Agroecosystem analysis in a semi-arid, small-scale farming system of Zimbabwe: A systems approach

Effective agroecosystem analysis is made difficult by the lack of suitable philosophical and methodological approaches and the difficulty of applying performance criteria, such as sustainability, to agroecosystem behavior. A research methodology for the analysis of agroecosystem behavior in data scarce environments was described. The approach was used to examine an agroecosystem in the mid-Zambezi Valley.; A questionnaire survey and RRA techniques were used to develop a general understanding of the needs, resources and constraints households faced in satisfying their needs. Thereafter an extended analysis was conducted with the objective of developing a computer simulation model of the agroecosystem. Five male villagers were elected and three women volunteered to act as village representatives (VRs) in the analysis. Data collection methods were developed and then used with the VRs to identify, and weight, by relative importance, major household needs and the production enterprises used to satisfy those needs as well as all inputs to and outputs from each production enterprise.; Household needs and yield probability density functions derived from the VRs were used to develop a computer simulation model of the agroecosystem. Up to 300 households were randomly placed on a raster based GIS image of the Masoka agroecosystem. Each cell of the landscape represented one acre. Households were allocated cells (fields) around their house sites and on soils adjacent to the Angwa River. The model simulated the productive activities of up to 300 households, updating variables in each cell and for each household once a season. Households could be allocated to one of four production strategies which determined the crops as well as the proportion of household land that was planted to each crop each year. Households could employ labor from other households to make up for deficits failing which their yields were reduced. A partial budget format was used to estimate returns to land, labor and initial investment. Soil erosion was modeled using the SLEMSA erosion model.; The effects of changing important model inputs and parameters on model response variables were examined. Factors that had notable effects on the proportion of deficit households, yields and returns to land, labor and initial investment were rainfall, the land area available and the cash needs required by households.