Modeling vegetation of the past: Integrating fossil pollen data with vegetation modeling

Background/Question/Methods

Dynamic global vegetation models (DGVMs) can be used to predict terrestrial ecosystem changes under future climate predictions. Data from paleoecological studies allows us to test the “predictive” ability of a DGVM (LPJ-GUESS) under past environmental conditions that are different from the present. The goal of the present study is to develop a framework for i) comparing process-based vegetation model simulation and fossil pollen data, ii) identifying the important driving factors of past vegetation change, and iii) providing a vehicle for testing hypotheses of past ecological events.

We simulated vegetation over the past 15,000 years using inputs of past environmental condition taken from a new transient climate simulation with the CCSM General Circulation Model (Liu, Z., et al., 2009, Transient Simulation of Last Deglaciation with a New Mechanism for Bølling-Allerød Warming. Science, 325, pp. 310-314) at five fossil pollen sites (Steel Lake, Tower Lake, Tannersville Bog, Cupola Pond, Lake Tulane) in eastern North America. We modeled changes in plant functional types (PFTs) for these sites, and compared the simulation with the observed fossil pollen record. We conducted sensitivity studies using records of past hydrological variability.

Results/Conclusions

Simulated pollen sequences at the five pollen sites in eastern North America are generally consistent with the pollen record. Significant mismatches indicate i) overrepresentation of local vegetation in smaller lakes, ii) species-specific or inaccurate parameterization of North American PFTs, and iii) species-specific or inaccurate correction factors for pollen productivity. The sensitivity studies show increasing ecosystem disturbance caused by changes in variability, and provide examples for understanding future predicted changes.

Our work provides a framework for similar model-data comparison, experiments for identifying important climatic factors, and hypothesis testing of past ecological events. Currently we are conducting experiments using this framework to explore hypotheses of the mid-Holocene hemlock decline and late-glacial no-analog vegetation communities.