Background/Question/Methods Modeling the processes of the past is society’s best hope for envisaging the future.  Many of the ecological conditions of the past are only known from climate proxies such as tree rings and sediments.  We show two new tree-ring based models defined by chemical and geographic processes that combine long-term and large-scale climate, fire, and tree growth data.  Low resolution (>50 years) variability in sea surface temperature (SST) are hypothesized to influence the growth of oak trees through changes in precipitation. North American mid-continental tree growth is coupled with sediment proxies of SST in the North Atlantic and Pacific Oceans over 14 k years.  SST records are used to estimate the growth (~56,000 tree rings) of 367 bur oaks (Quercus macrocarpa) preserved in the alluvial plain sediments of Midwestern agricultural ecosystems.  Additionally, we use tree, ocean, human population, and fire data in a stochastic-process model (Physical Chemistry Fire Frequency Model, PC2FM) to develop a low resolution 14 k yr simulation of fire frequency.  This model is based on the Arrhenius equation and calibrated with fire history data from 156 sites.  Large temporal and spatial variability in fire intervals is useful for understanding long-term climate processes that effect fire regimes over thousands of years.

Results/Conclusions Significant positive correlations over the last 14 k years were found between oak growth and SST in the northern North Atlantic (r = 0.55), the western Pacific (r = 0.52), and other SST locations.  Regression modeling (R² =0.52) indicates that Northern Hemisphere SST is the most significant predictor of growth at this mid-continental location (40 N, 93 W) during the post glacial period (14,000 - 6500 cal yr BP).  Interactions among the northern North Atlantic, and Pacific SSTs and differences in the northern and equatorial North Pacific were also significant.  Later (4000-1500 cal yr BP), these correlations reemerge, but are less robust (R² = 0.18). During the Thermal Optimum (6500 - 4500 cal yr BP) tree growth was not significantly correlated SSTs.  Low resolution long-term SST data and population data were applied to the PC2FM model.  Model simulation results show that mean fire intervals in the Prairie-Forest transition change from about 20 years at the peak of the Younger Dryas (12,000 cal yr BP) to 6 years about 300 years ago.  The effect of temperature on mean fire intervals was most important between 14,000 and 9,000 cal yr BP.