FAREAST: a forest gap model to simulate dynamics and patterns of eastern Eurasian forests

Aim  This paper describes the forest gap model, FAREAST, its testing and its application to simulating the distribution, composition and dynamics of forests in eastern Eurasia. Location  The FAREAST model is tested in north‐eastern China, initially for forests on the elevational gradient of Changbai Mountain, which is located on the border of the People's Republic of China and the Democratic People's Republic of Korea. Subsequently, the model is inspected regionally for other northern Chinese mountains and, finally, it is applied to predict subcontinental forest communities in the Russian Far East. Boreal larch ( Larix spp.) forests cover much of the 6 million km 2 of eastern Eurasia. Mixed broad‐leaved tree species/Korean pine ( Pinus koraiensis ) forests and spruce/fir ( Picea / Abies ) forests also occupy considerable areas. Methods  The model is tested using three types of information: (1) direct species composition comparisons between simulated and observed mature forests at the same locations; (2) forest type comparisons between simulated and observed forests along altitudinal gradients of several different mountains; and (3) comparison with forest stands in different succession stages of simulated forests. Results  Model comparisons with independent data indicate that the FAREAST model is capable of representing many of the broad features of the forests of north‐eastern China. After regional model validation in the north‐eastern region of China, geographical model applications were developed for the forests of the Russian Far East. In simulations at 31 different sites distributed across the entire Russian Far East and including a wide variety of natural forests, the model demonstrates an ability to reproduce observed vegetation pattern. The model simulations are correct with respect to our criteria for 23 of the 31 sites, and there are close results for three other sites. Among the five sites that are incorrectly predicted, four simulations can be corrected by adding a simple assumption to the model for permafrost effects on water balance. Main conclusions  Continental‐scale forest cover can be simulated using a forest gap model to represent individual–plant interactions with one another, and their environment, and with parameters that describe the biology of each tree species. It appears that such a model, validated relatively locally (in this case, in north‐eastern China), can then be applied over a much larger region. These results further imply that the Russian Far East forests can be regarded as a natural geographical expansion of north‐eastern Chinese forests. In both regions, forests share not only similar species compositions, but also similar underlying causes of forest successional dynamics.