Insert Table 1

Determination of drought tolerance

Drought tolerance was determined by vulnerability to xylem cavitation (P50) using the centrifuge method to induce cavitation in the xylem (Cochard et al., 2013, 2005). This advanced centrifuge technique creates centrifugal force that generates tension in the branch xylem vessels to induce cavitation in branch segment, thereby allowing measurement of xylem percentage loss of conductivity at set points of tension. Straight stems, 27 cm in length and with 6 mm basal diameter, were sampled and cut under water from the remaining 40 cm-long branch segments, placed on the custom-built rotor and spun at different velocities. To control for the artefact associated with the centrifuge, initial measurements were obtained at lower pressures (-0.5 MPa = 2378 rpm) that did not induce cavitation (López et al., 2019). The percent loss of conductance (PLC) at negative xylem pressure (tension) was automatically recorded through a step-wise increase (1000 rpm each) atca. 2 min stabilization time (Zhang et al., 2017) until 90-95% loss of conductivity was attained. At each new xylem pressure (tension), hydraulic conductance (Kh) was measured from 30 repeated measures. The PLC was computed as PLC = 100 × (1 − Kh/Kmax). The dependence of PLC on xylem pressure was used to generate vulnerability curves for each species and 50 % loss of conductance (P50) were obtained from slope of the curve using the fitplc R package (Duursma & Choat, 2017).

Statistical Analysis

Stem P50 difference between species were tested using a linear model (lm ), while differences between biome, life histories traits (resprouting ability and leaf forms), as well as interactions, were determined using a linear mixed effect model (lme4 ) R package (Bates et al., 2015) with species as a random variable. Residuals of models were inspected; appropriate transformations were conducted and extreme outliers were removed where necessary. ANOVA for mixed effects models was undertaken using Kenward Roger degrees of freedom approximation. Linear mixed effects model with species as random effect was used to explore predictors of cavitation resistance (P50). Posthoc Tukey tests were undertaken using the emmeans R package (Lenth, 2020) to determine which species and life histories are significantly different.