Phylogenetic signal in herbicide resistance evolvability, and the use of phylogenies to predict herbicide resistance risk

Abstract

The evolution of herbicide resistance in weeds is a problem affecting both food production and ecosystems. Numerous factors affect selection towards herbicide resistance making prediction of resistance evolution complicated. The potential for a species to evolve resistance is linked to reproductive and metabolic traits, which may give resistance evolvability a phylogenetic signal and allow phylogenies to be used to predict resistance risk. Here we show that proxies for "resistance evolvability" show correlations with phenotypic traits that evolve at the deep-time level, including an aquatic lifestyle and pollination mode. This results in these proxies having a significant phylogenetic signal, demonstrating a deep-time evolutionary signal at the clade level is present. Machine learning models using phylogenetic distances as predictor variables are able to predict resistance risk with a high accuracy. Fitting macroevolutionary models based on Ornstein Uhlenbeck processes to the resistance evolvability proxies identifies four resistance regimes, including a separation between monocots and eudicots, with the latter evolving resistance less rapidly than the former. These analyses demonstrate the value of phylogenetic comparative methods in predicting resistance risk in weed species that become more problematic, and such methods may provide an invaluable addition to the numerous efforts to predict resistance evolution.