AbstractQuestionNative species have the potential to provide productive, drought‐resistant communities for seeded rangelands and ecological restoration. Little is known, however, about how to identify multispecies mixtures with optimal levels of productivity and stress resistance from the thousands of possible community configurations. Here we examine if empirical models can be used to predict highly productive community configurations of seven native grasses and legumes in controlled conditions from the very large pool of possible communities, and which basic measure of community structure best predicts function.LocationGreenhouses in Saskatchewan, Canada.MethodsWe used a greenhouse experiment, where established communities varied in species and functional group richness, evenness, species and functional group identity, following a response surface design. We measured community productivity and evaluated the predictive power of a range of empirical models linking diversity and productivity.ResultsProductivity increased with increased functional dispersion, relative growth rate and decreased competitive effect. Selection effects were evident, with the abundance and occurrence of particular species or functional groups and plant traits also linked to increased productivity. Among the strongest predictors of productivity were the presence and abundance of perennial C3 grasses (particularly Pascopyrum smithii), likely because of the high early relative growth rate and strong competitive effect of those species.ConclusionsWe compiled and compared the ability of a range of empirical models to predict high‐productivity community configurations, and tested the accuracy of the best models in a confirmatory experiment. The relationship between predicted and observed productivity was significantly correlated in the confirmatory experiment, and demonstrates that under controlled conditions, basic measures of community structure can predict community function. This approach has potential, but variability within treatments may limit the accuracy of results. The models developed can be used as a screening tool, narrowing the search window for high functioning seed mixtures for use in ecological restoration.
Read full abstract