ABSTRACT
Standardized ring-width measurements from trees on spatial arrays of sites can provide continuous yearly paleoclimatic records, and this information can be transferred to estimates of climatic variables and mapped to reveal past spatial variations in climate (Fritts 1976, Fritts in press, Stockton et al. 1985). If the trees have the exact same response to climate, the ring-width indices theoretically need only to be averaged and plotted to reconstruct relative variations in climate (Schulman 1956; Fritts 1976; Schweingruber 1988). However, a variety of environmental conditions can limit growth, and the importance of limiting conditions can vary markedly from one season to the next depending upon the stage of growth, the conditions of the trees, microclimates and other factors of the site (Kramer and Kozlowski 1979; Fritts 1974, 1976; Fritts et al. manuscript). In addition, small differences in site factors can have large effects on limiting conditions particularly for species growing near their ecological limits (Fritts et al. 1965a, 1965b; Fritts 1969a). LaMarche (1982) suggests that differences in response to climate can be controlled by sampling trees from a particular habitat type such as arid or high-altitude sites. We propose that large differences in response still remain in selected chronology sequences from arid sites as evidenced by the large numbers of principal components (PCs) needed to reduce the chronology variance (Fritts 1976).
