BBCH-based phenological modeling of camelina under contrasting irrigation, nitrogen, and camelina–chickpea cropping patterns

Abstract Phenological responses of camelina to management were evaluated using days after planting (DAP) and cumulative growing degree days (GDD) at nine main BBCH stages. A two-year split-plot factorial experiment included two irrigation regimes, two nitrogen rates, seven camelina-containing cropping patterns, and three replications. For each stage, DAP was modeled from GDD and design covariates using multiple linear regression, quadratic regression, random forest (RF), support vector regression, and an artificial neural network. Supplemental irrigation was associated with stage-dependent differences in DAP and GDD, with lower values at BBCH 0–3 and higher values at BBCH 5–9; however, early-stage differences before irrigation application should not be interpreted as direct causal responses to irrigation. The higher nitrogen treatment (46 kg N ha⁻¹) was associated with greater DAP and GDD than the 11.5 kg N ha⁻¹ treatment at the observed stages; however, because the additional 34.5 kg N ha⁻¹ was applied only at 50 and 100 days after sowing, differences observed before these applications (particularly BBCH 0–3) should be interpreted as statistical differences among assigned nitrogen treatments rather than causal responses to the additional nitrogen. At BBCH 5, the nitrogen response depended on year and irrigation. Mixed cropping had the highest GDD throughout, whereas the lowest GDD occurred in sole camelina at BBCH 0–3 and in the 3:3 replacement pattern at BBCH 5–9. RF provided the best stage-specific predictive performance, with a macro-averaged performance of R² = 0.972, MAE = 0.23 d, and RMSE = 0.40 d, demonstrating strong agreement between predicted and observed phenological timing under the evaluated experimental conditions The resulting BBCH–GDD framework supports local phenological monitoring but does not support inferences about yield, product quality, resource-use efficiency, or economic performance.