<p>The increasing complexity of modern supply chains requires balancing efficiency with sustainability under dynamic conditions. This paper addresses the fertilizer distribution problem by formulating a five-dimensional green transportation model that jointly minimizes transportation cost, time, and carbon dioxide emissions in a dynamic, multi-objective, multi-item, multi-modal network. To capture the dynamic nature of real-world logistics, a hybrid predictive–prescriptive framework is proposed. In the predictive stage, a combination of artificial neural networks (ANN) is utilized. The first multi-output regression neural network forecasts key time-varying parameters, such as monthly transportation costs, time and emission factors. The second ANN forecasts an adaptive ensemble of predictive strategies based on their performance across varying environmental change intensities. The ANNs ensure reliable performance across training, validation, and testing sets. In the prescriptive stage, the forecast inputs are utilized in the multi-objective optimization problem, and optimized using the non-dominated sorting genetic algorithm (NSGA)-III, which generates a diverse Pareto front of trade-off solutions across economic and environmental objectives. Results from a real-world fertilizer distribution case study in Telangana, India, demonstrate that the proposed framework reduces total transportation costs by 5% and delivery time by 2% compared to a static cost-time minimization model. Moreover, this study also maintains the additional objective of minimizing emissions for the same period. The results highlight the effectiveness of integrating data-driven forecasting with dynamic multi-objective optimization for sustainable transportation planning over static models. The general formulation allows applicability beyond fertilizer logistics to other domains where uncertainty, multiple objectives, and sustainability targets are central.</p>