Climate-driven shifts in species’ distributions will lead to the assembly of new combinations of species, and these will engage in novel sets of interactions with consequences that are difficult to predict. Yet even within a constant species pool, climate change can alter the strength and direction of interactions among species, producing indirect effects that propagate non-linearly through the ecosystem. To address this critical source of uncertainty, researchers are working to incorporate interspecific interactions such as competition, consumption, and disease into predictive models of climate change impacts. But what level of ecological detail is sufficient? Can relevant interactions really be identified and their consequences bounded a priori? There is no identical ecological context from one site to the next, so acknowledging that interspecific interactions matter, we are challenged by the possibility that a single species with one basic physiology could experience some standard change in climate differently across its range due to slight (or dramatic) differences in the makeup of its surrounding community.
With hundreds of plant and animal species present in a typical ecological community, and each enmeshed in an intricate network of interrelations, building context dependency into predictive models presents a major challenge. Results from a long-term global change experiment in northern California suggest that the problem is very real. Plots of initially well-mixed grassland, when subjected to an experimentally extended winter rainy season, have grown increasingly dissimilar not just from other treatments but from each other as well. Within a single decade, a simple manipulation of one environmental variable, imposed identically over discrete plots of well-mixed grassland, has given rise to qualitatively distinct communities across a single meadow. Individual species, whether threatened native bunchgrasses or noxious invasive weeds, have followed directionally different trajectories based on fine-scale variation in biological and physical characteristics of their surroundings. Assuming such dynamics are common to other ecosystems beyond the study grassland, and the literature suggests they are, implications for species- and site-level predictions of climate change impacts are considerable.