Application of models in prediction of dairy cattle productivity in Mbeere district, Kenya

Abstract

Two models, Integrated Modeling Platform for mixed Agriculture Crop sysTem (IMP ACT) and Livestock Feeding Strategies Simulation Models (LIFE-SIM), used to assess farm characterization and prediction of milk yield were validated using data collected in smallholder dairy farms in Mbeere District, Kenya. Data were collected in arid and semi-arid environments under two smallholder dairy production systems in Evurore and Siakago Divisions. Data were collected using a structured questionnaire in a cross-sectional study from randomly selected 17 farms in each Division during the months of March and April, 2007. The data captured included: climate, land management, livestock management, labor allocation, farm's sales and expenses, and soil nutrient flow. A detailed longitudinal study followed in a random sample of six farms from each Division for three months between May and July 2007. The detailed study covered farms with lactating dairy cows, where data on breed, parity, lactation length, milk yield, feed types and their intake, calving intervals and biweekly weight changes were collected. The mean household size was 5.3 and 5.1 persons (median 5 and range 2-8) for Evurore and Siakago, respectively. The common dairy cattle breeds were Ayrshire and Friesian. The mean milk yield was 8.1 and 7.0 kg/cow/day for Siakago and Evurore, respectively with a range of 2.5-10.2 kg/cow/day. For the stall managed dairy animals, most farmers used crop residues and cut pastures as the main source of feed. The feeds were sourced from both within and outside the farms, and ranged from planted and natural forages, crop residues to commercial and home-made concentrates. In both Divisions, the basal feed resource was maize stover. However, methods of processing the stover prior to feeding varied, either fed whole in unprocessed form to chopping using either machete (panga) or chaff-cutter. Rhodes grass was the main ley pasture and was used by 17% Of the farmers, while natural pastures mainly consisted of Cynodon spp and Rhychelyntrum species. During the dry period, when various VI feed resources were in short supply, banana stems and cereal by-products (wheat and rice bran and pollard) were used to sustain the animals. Data from 33 mixed crop livestock farms in the cross-sectional study were used to validate IMPACT model. A snapshot baseline analysis of farm cash flow with IMPACT identified three main sources of income; crops, dairy and off-farm. Crop and/or dairy farming constituted the major source of income. Crops contributed 39% and 32% for Evurore and Siakago respectively while livestock contributed 35% and 58% in Evurore and Siakago respectively. Annual income from crops was estimated at KES 1.8± 0.21 and 1.7± 0.29 million for all the farms in Siakago and Evurore respectively whereas annual income from dairy farming was estimated at KES 3.3 ± 0.31 and 1.5± 0.08 million for Siakago and Evurore respectively. Off farm income was contributing 25% of income in Evurore but was insignificant source of income for farms in Siakago Division. The LIFE-SIM model predicted a mean lactation yield of 2351 ± 380kg equivalent to 7 kg/cow/day (305 days) which was below mean potential yield of 2649 ± 222kg. 70% of the cows had a deviation of less than 5% of the real recorded yield compared with the predicted, while 30% had a deviation of between 5- 10%. The model was therefore able to predict milk yield to a reasonable degree.