Figure 3.

Principal components describing symmetrical and asymmetrical shape variance. (A) Both abaxial-top and adaxial-bottom images were used in an Elliptical Fourier Descriptor (EFD) analysis to examine asymmetric shape variance. Abaxial-top images only were used in an analysis of symmetric shape variance, as mirror images were not needed to analyze asymmetry. Shown is the percent variance in asymmetric (left) and symmetric (right) analyses explained by each PC. Note that the amount of variance described by the first PCs (asymPC1: 58.4%, symPC1: 48.0%) is relatively high, and that collectively the first four PCs describe a large amount of the overall variance (asymPCs1–4: 88.3%, symPCs1–4: 88.8%). (B) Leaflet outlines representing ±2 standard deviations along asymmetric principal component axes. Particular attention is given to asymPC1 in this study, which explains asymmetric variance relating to overall bending of the leaf. (C) Leaflet outlines representing ±2 standard deviations along symmetric principal component axes. Particular attention is given to symPC1, which, like the other symPCs, describes symmetric shape variance relating to the distinctness of the petiole and distribution of laminar outgrowth along the proximal-distal axis.