Understanding the potential response of phenology to future warming through merging models and observations at multiple scales
The phenology of deciduous forests is poorly represented in the current generation of land surface models. In part this stems from a lack of agreement between studies about the mechanisms and pathways through which environmental forcings control spring onset and autumn senescence, and the resulting plethora of model formulations. A multitude of different factors (such as the timing of temperature effects, photoperiod, latitude, chilling, and dormancy requirements) all find some support in the literature. Reports also suggest that the controlling factors vary by species and perhaps by location or population. To date, an understanding of the nature of the controls of spring onset, and thus an effective predictive model, has remained elusive. Here, we examine the dominant controls of spring phenology in extra-tropical deciduous forests of the northern hemisphere. To do so, we use satellite observations of the timing of spring onset, merged with weather reanalysis data, eddy-covariance observations from a network of sites, and ground observations of bud burst for 27 species.
We find a convergent temperature sensitivity of spring phenology for the extratropical deciduous forests of the northern hemisphere. Using multiple modeling approaches, we show that the observed convergence between the temporal and spatial sensitivity of spring phenology to temperature suggests that the response of spring phenology to future climate warming may be accurately predicted for the majority of current extra-tropical winter deciduous forests. Our results imply a large potential expansion of the spring growing season under future climate warming. Our analysis suggests that this should increase rates of carbon uptake from the atmosphere, representing a negative feedback to climate change.