Strong temperature response in spring and autumn phenological transition dates in response to whole-ecosystem warming

Abstract:

Predicting vegetation phenology in response to changing environmental factors is key in understanding feedbacks between the biosphere and the climate. Experimental approaches that extend the temperature range beyond interannual climate variability can be used to constrain models to predict phenological changes under future climate scenarios. Here, we track and model spring and autumn phenological transition dates in a boreal Picea-Sphagnum bog in response to a gradient of whole ecosystem warming (up to +9°C) using digital repeat photography from the PhenoCam method.

Over the course of four years, warming treatments advanced spring green-up at a rate of 1.4 days per degree Celsius in all three different functional plant types (deciduous needle leaf, evergreen needle leaf, and an ericaceous shrub layer). Autumn senescence showed an even stronger response to warming (senescence delayed by 2.3 days per degree Celsius) and together with earlier leaf-out the growing season was extended up to 5 weeks in the warmest treatment. We applied a suite of 20 simple spring phenology models and one autumn model and assessed best model fit to observed data using the Akaike Information Criterion. For each plant function type, the best spring model included eight parameters and showed a difference between observed and modeled data of four to six days (based on root mean squared errors). The three parameter autumn model accounted for chilling degree days and predicted transition dates well with a root mean square error of 6 days.

We show with this modeling approach that the experimental warming treatment together with interannual variation in climate is a powerful tool to constrain phenology models making them more robust for phenology predictions under future climate scenarios.