2002 Denver Annual Meeting (October 27-30, 2002)

Paper No. 2
Presentation Time: 1:30 PM-5:30 PM

FUNCTION-FITTING VEGETATION PROXY RECORD PROFILES OF THE SANGAMON AND WISCONSIN EPISODES FROM MISSOURI AND ILLINOIS


CURRY, Brandon, Illinois State Geological Survey, University of Illinois, 615 E. Peabody Dr, Champaign, IL 61820, DORALE, Jeffrey A., Geology, Univ of Missouri, 101 Geology Building, Columbia, MO 65211 and HENSON, Robert A., Illinois Statistics Office, Univ of Illinois, Urbana-Champaign, 101A Illini Hall, 725 S. Wright St, Champaign, 61820, b-curry@illinois.edu

U-series dated d13C profiles from the Crevice Cave (SE Missouri) stalagmites are proxy records of C3 vs. C4-type vegetation spanning from 109.4 to 24.5 ka. A composite record of the overlapping stalagmites include 47 U-series ages, and 579 d13C values. This record was compared with the pollen diagram from Raymond Basin of Zhu and Baker (1995), a site on the Illinoian till plain 165 km NE of Crevice Cave. A stratigraphically constrained detrended correspondence analysis (DCA) yields eigenvalues of 0.54 and 0.26 for the first two axes which are likely proxies of temperature and effective moisture, respectively. The core from which the pollen record is derived has no finite age control. However, visual inspection of the d13C profiles from Crevice Cave and 2nd axis DCA pollen profile from Raymond Basin reveals striking similarities. Correlation of the most obvious shifts in values necessitates a novel interpretation of climate change. Pollen spectra dating from 76 to 71 ka indicate the vegetation grew under warm, moist conditions, conditions unique to the entire Sangamon Episode (last interglaciation). This correlation shows that the most temperate conditions of the entire last 130 ky occurred exactly during the 5a/4 transition when the Laurentide Ice Sheet was growing rapidly.

To statistically verify our visual correlation, we developed a three-step procedure. Step one uses an objective, iterative procedure in S-plus that searches for the maximum correlation between the two profiles. To improve the likelihood of finding a good correlation coefficient and decrease computation time, three starting values are computed by using maximum and minimum values of the age profile. The calculated sediment accumulation rate is fixed to avoid unreasonable jumps in the iterations. Step two uses the same procedures as step one, except that the three starting points were subjectively selected. Step three uses Monte Carlo simulations of randomly generated data for uncorrelated variables to determine if the correlations in the earlier steps are significant. The results indicate our profiles are significantly correlated with respective correlation coefficients of 0.55, 0.72, and 0.15 for the correlation with objective starting points, correlation with subjective starting points, and the best Monte Carlo result (out of 100 iterations).