Figure Legends
Figure 1. Relationships between 11 ‘Chardonnay’ leaf traits and soil bulk density. Trend lines correspond to statistically significant relationships (where p ≤0.05 for the slope parameter) between traits and bulk density, based on linear mixed effects models predicting traits as a function of soil bulk density (as a fixed factor) while accounting for plant identity (as a random factor; see Table S2 for full model diagnostics and fits). Also presented are marginalr 2 values (“Marg.r 2”) for each relationship, which represents the proportion of variation in a given trait explained by fixed factors alone (i.e., bulk density and model intercept), and conditionr 2 values (“Cond.r 2”), which represent the proportion of trait variation explained by fixed and random factors. Sample sizes for all models were 45 leaves, measured across 15 individual vines. Log-transformed trait values were used in models according to results presented in Table 1, and trait acronyms are presented in Table 1.
Figure 2. Principal Component Analysis (PCA) for seven ‘Chardonnay’ wine grape leaf traits measured in 2020 across a soil compaction gradient. Data point colours correspond to vine sampling rows, which are situated along a gradient of bulk density values, and dotted black lines represent 95% confidence ellipses for leaves across different rows. Planting row explained 39.6% of the multivariate trait variation evaluated here (p ≤ 0.001, and see Table S4 for full Permutational Multivariate Analysis of Variance results). Trait acronyms are presented in Table 1.
Figure 3. Relationships across four Leaf Economics Spectrum traits in ‘Chardonnay’ wine grapes. Colours correspond to sampling rows reflecting a soil compaction gradient, black solid trend lines correspond to the standardized major axis (SMA) regression model of a given bivariate trait relationship in ‘Chardonnay’, and dashed black trend lines represent convex hull models that encapsulate the two-dimensional trait space occupied by ‘Chardonnay’ leaves. Also shown are the data and SMA models for the same LES trait relationships observed among wild plants in the GLOPNET dataset (grey dashed trend lines and points). Details on all SMA models shown here are presented in full in Table 2. Trait acronyms are presented in Table 1.