Abstract: Accurate prediction of plants’ flowering onset date (FOD) is vital for maintaining ecosystem functions and boosting forestry economic gains. While the Spring Warming (SW) model is commonly used to predict flowering phenology, its traditional fixed setting of the heat accumulation threshold (HAT), measured by growing degree days (GDD), fails to account for the spatial variation in preseason thermal requirements reported in previous studies. This limitation reduces the accuracy of FOD predictions across large spatial areas. In this study, we hypothesized that the HAT in the SW model varies spatially with habitat-specific temperature due to thermal acclimation. To test this, we systematically quantified the spatial differences in HAT using observed FOD data of Robinia pseudoacacia, which is a keystone species for afforestation and a vital nectar source, from 58 stations across China between 1963 and 2008. We identified the key temperature variables influencing HAT variability and developed a simplified, spatially dynamic HAT scheme. The updated SW model, incorporating this variable HAT, was evaluated with cross-site FOD observations. Results showed significant variation of HAT across different climate zones. A geodetector analysis found that the mean temperature from February to May was the main factor driving HAT heterogeneity, supporting our hypothesis. Additionally, spatial factors such as elevation and longitude also contributed to HAT variation alongside thermal factors. Incorporating this spatially variable HAT, predicted from preseason temperatures, into the SW model significantly improved FOD prediction accuracy, decreasing the root mean square error (RMSE) by 11.91% compared to a model with a constant HAT. Future climate scenario predictions indicated that the SW mode with the fixed HAT underestimated FOD advances in warmer areas and overestimated the rate of change, especially when compared to the heterogeneous HAT model. Overall, we emphasize the importance of considering spatial thermal acclimation in broad-scale flowering onset predictions.

Key words: Flowering onset, Thermal acclimation,  , Spring warming model, Heat accumulation threshold, Spatial heterogeneity, Robinia pseudoacacia