Modelling trees' effect on maize in the Grevillea Robusta plus maize system in Central Kenya

In the Central Highlands on Kenya, Grevillea robusta (A. Cunn.) trees are commonly planted on maize (Zea mays L.) fields. This type of production system covers about 750 000 ha in the highland area around Mt Kenya. Knowing the influence of trees on maize yield would help to regulate the density and size distribution of the tree cover. The spatial distribution of trees varies from rather uniform to extremely aggregated patterns, calling for a spatial modelling approach. This study employed three different spatial competition index types to model the effect of trees on the maize yield at a given location. The maize yield was expressed as a function of competition index, which was calculated from the diameters and distances of trees. The data were collected from 14 maize fields, and included measurements on 857 trees and 2514 one-m(2) maize plots. The models indicated that high competition by trees decreases maize yields considerably, but at the field level the decrease is small with normal tree stockings (about 200 trees per hectare). Characteristic to the models was low degree of explained variance, which was partly due to high sampling error in maize yield measurement and low variation in competition (places with high competition by trees were seldom planted with maize). One of the maize yield models was combined with an earlier simulation program for G. robusta. The simulator allows one to predict the temporal development of any G. robusta - maize field in Central Kenya.