A comparison of two centrifuge techniques for constructing vulnerability curves: insight into the 'open-vessel' artifact.

A vulnerability curve (VC) describes the extent of xylem cavitation resistance. Centrifuges have been used to generate VCs for decades via static- and flow-centrifuge methods. Recently, the validity of the centrifuge techniques has been questioned. Researchers have hypothesized that the centrifuge techniques might yield unreliable VCs due to the open-vessel artifact. However, other researchers reject this hypothesis. The focus of the dispute is centered on whether exponential VCs are more reliable when the static-centrifuge method is used rather than the flow-centrifuge method. To further test the reliability of the centrifuge technique, two centrifuges were manufactured to simulate the static- and flow-centrifuge methods. VCs of three species with open vessels of known lengths were constructed using the two centrifuges. The results showed that both centrifuge techniques produced invalid VCs for Robinia because the water flow through stems under mild tension in centrifuges led to an increasing loss of water conductivity. In addition, the injection of water in the flow-centrifuge exacerbated the loss of water conductivity. However, both centrifuge techniques yielded reliable VCs for Prunus, regardless of the presence of open vessels in the tested samples. We conclude that centrifuge techniques can be used in species with open vessels only when the centrifuge produces a VC that matches the bench-dehydration VC.