6.4 Relationship between ΨL regulation and vulnerability to hydraulic failure
The most anisohydric species in our study possessed xylem that were more vulnerable to embolism than the more isohydric species. This pattern was consistent at both P12 and P50, where Q. alba (mean ΨL interquartile range 0.71\(\pm\) 0.04 MPa) embolism thresholds were consistently greater than the more isohydric L. tulipifera (mean ΨLinterquartile range 0.36 \(\pm\) 0.02 MPa) and A. saccharum (mean ΨL interquartile range 0.37 \(\pm\) 0.10 MPa), respectively. Across all stands, Ψsafety were smallest and often negative for Q. alba.The average slope of the regression between Ψsafety and ΨLinterquartile range were -3.80 (\(\pm\) 0.32) and -4.92 (\(\pm\) 0.40) for Ψsafety,P12and Ψsafety,P50, respectively and was consistently negative across 100 bootstrapped simulations (Fig. S4). Age-independent analysis of variation in Ψsafety across the aridity gradient revealed that the lowest Ψsafety often occurred in the more arid regions of our study (e.g., MO and NC_E); nevertheless, the largest differences were associated with species (Fig. S5 & Table S2). Overall, the degree of isohydricity was strongly linked to hydraulic safety across species and sites (R 2 = 0.57, p = <0.001 andR 2 = 0.61, p = <0.001 for Ψsafety,P12 and Ψsafety,P50, respectively), such that increasingly anisohydric behavior promoted greater risk for hydraulic damage (Fig. 7).
Estimated native embolism patterns were highly similar to Ψsafety (Fig. 8a). Q. alba had greater estimated native embolism than L. tulipifera and A. saccharum under field conditions (Species effect; Fndf,ddf = 162.559, p = 0.001, Table 4). We detected some intra-species differences associated with stand age across regions (Species*Age*Region effect; Fndf,ddf = 3.347, p = 0.039, Table 4). However, these differences were often inconsistent across regions (e.g., greater estimated native embolism with increasing stand age forQ. alba in IN and lower estimated native embolism with increasing stand age for Q. alba in NC_W, Fig. 8a) and may be due to non-overlapping ΨL sampling periods within chronosequences. Regardless, spatio-temporal effects (e.g., Species*Age*Region effect) were substantially more marginal than the large species effect (Table 4). Overall, increasing ΨL interquartile range across species and stands were strongly associated with a greater extent of estimated native embolism (R2 = 0.67, p = <0.001, Fig. 8b & Fig. S6). This relationship additionally coincided with a lower magnitude of hydraulic conductivity. For example, in NC_E, estimatedin-situ Kstem was 0.30 (\(\pm\) 0.133) kg m-1 s-1 MPa-1 forQ. alba while estimated in-situ Kstem for L. tulipifera and A. saccharum were 1.12 (\(\pm\)0.15) kg m-1 s-1MPa-1 and 0.62 (\(\pm\)0.04) kg m-1s-1 MPa-1, respectively.