ABSTRACT
Vulnerability curves (VCs) describe the loss of hydraulic conductance against increasing xylem tension, providing valuable insights about the response of plant water transport to water stress. Techniques to construct VCs have been developed and modified continuously, but controversies continue. We compared VCs constructed using the bench-top dehydration (BD), air-injection-flow (AI), pneumatic-air-discharge (PAD), optical (OP) and X-ray-computed microtomography (MicroCT) methods for tropical trees and lianas with contrasting vessel lengths. The PAD method generated highly vulnerable VCs, the AI method intermediate VCs, whereas the BD, OP and MicroCT methods produced comparable and more resistant VCs. Vessel-length and diameter accounted for the overestimation ratio of vulnerability estimated using the AI but not the PAD method. Compared with directly measured midday embolism levels, the PAD and AI methods substantially overestimated embolism, whereas the BD, MicroCT and OP methods provided more reasonable estimations. Cut-open vessels, uncertainties in maximum air volume estimations, sample-length effects, tissue cracks and shrinkage together may impede the reliability of the PAD method. In conclusion, we validate the BD, OP and MicroCT methods for tropical plants, whereas the PAD and AI need further mechanistic testing. Therefore, applications of VCs in estimating plant responses to drought need to be cautious.