干旱区梭梭生长的饱和水汽压差-根区土壤水分非线性阈值及协同效应定量分析OACHSSCD

Understanding the nonlinear responses of desert vegetation to climatic stress is crucial for predicting ecosystem dynamics under climate change.This study focused on Haloxylon ammodendron in the Junggar Basin from 1982 to 2020,aiming to quantitatively reveal the nonlinear response thresholds of its growth to vapor pressure deficit(VPD)and root zone soil moisture(SMrz),as well as their synergistic mechanisms,thereby providing a quantitative basis for precise water management and ecological restoration of Haloxylon forests in arid regions.By integrating piecewise regression models and generalized additive mixed models(GAMM),and based on topographic,climatic,and soil characteristics,the study area was divided into five subregions(east,south,west,north,and central).The 95％confidence intervals of VPD thresholds inhibiting Haloxylon growth were identified for each subregion,and the regulatory effect of SMrz on these thresholds,along with the synergistic effect between the two factors,was analyzed.The results showed that:(1)Significant VPD inhibitory thresholds existed in all subregions(Davies test,P＜0.05),with pronounced spatial variability.The lowest threshold was found in the northern subregion(0.6 kPa,95％CI:0.57-0.749 kPa),while the highest was in the western subregion(0.922 kPa,95％CI:0.896-0.94 kPa).(2)SMrz exerted a significant positive regulatory effect on the VPD threshold:the VPD threshold at the 75th percentile of soil moisture was increased by an average of 0.145 kPa compared with that at the 25th percentile.Moreover,the regulatory intensity exhibited remarkable spatial heterogeneity across different subregions,with the strongest regulation observed in the central subregion(the VPD threshold rising from 0.632 kPa to 0.780 kPa)and the weakest in the southern subregion(the VPD threshold increasing from 0.768 kPa to 0.881 kPa).(3)The GAMM model incorporating the interaction term of VPD and SMrz achieved the optimal explanatory power,with the deviance explained for the variation in the normalized difference vegetation index(NDVI)reaching 83.0％(R2=0.825).This performance was significantly superior to that of the VPD-only model(R2=0.531)and the additive model of VPD and SMrz(R2=0.647).Cross-validation results further indicated that the root mean square error(RMSE)of the interactive model was reduced by 12.4％and 3.8％compared with the latter two models,respectively,which meant the interactive model could more effectively capture the residual variation caused by the nonlinear relationships between the factors.This study clarified the nonlinear synergistic influence mechanism of VPD and SMrz on the growth of Haloxylon ammodendron,and the research findings provide an important scientific reference for the precise management and ecological restoration practices of Haloxylon ammodendron forests in arid regions.