Development and evaluation of reference feed intake models for meagre (Argyrosomus regius)

Abstract

The aim of the study was to identify a basic reference feed intake model to facilitate understanding of the impact of temperature as the key environmental factor, along with body weight and dietary composition. These basic models should provide the baseline for further research to advance precision farming practices and support efficient production for meagre (Argyrosomus regius). Several reference models, with different levels of complexity, were built using data from scientific publications and feeding tables, followed by evaluating 27 different models using various fitting methods. Each model's mean absolute percentage error was estimated through repeated 5-fold cross-validation (with n = 200 iterations). Models were divided into four categories based on the inclusion of temperature and diet composition parameters: simple feed-independent models, complex feed-independent models, simple feed-dependent models, and complex feed-dependent models. The best model from each category was identified, followed by an assessment of the overall best. Consistent with dynamic energy budget theory, models using a fixed body weight exponent of 2/3 demonstrated better fit. Feed-dependent models incorporating lipid levels outperformed feed-independent ones. Additionally, simpler models with temperature parameters effectively predicted feed intake at optimal temperatures, while more complex models predicted intake better above the thermal optimum. The reference models selected in this study can be applied to estimate feed requirements over time for production or experimental trials, benchmark feeding to isolate the effect of other variables, support growth models and alternative feeding tables, and provide decision support.