Characterizing farming systems around Kakamega Forest,Western Kenya,for targeting soil fertility–enhancing technologies

Kakamega district in Western Kenya represents the smallholder farming systems typical for much of the densely populated humid highlands in East Africa. A specific feature, however, is the presence of a protected forest reserve (Kakamega Forest National Park), covering some 20% of the district area. Year‐round crop production with little use of external inputs is resulting in declining soil fertility and crop yields. Technologies to counteract fertility constraints are rarely implemented, as they do not consider system diversity or farm‐specific characteristics. We surmised that farm type–specific targeting of technology options to address soil fertility–related production constraints would reduce the anthropogenic pressure on the resources of the adjacent Kakamega rainforest reserve. Based on Kenyan national census data, we selected 168 farms in physical proximity of the Kakamega forest and characterized them regarding production system and soil attributes. Cluster and principal component analyses identified five distinct farm categories. Three representative farms from each cluster group were subsequently selected to establish labor‐use patterns, draw resource‐flow maps, and determine NPK balances. Small subsistence‐oriented farms were most common (> 50%), with maize yields of 0.9 t ha^–1 (Cluster 1). Most farmers relied on the forest to provide fire wood, animal feed, and medicinal plants. Mixed farms, combining subsistence maize with industrial crops, were differentiated by soil type, with tea being grown on Ferralsol (Cluster 3), and sugar cane being grown on Acrisol (cluster 4). The dependence on forest resources was limited to animal grazing and the collection of feed stuff (Cluster 3), or the extraction of medicinal plants (Cluster 4). Only few farms showed a high degree of market orientation of the food‐crop production. These comprised either small farms with high investments in fertilizer and maize yields close to 2 t ha^–1 (Cluster 2), or larger farms (1.6–3.9 ha) with low fertilizer but high hired‐labor use (Cluster 5). Their reliance on forest resources was generally low. Resource flows showed mainly patterns of nutrient export in subsistence farms, and more complex flow patterns, involving several farm compartments, in the diversified farms. Partial nutrient balances were strongly negative for N and K, irrespective of soil or farm type. Soil‐fertility characteristics reflected the nutrient balances with generally low C and N in all farms on Acrisol, and low P in farms not applying mineral fertilizers or farmyard manure. The proposed typology is expected to improve the targeting of technologies addressing soil fertility–related production constraints, and to reduce the pressure on forest resources. This is of particular importance in the case of small‐scale subsistence and mixed farms close to the forest margin.