Classification and characterisation of livestock production systems in northern Tanzania

Abstract

Livestock keepers in sub-Saharan Africa face a range of pressures, including climate change, land loss, restrictive policies, and population increase. Widespread adaptation in response can lead to the emergence of new, non-traditional typologies of livestock production. We sought to characterise livestock production systems in two administrative regions in northern Tanzania, an area undergoing rapid social, economic, and environmental change. Questionnaire and spatial data were collected from 404 livestock-keeping households in 21 villages in Arusha and Manyara Regions in 2016. Multiple factor analysis and hierarchical cluster analysis were used to classify households into livestock production systems based on household-level characteristics. Adversity-based indicators of vulnerability, including reports of hunger, illness, and livestock, land and crop losses were compared between production systems. Three distinct clusters emerged through this process. The ethnic, environmental and livestock management characteristics of households in each cluster broadly mapped onto traditional definitions of 'pastoral', 'agro-pastoral' and 'smallholder' livestock production in the study area, suggesting that this quantitative classification system is complementary to more qualitative classification methods. Our approach allowed us to demonstrate a diversity in typologies of livestock production at small spatial scales, with almost half of study villages comprising more than one production system. We also found indicators of change within livestock production systems, most notably the adoption of crop agriculture in the majority of pastoral households. System-level heterogeneities in vulnerability were evident, with agro-pastoral households most likely to report hunger and pastoral households most likely to report illness in people and livestock, and livestock losses. We demonstrate that livestock production systems can provide context for assessing household vulnerability in northern Tanzania. Policy initiatives to improve household and community well-being should recognise the continuing diversity of traditional livestock production systems in northern Tanzania, including the diversity that can exist at small spatial scales.

Fig 1. Map of study area in northern Tanzania showing location of study villages in relation to main land classifications in Arusha and Manyara Regions (Map created using QGIS version 2.14.3.

Shape files from GADM; landcover raster data from Landsat (https://landsat.gsfc.nasa.gov/data/)).

Fig 2. Percent contribution of each group to the first (Dimension 1) and second (Dimension 2) factors derived from MFA performed on characteristics of livestock-keeping households in northern Tanzania.

Red (blue) dotted line represents the expected score if all domains contributed equally to the inertia on the first (second) factor (i.e. 100/8 = 12.5%).

Fig 3. Variable scores in relation to the first and second factors derived from MFA performed on characteristics of livestock keeping households in northern Tanzania.

Scores given to categorical (continuous) variables are shown in red (blue). Categorical (1 indicates presence of described characteristic; 0 indicates absence): CAT = Keep cattle; CH = Keep chickens; CR = Household grows crops; CRT = Grow crops for > 10 years; DO = Keep donkeys; ED = Household education to primary school or above; FI = Household consumed fish in past 3 days; IR = Iraqw ethnicity; LA = Latrine in household; ME = Meru ethnicity; MA = Maasai ethnicity; MAI = Grow maize; PI = Keep pigs; SMT = Small ruminant transhumance; VE = Household consumed vegetables in past 3 days; ZGCA = Zero graze cattle; ZGSM = Zero graze small ruminants. Continuous: AR = Village area; CD = Cattle density; CHD = Chicken density; CN = Household cattle number; CR = Local cropland % cover; EV = Enhanced vegetation index; GD = Goat density; GN = Household goat number; GR = Local grassland % cover; PID = Pig density; POD = Human population density; RA = Annual precipitation; SD = Sheep density; SN = Household sheep number.

Fig 4. Position of households on the first and second factors (Dimension 1 and 2) derived from the MFA performed on characteristics of livestock keeping households in northern Tanzania.

Households are shaded based on cluster membership.

Fig 5. Proportion of households in study villages assigned to each livestock production cluster in northern Tanzania in 2016.