EXPLOITING TEMPORAL VARIABILITY TO UNDERSTAND TREE RECRUITMENT RESPONSE TO CLIMATE CHANGE

Publisher Website
Google Scholar
Abstract
Predicting vegetation shifts under climate change is a challenging endeavor, given the complex interactions between biotic and abiotic variables that influence demographic rates. To determine how current trends and variation in climate change affect seedling establishment, we analyzed demographic responses to spatiotemporal variation to temperature and soil moisture in the southern Appalachian Mountains. We monitored seedling establishment for 10 years in five plots located along an elevational gradient of five dominant tree species:Acer rubrum, Betulaspp.,Liriodendron tulipifera, Nyssa sylvatica,andQuercus rubra. A hierarchical Bayes model allowed us to incorporate different sources of information, observation errors, and the inherent variability of the establishment process. From our analysis, spring temperatures and heterogeneity in soil moisture emerge as key drivers, and they act through nonlinear population demographic processes. We found that all species benefited from warmer springs, in particular the species found on dry slopes,N. sylvatica, and those dominant at higher elevations,Betulaspp. andQ. rubra. This last species also benefited from dry environments. Conversely,L. tulipifera, which is abundant on mesic sites, experienced highest establishment rates at high moisture. The mechanisms behind these results may differ among species. Higher temperatures are apparently more important for some, while dry conditions and reduced pathogenic attacks on their seeds and new seedlings have a large impact for others. Our results suggest that only communities found at higher elevations are in danger of regional extinction when their habitats disappear given the current climatic trends. We conclude that the recruitment dynamics of the communities where these species are dominant could be affected by minor changes in climate in ways that cannot be predicted using only climate envelopes, which use different variables and miss the nonlinearities.