Vulnerability curves by centrifugation: is there an open vessel artefact, and are 'r' shaped curves necessarily invalid?

Vulnerability curves using the 'Cavitron' centrifuge rotor yield anomalous results when vessels extend from the end of the stem segment to the centre ('open-to-centre' vessels). Curves showing a decline in conductivity at modest xylem pressures ('r' shaped) have been attributed to this artefact. We determined whether the original centrifugal method with its different rotor is influenced by open-to-centre vessels. Increasing the proportion of open-to-centre vessels by shortening stems had no substantial effect in four species. Nor was there more embolism at the segment end versus centre as seen in the Cavitron. The dehydration method yielded an 'r' shaped curve in Quercus gambelii that was similar to centrifuged stems with 86% open-to-centre vessels. Both 'r' and 's' (sigmoidal) curves from Cercocarpus intricatus were consistent with each other, differing only in whether native embolism had been removed. An 'r' shaped centrifuge curve in Olea europaea was indistinguishable from the loss of conductivity caused by forcing air directly across vessel end-walls. We conclude that centrifuge curves on long-vesselled material are not always prone to the open vessel artefact when the original rotor design is used, and 'r' shaped curves are not necessarily artefacts. Nevertheless, confirming curves with native embolism and dehydration data is recommended.