A machine learning-based framework for evaluating photosynthetic regulation and carbon sequestration capacity among shelter forest species

The middle section of the Yinshan mountains in Inner Mongolia, as a representative area of the Three-North Shelter Forest Program, faces critical challenges in enhancing ecosystem carbon sequestration functions. This study aims to investigate the diurnal variation characteristics of photosynthesis and water use efficiency (WUE) in arbor and shrub species in the Middle Section of the Yinshan Mountains, analyze the influence of environmental factors on photosynthetic regulation, and comprehensively evaluate the ecological adaptability and carbon sequestration functions of different tree species. For the first time, a combination of random forest regression models and cluster analysis is applied to provide scientific insights for improving shelter forest ecosystem functions.

Seven arbor species and five shrub species were selected in the middle section of the Yinshan mountains, Inner Mongolia. Photosynthetic parameters were measured using a CI-340 portable photosynthesis system. Spearman's correlation analysis, principal component analysis (PCA), random forest regression modeling, and hierarchical cluster analysis were employed to explore the photosynthetic mechanisms and ecological adaptability of the species.

The study revealed distinct diurnal variation patterns in photosynthetic activity among arbor and shrub species. Net photosynthetic rate (P_n) and WUE exhibited significant interspecific differences, with single-peak and double-peak patterns observed. Temperature, stomatal conductance, and transpiration rate were identified as key regulators of photosynthesis, particularly in coniferous species, which demonstrated higher WUE. Random forest regression analysis highlighted the critical roles of temperature and water management in photosynthetic processes. Cluster analysis further revealed marked differences in carbon sequestration capacity and ecological adaptability among species, offering critical guidance for species selection and carbon sequestration assessments.

This study underscores the significant differences in photosynthesis and WUE between arbor and shrub species in the middle section of the Yinshan mountains. Temperature, stomatal conductance, and transpiration rate were pivotal in photosynthetic regulation. Unlike traditional linear regression methods, the random forest model effectively captured the nonlinear relationships between environmental variables and photosynthetic rates. The hierarchical cluster analysis, applied for the first time in this context, identified distinct functional groups of species based on their carbon sequestration potential and ecological adaptability. These findings fill a critical gap in understanding photosynthetic mechanisms and carbon sequestration capacity within the Three-North Shelter forest program, providing a novel framework for quantitative carbon sequestration assessment and informing future afforestation and restoration strategies.