Adding value to field-based agronomic research through climate risk assessment: a case study of maize production in Kitale, Kenya
MetadataShow full item record
Dixit PN, Cooper PJM, Dimes J, Rao KPC. 2011. Adding value to field-based agronomic research through climate risk assessment: a case study of maize production in Kitale, Kenya. Experimental Agriculture 47 (2): 317-338.
Permanent link to this item: http://hdl.handle.net/10568/34955
In sub-Saharan Africa (SSA), rainfed agriculture is the dominant source of food production. Over the past 50 years much agronomic crop research has been undertaken, and the results of such work are used in formulating recommendations for farmers. However, since rainfall is highly variable across seasons the outcomes of such research will depend upon the rainfall characteristics of the seasons during which the work was undertaken. A major constraint that is faced by such research is the length of time for which studies could be continued, typically ranging between three and five years. This begs the question as to what extent the research was able to ‘sample’ the natural longer-term season-to-season rainfall variability. Without knowledge of the full implications of weather variability on the performance of innovations being recommended, farmers cannot be properly advised about the possible weather-induced risks that they may face over time. To overcome this constraint, crop growth simulation models such as the Agricultural Production Systems Simulator (APSIM) can be used as an integral part of field-based agronomic studies. When driven by long-term daily weather data (30+ years), such models can provide weather-induced risk estimates for a wide range of crop, soil and water management innovations for the major rainfed crops of SSA. Where access to long-term weather data is not possible, weather generators such as MarkSim can be used. This study demonstrates the value of such tools in climate risk analyses and assesses the value of the outputs in the context of a high potential maize production area in Kenya. MarkSim generated weather data is shown to provide a satisfactory approximation of recorded weather data at hand, and the output of 50 years of APSIM simulations demonstrate maize yield responses to plant population, weed control and nitrogen (N) fertilizer use that correspond well with results reported in the literature. Weather-induced risk is shown to have important effects on the rates of return ($ per $ invested) to N-fertilizer use which, across seasons and rates of N-application, ranged from 1.1 to 6.2. Similarly, rates of return to weed control and to planting at contrasting populations were also affected by seasonal variations in weather, but were always so high as to not constitute a risk for small-scale farmers. An analysis investigating the relative importance of temperature, radiation and water availability in contributing to weather-induced risk at different maize growth stages corresponded well with crop physiological studies reported in the literature.