Learning from the soil's memory : Tailoring of fertilizer application based on past manure applications increases fertilizer use efficiency and crop productivity on Kenyan smallholder farms

The large uncertainty in yield response to fertilizer application within smallholder cropping systems in sub-Saharan Africa (SSA) limits efforts aimed at intensifying crop production based on increased fertilizer application. We assessed the key field-scale cause of variability in maize (Zea mays) grain yield response to fertilizer nitrogen (N), phosphorus (P) and potassium (K) in the Sidindi area of Western Kenya based on past manure application, distance from the homestead, and clay and silt contents. We used data from nutrient omission trials conducted on 23 farms over seven consecutive cropping seasons covering the period 2013–2016, without changing treatments or plot location. Treatments included a control and PK, NK, NP and NPK. Accounting for past manure application increased the explained variability in maize yield, and yield response to N, P, and K application. Mean treatment maize grain yield in the control, PK, NK, NP and NPK treatments were 1.0, 2.2, 1.5, 2.9 and 4.5 t ha −1 at 88% dry matter respectively in fields without past manure application, and 2.4, 2.7, 4.4, 4.9 and 5.4 t ha -1 in fields which had received animal manure in at least two out of three seasons prior to the start of the trials. Mean maize yield response to N, P and K application was 2.3, 3.0 and 1.6 t ha −1 respectively in fields without past manure application, and 2.8, 1.1 and 0.6 t ha −1 in fields with past manure application. In the seventh cropping season, past animal manure application contributed a fertilizer equivalent of 28.3, 29.8 and 31.5 kg ha −1 of N, P and K, respectively. At both the onset and at the start of the last season, fields with past animal manure application had on average higher contents of SOC, available P and exchangeable K, yet differences were not always significant within treatments. Accounting for past animal manure application reduces crop fertilizer requirements for P and K as well as decreasing uncertainty in yield response to fertilizer. We conclude that the strong influence of past animal manure application on yield response to fertilizer application merits the inclusion of past manure application as a co-variate in analysis of yield response data from smallholder cropping systems of SSA.