Background/Question/Methods   The diversity and composition of forest communities will be altered by anthropogenic climate change. We examined the influence of interannual variability in spring temperature and drought on tree seedling demography using longterm seedling experiments at the Coweeta Long Term Ecological Research site, Macon County, NC, and the Duke Forest, Orange County, NC. These experiments have contributed to the doctoral research of a number of graduate students, including four of the authors (Bell, Hersh, Ibanez, and Mohan). We were interested in determining (1) whether various species of tree seedlings respond differentially to climate variability, (2) whether these responses are consistent with expectations based on distributional patterns (e.g. growth of northern species declines during years with higher spring temperatures) or species niche (e.g., growth of bottomland species declines during drought years), and (3) whether spatial variations in environment, such as might be caused by disturbance, interact with temporal variations in climate. We will develop a hierarchical Bayesian model for analyzing demographic data for more than 20,000 seedlings of 19 species, measured for 4-11 years. Preliminary analyses included a multivariate regression of diameter and height growth on environmental covariates and a logistic regression of seedling mortality on both environmental covariates and past growth.

Results/Conclusions   While most species grew faster in plots with high light and soil moisture, responses to climate varied greatly. Tree seedlings exhibited positive and negative diameter growth responses to drought and spring temperatures. Most species exhibited greater height growth during wet years with warm spring temperatures. There was no clear pattern regarding growth response to temperature or drought for northern vs. southern or upland vs. bottomland species, respectively. While species showed positive and negative growth effects of interactions between light and soil moisture, the interactions between light and the two climate variables were almost all positive, indicating that the climatic influences on growth may be amplified or lessened, depending on the direction of the main effects for each species. The probability of seedling mortality generally declined with increasing diameter growth for all species and mortality declined with increasing spring temperatures for most species. Even though tree seedling responses are conditional upon a variety of factors, these results indicate that both local and interannual variation in the environment will contribute to forest community response to climate change.