Both observations and theory suggest that forest ecosystems are prone to abrupt state shifts, particularly in semi-arid regions, or in regions with strong positive feedbacks linked to fire, pest outbreaks, or megafaunal herbivory.
The possibility of abrupt forest mortality in the mesic deciduous forests of eastern North America are less widely appreciated. However, several mesic tree species in eastern North America experienced repeated abrupt collapses in population abundances in the past, yet the causes of these collapses remain uncertain. Many papers point to enhanced hydrological variability, but the evidence remains unclear. Whether the mesic forests of the eastern US are at risk of abrupt mortality events this century is also unclear.
Key evidence comes from long (>10,000 year) records of forest composition based on fossil pollen assemblages extracted from lake and mire sediment cores. These data are available from the Neotoma Paleoecology Database, and we are developing new community models and statistical techniques to identify:
- Possible environmental and biological processes that might cause abrupt change
- Species, ecosystems, regions, and time periods characterized by high variance and high likelihood of abrupt change
- Times and regions characterized by high hydroclimatic variability
- The sensitivity of selected species to abrupt hydroclimatic thresholds and mortality thresholds.