Zinc contamination, combined with accelerating climate change, represents a major constraint to durum wheat cultivation in Algeria. Two contrasting Triticum durum varieties, Simeto and Ammar 6, were evaluated under four ZnSO4 concentrations using the raw replicate-level dataset provided for this revision (2 varieties × 4 doses × 3 biological replicates). The revised analysis integrates ANOVA, multivariate statistics, dose-response modelling and an exploratory CMIP6-linked vulnerability framework. The recalculated results confirm clear variety × treatment interactions for several traits. Ammar 6 maintained higher germination under zinc exposure and showed strong biomass accumulation, especially for aerial dry mass at D200, whereas Simeto showed a more complex trait-dependent response at D150 rather than a perfect D100–D150 plateau. Catalase activity was treated as a LOD/LOQ-censored low-level response: Ammar 6 at D0 and D100 was reported as <LOD, and values between LOD and LOQ were considered detected but below quantification and interpreted only qualitatively. The CMIP6/Bayesian component is presented as exploratory uncertainty propagation using equal central CSI weights (α = 0.5, γ = 0.5), not as a field-validated prediction. These findings support the use of raw-data-based varietal screening while emphasizing the need for larger replicate numbers and field validation under natural zinc gradients. The greenhouse experiment used three true biological replicates per variety x dose combination; the very small within-cell standard deviations are reported transparently and are not used to overstate certainty. The CMIP6/Bayesian component is therefore framed strictly as a hypothesis-generating vulnerability extrapolation, because zinc dose-response functions were not validated experimentally under temperature or precipitation stress.