5.4 Xylem embolism vulnerability curves
Vulnerability to hydraulic failure was estimated with cavitation-induced embolism curves. The relationship between the loss of hydraulic function and stem xylem water potential (Ψx) (MPa) was measured on stem tissues (n = 3–5) from 2–3 trees per species at each stand, resulting in 6–12 curves per species per stand, or 165 total curves. Vulnerability curves were generated using the air-injection technique (Sperry & Saliendra, 1994; Johnson et al ., 2016). Branches were harvested from the upper third of the canopy, and stem samples ~20 cm in length were collected from the terminal bud of felled branches. Samples were stored at 5 °C submerged in deionized water that was replenished daily and were measured within two weeks of collection.
We used a pressure flow meter (XYL’EM embolism meter, Bronkhorst, Montigny les Cormeilles, France) to measure stem hydraulic conductivity (Kstem ) (kg m-1s-1 MPa-1), and a pressure sleeve (Scholander Pressure Chamber model 1505D, PMS Instruments, Corvallis, OR, USA) to facilitate air-injection. Samples were rehydrated by flushing native embolism in submerged deionized water under vacuum for 24+ hours. Following rehydration, stem samples were exposed to positive air pressure in 0.5 to 1.0 MPa increments until >85% reduction of maximum Kstem was reached or the applied pressure approached instrument limitation. We then correctedKstem to 20 °C to account for changing viscosity of water with temperature (K20 ) (kg m-1 s-1 MPa-1). The percent loss of conductivity (PLC) (%) at a given applied pressure was calculated as:
\(PLC=100\ \times(1-\frac{K_{20}}{K_{\text{max\ }}}\ )\) (1)
where Kmax is temperature corrected maximumKstem when applied pressure = 0 MPa.
The relationship between PLC and Ψx was then fitted to the sigmoid function provided by Maherali et al ., (2006):
\(PLC=\ \frac{100}{\left[1+exp(\alpha\left(\Psi_{x}-\ b\right))\ \right]}\)(2)
where α and b are empirical coefficients determined using nonlinear curve fitting (MATLAB, The Mathworks Inc., Natick, MA, USA; v. R2018a). The fitted relationship was then used to calculate the Ψx at which 12% PLC (P12, MPa) and 50% PLC (P50, MPa) occurred. The P50 was set equal to the b parameter, and P12 calculated as 2/a + b , as described by Domec & Gartner (2001). The value P12, termed the air entry point, is an estimate of the xylem tension at which the resistance to air entry of pit membranes within the conducting xylem is overcome and cavitation and embolism begin.
The measurement and interpretation of the vulnerability curves was guided by extensive quality control to minimize sources of bias. Specifically, while the air-injection method remains the most popular technique for assessing vulnerability to embolism (Sperry & Saliendra, 1994; Johnson et al ., 2016), measurement artifacts from destructive sampling, such as the presence of open vessels, may over-estimate in-situ vulnerability (Martin-StPaul et al ., 2014). This bias may be particularly important for long-vesseled species like Q. alba (Cochard & Tyree, 1990). We therefore took multiple steps to minimize the presence of open vessels and to remove any curves that appeared to be affected by open vessel artifacts:
[1] First, we sampled young distal tissues from branch apices, which have relatively short vessels (Cochard & Tyree, 1990). WhileQuercus species can have vessels that extend to several meters in length, long vessels are less prevalent in young stems and distal branches (Cochard & Tyree, 1990; Fontes & Cavender-Bares, 2020). Thus, we collected only these tissue sections to increase the likelihood that xylem elements were short in length.
[2] Second, while many studies avoid open vessel artifacts by collecting branch samples that are twice the length of a reference average vessel length, we did not assume that our samples contained intact vessels. Instead, we directly tested for the presence of open vessels using an air-infiltration technique (Cochard et al.,2010). We discarded every stem that allowed low pressure air to freely pass through, indicating severed vessel end walls were present (Cochardet al., 2010). This was a labor-intensive step that required collecting a substantially greater number of stems than were ultimately used for vulnerability curves; however, it was necessary to ensure thatQ. alba samples had intact vessels.
[3] Third, we carefully considered the shape of the vulnerability curves and removed any that contained signatures of open vessel artifacts, noting that curves that are conspicuously ‘r’ shaped are likely affected by open vessel artifacts, and that ‘s’ shaped curves more accurately represent in-situ vulnerability (Torres-Ruizet al ., 2014; Skelton et al., 2018). We defined an ‘s’ shape curve as one that lost less than 7.5% of itsKmax as Ψx declined from 0 to −0.5 MPa and screened our dataset to use only these curves. We performed the analysis at alternative cutoff thresholds of 3%, 5%, and 10% loss of Kmax , but there were no noticeable effect on the results. Overall, including both ‘s’ and ‘r’ curves had little impact on characterizing embolism thresholds (Fig. S3). Nevertheless, ‘r-shaped’ curves for any species were not included for subsequent analyses, resulting in 40, 56, 26 suitable ‘s-shaped’ curves forL. tulipifera , Q. alba , and A. saccharum , respectively (or ~74% of the original 165 curves, Table S1).
5.5 Xylem anatomy
To understand how changes in xylem vulnerability are linked to variations in xylem anatomy, we measured vessel lumen area and vessel density on transverse sections (~40 µm width) extracted from stems used for embolism vulnerability measurements. Unfortunately, stem samples were unavailable from MO and for A. saccharum in the IN 85yo stand (Table S1).
Stem samples were softened by boiling in deionized water and sectioned by hand using a fresh razor blade (Schweingruber, 2007). Prior to analysis under the microscope, samples were oven-dried at 150 °C to reduce light refraction from water remaining in lumen areas. Slides from the NC_W and IN were imaged with a stereoscope and color camera at 150\(\times\)magnification (Leica M205F, Leica DFC310FX, Leica Microsystems, Heerbrugge, Switzerland). Vessel lumen area and density were then calculated using threshold balance manipulation and the analyze particle function of ImageJ v1.6 software (National Institutes of Health, USA) (Scholz et al ., 2013). Slides from the NC_E were photographed at 100\(\times\) and 200\(\times\) magnifications and analyzed using the Motic Images Advanced 3.2 software (Motic Corporation, Zhejiang, China).