To predict how climate variability affects a population, we need to know if different individuals have similar or divergent responses to climate variables such as rainfall and temperature. While it seems intuitive that small and large individuals might respond differently, current approaches do not carefully accommodate size-dependent climate effects. Stochastic matrix models are generally unconstrained; every transition can be affected by climate differently, so estimates cannot borrow strength from nearby size classes. Traditional IPMs are over-constrained because at most the slope and intercept in regressions on size would be environment-dependent. To understand how differently sized individuals respond to climate, and to compare responses across species, we need more data-driven approaches. Here we use functional linear models to investigate whether small and large individuals differ in the magnitude or direction of their responses to climate in long-termed mapped quadrat data from five U.S. states. We also asked if size or species identity is a better predictor of climate responses.
We find that in most species-sites, size influences the response of survival and growth to climate. In survival, good years for small individuals are also good years for large individuals, but small individuals are more sensitive to climate variation while large individuals consistently have high survival. In growth, the opposite is often true: good years for small plants can be bad years for large plants, but the magnitude of climate effects on relative growth rate is often similar. On average, the responses of large and small plants within a species tend to be highly correlated across years (0.87 for survival and 0.54 for growth), while responses of similar size classes in different species tend to be less correlated (0.19 for survival and 0.34 for growth). These results offer insight into the mechanisms by which populations persist and coexist in variable environments, and could be included in models of population or community dynamics.