Approximately half of plastic waste ends up in landfills, where fragmentation leads to the leakage of microplastics, nanoplastics, and petrogenic carbon back into ecosystems. However, the timeframe for plastic re-entry into the geological carbon cycle remains unknown. Using landfill-derived field data, we developed a model predicting fragmentation of various polymers into macroplastics, microplastics, fine microplastics, and nanoplastics. We find total waste plastic concentrations range from 85 to 414 mg g⁻¹, with microplastic, fine microplastic, and nanoplastic generation rates of 2–69, 0.5–36.8, and 0.04–1.9 mg per g of plastic, respectively. Plastic distribution depends more on landfill depth than disposal age. Polyethylene terephthalate fragments faster than polypropylene or polyethylene. Our model predicts peak microplastic and fine microplastic fractions within 157–382 and 412–2118 years, respectively, with approximately half of the plastic-derived carbon available for geological cycling in 80–208 years. This research helps clarify the environmental fate of pervasive plastic pollution.

Biological treatment technologies (such as anaerobic digestion, composting, and insect farming) have been extensively employed to handle various degradable organic wastes. However, the inherent complexity and instability of biological treatment processes adversely affect the production of renewable energy and nutrient-rich products. To ensure stable processes and consistent product quality, researchers have invested heavily in control strategies for biological treatment, with machine learning (ML) recently proving effective in optimizing treatment, predicting parameters, detecting disturbances, and enabling real-time monitoring. This review critically assesses the application of ML in biological treatment, providing an in-depth evaluation of key algorithms. This study reveals that artificial neural networks, tree-based models, support vector machines, and genetic algorithms are the leading algorithms in biological treatment. A thorough investigation of the applications of ML in anaerobic digestion, composting, and insect farming underscores its remarkable capacity to predict products, optimize processes, perform real-time monitoring, and mitigate pollution emissions. Furthermore, this review outlines the challenges and prospects encountered in applying ML to biological treatment, highlighting crucial directions for future research in this area.