Artificial neural network-based estimation of physiological, biochemical, and nutrient parameters in durum wheat under NaCl and biostimulant treatments

Background: Durum wheat (Triticum durum L.) productivity is strongly limited by salinity stress, particularly during early growth stages, due to disruptions in growth, water relations, and nutrient uptake. Seaweed extracts (SWEs), especially those derived from Ascophyllum nodosum, are widely used as biostimulants to enhance stress tolerance; however, their effects on durum wheat under salinity remain insufficiently characterized. In parallel, artificial neural networks (ANNs) provide effective tools for modeling complex plant responses to environmental stress. Results: Salinity significantly reduced growth and physiological parameters, including biomass, chlorophyll content, and relative water content. SWE applications (2 and 4 g L⁻¹) effectively mitigated these negative effects. Biochemical traits such as proline accumulation, total phenolic content, and total antioxidant capacity were markedly enhanced under salinity. SWE treatments also improved macro- and micronutrient uptake in roots and shoots. ANN models successfully predicted multiple plant traits with high accuracy (R² > 0.90 for several key parameters). These models were implemented in a web-based R Shiny application to enable real-time prediction of plant responses. Conclusions : SWE application alleviates salinity-induced stress in durum wheat by improving growth, antioxidant capacity, and nutrient acquisition. The integration of ANN modeling with experimental data provides a reliable and practical approach for predicting plant responses, supporting artificial intelligence-assisted strategies for sustainable wheat production under saline conditions.