Interactions at the individual scale control forest response to climate change

Background/Question/Methods The recent global increase in forest mortality episodes could not have been predicted from current vegetation models that are calibrated to regional climate data.  Physiological studies show that mortality results from interactions between climate and competition at the individual scale.  Models of forest response to climate do not include interactions because they are hard to estimate and require long-term observations on individual trees obtained at frequent (annual) intervals.  A new demographic approach provides estimates of climate-competition interactions in two critical ways, 1) among individuals, as a joint distribution of responses to combinations of inputs, such as resources and climate, and 2) within individuals, due to allocation requirements that control outputs, such as demographic rates.  Dynamic inverse prediction (DIP) was implemented to identify variables most critical for individual responses.

Results/Conclusions Application to 20 years of data from climate and competition gradients shows that interactions control forest demographic responses, and their omission from models leads to inaccurate predictions.  Growth, maturation, fecundity, and survival  inferred from 40,000 individual trees across climate gradients from northern Hardwoods to SE Piedmont spanned several extreme droughts.   Competition varied among individuals due to canopy manipulation.   Species most vulnerable to increasing aridity are not those that show the largest growth response to precipitation, but rather depend on interactions with the local resource environment. Aggregating individual DIP results identifies the species at greatest risk from drought, having sensitivities as large as the risks from light competition.  These include Tsuga, Cercis, Fagus, Pinus taeda, P. rigida, and P. echinata. Individual interactions further identify where and how landscape variation in moisture mediates responses to climate.    This first assessment of regional species vulnerability that is based on the scale at which climate operates, individual trees competing for carbon and water, supports predictions of potential savannification in the southeastern US.