How do leaf nutrients orchestrate citrus growth? An artificial intelligence and machine learning approach to decoding complex interactions

The integration of artificial intelligence (AI) and machine learning (ML) into agricultural science offers unprecedented potential to decode nutrient-driven growth processes in perennial crops. This study establishes a computational framework to unravel the complex, non-linear relationships between leaf mineral nutrients and crown diameter—a critical growth parameter in citrus nursery systems. Using a dual-metric feature selection approach (correlation + mutual information), we identified seven key minerals (N, P, K, S, Ca, Fe, Zn) governing seedling vigor, while Mg, Mn, and Cu exhibited no significant associations with growth. Model benchmarking revealed a distinct performance hierarchy: GRNN achieved optimal accuracy (training R² = 0.94, testing R² = 0.64), closely followed by RF, with both significantly outperforming MLP, SVR, and MLR. GRNN's superior capability in capturing nutrient synergies and antagonisms (e.g., K-P-S enrichment vs. Zn inhibition) enabled genetic algorithm optimization to enhance crown diameter by 45.6% through increasing N, K, S, P, Ca, and Fe while decreasing Zn. This is the first study to apply GRNN for modeling citrus nutrient-growth interactions, demonstrating its superior capacity to resolve nonlinear synergies and antagonisms compared with conventional ML approaches. These results validate that crop growth is governed by nutrient equilibrium rather than isolated sufficiency thresholds . Our GRNN-GA framework bridges computational intelligence and agronomy, providing actionable strategies for precision nutrient management in citrus nurseries.