The effects of management history and landscape position on inter-field variation in soil fertility and millet yields in southwestern Niger

Abstract

The effects of human management on the inter-field variability of soil fertility and crop yields were studied in southwestern Niger. Chemical fertility (organic C, total N, available P, K, exchangeable bases, pH) of the surface soils of 181 fields dispersed within two agropastoral territories covering 412km2 was measured. Analyses of crop yields demonstrated that inter-field variation in chemical fertility more significantly affected yields than the topographic position of fields or the physical properties of their soils (soil depth and texture). Regression analysis of soil fertility parameters showed that chemical fertility is generally more influenced by the distance from nearest village than by topographic position or soil physical properties. Soil fertility declines with distance from nearest village in a negative curvilinear fashion consistent with previous work describing infield and outfield patterns of soil fertility. However, when recent management variables for monitored fields were included in regression models (average surface fraction of field that was manured over previous eight years and number of years cropped over previous twelve years), average manuring fraction was found to better predict field-level soil fertility variation than distance to nearest village for all macronutrients except phosphorus. This illustrates the importance of contemporary patterns of livestock presence and manuring across agropastoral landscapes—patterns that are not solely oriented spatially with respect to proximity to permanent villages but to pastoral encampments, livestock corridors and water points. The availability of phosphorus, a conservative nutrient within the soil system, more reflects historic nutrient inputs which are higher as one approaches villages. Therefore, phosphorus availability remains negatively correlated with distance to nearest village even when contemporary management variables are included in regression models. Soil pH shows a more complex pattern being positively affected by manure inputs but negatively associated with the duration of cultivation. Implications for commonly-used assessments of nutrient budgets and landscape-scale fertility gradients (infield—outfield model) are discussed.