A transiting temperate-subtropical mixed forest: carbon cycle projection and uncertainty

Terrestrial ecosystems respond to climate change in various ways, making it crucial to improve our understanding of these dynamics and uncertainty in projections. Here, we investigate how the species composition in a temperate-subtropical mixed forest on Jeju Island, South Korea, would change by 2099 and analysed the resultant effects on phenological timings and carbon flux using an individual cohort-based model—the ecosystem demography biosphere model version 2. We use the analyses of variance to decompose the contribution of model parameters (four sets) and climate inputs (four global climate models under four representative concentration pathway (RCP) scenarios) to the total uncertainty in the leaf area index (LAI) and net ecosystem productivity (NEP) projections. We find that with increases in temperature, photosynthetically active radiation, and vapour pressure deficit, the dominance of subtropical species will gradually increase by approximately 11%, from 30.2% in 2013 to 41.1% by the end of this century, yet there was a large variation in the projections depending on the model parameter and climate inputs. We also show the increases in the LAI and length of growing season by the end of this century, resulting in an increased NEP at the rate of up to 62.7 gC m⁻² yr⁻¹ per decade under the RCP8.5. The uncertainty in the LAI projection was largely due to the model parameter (and its interaction with climate inputs); however, the uncertainty contribution of climate models is as large as the emission scenario in the NEP projection. This study highlights the importance of identifying uncertainty sources for a robust projection of terrestrial ecosystem and carbon cycle.