Temperature variations impacting leaf senescence initiation pathways alter leaf fall timing patterns in northern deciduous forests

Abstract

Simulating the timing of leaf fall in large scale is crucial for accurate estimation of ecosystem carbon sequestration. However, the limited understanding of leaf senescence mechanisms often impedes the accuracy of simulation and prediction. In this study, we employed the advanced process-based models to fit remote sensing-derived end dates of the growing season (EOS) across deciduous broadleaf forests in the Northern Hemisphere, and revealed the spatial pattern associated with two leaf senescence pathways (i.e., either photoperiod- or temperature- initiated leaf senescence) and their potential effects on EOS prediction. The results show that the pixel-specific optimum models effectively fitted all EOS time series. Leaf senescence in 67.6 % and 32.4 % of pixels was initiated by shortening daylength and declining temperature, respectively. Shortening daylength triggered leaf senescence occurs mainly in areas with shorter summer daylength and/or warmer autumns, whereas declining temperature induced leaf senescence appears primarily in areas with longer summer daylength and/or colder autumns. The strong dependence of leaf senescence initiation cues on local temperature conditions implies that the ongoing increase in autumn temperature has the potential to alter the leaf senescence initiation, shifting from temperature cues to photoperiod signals. This shift would occur in 26.2-49.6 % of the areas where leaf senescence is initiated by declining temperature under RCP 4.5 and 8.5 scenarios, while forest areas where leaf senescence is induced by shortening daylength may expand northward. The overall delaying of the currently predicted EOS would therefore slow down by 4.5-10.3 % under the two warming scenarios. This implies that the adaptive nature of plants will reduce the overestimation of changes in carbon exchange capacity between ecosystems and atmosphere. Our study offers novel insights into understanding the mechanism of leaf senescence and improving the estimation of autumn phenology and ecosystem carbon balance in the deciduous broadleaf forests.

Keywords: Autumn phenology prediction; Deciduous broadleaf forests; Leaf senescence; Photoperiod; Process-based modeling; Temperature.