Evaluating vegetation carbon storage by primary forests in Sweden using LPJ-GUESS

Sammanfattning: Over the past 200 years, the structure of forests in Sweden has changed drastically, with forestry becoming the dominant land use. This has led to the loss of primary forests, which has major impacts on different ecosystem services, including carbon storage. Primary forests are unique ecosystems that are untouched by humans and have been sequestering carbon for centuries.The effects of this shift from no land use to land use on the vegetation carbon storage is poorly understood. The difference in carbon storage between primary and managed forests could give an indication of the effects of land use. LPJ-GUESS is a dynamic vegetation model that can estimate both potential natural vegetation and managed ecosystems. However, the ability of LPJ-GUESS to simulate potential natural vegetation has not been evaluated. Here, a unique dataset on 11 primary forests in Sweden was used to evaluate the potential natural vegetation. The vegetation carbon storage and different aspects of primary forest structure were investigated using a regression analyis and compared with bootstrapped field data. The results showed that LPJ-GUESS overestimated carbon storage, but the adjustement of the bole height ratio to 0.25, the disturbance interval to 143 years and the leaf longevity to 7 years improved the model performance. With these improvements, the model could accurately explain 40% of the variation in the field data. The improvements however negatively affected the maximum tree height and further overestimated carbon storage in spruce trees. Furthermore, the same postive results to the adjustments of the parameters were not found for primary forest data form the Swedish national forest inventory. The initial overestimation of the modelled vegetation carbon storage could be explained by the simulation of very thin trees and the inclusion of grass in the vegetation carbon storage. The improvements had a good effect on most investigated structural parameters, but the changes within parameters, especially the leaf longevity for pine, were found to impact the tree type composition. In conclusion, the results showed that LPJ-GUESS could simulate potential natural vegetation moderately well and that this method could thus be used to more efficiently estimate vegetation carbon storage in primary forests. This method could further be used to contrast carbon storage in primary forests with managed forests and the results have a general relevance when simulating natural vegetation in LPJ-GUESS.