HOLOCENE OAK TREE-RING CHRONOLOGY DEVELOPMENT AND ANALYSIS IN THE AGRICULTURAL LANDSCAPE OF THE MIDWESTERN UNITED STATES

Recent research on the dendrochronology and abundance of ancient oak (Quercus spp.) collected from Midwestern streams and riparian forests indicates that the construction of millennium-length tree-ring chronologies is possible in the agricultural heartland of the United States. Over 100 oak logs recovered from stream and sediments have been 14C and tree-ring dated and range in age from modern to 13,818 years B.P. The temporal distribution of logs confirms the presence of oak in each thousand-year class from present to 14,000 years ago suggests the possibility of constructing a continuous Holocene plus length tree-ring chronology. These tree-ring chronologies will provide climate information for periods during the Holocene that have few climate proxies in the Midwest. The oak chronology region is surrounded by hundreds of miles of the most economically important agricultural region in the United States. Ancient oak wood, that grew thousands of years ago, is often found washing out of stream banks 7 m below soybean and corn fields.

As we cumulate cross-sections from trees that grew contemporaneously, we construct floating 14C and tree-ring dated ring-width chronologies. Cross-dated groups of ring-width series are being analyzed for periodicity and other climate related information that do not require absolute tree-ring dates. Here, we examine these proxy climate series to identify the presence and temporal variability of climate forcing, such as the El Nino/Southern Oscillation. How this cycle behaves during periods of low anthropogenic atmospheric input is a growing question. We cross-dated a temporal cluster of five 14C dated oak logs, which had a mean 14C date circa 2480 years B.P. Ten relatively dated tree-ring series (2 radii per sample, mean r2 among all radii=0.46) were used to construct a 268-year long mean chronology. We double-detrended individual ring-width series by first using a negative-exponential curve and then fitting a 32-year spline. The chronology mean sensitivity was 0.18 and sub-sample signal reached the 0.85 level with seven series. Spectral analysis of ring-width series indicates a peak in variance at 5.9 years. Fisher-Kappa and Bartlett Komogorov-Smirnov tests indicate this peak is significantly different from the white noise spectrum at the 0.01 and 0.001 probability level.