6.1 Spatio-temporal variation in embolism vulnerability
We found little variation in embolism vulnerability across stands, though clear species differences were observed. Across all stands, mean embolism thresholds were markedly different across species (Fndf,ddf = 149.87,p = 0.001 for P12, and Fndf,ddf = 169.62,p = 0.003 for P50, Fig. 3). At the P50 threshold however, we detected some variability arising from the interaction between species and age (Fndf,ddf = 18.88, p = 0.017, Table 2) and age and region (Fndf,ddf = 21.312,p = 0.016, Table 2). Specifically, we found that A. saccharum P50 differed between young (15yo) and intermediate (35yo) stands, although embolism vulnerabilities were invariant across stand ages for Q. alba and L. tulipifera (Fig 4a). Additionally, across all species, young stands (15yo) had more vulnerable xylem in the mesic NC_W chronosequence stands than in the drier IN and NC_E stands; however, this pattern was not observed for the 35yo and 85yo age classes (Fig 4b). In general, Q. alba had the most vulnerable xylem whileA. saccharum had the least (Fig. 3). Mean P12 across all stands were −1.09 MPa (SE = 0.06), −1.65 MPa (SE = 0.10), and −2.75 MPa (SE = 0.20) and mean P50 was −2.72 MPa (SE = 0.09), −3.91 MPa (SE = 0.12), and −4.77 MPa (SE = 0.18) for Q. alba, L. tulipifera, and A. saccharum , respectively.
6.2 Relationship between xylem anatomy and embolism vulnerability
Xylem anatomy varied considerably between diffuse-porous L. tulipifera and ring-porous Q. alba . Tissues of L. tulipifera consistently had smaller mean vessel lumen area (Fndf,ddf = 169.953, p = <0.001, Fig. 5) and larger mean vessel densities (Fndf,ddf = 270.636, p = <0.001, Fig. 6) than Q. alba . Mean vessel lumen area was not influenced by local climate or age (region, age, or interactions NS, Table 3).
Xylem anatomy had moderate explanatory power for tissue-level embolism vulnerability. Across species, stems with larger vessel lumen area (Fig. 5d) and lower vessel densities (Fig. 6d) approached 50% loss of hydraulic function at lower Ψx (R 2 = 0.335, p = <0.001 for lumen area, and R 2 = 0.268, p = <0.001 for vessel density). Patterns with P12 were similar, but generally weaker than in relation to P50. Specifically, tissues with larger mean vessel lumen area tended to approach 12% loss of hydraulic function at lower Ψxrelative to tissues with smaller mean lumen area (R 2 = 0.112, p = 0.005, Fig. 5b). Tissues with greater vessel densities were generally more embolism-resistant at P12 (R 2 = 0.08, p= 0.019, Fig. 5b). However, this pattern was contradicted by Q. alba , where stems with greater vessel densities were more vulnerable to 12% loss of hydraulic function (R 2 = 0.256,p = 0.002, Fig. 6b).