Wednesday, August 8, 2007 - 8:00 AM

COS 72-1: Modeling late-glacial no-analog climates with expanded response surfaces

Leila M. Zajac, Jack W. Williams, and Rick Nordheim. University of Wisconsin-Madison

Climate and vegetation reconstructions based on paleo-ecological data for the late-glacial period are challenged by climates and vegetation assemblages with no modern analog. Standard empirical reconstruction techniques based on space-for-time substitutions fail because modern vegetation-climate relationships do not fully encompass late-glacial climates.

In order to model late-glacial climates, we developed a new variant on response surfaces, called expanded response surfaces, which attempt to estimate pollen-climate relationships for climates outside the modern climate domain. We derived the response surfaces from a new dataset of >4500 North American surface pollen samples, and modeled the pollen relative abundances relative to mean winter and mean summer precipitation and temperature for twenty late-glacial taxa common to three late-glacial pollen records from northeastern Illinois. We assume that relative pollen abundance responds follow a unimodal, symmetrical distribution to climate variables. All of the taxa in the dataset show this symmetrical, unimodal pattern; however some taxa have distributions that are truncated at high abundances by the edge of the modern climate envelope. We assume that species with truncated distributions have fundamental niches that extend beyond the current realized climate space.

We statistically identified the mode of each taxa’s response surface by gridding the data and finding the area of highest mean pollen abundance. We then identified taxa with response surfaces truncated by the modern climate envelope. We expanded truncated response surfaces by symmetrically mirroring the non-truncated portion around the mode. We tested the accuracy of the method by using one set of taxa to reconstruct the regional late glacial climate, applying this reconstructed climate to generate pollen abundances for a second set of taxa, and comparing the reconstructed pollen abundances to the actual fossil pollen abundances. Our results show good agreement; indicating that the method can reproduce climate-driven changes in vegetation communities even during periods of no-analog climates.