Soil type, management history and current resource allocation: Three dimensions regulating variability in crop productivity on African smallholder farms

Soil fertility varies markedly within and between African smallholder farms, both as a consequence of inherent factors and differential management. Fields closest to homesteads (homefields) typically receive most nutrients and are more fertile than outlying fields (outfields), with implications for crop production and nutrient use efficiencies. Maize yields following application of 100 kg N ha(-1) and different rates and sources of P were assessed on homefields and outfields of smallholder farms in Zimbabwe. Soil organic carbon, available P and exchangeable bases were greater on the homefields than outfields. In each of three experimental seasons, maize yields in homefield control plots were greater than in the outfields of farms on a granitic sandy and a red-clay soil. Application of mineral N significantly increased maize yields on homefields in the first season (2.1-3.0 t ha(-1) on the clay soil and 1.0-1.5 t ha(-1) on the sandy soil) but the effects of N alone were not significant on the outfields due to other yield-limiting factors. Greatest yields of about 6 t ha(-1) were achieved on the clayey homefield with 100 kg N ha(-1) and 30 kg P ha(-1) applied as single super phosphate (SSP). Manure application gave greater yields (3-4 t ha(-1)) than SSP (2-3 t ha(-1)) in the sandy homefield and in the clayey outfield. Maize did not respond significantly to N, dolomitic lime, manure and P on the sandy outfield in the first and second seasons. In the third season, manure application (similar to 17 t manure ha(-1) year(-1)) on the sandy outfield did result in a significant response in grain yields. Apparent P recovery in the first season was 55-65% when P was applied at 10 kg ha(-1) on the clayey homefield (SSP), clayey outfield (SSP and manure) and sandy homefield (manure) with apparent P recovery less than 40% when P was applied at 30 kg ha(-1). On the sandy outfield, P recovery was initially poor (<20%), but increased in the successive seasons with manure application. In a second experiment, less than 60 kg N ha(-1) was required to attain at least 90% of the maximum yields of 2-3 t ha(-1) on the sandy homefield and clayey outfield. N use efficiency varied from > 50 kg grain kg(-1) N on the infields, to less than 5 kg grain kg(-1) N on the sandy outfields. Apparent N recovery efficiency by maize was greatest at small N application rates with P applied. We conclude that blanket fertilizer recommendations are of limited relevance for heterogeneous smallholder farms. Targeted application of mineral fertilizers and manure according to soil type and past management of fields is imperative for improving crop yields and nutrient use efficiencies. (c) 2007 Elsevier B.V. All rights reserved