In recent decades, the global climate has changed significantly. The climate in Northwest China became warm-wet, especially in the Tianshan Mountains. In order to explore the response of tree growth to recent climate change, the two dominant trees species, Picea schrenkiana Fisch. et Mey. and Larix sibirica Ledeb., were studied with the dendrochronological method in the western Tianshan Mountains (WT) and the eastern Tianshan Mountains (ET). Our results showed that: (1) The tree growth of four sample sites in the WT significantly increased in recent decades, while the trees in the three sample sites in the ET significantly decreased. (2) In the WT, except for the Manas site, the tree-ring chronologies of the other three sites were significantly positively correlated with the mean annual minimum temperature. Tree-ring chronologies in the WT, except for Bangfanggou site, were significantly positively correlated with annual precipitation. In the ET, only the tree chronology of L. sibirica in the Balikun site was significantly negatively correlated with the annual temperatures, including the mean minimum, mean and mean maximum temperature. (3) The proportion of trees with a significant upward growth trend at each site decreased from west to east, and the proportion of trees with a significant downward growth trend at each site increased from west to east along the whole Tianshan Mountains. (4) The correlation of tree-ring chronologies with the annual temperature and annual precipitation was not stable during the study period. Warm-humidification promoted the growth of trees in the WT but inhibited tree growth in the ET, which may be exacerbated drought stress in the ET where the increase in precipitation was not enough to offset the increased evapotranspiration potential caused by warming.

The increasing frequency of both singular and compound extreme climate events driven by global warming has profoundly impacted terrestrial ecosystems. Using machine learning-based Random Forest algorithms and moving correlation analysis, this study quantifies the impacts of extreme climate indices (ECIs) on two ecological indicators (EIs), the NDVI and GPP, from 1982 to 2019. The results reveal that singular extreme climate events exert a more pronounced influence on ecosystems across Southwestern China (SWC) than compound ones. Specifically, the NDVI and GPP exhibited strong correlations with summer days (SU) and diurnal temperature range (DTR), with SU contributing positively (weight = 0.275 for the GPP and 0.238 for the NDVI) and DTR negatively (weight = 0.107 for the GPP and 0.130 for the NDVI). Regional analyses highlighted distinct spatial patterns: in mid–high-altitude areas (>1 km), including the Hengduan Mountains (HDMs) and Yunnan–Guizhou Plateau (YGP), extreme temperatures and precipitation significantly promoted vegetation growth, with rainfall day index (RDI), frost days (FD), extreme temperature index (ETI), SU, and DTR all having a strong influence (>0.1) on the GPP and NDVI. These areas showed strong adaptability to extreme climate, benefiting overall vegetation health. In contrast, ecosystems in low-altitude areas (<1 km) showed more variable responses. The Guangxi Basin (GXB) exhibited strong resistance to ECIs, with vegetation being almost unaffected by extreme precipitation and benefiting from continuous warming. Only consecutive wet days (CWD) and FD were significantly negatively correlated with EIs (p < 0.05), and their correlation weights were low (weights = 0.043 and 0.013). However, the vegetation in the Sichuan Basin (SCB) is more susceptible to climate extremes, which have particularly strong effects on the NDVI. SU, tropical nights (TR), ETI, and growing season length (GSL), which have positive effects on EIs in mid–high-altitude areas, show extremely significant negative correlations in the SCB (p < 0.001), and their weights account for one-third of the total (weights = 0.15, 0.11, 0.061 and 0.012, respectively). These findings underscore the heterogeneous responses of ecosystems to ECIs and emphasize the need for region-specific strategies in ecosystem management and disaster prevention amid climate change.

Compound extreme climate events (CECEs) are intensifying under global warming, yet their ecological impacts and interaction with land system changes remain unclear. This study employs a copula-based framework to construct compound extreme climate indices (CECIs), including the warm-wet index (CWWI) and cold-wet index (CCWI), in China's Loess Plateau (LP). Combined with ecological indicators (EIs, NDVI and GPP included) and land use data, the spatiotemporal dynamics of CECIs and their ecological effects were assessed via multiple linear regression during 2000–2019. CWWI increased significantly (p<0.01), indicating regional warm-humidification. Vegetation exhibited strong seasonal responses to CECIs: 86.25% of the area showed a positive correlation between spring GPP and CWWI, and 55.81% for NDVI. Spring CWWI significantly explained 71.88% of summer GPP and 40.63% of NDVI variation, indicating robust ecological memory. CECIs outperformed single indices (precipitation/temperature) in over 40% of regions, particularly for GPP, where explanatory power improved by up to 0.15 in R² values. Although human activities played a leading role in several zones undergoing intensive ecological restoration, compound hydrothermal variability remained the key driver of vegetation growth and productivity. Land use conversion, mainly from cropland to grassland, further enhanced vegetation cover and modulated local climate, forming a feedback loop that strengthens the coupling between ecosystem processes and CECEs. Our study highlights the effectiveness of compound indices in capturing ecological stress and provides insights into seasonal and lagged vegetation responses under climate extremes. The proposed framework offers guidance for ecological restoration and climate adaptation in semi-arid and climate-sensitive regions.