Objective This study investigates how radial growth of Picea schrenkiana responds to vapor pressure deficit (VPD) across spatial gradients and among individuals, identifies the relative contributions of tree age and environmental factors, and disentangles the immediate versus lagged effects of VPD, thereby providing a theoretical basis for predicting mountain forest dynamics under climate warming and drying.

Method Using tree-ring width chronologies and climate data from P. schrenkiana stands across the Tianshan Mountains, we compared site-level mean chronologies among different study areas to reveal spatial variation in VPD response. At the individual-tree scale, generalized additive mixed models (GAMMs) were employed to quantify the response intensity of radial growth to VPD in the current and previous years, and random forest models were used to evaluate the relative contributions of tree age and environmental factors.

Result (1) From 1980 to 2023, all four study regions experienced significant warming (P < 0.01), while summer VPD increased highly significantly (P < 0.001). The regions showed marked differences in hydrothermal conditions, with Hami and Banfanggou being relatively warm and dry, whereas Wusu and Jimsar were relatively humid. (2) Radial growth of Picea schrenkiana and its climatic responses showed clear regional differentiation. In the warm-dry zone (Hami and Banfanggou), the tree-ring width index (RWI) was significantly negatively correlated with summer VPD (P < 0.01), whereas responses in the relatively humid zone (Wusu and Jimsar) were weaker or non-significant. (3) At the regional scale, for current-year VPD responses, the mean β values in the warm-dry zone were −0.041 and −0.055 in Banfanggou and Hami, respectively, indicating relatively strong negative responses, whereas the mean β values in the relatively humid zone (Wusu and Jimsar) were both close to zero. For previous-year VPD responses, negative responses were further strengthened in the warm-dry zone, with the proportion of individuals showing negative responses increasing to 68%–69%, indicating a stronger lagged negative effect of VPD on radial growth under warm-dry conditions. (4) At the individual-tree scale, response strategies showed marked heterogeneity. Based on the positive and negative combinations of GAMM response coefficients (β), four response types were identified. The warm-dry zone (Hami and Banfanggou) was dominated by the double-negative type (47.1% and 51.7%, respectively), with relatively concentrated response directions, whereas the relatively cold-humid zone (Wusu and Jimsar) showed a more balanced distribution of response types, with both positive and negative responses, indicating that regional climatic background not only affects overall response intensity but also regulates individual response strategies. (5) Random forest analysis showed that tree age had the highest importance (IncMSE = 0.00579), followed by elevation and climatic factors, whereas precipitation and topographic factors such as aspect and slope contributed relatively little, indicating that tree age is the primary factor underlying individual differences in VPD growth sensitivity.

Conclusion At both regional and individual-tree scales, the VPD growth response of P. schrenkiana is jointly regulated by regional hydrothermal gradients and individual attributes. Tree age constitutes the intrinsic basis determining VPD growth sensitivity, while regional hydrothermal background modulates the expression of individual response strategies. The combined effect of immediate and lagged VPD responses further exacerbates the risk of forest decline in warm-dry zones. Incorporating tree age structure and regional habitat type into forest dynamic prediction models can improve the assessment of mountain forest growth response and ecosystem stability under climate warming and drying.