Analytic Hierarchy Process Ranking of Durum Wheat Management Scenarios in a Mediterranean Environment: A Simulation-Based Multi-Criteria Analysis

The Analytic Hierarchy Process (AHP) was applied to six agronomic scenarios for durum wheat (Triticum durum Desf.) in the Capitanata plain (Apulia, southern Italy), combining three sowing dates (15 October, 1 November, 15 November) with two water regimes (rainfed; supplemental irrigation at flowering, 13–41 mm season⁻¹). Five performance indicators — grain yield (4983–5722 kg ha⁻¹), CO₂-equivalent emissions (CO₂_eq, 1190–1214 kg ha⁻¹), carbon footprint (CFP, 0.19–0.24 kg CO₂ eq kg⁻¹), total water consumption (TotW, 5555–6387 m³ ha⁻¹) and water footprint (WFP, 1.08–1.18 m³ kg⁻¹) — were derived from AquaCrop-GIS simulations coupled with cradle-to-gate life cycle assessment. A symmetric weight scheme (w = 0.60 for the dominant criterion, w = 0.10 for each of the remaining four) defined six decision profiles, with all pairwise comparison matrices perfectly consistent (CI = 0, CR = 0). The rankings revealed a systematic inversion between absolute indicators (CO₂_eq, TotW) and ratio indicators (CFP, WFP): under absolute-metric profiles, the lowest-yielding scenario (4983 kg ha⁻¹) paradoxically ranked first because reduced productivity mechanically lowered per-hectare resource consumption. Under ratio-metric and balanced profiles, early-November sowing consistently led the rankings, combining the lowest carbon footprint (0.19 kg CO₂ eq kg⁻¹) and the lowest water footprint (1.08 m³ kg⁻¹) among the six scenarios. Switching point analyses quantified the weight thresholds at which leadership changed: w₁ = 0.20 (S1 → S2) and w₁ = 0.60 (S2 → S5) along the Yield axis, w₃ = 0.60 (S2 → S5) along the CFP axis, with rainfed early-November sowing retaining leadership across the full range of WFP weights. The AHP procedure was also applied to the 72 simulation replicates spanning all combinations of soil profile, climatic cell and cropping year in the 2013–2023 dataset, providing the empirical rank distribution for each scenario under each profile and extending the mean-based analysis to the full pedo-climatic variability of the region.