USING A FOREST STAND SIMULATION-MODEL TO EXAMINE THE ECOLOGICAL AND CLIMATIC SIGNIFICANCE OF THE LATE-QUATERNARY PINE SPRUCE POLLEN ZONE IN EASTERN VIRGINIA, USA

Full-glacial pollen records from southeastern United States are composed primarily of pine and spruce, with lesser amounts of fir, birch, and oak. A simulation model of forest dynamics was used to reconstruct the composition and structure of these forests on the Delmarva Peninsula of Virginia, where pollen data were available to test the model, and climate and soils data were available to drive the model. Reconstructed annual precipitation and summer air temperature were consistent with modern analog estimates from the pollen record. Annual precipitation was also consistent with climates simulated by atmospheric general circulation models, but summers were colder. Correcting these simulated climates for possible errors resulted in summer air temperature consistent with our estimate. However, two alternative parameter sets relating simulated tree growth to air temperature sums precluded robust forest reconstructions. With one parameter set, the species dominating the simulated forests were not consistent with the pollen record. The other parameter set produced forests more consistent with paleoecological data, indicating that the climate was correct. These differences in simulated forest composition reflected inadequacies in the parameterization of air temperature effects in forest models. © 1990.