Paleobotanists have developed numerous methods for quantifying correlations between leaf form and climate, so that fossil leaves can be used to estimate past terrestrial climate. Approaches currently used can be broken into two classes: those using categorical, primarily qualitative shape characters (such as round base) and those describing shape as ratios or other abstract mathematical descriptors (such as perimeter-to-area ratio). Both classes of approach have limitations and are hard to compare to one another. We show how eigenshape analysis provides both interpretable characters and measures of continuous variation, thus potentially overcoming the limitations of both current approaches. We applied eigenshape analysis to a sample of 394 nonlobed leaves from extant assemblages to empirically identify independent leaf shape characters. This resulted in the identification of 20 shape characters. Of these, 13 reflected variation due to broad-scale asymmetry as well as asymmetry localized to the tip, base, and lamina-petiole junction. The remaining seven characters, in order of variance explained, corresponded roughly to base shape, tip angle, location of widest portion of the leaf (ovate to obovate variation), extent of widest portion of the leaf (elliptic to oblong variation), tip shape (blunt to drip tip), ovate to obovate variation localized to the middle of the leaf, and extension of the lamina along the petiole at the lamina-petiole junction. All 7 of these characters and only 1 of the 13 asymmetry characters had an association with climate. These characters corresponded to five of the seven broad categories of Wolfe's Climate-Leaf Analysis Multivariate Program shape descriptors, excluding only tooth shape and lobing. The results of this study indicate that eigenshape-generated measures of leaf shape merit further exploration.

You do not currently have access to this article.