Drought vulnerability of trees and other woody plants is much debated in the context of climate change, which creates a high interest in understanding plant water relations. The role and functioning of internal water storage is crucial, but still insufficiently understood. Drought vulnerability is typically assessed by considering loss in conductivity in function of decreasing xylem water potential, in a so-called 'vulnerability curve'. The xylem water potential at which a certain percentage of

more »

... conductivity is lost (usually 50%) gives an indication of the vulnerability to cavitation. In a 'desorption curve', we can examine the release of water from internal storage tissues with decreasing water potential. Both curves are very valuable, but rely on a sequence of manual measurements (xylem water potential, hydraulic conductivity and water content) and are time-consuming. Therefore, we propose a new type of vulnerability curve that is based on continuous measurements of diameter shrinkage and ultrasonic acoustic emissions (UAE). We monitored weight loss, xylem diameter shrinkage and UAE and measured xylem water potential during the dehydration of excised branches of Vitis vinifera L. 'Johanniter'. The vulnerability curves could be interpreted in terms of water loss in elastic and inelastic tissues. The proposed method can be a tool to assess hydraulic capacitance and conductivity of the xylem. systeme, Icking, Germany). One branch segment was placed on the weighing scale without any sensor and stripped of all leaves. The other three branches were equipped with an UAE sensor in the middle of the segment using a clamp (as in Rosner et al. (2009)). Each second, the maximum signal amplitude (APK) was logged and only the values larger than 35 dB (background noise + 5 dB) were retained. We installed an LVDT next to the UAE sensor on the xylem, where the bark was carefully removed and Vaseline was applied to prevent water loss through the bare xylem. Before and after the dehydration experiment, a wood sample of about 5 cm was taken from the end of each branch segment and oven dried until constant weight. With these samples, initial and final water content, and green and dry wood density, were calculated for each branch segment. The course of the water content of each branch segment was estimated by rescaling the continuous weight data to the range of water contents between initial and final water content in the samples. Volumetric water content (VWC) was calculated by multiplying mass fraction with initial dry wood density. During the first 36 hours, xylem water potential was determined each three hours on excised leaves of the three equipped branch segments using the pressure bomb (PMS Instrument Company, Corvalis, OR, USA). RESULTS AND DISCUSSION Meaningful transition points in new desorption-vulnerability curve We displayed all measurements as a function of xylem diameter shrinkage ( Fig. 1) . The presented data are gathered during the first 84 h of dehydration. Two transition points characterize the dehydration process when considering the ensemble of xylem shrinkage, water loss and cumulative UAE. The dehydration started with a continuous decrease in VWC with increased xylem shrinkage, which corresponded to water loss from elastic tissues. After an initial quiet period, cumulative UAE started to increase gradually. The slope change around 0.37 mm xylem shrinkage is the transition between a phase with little cavitation and a phase with sharply increasing cavitation, also observed by Tyree and Yang (1990). This is associated with an increase in water loss per unit decrease in xylem diameter. Up to the point of strong cavitation increase, xylem shrinkage was proportional to total water loss. Water loss by living cells in the xylem is related to xylem shrinkage in this phase, in the same way as water loss in the bark is proportional to bark shrinkage (Zweifel et al., 2001). In the second phase, reduced xylem shrinkage compared to the amount of water loss implies that more water is released by cavitation. After the second transition point (0.68 mm xylem shrinkage), UAE activity is lower. During cavitation, water in the lumen of xylem vessels is replaced by air. When all free water is replaced by air, the fibre saturation point (FSP) is reached. A further reduction in water content below FSP is associated with cell wall shrinkage (Roderick and Berry, 2000) and results in an abrupt change in physical properties of the wood (Roderick and Berry, 2000) with, for example, a strong increase in the velocity of ultrasonic acoustic waves (Sakai et al., 1990). This means that more and more noise from cracks, associated with strong deformation stresses that arise from cell wall shrinkage (Sakai et al., 1990; Wolkerstorfer et al., 2012), will reach the UAE sensor. Wolkerstorfer et al. (2012) could distinguish cracks from cavitation hits using the amplitude of the signal. Similarly, we also observed a decrease in 10-min average amplitude near the end of the cavitation phase (Fig. 2) . During the final dehydration phase, the amplitude is variable and reaches also high values, probably due to the strong increase in wave velocity. The observations suggest that the second transition corresponds to the fibre saturation point. The volumetric water content at FSP (80 kg m -3 ), estimated with the formula of Roderick and Berry (2000) , is indeed reached near the transition (Fig. 1 ). In this formula (Roderick and Berry, 2000), α f stands for mass fraction of water over oven-dry wood. The volumetric water content is obtained by multiplying α f with the initial basic density of wood (ρ b = 158.5 kg m -3 , mass of oven-dry wood divided by fresh volume). The calculated FSP is an integration over the whole segment, but is only gradually reached as the cavitation phase is completed. This explains why the maximum UAE amplitude and the observed transition occurs before FSP is reached. Diameter shrinkage on the x-axis In standard desorption and vulnerability curves, dehydration processes (water loss, cavitation or conductivity loss) are related to water potential. We chose to use xylem diameter on the x-axis for two reasons. First, the water transport and storage model (Steppe et al., 2006; Steppe and Lemeur, 2007), developed at the Laboratory of Plant Ecology, UGent, Belgium, uses diameter variations as continuous output variable. Robust and reliable sensors are available to measure diameter variations continuously and across long periods in the field. It would, hence, be very useful if plant processes that occur during drought stress could be related to this easily available variable. Second, the automated and continuous measurements of diameter shrinkage greatly facilitate the procedure to obtain useful desorption or vulnerability curves. Transitions are also difficult to designate in a graph when only point measurements, each with a certain measurement error, are available. During the first 36 h of the dehydration experiment, we could measure xylem water potential on detached leaves. Because the number of leaves per branch was limited and because measurements could only be conducted on non-wilted leaves, the range of ψmeasurements was restricted to the first phase and the beginning of the second phase. Here, xylem water potential decreased linearly with increasing xylem shrinkage ( ψ = −0.8 + 3.7⋅ XylemShrinkage , R 2 =0.95). Outside this range, the linear relationship may not hold. Xylem diameter shrinkage was also used by Hölttä et al. (2005) to replace xylem potential in a field experiment. They plotted cumulative UAE versus maximum xylem shrinkage of Pinus sylvestris during two consecutive days and called it a "vulnerability curve". Hydraulic capacitance and conductivity of the xylem According to Zweifel et al. (2001), over-bark diameter shrinkage is proportional to water loss from the elastic tissues of the bark. We assume that xylem shrinkage is also proportional to water loss from elastic tissues (parenchyma) in the xylem up to the FSP. The cumulative UAE up to the FSP is likely to be proportional to the amount of water released by cavitation. If these assumptions are correct, the progress along the x-and yaxis gives an idea of the relative contribution of the living and dead xylem tissue to the hydraulic capacitance of the xylem. Water loss from the water conducting xylem elements implies a reduction in hydraulic conductivity, so cumulative UAE is also a measure for reduction in hydraulic conductivity (Lo Gullo and Salleo, 1993; Hacke et al., 2000). CONCLUSIONS We propose a new type of vulnerability curve measured by continuous plant sensors, which may help to assess the different components of the hydraulic capacitance (internal water reserves), and the vulnerability to cavitation in terms of hydraulic conductivity. We think that this automated and easy method can be valuable as screening procedure when we want to compare water relations in different environments, in different species and varieties, or throughout the same plant. Further experiments will be done to verify the validity of our new desorption-vulnerability method.