[en] In response to increasing food security and sustainability challenges, soilless cultivation systems have emerged as a promising alternative to conventional agriculture. Among these, aquaponics stands out for integrating fish farming and plant production within a circular nutrient-recycling loop. However, reconciling the contrasting thermal requirements of trout, a key species in Walloon aquaculture, and nitrifying bacteria remains a major challenge.
This study provided a comprehensive evaluation of a trout-based decoupled aquaponic system, assessing its agronomic and phytochemical performance on basil (Ocimum basilicum) and developing artificial intelligence (AI) tools for optimization. The experiment was carried out over a 60-day growth cycle in a semi-controlled greenhouse using three NFT cultivation tables: two supplemented with mineral fertilizers at different electrical conductivities and one non-supplemented control.
Climatic parameters and physicochemical variables were continuously monitored. Growth indicators fresh and dry biomass, leaf area, and relative growth rate) were measured to assess vegetative performance, while phytochemical analyses (proline content, mineral composition, and essential oil profiling via GC-MS and GC-FID) revealed significant differences between treatments.
Finally, Random Forest and XGBoost models were developed from experimental data to predict basil growth and optimize nutrient management. SHAP analysis was used to identify the most influential variables and propose data-driven management scenarios for optimizing trout-based aquaponic systems.
