@article{Shen2026, 
author = {Feixue Shen and Lin Yang and Xiuqiang Peng and Dianpeng Li and Chenconghai Yang and Yue Pu and Mao Guo and Yuru Yan and Chenghu Zhou},
title = {Anthropogenic disturbance decouples coastal soil organic carbon and bulk density},
year = {2026},
journal = {Environmental Science and Ecotechnology},
volume = {32},
keywords = {Coastal soil, Vertical sampling, Spatial heterogeneity, Digital soil mapping, Machine learning},
url = {https://www.sciopen.com/article/10.1016/j.ese.2026.100728},
doi = {10.1016/j.ese.2026.100728},
abstract = {Coastal blue carbon ecosystems play a pivotal role in mitigating global climate change through rapid sediment burial and efficient carbon preservation. However, accurate carbon accounting is severely hindered by the systematic neglect of soil bulk density (BD) variations and the uncritical reliance on terrestrial-derived pedotransfer functions, masking hidden uncertainties in regional carbon stock assessments. Here we present a high-resolution, multi-depth assessment of soil organic carbon (SOC) content and BD across a 1-m vertical gradient in the intensively managed coastal zone of Jiangsu Province, China. Machine learning frameworks reveal a striking spatial and vertical decoupling between SOC and BD driven by divergent environmental controls: SOC content responds to soil depth and ocean salinity, whereas BD aligns with hydro-geomorphic distance to the coast. Vegetation mediates a tight vertical negative coupling (p < 0.001) in natural salt marshes, whereas anthropogenic activities decouple this relationship in croplands, restricting standard pedotransfer function predictability in all layers (R2 ≤ 0.22). Across all layers, conventional spatial estimation models yield high baseline prediction uncertainty (RMSE = 0.22 g cm−3). These findings demonstrate that independent, high-resolution BD profiling is indispensable for valid blue carbon verification. Our results establish a transferable baseline to optimize depth-specific sampling and refine global coastal carbon accounting models under intensifying human pressures.}
}