A grand challenge for humanity is predicting how climate change will alter the functioning of the biosphere and the services it provides. Current predictive models suffer from two main limitations. First, most models assume that plant physiology is driven by ambient macroclimate, but this is almost never accurate. Instead, plant traits operate via energy budgets to decouple macroclimate from the microclimates at which physiological processes occur. For example, plant leaves and forest canopies buffer variation in ambient temperature, which promotes maintenance of plant temperatures near metabolic optima and provides thermal refugia in variable climates. Second, most models are based on tissue, organ, or organismal physiology, so their predictions do not easily map to higher levels of organization (such as communities and ecosystems) that experience spatial variation in microclimate and physiological traits. For example, while climate has a well-established influence on rates of leaf-level carbon assimilation, it has a much weaker influence on rates of ecosystem-level carbon assimilation.