Simulating the diameter growth responses of Larix gmelini Rupr. and Betula platyphylla Suk. to biotic and abiotic factors in secondary forests in Northeast China

AbstractKey messageThe diameter growth of Dahurian larch (Larix gmelini Rupr.) and white birch (Betula platyphylla Suk.) species in secondary forest of Northeast China was not only influenced by biological factors such as tree size and stand characteristics, but also significantly affected by topographic and climatic factors such as temperature and precipitation. It is necessary to consider the abiotic factors in simulating the diameter growth.ContextClimate change, such as global temperature rise, increased frequency of extreme weather events, and rising sea levels, has put forest ecosystems in an unstable state and has an impact on species composition, growth harvest, productivity and other functions of forests. And this impact varies in climate scenarios, regions and forest types.AimsTo gain a comprehensive understanding of the adaptation for key species to their environment in secondary forests in Northeast China, the diameter growth responses of Dahurian larch and white birch to biotic and abiotic factors were simulated to assess the effects of climate on diameter growth.MethodsChina’s National Forest Continuous Inventory (NFCI) data from 2005 to 2015 were used to develop linear mixed-effects diameter growth models with plot-level random effects, and leave-one-out cross-validation was applied to evaluate the developed models. At the beginning of modeling, correlation analysis and best-subset regression were used to analyze the correlation between the diameter increment and the biotic and abiotic factors.Results(i) Sorting the categories of predictors in descending order based on the relative importance of the significant predictors, diameter growth of Dahurian larch was affected by competition, tree size, topographic conditions, stand attributes, diversity index, and climate factors, while the white birch species was affected by competition, tree size, stand attributes, climate factors, diversity index, and topographic conditions; (ii) the plot-level mixed-effects model, which achieved better fit and prediction performance than did basic linear models of individual-tree diameter growth in the cases of prediction calibration, was preferable for modeling individual-tree diameter growth; (iii) the prediction accuracy of the mixed-effects model increased gradually with increasing size of calibration sample, and the best sampling strategy was the use of nine random trees to calibrate and make predictions with the mixed-effects model for the larch and birch