Terrestrial ecosystem models are fundamental to forecasting ecological responses to 21st-century global change drivers, yet their predictions are alarmingly divergent and their ability to simulate decadal- to centennial-scale processes remains largely untested. Moreover, many future ecological simulations begin with an assumed pre-settlement baseline of “potential vegetation” that is stable and in equilibrium with climate — an assumption that is likely invalid. Paleoecological and paleoclimatic records over the last 2000 years suggest that forests from New England to the Upper Midwest were characterized by broad scale compositional shifts, including regional responses to abrupt large-scale droughts and the immigration of new species. Because of the slow biotic and abiotic responses in forests and internal feedbacks, such processes can affect ecosystem dynamics for centuries. The rich paleo-datasets documenting these past dynamics offer an unexploited opportunity to better test and parameterize ecosystem models.
PalEON Project Goals
PalEON (the PaleoEcological Observatory Network) is an interdisciplinary team of paleoecologists, ecological statisticians, and ecosystem modelers. Our goal is to reconstruct forest composition, fire regime, and climate in forests across the northeastern US and Alaska over the past 2000 years and then use this to drive and validate terrestrial ecosystem models. We will develop a coherent spatiotemporal inference framework to quantify trends and extreme events in paleoecological and paleoclimatic time series. Variables such as forest composition, fire regime, and moisture balance will be inferred from corresponding paleoecological proxies, with rigorous estimates of uncertainty.
These datasets will be applied to improve terrestrial ecosystem models in two contexts. First, we will develop specific data products, such as high- resolution settlement-era forest composition maps from witness tree and General Land Office data, that can be used to drive ecosystem models. PalEON will develop formal data assimilation tools that will allow the models we use to forecast on centennial scales to be informed by decadal- to centennial-scale data. Second, we will develop data products for the purpose of model validation (e.g. fire-frequency reconstructions from sedimentary charcoal data). These long-term validation datasets will help us assess the ability of these models to capture past dynamics correctly, and will help us understand why their future projections are so divergent.
- Jason McLachlan (University of Notre Dame)
- Chris Paciorek (University of California – Berkeley)
- Michael Dietze (Boston University)
- Steve Jackson (USGS and University of Arizona)
- Jack Williams (University of Wisconsin)
- Amy Hessl (West Virginia University)
- Neil Pederson (LDEO, Columbia University)
- Jenn Marlon (Yale University)
- Mevin Hooten (USGS and Colorado State University)
- Phil Higuera (University of Idaho)
- Dave Moore (University of Arizona)