New insights on the link between phenology and productivity of temperate and boreal broadleaf deciduous forests across the globe

Shifts in the timing of key phenological phases in plants have been indentified as useful indicators to track the impact of ongoing climate change. Phenological responses themselves provide important feedbacks to the climate system as they exert control over most above- and belowground ecosystem processes and productivity. Recent advances in near-surface remote sensing enable automated and consistent observation of vegetation status at spatial and temporal scales suitable for integration with eddy covariance measurements. This allows us to gain phenological understanding of biosphere-atmosphere energy, water vapor and trace gas exchanges and thus productivity. Several previous studies have investigated the link between key phenological phases in temperate and boreal forests and their productivity across sites and years. Most of these studies focused on shifts in the timing of spring phenological events such as budburst. Few studies, however, have investigated the importance of the rates of changes across key phenological phases such as leaf development and senescence in spring and autumn, respectively, as explanatory variables of observed interannual and across-site variability in forest productivity. To shed light on this question we first optimized and evaluated a popular bioclimatic index (the growing season index of Jolly et al., 2005) that integrates known controls on temperate and boreal forest canopy development (i.e., air temperature, photoperiod, soil water balance) with multiple years of digital repeat photography at different sites from the PhenoCam network. Next, using meteorological and eddy covariance measurements from 24 temperate and boreal broadleaf deciduous forest sites across the globe (113 site years) as provided by the FLUXNET 'La Thuile' data set, we characterized continuous canopy development with the optimized bioclimatic index and derived spring and autumn phenological dates and rates and annual integrals of net ecosystem productivity (NEP) and gross ecosystem productivity (GEP). Preliminary results suggest that growing season length together with rate of senescence might be more important to explain observed interannual and especially across-site variability in NEP and GEP than the timing of budburst.