Nonlinear effects of forest landscape patterns on carbon storage and their driving factors in Hunan Province

To explore and evaluate the nonlinear response relationship and driving mechanisms between forest landscape patterns and carbon storage in Hunan Province, and to optimize forest management and enhance regional carbon sequestration capacity.

Based on land use data from 2000 to 2020 in Hunan Province, forest landscape patterns were assessed using the Fragstats and moving window method. The Invest model was used to evaluate the spatiotemporal variation characteristics of carbon storage. On this basis, the random forest machine learning method was applied to clarify the nonlinear impact of forest landscape patterns on regional carbon storage. Finally, the XGBoost-SHAP model was employed to analyze the driving intensity of natural and socio-economic factors on carbon storage.

1) From 2000 to 2020, forest patch density in Hunan Province increased from 1.2 to 1.4, while the interspersion and juxtaposition index increased from 55.7 to 59.5. This indicates intensified forest landscape fragmentation and more pronounced spatial interspersion among different landscape types; 2) The total carbon storage in Hunan Province decreased from 2.097 billion tons to 2.086 billion tonnes, with the spatial distribution of carbon storage being extremely uneven, concentrated mainly in the southern and western regions of Hunan. The total carbon storage in the Changsha-Zhuzhou-Xiangtan urban agglomeration continued to decline; 3) Shannon's diversity index, patch density, and contagion index positive correlations with total carbon, aboveground, and soil carbon storage; 4) GDP, temperature, and population density are the main factors driving the changes in unit carbon storage, with GDP having the most significant positive driving effect (SHAP = 0.61).

This study emphasizes that optimizing forest landscape structure and increasing economic investment can sustainably enhance regional carbon storage capacity. The findings provide a scientific basis for understanding the dynamic relationship between forest landscape pattern evolution and carbon storage, as well as for forest management practices.