Abstract
Introduction Seasonal canopy changes in subtropical evergreen broad-leaved forests are subtle, making it challenging to quantify phenological dynamics and their climatic drivers. Methods Using phenology camera ( PhenoCam ) imagery from evergreen broad-leaved trees in Jiangxi Province, China, collected in 2024, together with environmental variables spanning 2023 and 2024, four vegetation indices - the Normalized Difference Vegetation Index ( NDVI ) and RGB-based chromatic indices, including Green Chromatic Coordinate ( G cc ), Red Chromatic Coordinate ( R cc ), and the derived Red-Green Vegetation Index ( RGVI ) - were extracted to evaluate their performances in resolving phenophases and to identify environmental controls. Multiple meteorological variables were reduced to a small set of independent climatic variables using principal component analysis, and related to latent phenological indicators via Pearson correlation analysis. Phenological transition dates were estimated by fitting a double Logistic model to the time series data, after which the effects of lagged environmental variables on each transition were quantified. Results The analysis shows that R cc and RGVI most closely tracked seasonal climate variation and outperformed NDVI and G cc for phenophases detection. Canopy dynamics were primarily associated with radiation, air temperature, moisture availability and atmospheric pressure. Based on RGVI double logistic fitting, the growing season commenced in early April (day-of-year ( DOY ) 98), peaked by late April ( DOY 116) and commenced a significant decline in early December ( DOY 340), spanning a growing-season length of 242 days. The timing of phenological events showed clear carry-over effects: prior-season environmental anomalies exerted lagged influences on subsequent canopy development, after certain threshold conditions were exceeded. Discussion Overall, our findings clarify that near-ground PhenoCam s provide sensitive, scalable indicators of evergreen canopy dynamics in the subtropics. Red-Green Vegetation Index offers reliable phenophase detection, and incorporating cross-season lag effects will improve the understanding of phenological mechanisms in evergreen ecosystems.