Food security in a changing world : disentangling the diversity of rural livelihood strategies across Uganda

Abstract

Climate change increasingly challenges smallholder farming and our ability to achieve Sustainable Development Goal 2 (Zero Hunger) in sub-Saharan Africa. Agricultural interventions are needed that aim at improving the food insecurity of the most vulnerable rural households. Interventions must fit the local context of a diverse population of rural households, and a key challenge is to identify which kinds of interventions work in which regions and for which households. Micro-level information can account for this diversity, but is an underused source of information for planning of interventions at national and sub-national levels. In this thesis, I explored how micro-level information from cross-country household survey data can be used for effective planning of interventions. A further research aim was to understand within-country patterns of livelihood strategies in relation to food security and vulnerability to climate change of rural households in Uganda. Cross-country household data from the World Bank Living Standard Measurement Survey – Integrated Surveys on Agriculture (LSMS-ISA) were used to 1) aggregate household level information to higher levels (e.g. districts, regions, livelihood zones), 2) spatially interpolate household level information and 3) identify hotspot areas of household vulnerability. I used data that I collected from two sites in Uganda for an in-depth analysis on current coping strategies of households for climate and price variability. Household food security was approximated using a food availability indicator that quantified the contribution of livelihood activities to household food availability. Livelihood strategies of rural households across Uganda varied with household food availability. They changed from subsistence-oriented on-farm activities to market-oriented on-farm and off-farm activities as household food availability increased. Aggregation revealed spatial differences in food availability and livelihood activities. However, a geostatistical interpolation approach showed that local variability in food availability and livelihood activities was often larger than variability across larger areas. These findings stress that the large diversity in livelihood activities within any given area must be recognised in decision making at higher levels. Climate change scenarios were linked to the household livelihood activities to identify hotspot areas of vulnerable households in a country-wide assessment of climate change impacts on crop suitability. Groups of crop-related adaptation options were determined per hotspot area. Adaptation options related to temperature were suitable in the north, while drought-related adaptation options were more suitable in the southwest of Uganda. An in-depth analysis indicated that few ex-ante coping strategies were applied under current climate and price variability. Such coping strategies mostly required little financial investment such as switching crops, which was common for households with more land available. Households tended to react to shocks rather than taking preventive action. Better-off households compensated for crop losses by selling livestock or relying more on off-farm income, while the poor and food insecure lacked the resources to do so. These findings suggest that lack of resources can prevent households from adapting to climate change, even when adaptation options are useful from an agronomic perspective. Therefore, contextualised research is needed to understand local barriers to adoption, so that adaptation options can be tailored to local contexts and underpinned by enabling policies and institutional arrangements. Current top-down approaches to planning interventions ignore local diversity of livelihood strategies and food security. However, my results demonstrate that food security and vulnerability tend to be locally driven with large variability at small scale. Therefore, I propose a three-step approach for using micro-level information for multi-level planning. Step 1 disentangles livelihood diversity using cross-country household surveys. Step 2 locates important production activities (Pathway 2a) or vulnerable households and suitable adaptation options (Pathway 2b). Step 3 uses site-specific household surveys to assess which interventions work for which groups of households in the local context. This approach adds to existing approaches by generating spatially-explicit and quantitative information on livelihood activities for food availability and on household vulnerability, while accounting for the diversity of households within and across areas. It enables the exploration and tailoring of intervention options under different future scenarios. In this way, my work contributes to identifying pathways to achieve zero hunger by 2030 in sub-Saharan Africa.