Hyperaccumulators and herbivores-a Bayesian meta-analysis of feeding choice trials

The leading hypothesis for the evolution of the metal hyperaccumulation trait in plants is as a defense against herbivores. A central piece of evidence expected for this hypothesis is that plants benefit from herbivores being deterred from eating high metal tissues. While many studies have investigated whether or not herbivores are deterred by high metal feeds, there has been no quantitative synthesis of these studies. We performed a Bayesian meta-analysis of 31 feeding choice trials from ten published studies, where invertebrates were offered diets of plant tissue from hyperaccumulating species with high and low metal concentrations. Results of individual trials ranged from distinct preference to distinct aversion for high metal diets. The overall mean effect was for herbivore aversion to high metal diets, whether we used fixed or random effects. However, random effect models were better supported than fixed effect models, indicating there was much real variation between trials. This variation could be attributed partly to each of herbivores, plants, studies, and metals, with herbivores being the greatest source of variation. On average, high metal diets deterred insects but not gastropods, which is supported by other research of metal tolerance and sequestration by gastropods. This suggests that the evolution of hyperaccumulation may have differing selective pressures depending upon the suite of herbivores the plants are naturally exposed to. Future studies should give greater consideration to the selection of herbivores and plants tested.