Fig 1.
Initial (hypothetical) path model of the physiological and morphological traits.
The grey part is the path model from Belluau and Shipley (2017) [10]. Solid lines are positive causal relationships while dashed lines are negative. The path coefficients are not reported here except for non-significant links (ns). This initial (hypothetical) path model was statistically rejected (Satorra-Bentler robust Chi-square = 85.051; 47 df; p = 0.0168; Comparative Fit Index (CFI) = 0.671; Root Mean Square Error of Approximation (RMSEA) = 0.180).
Table 1.
Species list containing soil hydrological classification, life cycle, orders and families.
Table 2.
List and description of the soft and hard traits used in the analysis.
Fig 2.
Path analysis of the physiological and morphological traits.
There is no significant misfit between the empirical data and the causal structure specified by the model (Satorra-Bentler robust Chi-square = 40.795; 42 df; p = 0.782; Comparative Fit Index (CFI) = 1.00; Root Mean Square Error of Approximation (RMSEA) = 0.00), Akaike information criterion (AIC) = 804.037. All path coefficients are significantly different from zero, otherwise the paths’ significativity are specified in the diagram. The R2 are the percentage of variance explained by the causal variables. Values on the lines are the path coefficients between the causal variable and the caused variable. Solid lines are positive causal relationships and dashed lines are the negative ones. Thickness of the lines is proportional to the strength of path coefficients. For all traits, we used the average value of five individuals per species. For all leaf traits, we excluded the petioles because petioles don’t belong to the same function as the leaf blade (i.e. support for petioles and acquisition for leaf blade). Leaves were chosen to be representative of the average mature leaf into one individual and we avoided newly formed leaves or ones showing any sign of senescence.
Table 3.
Cumulative link model analysis of the hard and soft traits as predictors of the habitat wetness preferences.
Table 4.
Stepwise backward linear regression, based on AIC values, of each of five hard traits (Awilt, gswilt, Ψwilt, gsmax, Amax) on a linear combination of five soft traits (leaf dry matter content (LDMC, g g-1), specific leaf area (SLA, m2 kg-1), leaf nitrogen content (LNC, mg g-1), stomatal area (stomarea, (μm) and specific root length (SRL, m g-1)).
Fig 3.
Box plots showing the differences between species means of the soft trait values.
Traits are measured at field capacity and grouped by species according to affinity for habitat wetness. "D" (species typical of "dry" soils), "I" (species typical of "intermediate" soils), "W" (species typical of "wet" soils). Non-parametric (Kruskal-Wallis) 1-way ANOVAs did not detect any significant differences between the three species groups for any of these five traits.