Differentiated responses of larch forests to climate change along latitudinal and elevational gradients in Northeast China

Rapid climate change is reshaping forests globally, yet its impacts vary along environmental gradients, complicating predictions of forest responses. Using four decades of satellite NDVI and gridded climate data, we examined patterns of larch forest changes in Northeast China, an area experiencing a rapid warming-drying trend, along latitudinal and elevational gradients. Larch forests of this region showed an overall greening trend, but the pattern was highly structured. Weak greening and localized browning were concentrated in low-latitude, low-elevation sectors, whereas mid- to high-elevation areas and higher latitudes were characterized by stability and patchy greening. The rate of NDVI change increased with elevation at mid- and high latitudes but declined with elevation at lower latitudes. Latitudinal patterns of NDVI change were strongly elevation-dependent, with the weakest trends occurring at intermediate latitudes in low-elevation areas, a steady northward increase at mid elevations, and the strongest positive responses at intermediate-to-high latitudes in high-elevation areas. In lower latitudes and elevations, NDVI was primarily associated with temperature seasonality, growing-season maximum temperature and annual precipitation. In higher latitudes and elevations, vapour pressure deficit, precipitation seasonality and solar radiation showed the strongest influences with greening maintained within a narrow window of moderate atmospheric demand. These findings suggest that the observed spatial patterns are consistent with shifts in hydrothermal constraints along climatic gradients, alleviating energy limitation in colder environments while intensifying atmospheric and hydrological stress in warmer conditions. By identifying spatially varying climate threshold-like patterns and climate-sensitive zones, this study provides an important basis for informing climate-adaptive forest management.