Trends and uncertainties in budburst projections of Norway spruce in Northern Europe.

Cecilia Olsson,Johan Lindström,Stefan Olin,Anna Maria Jönsson

doi:10.1002/ece3.3476

Abstract

Budburst is regulated by temperature conditions, and a warming climate is associated with earlier budburst. A range of phenology models has been developed to assess climate change effects, and they tend to produce different results. This is mainly caused by different model representations of tree physiology processes, selection of observational data for model parameterization, and selection of climate model data to generate future projections. In this study, we applied (i) Bayesian inference to estimate model parameter values to address uncertainties associated with selection of observational data, (ii) selection of climate model data representative of a larger dataset, and (iii) ensembles modeling over multiple initial conditions, model classes, model parameterizations, and boundary conditions to generate future projections and uncertainty estimates. The ensemble projection indicated that the budburst of Norway spruce in northern Europe will on average take place 10.2 ± 3.7 days earlier in 2051–2080 than in 1971–2000, given climate conditions corresponding to RCP 8.5. Three provenances were assessed separately (one early and two late), and the projections indicated that the relationship among provenance will remain also in a warmer climate. Structurally complex models were more likely to fail predicting budburst for some combinations of site and year than simple models. However, they contributed to the overall picture of current understanding of climate impacts on tree phenology by capturing additional aspects of temperature response, for example, chilling. Model parameterizations based on single sites were more likely to result in model failure than parameterizations based on multiple sites, highlighting that the model parameterization is sensitive to initial conditions and may not perform well under other climate conditions, whether the change is due to a shift in space or over time. By addressing a range of uncertainties, this study showed that ensemble modeling provides a more robust impact assessment than would a single phenology model run.

Highlights

Compared with the global average, climate warming is expected to be higher during winter months, and more pronounced further north and in mountainous regions, such as in the Alps (IPCC, 2013)
The variations captured by the ensemble projections were primarily caused by differences among phenology model classes (MC), secondly, by the initial conditions used for model parameterization (IC), and lastly, by climate model data (BC)
We applied (i) Bayesian inference to estimate model parameter values to address uncertainties associated with selection of observational data, (ii) climate data selection to identify a subensemble of climate model data representative of a larger dataset, and (iii) ensembles modeling over multiple initial conditions, model classes, model parameterizations, and boundary conditions to generate future projections and uncertainty estimates

Summary

Introduction

Compared with the global average, climate warming is expected to be higher during winter months, and more pronounced further north and in mountainous regions, such as in the Alps (IPCC, 2013). Plant spring phenology is highly tuned to winter and spring temperatures and is a good indicator of climate change. Many plants have responded to the recent warming by becoming active earlier in the year, but the climate change response varies among species and locations and depends on the time period considered (Ahas, Aasa, Menzel, Vg, & Scheifinger, 2002; Menzel & Fabian, 1999; Menzel et al, 2008). As frost hardiness in spring is negatively related to growth activity (Westin, Sundblad, Strand, & Hällgren, 2000), an earlier onset of the growing season may increase the risks and severity of frost damage during late spring cold spells (Jönsson & Bärring, 2011). For commercially important species like Norway spruce, for which large differences in phenology traits exist among provenances, comprehensive cultivation research is carried out to identify traits favorable in a warmer climate (e.g., Skrøppa & Steffenrem, 2016; Westin et al, 2000)

Objectives

Findings

Discussion

Conclusion