Source Apportionment Methods for Soil Heavy Metals: Principles and Optimal Scenarios

Accurate source apportionment of soil heavy metals (HMs) is critical for targeted pollution mitigation and ecological remediation. This review systematically synthesizes and compares five mainstream source apportionment approaches—receptor models (positive matrix factorization, PMF; absolute principal component score-multiple linear regression, APCS-MLR; UNMIX model), stable isotope tracing, and random forest (RF)-based machine learning—to provide researchers with a comprehensive methodological framework. The methodology includes a systematic literature review, comparative analysis of methodological principles, and synthesis of representative case studies from diverse geographical contexts. The core principles, evolutionary paths, typical use cases (e.g., industrial zones, agricultural fields, regional surveys), and inherent limitations are synthesized for each method. A practical decision framework linking research contexts (study objectives, spatial scales, data availability) to optimal method selection, along with guidelines for multi-method integration, is proposed. This review provides actionable guidance for researchers and practitioners in selecting appropriate methods for specific pollution scenarios, ultimately supporting more effective environmental management and policy development.