Extended duration of the budburst period under future climate warming: insights from a model

Spring phenology is a key indicator of the terrestrial ecosystems’ response to climate change. However, most phenological studies only focus on the analysis of the average date of a particular phenological event in tree populations, and largely overlook the variability of this date within the populations, resulting in large uncertainties in projecting phenological change and the stability of community under ongoing climate warming. Here, we constructed a model able to simulate the within-population variability (WPV) of budburst dates in tree populations using budburst data observed from 2000 to 2021, and we used the model to evaluate the response of WPV to climate warming in five temperate deciduous tree species (Carpinus betulus, Quercus petraea, Fraxinus excelsior, Fagus sylvatica and Castanea sativa). The WPV model received support for all five species, with a RMSE of 8.6 ± 2.9 days over validation data, which is near the observation resolution. Retrospective simulations using past climate suggested that the beginning (i.e., date at which 20 % trees burst their buds, BP20) and end (i.e., date at which 80 % trees burst their buds, BP80) of budburst in the population advanced over 1961-2021 of 1.3 ± 0.4 days decade^-1 and 1.4 ± 0.4 days decade ^-1, as a consequence of climate warming. However, the duration of the budburst period (DurBB, time interval between BP20 and BP80) did not change significantly. Using three climate models, we found BP20 and BP80 to occur later by 3.1 ± 1.3 days decade^-1 and 3.8 ± 1.5 days decade^-1 in populations of Quercus, Fraxinus and Carpinus along the 21^st century, which was caused by insufficient chilling accumulation, contrasting with a continuous trend towards earlier budburst by 0.9 ± 0.6 days decade^-1 and 0.5 ± 0.7 days decade^-1 in Fagus and Castanea. Importantly, the duration of the budburst period (DurBB) in the population was projected to increase in the future, especially for Quercus and Fraxinus, due to a stronger temperature sensitivity of the end of budburst in the population. Furthermore, our model suggests modifications at the community scale, with shifts in the budburst sequence for some species. Our work provides a novel model, simulating the continuity of budburst in tree populations in spring. This phenological model can be adapted to the study of other stages of the tree phenological cycle, which are all of continuous nature in tree populations (e.g., leaf senescence, wood formation etc.). Furthermore, based on this approach, our study projects a delayed, and extended duration of budburst in the population under climate warming for two out of the five species investigated. If confirmed in natura, these differential changes in budburst duration could influence the competition among species in forest communities.