Spatial and density-dependent multilevel selection on weed-infested maize

Artificial group selection has long been proposed as a useful method for crop breeding, yet the possibility that group selection occurs naturally in agroecosystems has not been explored. Due to natural and/or artificial selection, the fitness of an individual can depend on both the individual's traits, the traits of neighboring individuals, and on group's emergent traits (as density). This process is defined as multilevel selection type I. Using contextual analysis, I detected significant multilevel selection type I at the individual and group levels on weed-infested maize (Zea maysL.) genotypes with different spatial patterns and densities. In general, uniformity promoted multilevel selection type I, but this response varied within maize varieties and years. The results herein presented show that crop productivity and weed suppression increased at high density-uniformity conditions, but this also depends on variety. Multilevel selection type II is defined as the differential reproduction of entire groups, a process that in agronomic settings is implemented by humans. A reduced phenotypic variation in the angle of insertion of the oldest living leaf at harvest was identified as a group attribute to be selected for artificial multilevel selection type II. This trait experienced stabilizing selection at the individual level, which significantly increased crop productivity.