Multispecies Mixtures: An Individual-Centered Quantitative Genetic Framework for Complex Plant Neighborhoods

Abstract

Mixing crop species or varieties in the same field can raise yield and stability, but performance varies widely among combinations that breeders cannot yet predict. What makes a good neighbor is partly heritable, and quantitative genetics models this as an indirect genetic effect describing a genotype's contribution to its neighbors' phenotypes. Here, we extend existing models with a complementary approach based on continuous neighborhoods, in which diverse genotypes of two species are mixed. In a continuous neighborhood design, several genotypes of each species are grown in a single trial, with every plant georeferenced. Each plant's phenotype is decomposed into direct genetic effects, indirect genetic effects from neighbors of both species, and environmental effects. Breeding values are then assembled for each genotype by combining its three genetic effects. We derived analytical expressions for the variances and covariance of these effects, validated the model, and evaluated its statistical properties through simulations spanning a wide range of parameter settings. For the same number of plants, the continuous neighborhood design estimated indirect effects far more accurately than a pairwise design and, in addition, allowed indirect environmental variance to be estimated. Applied to a wheat-alfalfa experiment comprising 3,840 plants, from 181 wheat pure lines and 106 alfalfa families, the model attributed 8.6% and 8.2% of the variance in wheat grain number and alfalfa biomass to direct genetic effects respectively, and 4.0% and 4.6% to indirect genetic effects from alfalfa. The framework therefore offers an individual-entered basis for analyzing multispecies neighborhoods and their breeding potential.