Detecting and Predicting Climatic Variation from Old-Growth Baldcypress

Tree-ring data can extend back in time for thousands of years allowing researchers to reconstruct certain environmental factors that have left an imprint or signal in the tree-ring record. Typically, these factors include reconstructions of annual precipitation or temperature for months or seasons to which a particular tree species is sensitive. Over the last several decades, scientists have used tree-ring records in novel ways to investigate the timing and extent of such natural phenomena as volcanoes (Baillie and Munro, 1988), earthquakes (Sheppard and Jacoby, 1987), El Nina/southern oscillation (Stahle and Cleaveland, 1993), fire (Swetnam 1993), carbon dioxide (CO,) (Graybill and Idso, 1993), and synchronous landscape-level disturbances (Reams and Van Deusen, 1993) by recognizing the possibility that various signals may be recorded in the growth record of trees, depending on microsite characteristics, geographic location, and disturbance history (Fritts 1976). Climate reconstruction from tree-ring data involves establishing a relationship between the tree-ring variable(s) and some measure of climate. The uniformitarian assumption (Fritts, 1976) is then called upon to allow for using this established relationship to reconstruct the climate variable during the period before climate measurements were available. Weather stations in the United States were rarely in place prior to approximately 1860, but many living trees provide data for centuries before that time. Thus, the motivation to use tree-ring derived variables to reconstruct climate is clear. The usual procedure involves fitting a regression equation that uses climate as