European Journal of Sustainable Agroecosystems

Background: Integrated crop–livestock systems (ICLS) represent a cornerstone of sustainable agricultural intensification, promoting dynamic nutrient cycling, biodiversity, and ecosystem resilience. However, optimizing multi-enterprise interactions requires sophisticated decision-making systems to manage spatial-temporal variability and balance nutrient budgets.

Objective of the review: This article provides a comprehensive evaluation of how data science, artificial intelligence (AI), machine learning (ML), and predictive analytics are deployed within ICLS to optimize fertilizer recommendations and precision nutrient management.

Sources of literature: A systematic search was executed across primary scholarly databases—including Scopus, Web of Science, PubMed, ScienceDirect, SpringerLink, Google Scholar, and IEEE Xplore—synthesizing high-impact peer-reviewed literature published between 2015 and 2025. Major findings: ** The synthesis reveals that ML algorithms (e.g., Random Forest, Deep Learning, Support Vector Machines) integrated with remote sensing and Internet of Things (IoT) Sensors improve nutrient-use efficiency by up to 28%, reduce synthetic fertilizer reliance, and decrease nitrogen runoff by 35%.

Research gaps: Key unresolved challenges include the low explainability of deep learning architectures ("black-box" limitations), severe spatial-temporal data heterogeneity across diverse agro-ecological zones, data ownership conflicts, and a scarcity of standardized multi-enterprise data fusion frameworks that dynamically map livestock metabolic outputs to crop agronomic demands.

Conclusion: Predictive analytics transforms static nutrient management into dynamic, context-aware decision support ecosystems. Overcoming current methodological bottlenecks through Explainable AI (XAI) and edge-computing infrastructure will accelerate farm-level adoption, aligning intensive production with ecological boundaries to achieve climate-resilient global food security.