TURNING BACK THE CLOCK: RECONSTRUCTIONS OF HISTORICAL FOREST COMPOSITION, STEM DENSITY, AND BIOMASS USING PUBLIC LAND SURVEY, FOSSIL POLLEN DATA, AND STEPPS

North American forests have been transformed by human action, challenging both paleoecologists and global change ecologists who use contemporary ecosystems as a baseline for past or future inferences. As part of the Paleoecological Observatory Network (PalEON) project, we present new gridded (8x8km) reconstructions of historic forest composition, stem density, and biomass for 28 tree taxa in northern US forests for the pre-settlement era (ca. 1700 to 1890 CE), based on Public Land Survey (PLS) data. These PLS-based reconstructions include multiple corrections for potential surveyor biases, including spatially varying correction factors to accommodate sampling design, azimuthal censoring, and biases in tree selection. We employ these pre-settlement vegetation reconstructions, in combination with networks of fossil pollen records drawn from the Neotoma Paleoecology Database (www.neotomadb.org), to calibrate the STEPPS pollen-vegetation model and reconstruct forest composition in the northern US over the last two millennia. STEPPS, as a hierarchical Bayesian model, includes estimates of uncertainty in parameter estimation and state variable reconstruction. We apply the pre-settlement datasets to map the distributions of lost forests (pre-settlement forests with no current analog) and novel forests (modern forests with no past analogs). We demonstrate shifts in tree-climate relationships over the last two centuries due to historic land use and climate change. The next stage of work is to use these paleovegetation reconstructions to initialize, validate, and improve the simulations of terrestrial ecosystem models; this work is underway.