Spatial distribution of lead concentration in peri-urban soil: Threshold and interaction effects of environmental variables

• Published in: Geoderma Vol. 429; p. 116193
• Main Authors: Wu, Zihao, Chen, Yiyun, Yang, Zhen, Liu, Yaolin, Zhu, Yuanli, Tong, Zhaomin, An, Rui
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.01.2023
• Subjects: altitude; China; decision support systems; Environmental variable; food safety; gasoline; Gradient boosting decision tree; industrialization; Interaction effect; lead; Lead concentration; population density; soil; soil quality; Spatial distribution; Threshold effect; total nitrogen; urbanization; China; Threshold effect; Spatial distribution; Lead concentration; Gradient boosting decision tree; Interaction effect; Environmental variable
• ISSN: 0016-7061; 1872-6259
• DOI: 10.1016/j.geoderma.2022.116193

•Environmental variables have threshold and interaction effects on Pb concentration.•Gradient boosting decision tree is used to explore the two effects.•Threshold effects exhibit the effective ranges of covariates on Pb accumulation.•The two-way and three-way interaction effects of covariates on Pb are visualized.•Industrialization and urbanization are the most important sources of Pb contamination. Understanding the spatial distribution of soil lead concentration in peri-urban soil and its underlying mechanism is critical to controlling lead contamination, ensuring food safety, and maintaining sustainable urban development. Although the linear and nonlinear relationships between Pb and environmental variables have been extensively studied, the threshold and interaction effects of environmental factors on Pb content are often ignored. This study aims to investigate the threshold and interaction effects and obtain the spatial distribution of Pb content via gradient boosting decision tree. The peri-urban area of Wuhan City, China, was selected as the study area, where soil health was threatened by agricultural production and urbanization and industrialization process. Results showed that the 14 covariates can explain 61.9 % variation of Pb content, in which anthropogenic factors played a major role. Specifically, the Pb content was mainly affected by distance from the industrial enterprises and gasoline stations, population density, elevation, and total nitrogen, which denoted that Pb contamination was mainly originated from industrial production, gasoline stations, urbanized population, and fertilization. All environmental variables had threshold effects on Pb, and the relationship between Pb and these variables mutated when the threshold was exceeded. The results of threshold effects exhibited that Pb was elevated in locations near industrial enterprises (<4 km) or gasoline stations (<1.5 km), with a dense population (>5 person/hectare), and with low elevation (<25 and 45–65 m). The two-way and three-way interaction effects of covariates on Pb content was determined, indicating that the influence of environmental variables on Pb concentration is a complex interaction rather than a simple superposition. These findings highlight the consideration of threshold and interaction effects to understand the underlying processes that lead to Pb contamination, and have important implications for sustainable urban development.