Improving feed intake predictions in aquaculture: Integrating dissolved oxygen, body weight, and temperature for Atlantic salmon and meagre

Mathematical models serve as essential tools to optimize feeding strategies, enhance feed efficiency and promote more sustainable aquaculture production. While previous studies primarily aimed to develop feed intake (FI) models based on body weight and temperature, the effect of dissolved oxygen (DO) — a critical environmental factor influencing FI—has often been overlooked. This work aimed to develop FI models for Atlantic salmon and meagre incorporating three main factors (i) body weight, (ii) temperature and (iii) dissolved oxygen. Oxygen-independent models were developed using data sourced from the literature, while Oxygen-dependent models utilized experimental trial data. The Akaike Information Criterion was used to assess and compare the relative quality of the Oxygen-dependent models. To evaluate the models’ performance, an initial analysis was conducted by plotting the model predictions with DO levels from 0 to 10 mg/L. In addition, an independent dataset for each species was used to evaluate the developed models' performance, and model accuracy for each FI model was assessed via the mean absolute percentage error (MAPE). Results show that DO levels influence FI and growth in both species. For Atlantic salmon, Oxygen-dependent models significantly improved FI predictions, whereas for meagre, the Oxygen-independent model provided the most accurate FI predictions. These findings emphasize the importance of incorporating DO into FI models for species where it plays a significant role, while further research is needed to clarify the putative role of DO in meagre FI.