Abstract
Crop rotation optimization requires resolving how rotation patterns can be designed to jointly regulate soil microbial diversity, environmental responses, and agronomic performance under heterogeneous and dynamic conditions. This review develops an AI-enabled framework for crop rotation optimization by integrating rotation-pattern optimization, soil microbial community intelligence, and environmental response prediction and decision support. We first synthesize the ecological mechanisms through which crop rotation reshapes soil microbial communities and identify the key biological and environmental variables that constrain rotation performance. We then critically examine the application of machine learning, deep learning, and optimization algorithms to rotation-pattern design, microbial community analysis, and prediction of crop–soil–environment interactions, with emphasis on how heterogeneous data can be transformed into actionable rotation decisions. Representative applications further demonstrate the potential of data-driven approaches to identify high-dimensional rotation–microbiome–environment relationships and support adaptive management. Key challenges remain in integrating multimodal and longitudinal datasets, improving model interpretability and cross-region generalizability, and translating algorithmic predictions into robust field-scale decisions. Overall, this review establishes an AI-driven crop rotation–microbiome–environment optimization framework that shifts crop rotation research from experience-based pattern selection toward data-informed, adaptive, and mechanism-aware decision-making, providing a conceptual foundation for precision management of soil biodiversity and sustainable agroecosystems.
