Ensemble empirical mode decomposition for analyzing phenological responses to warming

Numerous studies conducted over the past decade have revealed that plant phenophases have shifted in many temperate ecosystems. Although the consensus is that these shifts reflect plant responses to rise in temperatures, we have yet to match unequivocally the phenological and temperature trends. More importantly, little is known about warming's effects on and contributions to phenophase variability. The key to accomplishing both tasks lies in a proper separation of a trend from natural variability. Based on ensemble empirical mode decomposition, this study shows that, over the past 30 years, the advancing trends in the first flowering dates (FFD) of apple (Malus domestica) in Austria and blackthorn (Prunus spinosa) in Germany unequivocally correspond to the respective regional winter/spring warming trends. The variability of both FFD series before 1981 was almost entirely due to natural variability. In contrast, warming since 1981 contributed 4% and 21% toward the total phenophase variability of apple and blackthorn, respectively. Furthermore, while contributing to both the temperature and FFD overall variability, recent warming also lowers the FFD natural variability by modulating the temperature natural oscillation amplitudes. Thus, warming can affect both the timing and the natural variability of a phenophase development. As concurrently shifting the timings and reducing the natural variability of phenophase developments may have important ecological and evolutionary consequences, it will be of great interest and importance to examine whether the conclusion holds for other phenophases and species in various regions as well. The introduced approach will be a valuable tool for answering the question. (C) 2014 Elsevier B.V. All rights reserved.