Classification of low-resource livestock producers in the North West Province

Abstract

A survey of livestock producers (n= 117) was conducted in the Taung district of the North West Province in order to identify target groups for livestock improvement among farming communities.The data recorded included livestock numbers and herd structure (young and mature), off-take, purchases of livestock, consumption of own stock, milk and purchase of red meat for home consumption. Data on feeding management and adoption of supplementary feeding practices were also recorded. Data were analysed using multi-variate statistics. The first step was to apply a variable reduction technique (ACECLUS) on livestock numbers and production indices in order to generate canonical variables, three of which were used in an hierarchical cluster analysis ( Wards method) to decide on the appropriate number of clusters. The pseudo-F, pseudo-t 2 and the cubic clustering criterion were all unanimous for a 4-cluster solution which accounted for 70% of the variation. However the clusters so developed were further finely adjusted using a non-hierarchic al procedure, FASCLUS (SAS, 1987). The validity of these clusters (recommendation domain) was established using canonical discriminant analysis on variables not used in clustering. Using Chi-square statistics, it was then finally established whether these groups of respondents differ in their management practices, perceived potential of supplementary feeding and livestock production constraints. Results and Discussion Cluster 1 contained 9.4 % of respondents, recorded the highest lamb holdings, sheep off-takes and also highest sheep production index (PI; Table 1). This group also recorded intermediate levels of kids, goat off-take and goat PI. This profile is typical of small ruminant producers. Cluster 2 contained 6.8 % of the respondents, is characterized by intermediate levels of lambs, sheep off-take and sheep PI. This profile is typical of producers engaged in moderate sheep farming. Cluster 3 comprised 73.5 % of the respondents, who have low levels of livestock, off-takes and consequently low production indices. This typifies a subsistence group of producers. Cluster 4 comprised 10.3 % of respondents who recorded the highest number of kids, goat off-take and goat PI, which is typical of goat producers. The number of calves, chicken off-take, cattle production index did not differ among the clusters. The following hypotheses were tested to validate the typology developed in table 1: (a) Cluster 1 producers with highest number of lambs must have the highest levels of sheep value and are unlikely to purchase beef or mutton because part of sheep off-take is for home consumption (b) Cluster 4 producers with highest kids holdings should have highest levels of goat value, and goat milk consumption and are unlikely to purchase beef, mutton or goat meat. According to Hair et al. (1992) variables with loadings of absolute values greater than 0.3 are significant discriminants. The first two discriminant functions accounted for 96% of the variation, thus suggesting that a reasonable classification has been achieved. Function 1 loaded significantly for sheep value (0.95), donkey value (0.25), beef purchase (-0.29) and mouton purchase (-0.30) and group means for this function validates and