Data analyses
Owing to the contrasting soil-water conditions and water-use strategies
along the DGW gradient in spring and summer (Wu, Zheng, Li, et al.,
2019), all data were grouped into two growth stages, namely the early
growing season (EGS) from April to June, which included spring and early
summer, and late growing season (LGS) from July to September, which
included late summer and autumn. Descriptive statistics were used to
calculate the means and standard errors of each subset. A linear mixed
effect model (LMEM) was used to evaluate the effects of growth stage on
the hydrological drought (different depth to groundwater) response of
the studied species using “lme4” package (Bates et al., 2014) and
running on 3.6.3 version on R. Differences among hydraulic traits,
π100, Ψ TLP, osmotic substance
contents, assimilation branch areas, and growth rates at different sites
within the same growing season were also tested using one-way analyses
of variance (ANOVA) with Tukey’s-HSD post-hoc test. To compare the
differences of traits in the EGS and LGS, independent-samplest -tests were used. Linear regressions were carried out between
hydraulic properties, P-V curve parameters, Huber values, growth rates,
and DGWs. Figures were prepared using Origin 8.0 (Origin Lab Corp.,
Northampton, MA, USA) and R ver. 3.6.3 (R development Core Team 2016).
Data analyses were conducted using R ver. 3.6.3 and IBM SPSS Statistics
software (Ver. 22, Armonk, NY, USA).