Community Performance Curves predict ecosystem stability despite interactions

Traditional biodiversity–stability theory has long relied on static trait descriptors, such as the mean and variance of species’ thermal optima, to predict ecosystem stability. However, these trait moments often fail to capture the dynamic and nonlinear ways in which species respond to environmental fluctuations. This study proposes a paradigm shift: moving from trait-based summaries to community performance curves (CPCs) as mechanistic predictors of ecological stability.Performance curves describe how intrinsic growth rates vary across environmental gradients, integrating multiple traits into a single function. By aggregating these curves across species, CPCs quantify the community’s collective response to environmental change. The variability of CPCs, measured via their coefficient of variation, emerges as a powerful predictor of community stability, outperforming traditional trait metrics even under strong interspecific interactions.This framework introduces the concept of performance curves as “meta-traits”: emergent organismal characteristics formed from coordinated traits that jointly shape ecological performance. Unlike trait moments, CPCs retain information about curve breadth, asymmetry, and overlap, key features that determine compensatory dynamics and response diversity. Communities with broad and complementary performance curves exhibit greater stability, while those with narrow, overlapping curves are more vulnerable to environmental variability.Using multispecies Lotka–Volterra simulations and simulated temperature fluctuations, we show that the community performance curve explains between 90% to 70% of variation in community stability, outperforming trait-based metrics and maintaining high predictive power even as interspecific interaction strength increases. An empirical test with ciliate microcosms corroborated these findings: greater variability in the community performance curve was associated with greater variability of total community biomass variability, whereas, trait moments were weaker predictors. Moreover, informed CPCs, those weighted by the actual frequency of environmental states, further improve predictive power, highlighting the importance of environmental autocorrelation and Jensen’s inequality.In sum, this work reframes biodiversity–stability theory around community-level performance, offering a more mechanistic, predictive, and ecologically grounded approach to understanding how ecosystems respond to environmental change.