Congratulations to undergraduate student, Finley Shinnick, who was selected to be an oral presenter for the 2026 Colleges of Science and Engineering Joint Annual Meeting (COSE-JAM). Finley presented her work with Hannah and Jason on “Improving Model Reconstructions of Vegetation Signal.”
Abstract: Information on how ecosystems responded to climate events in the past informs predictions of how present ecosystems may react to future climate changes. Pollen grains preserved in lake sediment at measurable depths capture this historical vegetation response across thousands of years. The Spatio-Temporal Empirical Prediction from Pollen in Sediments (STEPPS) model integrates hundreds of these pollen records using a Bayesian statistical framework to reconstruct vegetation across large regions with time steps of centuries. STEPPS is useful at capturing long-term ecosystem boundary shifts in response to climate events, including in the Upper Midwest, but predicting future vegetation change requires knowing how short-term events, like droughts and wild fires, contribute to long-term trends. In this project, we begin updating the STEPPS model framework to reconstruct local vegetation at decadal intervals using finer resolution pollen data. We first corrected the STEPPS coding framework to be compatible with modern software and capable of handling a flexible number of lake sites. As an initial step toward reconstructing vegetation using our finer-resolved lake sites, we ran STEPPS within single-grid cells. We found that vegetation trends using reduced, finer resolution lakes remain consistent following STEPPS reconstruction. This suggests that modeling finer temporal resolution using fewer, better sampled lakes captures the same vegetation process modeled by the full STEPPS model. This is a promising first step towards incorporating the finer-resolution vegetation process into the larger STEPPS model and improving our understanding of natural disturbance’s contributions to long-term ecosystem change.