Continental-scale parameterization and prediction of leaf phenology for the North American forests

Aim Leaf phenology regulates multiple aspects of plant vital activities and provides feedback to climate change. Despite its importance, an effective parameterization method to predict continental-scale leaf phenology has been elusive. Here, we aimed to develop a new parameterization method using local climatic conditions instead of species or plant functional types to calibrate the phenology parameters of forests. Location North America. Time period 1985-2016. Major taxa studied Forests. Methods We used the average temperature and photoperiod from the first day to the phenological date in a given year to express the climatic conditions. Using this parameterization method and the typical phenology models, we examined the predictions of start (SOS), end (EOS) and length (GSL) of the growing season with observations from 114 phenology cameras in North American forests. We extrapolated to the continental scale using satellite-derived leaf area index product as a basis to predict the distribution and changes of forest leaf phenology. Results We found that the optimum parameters could explain the 81%, 84% and 88% variation in SOS, EOS and GSL of the plot-scale phenology, respectively. Continental-scale results over the past decades showed that EOS was delayed by 1.6 days/decade and that GSL increased by 2.2 days/decade, but that there was no significant trend for SOS. Main conclusions Our study provides strong empirical evidence that the phenology parameters vary with local climatic conditions, and this new parameterization method has potential application for predicting and evaluating future changes in continental-scale phenology.

Automated summary

This study developed a new parameterization method to predict continental-scale leaf phenology and validated it using forest observations in North America. The results showed that phenology parameters vary with local climatic conditions, and over the past decades, the end of the growing season was delayed and the length of the growing season increased, but the start of the growing season did not show a significant trend.