Species-specific phenological indicators reveal topographic–climatic associations with evergreen-tree phenology in subtropical hilly regions

Evergreen-tree phenology is an important ecological indicator of forest responses to climate change and local environmental heterogeneity. In subtropical hilly regions, frequent cloudy and rainy conditions, complex terrain, and weak seasonal greenness signals hinder species-level phenological retrieval and interpretation. Here, we evaluated the ecological indicator value of the start of season (SOS), end of season (EOS), and length of season (LOS) for eight representative evergreen tree species in Xiamen, Fujian Province, China. We reconstructed 8-day, 30-m normalized difference vegetation index (NDVI) time series from Landsat 8 and MODIS observations during 2015–2020 using the gap-filling–Savitzky–Golay (GF-SG) method, derived phenological metrics, and analyzed topographic–climatic associations using mixed-effects piecewise structural equation models (SEMs). Independent withheld-Landsat validation of reconstructed NDVI showed R 2 values of 0.581–0.671 during June–August and 0.710–0.770 during other months. SOS and EOS shared a broadly similar coastal–hilly gradient, with earlier dates in southeastern coastal and lowland areas and later dates toward the northwestern hills; LOS generally increased toward the hills. EOS exhibited broader interspecific distributions and greater interannual variability than SOS. For SOS, precipitation-effect estimates were predominantly negative across species, whereas temperature associations varied more strongly in direction and magnitude. For EOS, temperature-effect estimates were positive across all eight species, whereas precipitation associations showed greater interspecific heterogeneity; aspect showed the most directionally consistent topographic association with EOS. These patterns support EOS as a comparatively sensitive ecological indicator of species-level differentiation and late-season environmental variation in subtropical evergreen forests.