Selection for improved yield in inter-specific mixtures or intercrops.

A simple recurrent selection scheme using randomly constructed intercrop or mixture treatments for the mutual improvement of two or more species is described. Several possible selection criteria are available for each species, including its own yield, those of each of its associates, and combinations of these. Statistical expressions are developed to describe the expected gain in economic yield in each species given selection for one or more species on the basis of any common criterion, and it is shown that the gains from all responding species and applied selection criteria are additive. Negative correlations between the direct and associate effects of genotypes favour the selection of whole mixtures. Selection indices of several species yields can be applied either on an individual species basis or to whole mixtures, but only in the latter case can the statistics required for the calculation of the optimum index be estimated from the data provided by the trial. Some general properties and possible long term effects of different selection methods are discussed.