DEEP TILLAGE, LIME, MICRONUTRIENTS AND ORGANIC INPUTS REHABILITATE POORLY RESPONSIVE SOILS IN SMALLHOLDER MAIZE (Zea mays) AND SOYBEAN (Glycine max) CROPPING SYSTEMS OF WESTERN KENYA

Abstract

Maize (Zea mays) and soybean [Glycine max (L.) Merr] grain yields in smallholder farms of western Kenya are persistently low, particularly in soils that respond poorly to fertilizer application. A multi-locational trial was conducted in 25 sites in Siaya and Kakamega counties of western Kenya to determine the (1) occurrence of soils that do not respond to application of optimal rates of N, P and K fertilizers, (2) soil properties underlying the poor responsiveness and (3) effects of selected rehabilitative measures on grain yield responses, striga (Striga hermontica) weed infestation, soybean leaf rust severity and maize rhizosphere colonization by the arbuscular mycorrhiza fungi (AMF). Based on the hypothesis that soil texture and gradient variations contribute to poor responses to fertilizers, soils of those sites were categorized into four domains; (i) Domain 1 ≥50% silt + clay and ≥5% slope; (ii) Domain 2 ≤50% silt + clay and ≥ 5% slope; (iii) Domain 3 ≥50% silt + clay and ≤5% slope; and (iv) Domain 4 ≤50% silt + clay and ≤5% slope. A trial comprising six treatments was installed in each site using a one-farmer one-replicate design. Soils responsive to N, P and K application were distinguished from the poor responsive using two boundary line indicators, viz., 1:1 and 1:2 fitted on scatter plots of maize grain yields obtained from control against fertilized plots. Statistical analysis was done using the linear mixed effects model (LMM) in R. Two response categories were obtained based on the soil domains i.e. responsive soils (60 % of the total sites) and poorly responsive (40% of the total sites). Further, using the cropping season to classify soil responsiveness, three categories are observed i.e. responsive soils (54%), poorly responsive soils (30%) and non-responsive soils (16%). The spatial analysis revealed that there was a bigger ratio of poorly responsive soils in Siaya county compared to Kakamega county. Statistical analysis showed that K was significantly different between the poor and responsive soil categories, while the general trends showed sulphur (S) to be lower in poorly responsive soils than in the responsive soil. Various rehabilitation scenarios increased maize (6.30 t ha-1) and soybean (2.04 t ha-1) yields as compared to respective N, P and K yields at 0.33 and 0.23 tha-1, suppressed striga weed, reduced severity of soybean leaf rusts and increased the colonization of AMF by 5-10%. These results indicate that poorly responsive soils occur both at spatial and temporal scales and other factors apart from the supplied nutrients limit maize production in such soils. Use of amendments such as lime, organic matter, micronutrients and deep tillage can salvage the situation. However, further research is indispensable to identify the overriding factors that limit maize production in poorly responsive soils.