Soil total phosphorus plays an important role in soil fertility, plant growth, and bioge-ochemical cycles. This study aims to determine the spatial distribution characteristics of soil total phosphorus and identify its main influencing factors in the study area, thereby providing a basis for the scientific management of soil total phosphorus. Here, we conducted a comprehensive analysis by combining classical statistical analysis, ge-ostatistics methods, Pearson correlation analysis, one-way analysis of variance (ANOVA), and structural equation modeling (SEM) to explore the spatial distribution patterns of soil total phosphorus and its influencing factors. The results showed that soil total phosphorus in the study area ranged from 161.00 to 991.00 mg/kg, with an average of 495.71 mg/kg. Spatially, soil total phosphorus exhibited a patchy distribu-tion pattern, with high values primarily concentrated in cultivated areas along rivers and low values mainly located in forested areas in the southeastern and central re-gions. Additionally, the nugget effect of soil total phosphorus was 71.5%, indicating a moderate level of spatial variability. The Pearson correlation analysis revealed that soil total phosphorus content was significantly correlated with multiple factors, including land use types, soil parent material, distance from settlements, slope, and soil pH. Based on these findings, we employed ANOVA to analyze the impacts of various fac-tors. The results indicated that soil total phosphorus content showed significant differences under the influence of different factors. Subsequently, we further explored in depth the action paths through which these factors affect soil total phosphorus us-ing SEM. The SEM results showed that the absolute values of the total effects of the influencing factors on soil total phosphorus, ranked from highest to lowest, were as follows: land use types (0.499) > soil parent material (0.240) > distance from settle-ments (0.178) > slope (0.161) > elevation (0.127) > soil pH (0.114) > normalized differ-ence vegetation index (0.103). These findings provide a scientific foundation for the effective management of soil total phosphorus in similar study areas.

Mapping soil properties in sub-watersheds is critical for agricultural productivity, land management, and ecological security. Machine learning has been widely applied to digital soil mapping due to a rapidly increasing number of environmental covariates. However, the inclusion of many environmental covariates in machine learning models leads to the problem of multicollinearity, with poorly understood consequences for prediction performance. Here, we explored the effects of variable selection on the prediction performance of two machine learning models for multiple soil properties in the Haihun River sub-watershed, Jiangxi Province, China. Surface soils (0–20 cm) were collected from a total of 180 sample points in 2022. The optimal covariates were selected from 40 environmental covariates using a recursive feature elimination algorithm. Compared to all-variable models, the random forest (RF) and extreme gradient boosting (XGBoost) models with variable selection improved in prediction accuracy. The R² values of the RF and XGBoost models increased by 0.34 and 0.47 for the soil organic carbon, by 0.67 and 0.62 for the total phosphorus, and by 0.43 and 0.62 for the available phosphorus, respectively. The models with variable selection presented reduced global uncertainty, and the overall uncertainty of the RF model was lower than that of the XGBoost model. The soil properties showed high spatial heterogeneity based on the models with variable selection. Remote sensing covariates (particularly principal component 2) were the major factors controlling the distribution of the soil organic carbon. Human activity covariates (mainly land use) and organism covariates (mainly potential evapotranspiration) played a predominant role in driving the distribution of the soil total and soil available phosphorus, respectively. This study indicates the importance of variable selection for predicting multiple soil properties and mapping their spatial distribution in sub-watersheds.