As global biodiversity loss driven by land use and land cover (LULC) changes have become a pressing challenge, forest management plays a central role in both causing and mitigating this trend. One area that remains insufficiently explored is the spatiotemporal response of bird populations to forest dynamics. This study investigates forest dynamics and landscape pattern changes in Yunnan Province, China, from 2005 to 2020, and assesses their impacts on the habitats of 15 bird population categories and 81 species. By integrating species distribution models (SDMs) and the Mann-Kendall (MK) trend test, this study reveals the complex coupling mechanisms between forest and bird. As forest area expansion is accompanied by the fragmentation of shrubland and grassland, leading to reduced habitat for closed shrubland (CS) and migratory birds (M), the opposite effect is observed in the fragmentation of agricultural land. Positive effects of forest expansion were observed in areas with high pre-existing forest cover (82.64 %) and limited built-up land (0.09 %). However, a negative correlation was found when the original forest cover reached 88.31 %, underscoring a threshold-dependent, non-linear relationship between LULC composition and bird habitat suitability. Additionally, Fractional Vegetation Cover (FVC), and Leaf Area Index (LAI) effectively assess bird habitat quality, though they are less reliable for certain species, such as raptors. This study provides insights and theoretical guidance for other regions implementing forest management to mitigate biodiversity loss risks.
Low-carbon transition has become a key orientation for urban renewal, and urban green spaces (UGS) in high-density built-up areas function as near-source carbon sinks closely coupled with high-emission spaces. However, UGS carbon sink performance exhibits pronounced spatiotemporal heterogeneity shaped by built-environment variability and seasonal processes, while existing studies still lack an integrated framework linking spatial differentiation, seasonal dynamics, and underlying mechanisms. The local climate zone (LCZ) scheme provides a standardized, fine-grained classification that integrates urban morphology, land cover, and local environmental conditions. To address this gap, this study applied the LCZ framework to UGS carbon sink analysis in central Beijing and developed a comprehensive ”spatial stratification–carbon sink estimation–mechanism identification” framework. Net ecosystem productivity (NEP) was estimated using an optimized CASA model coupled with soil heterotrophic respiration, and multi-source data, machine learning, SHAP analysis, and pathway modeling were integrated to identify key drivers and nonlinear mechanisms. NEP varied significantly across LCZ classes, generally decreasing from natural LCZ classes to high-density built LCZ classes. High-density built LCZ classes were more sensitive to seasonal environmental stress, showing reduced carbon sink performance during the growing season and even temporary shifts to carbon sources. Vegetation condition was identified as the proximal and direct driver of NEP, whereas hydrothermal conditions, urban morphology, topography, and human activity intensity shaped its spatiotemporal variability mainly through nonlinear and indirect pathways. These findings demonstrate that LCZ provides an effective stratification framework for explaining UGS carbon sink differences and supports targeted carbon enhancement and spatial optimization, of existing green spaces.