Which Recurrent Selection Scheme To Improve Mixtures of Crop Species? Theoretical Expectations

In a context of increasing environmental challenges, there is an emerging demand for plant cultivars that are adapted to cultivation in species mixture. It is thus pressing to look for the optimization of selection schemes to grow species mixtures, and especially recurrent selection schemes which are at the core of the improvement of many plant species. We considered the case of two populations from different species to be improved by recurrent selection for their performances in mixture. We set up an analytical model of performances in mixture. We expressed the expected responses of the performances in mixture to one cycle of selection in the case of a Reciprocal Mixture Ability selection scheme and of two parallel selection schemes aiming to improve General Mixture Abilities or performances in pure stands. We numerically compared these selection schemes when half-sib or topcross progeny families of selection candidates are tested in mixture. Selection in pure stands appeared efficient within a limited range of genetic correlations between pure stand performance and mixture model effects. The Reciprocal Mixture Ability selection scheme was expected to be less efficient than parallel selections for General Mixture Ability in some situations. The last option enables to control the ratio of expected responses of species contributions to the mixture performance without bias when using selection indices. When more than two species are be improved for their performances in mixture, the advantage of parallel selections for General Mixture Ability is even more marked, providing that compensation trends between species are not too prevalent.