Means and extremes: building variability into community‐level climate change experiments

Experimental studies assessing climatic effects on ecological communities have typically applied static warming treatments. Although these studies have been informative, they have usually failed to incorporate either current or predicted future, patterns of variability. Future climates are likely to include extreme events which have greater impacts on ecological systems than changes in means alone. Here, we review the studies which have used experiments to assess impacts of temperature on marine, freshwater and terrestrial communities, and classify them into a set of 'generations' based on how they incorporate variability. The majority of studies have failed to incorporate extreme events. In terrestrial ecosystems in particular, experimental treatments have reduced temperature variability, when most climate models predict increased variability. Marine studies have tended to not concentrate on changes in variability, likely in part because the thermal mass of oceans will moderate variation. In freshwaters, climate change experiments have a much shorter history than in the other ecosystems, and have tended to take a relatively simple approach. We propose a new 'generation' of climate change experiments using down-scaled climate models which incorporate predicted changes in climatic variability, and describe a process for generating data which can be applied as experimental climate change treatments.