How do dairy cow diets in the tropics change as systems intensify?

Smallholder dairy production is widespread in Low and Middle Income Countries often acting as an important source of cash income as well as contributing to household nutrition through regular supply of dairy products. Dairy systems vary from backyard production involving one or two cows to semi-commercial operations supplying urban markets. Dairy production is intensifying in response to growing demand for dairy products from a burgeoning urban population. Feed supply is a perennial constraint to the productivity increases that will be necessary to meet this growing demand. Smallholder dairy cows are often fed opportunistically and it is difficult to accurately determine the amount and composition of feed consumed by dairy cows. Improved estimates of “feed baskets” would help in projecting likely feed demand as systems intensify and suggest ways in which farming systems are likely to evolve to meet this demand. Existing household datasets including from the World Bank LSMS survey, the RHoMIS tool, FEAST and from other publically available datasets published by the International Livestock Research Institute provide fragmented information on dairy feed baskets. Further data can be accessed through the Livestock Data for Decisions (LD4D) Community of Practice. These datasets also include a range of other biophysical and socio-economic indicators that are useful to characterize dairy systems. This project will use data mining, machine learning and econometric approaches to develop an understanding of the broad determinants of dairy cow diets based on a series of “learning data sets”. These relationships will then be used to predict dairy cow diets in new locations and the validity of these estimates across intensification gradients will be assessed through household survey. The project is likely to work in locations where ILRI has strong dairy data resources (e.g. India, Ethiopia, Kenya, Tanzania).