Source Apportionment of Heavy Metal(loid)s in the Surface Soils of Cerrito Blanco, Mexico: A Comparative Study of Three Receptor Models (APCS-MLR, PMF and UNMIX Model)

It is essential for the monitoring and conservation of soil environments to have a comprehensive awareness of the source characteristics including that of heavy metal (loid)s. This study focuses on a semi-arid mining area in Mexico where heavy metal (loid)s originated from past mining and rapid industrialisation activities, which have degraded the soil environment as well as the security of agricultural products. In this study, Cerrito Blanco, which is part of an abandoned mining region in Matehuala, Mexico was identified as the principal source of contaminants. It was also observed that the contributions of the contaminants to the overall pollution could be calculated by using a combination of multivariate statistics and receptor models. The three receptor models such as APCS-MLR, PMF, and UNMIX were used and mutually compared to improve the accuracy and quantitative assessment of source contributions. A total of eleven heavy metal(loid)s were selected for this study, out of which the mean concentrations level of As, Fe, Mn, and Zn exceeded their reference limit values. The spatial distribution mapping revealed the distribution patterns and significant effects on concentrations of heavy metal(loid)s in surface soil. APCS-MLR identified three potential sources with contribution rates of 18.16% (groundwater source), 57.33% (past mining and industrialisation), and 24.51% (natural source) respectively. Two models, namely, PMF and UNMIX were employed to establish the contributions from common pollution sources. The contributions from four common sources (groundwater, past mining and industrialisation, natural source, and human activity) contributed 15.57, 42.86, 36.06, and 5.51% according to the PMF model, but 14.73%, 45%, 31.91%, and 8.36% respectively by the UNMIX model. The results of three receptor models showed that heavy metal(loid)s concentrations were mostly influenced by past mining and industrialisation activities. As a result, the identification of the potential sources of heavy metal(loid)s performed better using the APCS-MLR model than PMF and UNMIX model due to its higher R2 value (0.90–1) and lower P/M error (1–1.15). To achieve more reliable and objective conclusions of source apportionment, it was recommended that multiple receptor models be applied.