In eastern Kenya, although the traditional livestock grazing is still important, increased human population pressure has caused a swing towards production of food crops. The distribution of 600-800 mm annual rainfall over two growing seasons results in a high risk of crop failure in both, but - in spite of this constraint - land over-utilisation has resulted in declining yields and accelerated soil erosion. ACIAR Project 8326 initially featured socioeconomic studies of the prospects for alleviating these problems through improved technology and the evaluation of introduced forage legumes that might increase soil nitrogen more than indigenous grain legumes. Greatly increased awareness of the importance to farm management of risk of inadequate rainfallresulted from that project

The present project will continue work on climatic risk and also have a major emphasis on soil water and fertility. In the climatic risk component it will use a maize-production simulation model (which was calibrated and tested for this region in Project 8326) to estimate the variation in maize yields, using historical weather data as inputs. This will enable both improved quantification of climatic risks in the existing farming system and testing of the consequences of changes in management and crop genotype. Thus it will identify optimum strategies for avoiding risks and improving average yields - for example, profitable fertiliser rates for a given risk of crop failure.

Using the same historical weather data, the team will make early predictions of the probable eventual merit of a given season. The farmer can then take appropriate action to increase yield potential by applying inputs or to minimise losses by withholding them. Critical testing of this promising notion has already begun in Australia, and will conclude in Kenya with on-farm testing of possible 'response' practices. The locally calibrated crop model(s) should also forecast probable yields - valuable information for marketing or drought relief.

In the semi-arid tropics, sesquioxidic soils are especially prone to three major problems when cultivated - low efficiency of recharging soil water storage, high erosion rates and poor crop establishment - and research in the second component concerns management strategies for alleviating these. It will assess the technical and economic feasibility of the various elements of a strategy involving minimum or no tillage and retention of a mulch cover, comparing them with existing practices. Technical aspects will be conducted on research stations, while those depending on realistic farm conditions and comparisons with existing practices will use farms. For both types, continuous recording of rainfall and runoff will permit analysis of water and soil loss on a weather-event basis, with the prospect of improving models for run-off and erosion prediction.

The third component of the project concerns soil fertility management. It will seek to better quantify nutritional constraints (especially nitrogen), to assess the economics of commercial fertiliser relative to inadequate rainfall risk and to explore opportunities for more effective use of manure and legumes. The research will be conducted primarily on farms to ensure realistic conditions. A high priority is to determine how fertility inputs to a failed crop affect the next crop, and a model of mineral nitrogen supply currently being developed in the Australian component will be applicable to this study.

Research in the Australian component addresses the same three themes, developing tools and concepts relevant to Africa to facilitate progress of the work in Kenya.

There are no fixed boundaries between themes. Research will be integrated in both Kenya and Australia. On-farm research, in particular, will involve agronomists, soil scientists, hydrologists and economists.

The project has enormous potential benefits, but will involve necessary compromises between adequacy and achievability.