. 

Gujarati D N 1995 Basic Econometrics. 3rd Edition. McGrraw-Hill Company, New York. 

Heltberg R, Siegel P B and Jorgensen L S 2009 Addressing human vulnerability to climate change: Toward a ‘no-regrets’ approach. Global Environmental Change, 19: 89-99. 

Hitchock R K 1979 The traditional response to drought in Botswana. In; Hinchey M T (ed.), Botswana Drought Symposium, Botswana Society, Gaborone, Botswana. 

Hulme M, Dougherty R, Ngara T, New M and Lister D 2001 African climate change: 1900–2100. Climate Research 17. 

IPCC (Intergovernmental Panel on Climate Change) 2001 Climate change 2001: Impacts, adaptations, vulnerability, Contribution of working Group II to the Third Assessment Report of the IPCC, Geneva: UNEP/WMO. 

Kabat P, Schulze R E, Hellmuth M E and Veraart J A 2003 Coping with impacts of climate variability and climate change in water management: A scoping paper. DWC-Report no. DWCSSO-01 International Secretariat of the Dialogue on Water and Climate, Wageningen, Netherlands. 

Keijsper V 1993 Livestock development in Botswana: Consequences for the rural poor. Africa Spectrum, 28: 323-339. 

MFDP ( Ministry of Finance and Development Planning)  2008 Annual poverty monitoring report 2007/08. Government Printers, Gaborone, Botswana.  

Mogotsi K, Nyangito M M and Nyariki D M 2011a The perfect drought? Constraints limiting Kalahari agro-pastoral communities from coping and adapting. African Journal of Environmental Science and Technology, 5:168-177. 

Mogotsi K, Nyangito M M and Nyariki D M 2011b Drought management strategies among agro-pastoral communities in non-equilibrium Kalahari ecosystems. Environmental Research Journal 5: 156-162. 

Mogotsi K, Moroka A B, Sitang O and Chibua R 2011c Seasonal precipitation forecasts: Agro-ecological knowledge among rural Kalahari communities. African Journal of Agricultural Research 6: 916-922. 

NPAD (National Policy on Agricultural Development) 1991 Ministry of Agriculture, Government of Botswana, Gaborone. 

Sen A 1981 Poverty and famines: An essay on entitlement and deprivation. Oxford University Press. 

Setshwaelo L L 2001 A study of grasslands/livestock vulnerability and adaptation to climate change, Botswana. 

Vetter S 2009 Drought, change and resilience in South Africa’s arid and semi-arid rangelands. South African Journal of Science, 105: 29-33. 


Received 4 August 2012; Accepted 15 September 2012; Published 1 October 2012

Go to top