Abstract

The CSM-CROPGRO-Lentil model (DSSAT v4.8.5) was calibrated and validated for 43 ICARDA lentil genotypes across three growing seasons (2019–2022) under semi-arid rainfed conditions in southeastern Türkiye. Calibration against the 2019–2020 season yielded excellent performance for phenological traits (anthesis date nRMSE = 0.8%; maturity date nRMSE = 1.2%), thousand-seed weight (nRMSE = 1.6%), and good performance for grain yield (nRMSE = 19.3%; d = 0.935) and biological yield (nRMSE = 11.1%; d = 0.941). Phenological simulations remained robust (anthesis and maturity dates nRMSE = 6.8% and 1.6%), but grain and biological yield simulations failed (nRMSE = 43.0–74.7%) with opposing bias directions in both validation seasons. Spearman rank correlation collapsed to non-significant levels in both validation years (rho = −0.020 and +0.263), indicating a loss of genotype-ranking capacity, in which only three accessions (G37, G3695, and G21151) maintained simulation error below 40% across both drought seasons. Model bias was not significantly associated with PEG-derived drought tolerance classifications, based on composite stress tolerance index-based reclassification with balanced group sizes (n = 5 vs. n = 5; Mann–Whitney p ≥ 0.548), suggesting that limitations in species-level water stress response functions may have contributed to model failure, although the potential influence of interannual differences in soil hydraulic properties cannot be excluded. Late-maturing genotypes were simulated significantly more accurately than early and mid-maturing genotypes in 2021–2022 (Kruskal–Wallis p = 0.013), while large-seeded genotypes showed disproportionately greater underestimation in 2020–2021 (rho = −0.410, p = 0.006). These findings demonstrate that CROPGRO-Lentil reliably simulates phenological diversity under favorable conditions but requires genotype-specific drought stress parameterization and improved seed-filling dynamics to support genotype evaluation under increasingly variable semi-arid moisture regimes.

IPC Classification

Keywords