Zhenyi Jia

Coastal wetlands play a crucial role in carbon sequestration, yet the invasion of Spartina alterniflora (SA) significantly alters the cycling and sequestration of soil organic carbon (SOC) in coastal wetlands. Nevertheless, the potential underlying mechanisms governing the dynamics of SOC in coastal wetlands following SA invasion remain poorly understood. Here, we investigated the impacts of SA invasion on the dynamics and potential sequestration mechanisms of SOC in the Hangzhou Bay Estuary Wetland, China. Compared to the bare flat (BF), SOC and its fractions in 0–20 cm increased by 1.37–2.24 times after 8 years of SA invasion. Variance partitioning analysis indicated that the combined effects of soil physicochemical properties, soil carbon cycle-related enzymes, and vegetation type were the primary drivers of SOC and its fractions. Redundancy analysis revealed significant positive correlations between SOC and key soil physicochemical properties and enzymes, including sucrase, clay particles, total nitrogen, ammonium nitrogen, and β-glucosidase. Structural equation modeling demonstrated that SA invasion was associated with significant alterations in soil physicochemical properties and positively correlated with both stable and labile carbon fractions, or indirectly linked to these fractions through carbon cycle-related enzymes, thereby substantially positively contributing to SOC. This study supports the hypothesis that the invasion of SA affects the linkage pathway of SOC sequestration and offers valuable guidance for carbon sequestration strategies of coastal wetlands.

The accurate identification of pollutant sources and their spatial distribution is crucial for mitigating soil heavy metals (SHMs) pollution. However, the receptor model struggles to effectively categorize pollutant sources and pinpoint their locations and dispersion trends. We propose a novel comprehensive framework that combines a receptor model, random forest (RF), affinity propagation (AP) algorithm, and bivariate local indicator of spatial association (BLISA), to optimize the traditional approach for tracing SHMs sources in industrial regions. We apportioned SHMs sources using a receptor model combined with RF, while BLISA combined with AP methods were employed to accurately locate the source areas and identify their dispersion tendencies. The results revealed that SHMs originated from mixed sources of equipment manufacturing agglomeration and agricultural activities (59.0%), geological background (30.5%), and emissions from heavily-polluting industries (10.5%). The pollution sources of soil Cd and Pb were located near specific industries, showing characteristics of multi-site concurrent pollution diffusion influenced by their proximity to industrial sites. The spatial distribution of Cr, Cu, and Zn sources was concentrated in high-density urban industrial areas, transitioning from point to nonpoint sources, with diffusion patterns influenced by the spatial agglomeration effect of industries. Our enhanced framework accurately identifies the location of SHMs sources and their dispersion tendencies, thereby improving regional soil pollution management.

Accurately mapping the spatial distribution and variation of soil organic matter (SOM) is of great significance for guiding regional soil management. However, the applicability and prediction performance of machine learning techniques in dryland agroecosystems still needs to be further studied. In this study, we collected a total of 733 topsoil samples from the farmland in Xiao County, Anhui Province, which is a typical dryland agroecosystem in the Huang-Huai-Hai Plain. Then, the environmental covariates were selected, and the ordinary kriging (OK), multiple linear stepwise regression (MLR), regression kriging (RK), radial basis function neural network (RBFNN), and random forest (RF) models were conducted to map the SOM content, and the optimal model was ascertained. The results demonstrated that the alkali-hydrolyzable nitrogen (26.11%), available potassium (17.73%), mean annual precipitation (13.26%), and pH (11.80%) were the main controlling factors affecting the spatial distribution of SOM in the study area. Meanwhile, the introduction of environmental covariates can effectively improve the SOM prediction accuracy, and the RF model (R² = 0.48, MAE = 2.38 g kg⁻¹, MRE = 12.99%, RMSE = 3.14 g kg⁻¹) has a better performance than the RFBNN, MLR, RK, and OK methods. Although there are local differences in the spatial distribution of SOM predicted by the five methods, the overall spatial distribution of SOM was characterized by the low concentration area (13.44–20.00 g kg⁻¹) distributed in the central and northwest of study area, and the high concentration area (24.00–28.95 g kg⁻¹) distributed in the southeast. Overall, our study demonstrated that machine learning-based models could accurately predict the SOM content in dryland agroecosystem, and the produced maps function as baseline maps for sustainable agricultural management.