Drought amplifies the impact of vegetation phenology on wildfire in Guizhou Province, China

Abstract

Vegetation phenology is highly sensitive to climate change and may influence wildfire by regulating fuel accumulation and fuel moisture. However, under increasingly frequent extreme droughts, clear evidence is still lacking regarding how phenological changes affect wildfires across seasons and whether drought amplifies this influence. Focusing on Guizhou Province, China, where vegetation restoration succeeded but wildfires remain frequent, we compiled multi-source remote-sensing and meteorological data from 2001 to 2023. We combined interpretable machine learning with structural equation modeling to identify key drivers of wildfire occurrence and disentangle “drought–phenology–wildfire” pathways. Results show that vegetation phenology in Guizhou shifted toward an earlier start of the growing season (SOS), a delayed end of the growing season (EOS), and a prolonged length of the growing season (LOS). Wildfire exhibits pronounced spatial clustering and strong seasonality: 1748 fires occurred during 2001–2023, with a cumulative burned area of 7970 km². Winter dominated burned area (5722 km², 71.8%), followed by spring (1982 km², 24.9%). Delayed SOS was associated with increased spring and winter burned area (BA), while delayed EOS and prolonged LOS were associated with greater BA in the following spring and summer, respectively. The results of structural equation modeling suggest that vegetation phenology may influence wildfire in Guizhou primarily by altering fuel accumulation and availability. Further analyses indicate that drought significantly strengthens phenological effects on wildfire intensity, with stronger phenology–burned area relationships under drought than under non-drought conditions. Overall, our findings support the coordinated management of ecological restoration and wildfire risk in Guizhou.