LAUSR

Abstract Autumn phenology is important in determining the growing season length and controlling carbon and energy exchanges in terrestrial ecosystems. However, our knowledge on the interaction processes of vegetation autumn phenology and climate changes remains limited, especially for herbaceous plants. In this study, we comprehensively analyzed the responses of autumn phenology of grassland vegetation to climate changes by using ground-observed brown-down date records of 15 grass species and daily temperature, precipitation, and day length data at six stations. Aside from conducting correlation analysis, we also simulated the brown-down date with a newly developed model by incorporating the effect of drought stress (CDDP) into the traditional chilling-degree-days (CDD) model and compared it with the CDD model. Another revised CDD model included the effect of day length (CDDD) and null model (multiyear average, NM). The statistical results showed a predominant significant negative correlation between the brown-down date and previous temperature/day length in 27.3%/40.9% of site species but a predominant significant positive correlation between the brown-down date and previous precipitation in 54.6% of site species. The opposite effects of previous precipitation and previous temperature/day length on the brown-down date were induced by local thermal–moisture conditions. With regard to the modeling results, the CDDP model was selected as the optimal model for 73% of site species with insufficient water supply in preseason, while the CDD model was selected as the optimal model for 18% of site species with a relatively wet but cold preseason. The CDDD model was selected as the optimal model for only two cases. The average estimation error based on the CDDP model (7.4 days) was lower by 2.0/1.5/1.7 days than that based on the CDD/CDDD/NM model. Overall, this study comprehensively demonstrated the important role of water availability in controlling the autumn phenology process of herbaceous plants.