Multi-criteria decision analysis for sustainable agriculture: A critical review, method-selection framework, and future research directions

CONTEXT Sustainable agriculture requires decisions across economic, environmental, social and technical objectives. Multi-criteria decision analysis (MCDA) can make these trade-offs explicit, but method choice can materially shape recommendations. OBJECTIVE To critically compare established and recent agricultural MCDA approaches, assess their fit to different decision problems, and develop a problem-driven framework for method selection, validation and future AI-enabled applications. METHODS We conducted a critical, selective review of foundational MCDA sources, established agricultural applications, recent domain reviews and publications available through August 2026. The review primarily addresses multi-attribute decision-making (MADM) for discrete alternatives and includes selected multi-objective decision-making (MODM) approaches where agricultural decisions require resource allocation under constraints. Methods were grouped by decision logic and applications by farm management, land/resource assessment, agri-food supply chains and digital or precision agriculture. Comparison focused on decision output, evidence, criteria interdependence, stakeholder involvement, compensation, uncertainty, resource constraints and robustness. RESULTS AND CONCLUSIONS Four recurring issues emerged: justification of weights, ranking stability, treatment of uncertainty and transparency of stakeholder input. The proposed framework links method choice to required output, available evidence, interdependence, stakeholder setting, compensation rules and constraints, with robustness testing as an integral validation step. AI can strengthen prediction and scenario generation, while MCDA retains a distinct role in making preferences, thresholds and trade-offs explicit. The value of MCDA therefore depends less on methodological novelty than on the fit between the decision problem, evidence, assumptions and validation. SIGNIFICANCE The review moves beyond cataloguing MCDA techniques by linking problem characteristics to method choice and validation requirements. It provides a practical basis for more transparent, robust and explainable decision support in sustainable agriculture, including AI-enabled systems.