Assessment of vegetation vulnerability in floodplain wetlands: A perspective from carryover effect of seasonal growth under various extreme hydrological scenarios

• Exogenous and endogenous memory jointly determine wetland vegetation growth. • A multi-feature decision tree was proposed for mapping floodplain wetlands. • Hydrological variability has lagged and cumulative effects on floodplain wetlands. • Seasonal VGC indicates vegetation vulnerability in diverse scenarios. • Wetland vegetation vulnerability varies across extreme hydrological scenarios. Floodplain wetlands, which are critical for ecosystem health and human well-being, are increasingly threatened by intensified hydrological variability and extreme hydrological events. However, it remains unclear how floodplain wetlands respond to these hydrological changes. Here, from the perspective of both endogenous and exogenous memory of vegetation, we explored the response of Poyang Lake Wetland (PYLW) to multi-timescale hydrological dynamics. First, we applied a dynamic threshold method to extract land surface phenology from 2011 to 2020, subdividing the year into four sub-seasons. Next, based on wetland vegetation mapping, the Carnegie-Ames-Stanford Approach (CASA) was used to simulate monthly net primary productivity (NPP). Then, with the NPP and inundation frequency time-series data, we assessed the time-lagged and cumulative response of PYLW vegetation to hydrological variability (exogenous memory) through Pearson rank correlation analysis. Subsequently, we employed partial correlation analysis, with the control of critical temporal hydrological variability, to evaluate the seasonal vegetation growth carryover (VGC) effect (endogenous memory). Finally, we proposed to use the seasonal VGC effect for modelling vegetation vulnerability under various extreme hydrological scenarios. The results reveal that the time-lagged and cumulative effects of hydrological variability on vegetation growth in PYLW reached the peak averagely after 6.51 and 7.08 months, respectively. The extreme hydrological scenarios in PYLW were categorized into three types of flood-only, flood-after-drought, and flood-before-drought. In the flood-after-drought scenario, vegetation generally showed high vulnerability, and the most vulnerable vegetation type varied across different scenarios. Our findings provide effective support for vegetation restoration, hydrological management, and biodiversity conservation in floodplains.