Effect of the Grain-for-Green Program on forest fragmentation

Fragmentation profoundly alters the character and pattern of forests, influencing biodiversity in the context of global climate change. Forest area density (FAD) is currently considered the most effective method for characterizing forest fragmentation. Therefore, reasonable and accurate quantification of FAD and its relationship with forest cover changes are essential for evaluating forest fragmentation in the Loess Plateau under the implementation of the Grain-for-Green Program (GGP). Based on land use/cover (LULC) in 1998-2022, we integrated the Random forest (RF) algorithm and the conversion of land use and its effects at small regional extent (CLUE-S) model to derive LULC in 2030. Subsequently, we proposed a multi-scale window threshold to evaluate forest fragmentation by constructing nonlinear models. Furthermore, we explored the relationship between FAD and forest cover changes by spatial regression models. The results are as follows: (1) in 1998-2030, the main conversion of LULC was of the cropland to broad-leaf forests and grassland. (2) The 43 x 43 window served as the optimal multi-scale threshold to assess forest fragmentation. (3) The GGP slowed down forest fragmentation. (4) The degree of fragmentation in broad-leaf forests was higher than that in coniferous forests. (5) Spatial error model (SEM) was the most appropriate for establishing the spatial relationship between forest fragmentation and forest cover changes. In conclusion, since the implementation of the GGP, forest cover changes had a significant impact on forest fragmentation and its multi-scale threshold. This research provides a basis for the reasonable configuration of land resources in the Loess Plateau and novel methods for studying forest cover changes and fragmentation.