Decomposing sources of uncertainty in climate change projections of boreal forest primary production

We are bound to large uncertainties when considering impacts of climate change on forest productivity. Studies formally acknowledging and determining the relative importance of different sources of this uncertainty are still scarce, although the choice of the climate scenario, and e.g. the assumption of the CO2 effects on tree water use can easily result in contradicting conclusions of future forest productivity. In a large scale, forest productivity is primarily driven by two large fluxes, gross primary production (GPP), which is the source for all carbon in forest ecosystems, and heterotrophic respiration. Here we show how uncertainty of GPP projections of Finnish boreal forests divides between input, mechanistic and parametric uncertainty. We used the simple semi-empirical stand GPP and water balance model PRELES with an ensemble of downscaled global circulation model (GCM) projections for the 21st century under different emissions and forcing scenarios (both RCP and SRES). We also evaluated the sensitivity of assumptions of the relationships between atmospheric CO2 concentration (Ca), photosynthesis and water use of trees. Even mean changes in climate projections of different meteorological variables for Finland were so high that it is likely that the primary productivity of forests will increase by the end of the century. The scale of productivity change largely depends on the long-term Ca fertilization effect on GPP and transpiration. However, GCM variability was the major source of uncertainty until 2060, after which emission scenario/pathway became the dominant factor. Large uncertainties with a wide range of projections can make it more difficult to draw ecologically meaningful conclusions especially on the local to regional scales, yet a thorough assessment of uncertainties is important for drawing robust conclusions.