The first two dimensions of the PCA resolved approximately 70.6 % of
the total variance in traits, with the x axis explaining 44.7 % and the
y axis explaining 25.9 % (Fig. 6 ). The x axis opposed the
maximum height to leaf dry weight and leaf area traits, while the y axis
opposed total biomass and reproduction to survival. Interestingly,
traits did not clearly share high correlation across axes. PCA
clustering revealed clear differences in trait response between species.L. corniculatus segregated largely to the negative direction of
the first dimension, whereas D. glomerata and P.
lanceolata segregated largely in positive and negative second dimension
respectively. In contrast, climate was not clustered following a PCA
analysis – with all climate level confidence interval ellipsoids
overlapping even at the 5 % level.