Stem physiological measurements
For each species, stem hydraulic conductivity was measured on eight stem segments from different healthy trees in July 2018. The long (c.a. 1 m), straight and sun-exposed terminal branches with diameter ranging from 6 to 8 mm were sampled at predawn. The end of samples was recut under water and transported to the laboratory immediately. The samples were wrapped with black plastic bag and kept the cutting end submerged in the water during transportation. Upon arrival at the laboratory, a stem segment about 15 cm in length was excised from each of the originally sampled branch. After peeling off the barks and shaving both ends of the segments with a sharp razor blade, segments were connected to a tubing apparatus consist with a 0.22 μm filter and a 50 cm hydraulic head to allow the degassed and filtered 20 mmol L-1 KCL solution to flow through the segments. The stem native hydraulic conductivity (K h, kg m s-1 MPa-1) was calculated as follows:
\(K_{h}=\frac{J_{v}}{\left(\frac{P}{L}\right)}\) (6)
where Jv (kg s-1) is flow rate through the segments and ΔP/ΔL is the pressure gradient across the segment (MPa m-1).
The leaf-specific hydraulic conductivity (K l, 10-4 kg m-1 s-1MPa-1) and sapwood-specific hydraulic conductivity (K s, kg m-1s-1 MPa-1) of each segment were calculated as K h divided by leaf area (LA) and sapwood area (SA), respectively. Branch segments used for the conductivity measurements were perfused by Methyl blue dye under a hydraulic head of 50 cm, the average value of dyed area of both ends of segments was determined as SA. The transverse cross-sections were scanned and the images were analyzed using ImageJ software (US national Institutes of Health, Bethesda, MD, USA) to calculate areas of stained xylems. Meanwhile, distal needles of each segment were scanned to calculate leaf area (LA) and then oven-dried at 75 °C for about 48 h to get constant dry mass to calculate leaf mass per area (LMA, g cm-3). The leaf area to sapwood area ratio (LA/SA, m2cm-2) of each species was calculated as LA divided by SA. The segments used for hydraulic conductivity were then used to measure wood density (WD, g cm-3) by the water displacement method.
Stem hydraulic vulnerable curves were constructed using the centrifugal force method (Alder, Pockman, Sperry, & Nuismer, 1997) on six stem segments of 14.2 cm in length for each species. After measuring the native hydraulic conductivity, the segments were fixed into a high-speed centrifuge (Model 20K, Cence Instruments, Changsha, China) equipped with a customized rotor (‘Sperry rotor’). The vulnerability to embolism of each segment was determined by measuring the decrease of hydraulic conductivity in response to a stepwise increase in xylem tension generated by spinning. Vulnerability curves were fitted using sigmoidal models (Fig. S1) and the pressure at 50% loss of conductivity (P 50, MPa) was calculated (Pammenter & Vander Willigen, 1998).