Clustering reveals cavitation-related acoustic emission signals from dehydrating branches

Abstract

The formation of air emboli in the xylem during drought is one of the key processes leading to plant mortality due to loss in hydraulic conductivity, and strongly fuels the interest in quantifying vulnerability to cavitation. The acoustic emission (AE) technique can be used to measure hydraulic conductivity losses and construct vulnerability curves. For years, it has been believed that all the AE signals are produced by the formation of gas emboli in the xylem sap under tension. More recent experiments, however, demonstrate that gas emboli formation cannot explain all the signals detected during drought, suggesting that different sources of AE exist. This complicates the use of the AE technique to measure emboli formation in plants. We therefore analysed AE waveforms measured on branches of grapevine (Vitis vinifera L. 'Chardonnay') during bench dehydration with broadband sensors, and applied an automated clustering algorithm in order to find natural clusters of AE signals. We used AE features and AE activity patterns during consecutive dehydration phases to identify the different AE sources. Based on the frequency spectrum of the signals, we distinguished three different types of AE signals, of which the frequency cluster with high 100-200 kHz frequency content was strongly correlated with cavitation. Our results indicate that cavitation-related AE signals can be filtered from other AE sources, which presents a promising avenue into quantifying xylem embolism in plants in laboratory and field conditions.

Acoustic Emissions to Measure Drought-Induced Cavitation in Plants

“…One reason for the higher number of signals observed by Vergeynst et al [80] could be overlapped between adjacent cluster types, so that part of the cavitation-related signal type possibly originated from other co-occurring AE sources, such as micro-fractures or water drainage. This overlap might be caused by different attenuation of the AE signals dependent on their frequency.…”

“…Given that PLC should by definition be linked to cumulative AE, and that, according to Aggelis et al [81], a specific AE-inducing mechanism results in an AE activity, Vergeynst et al [80] recommended determining the VC endpoint by the end of the AE activity peak, which mathematically corresponds to the local maximum of the third derivative of the curve of cumulative AE versus time (Figure 3).…”

“…In measurements on grapevines, the number of cavitation-related AE signals exceeded the number of vessels in the branch by one or two size orders [33]. Moreover, the amplitude Vergeynst et al [80] proposed clustering to determine cavitation-related AE in angiosperms. By using state-of-the-art techniques from material sciences, finite element modelling and an automated clustering algorithm [85,95,96], in combination with broadband point-contact sensors, they were able to extract the AE that originated from cavitation during dehydration of grapevine branches.…”

“…The flat frequency response of these sensors in a wide frequency range (20-1000 kHz) allows differentiation between different AE sources. Instead of relating waveform features of AE signals with cavitation, Vergeynst et al [80] recommended to use the frequency features of AE signals. Following Sause and Horn [97], they extracted from the frequency spectrum, peak frequency (PF), frequency centroid (FC), weighted PF (WPF, geometric mean of PF and FC) and the partial powers of the following frequency ranges: 0-100 kHz, 100-200 kHz, 200-400 kHz and 400-800 kHz ( Figure 5).…”

“…When using the AE technique on living plants in lab or field conditions, Vergeynst et al [80] use a broadband point-contact sensor (20-1000 kHz) with flat frequency response in order to differentiate between the different AE sources. Based on preliminary analysis and clustering, a minimum set of signal frequency features may be selected (weighted peak frequency, frequency centroid and partial powers) that enable discrimination between different AE sources.…”