Quantifying driving factors of vegetation carbon stocks of Moso bamboo forests using machine learning algorithm combined with structural equation model

Moso bamboo (Phyllostachys pubescens) is widely distributed in subtropical China and plays an important role in carbon cycling in terrestrial ecosystems. Knowledge of the main driving factors that affect aboveground carbon stocks in Moso bamboo forests is needed to increase carbon sequestration potential. We used a large-scale database from national forest continuous inventory from 2004 to 2014 in Zhejiang, China, and combined Random Forest analysis (RF) with structural equation modeling (SEM) to quantify the contribution of main driving factors on vegetation carbon stocks in Moso bamboo forests, and to evaluate the direct and indirect total effects of the main driving factors on aboveground carbon stocks, as well as to investigate changes in the standardized total effects from 2004 to 2014. The RF model explained 84.9% of the variation in 2004, 78.8% in 2009, and 82.2% in 2014. The SEM that included the average age, average diameter at breast height (DBH), culms density, mean annual temperature (MAT), and mean annual precipitation (MAP) explained 92.5% of the variation in 2004, 88.2% in 2009, and 88.6% in 2014. The results showed that average age, average DBH, culms density, MAT, and MAP were the most crucial driving factors of vegetation carbon stocks in Moso bamboo forests. The values of standardized total effects of the main driving factors showed that the average age, average DBH, and culms density had positive effects on vegetation carbons stocks, whereas MAT and MAP had negative effects. Furthermore, the positive effects of average age on the increase of vegetation carbon stocks increased significantly, the negative effects of MAT increased with the increasing MAT, but the negative effects of MAP decreased with the increasing MAP. Overall, our study provided new insights into the sensitivity and potential response of carbon sequestration in Moso bamboo forests to structural development and climate change in Zhejiang Province.