Introduction
The fish farming industry needs instruments that can monitor in real time fish health and welfare objectively, without disturbing the fish or interfering with the daily management. In this context, precision livestock farming is gaining increased attention to enhance animal welfare, but also to enhance production and environmental sustainability (Halachmi et al., 2019; Brijs et al., 2021). In this study, we monitored European sea bass (Dicentrarchus labrax) implanted with acoustic transmitter which is measuring fish depth and swimming activity, a proxy of energy expenditure at the KEFISH farm at Argostoli (Greece) over ~4 months. In addition, environmental sensors were deployed in the farm cages to monitor environmental variables (temperature, oxygen concentration, salinity and turbidity). The aim of this work was to investigate the possibility to model key performances indicators (KPIs; i.e., growth performance and fish mortality) of sea bass using the physiological and environmental parameters recorded using a wireless sensors system, and so to investigate if the use of such sensors may help management procedure, enhance fish welfare and production.
Materials and methods
The environmental and physiological sensors were deployed in the KEFISH farm for a monitoring period of 134 days (from the 09.20.2021 to 01.31.2022). Over the experiment duration, morphometrics measurements were carried out by the farmer when fishing (n=7 sampling times) allowing the estimation of the weight gained by fish per day over the trial. Mortality events (count of dead fish) were also recorded by the farmer every day. Physiological sensors used in the study were acoustic accelerometer tags with ’tailbeat mode’ algorithm, allowing to measure fish acceleration (m s-2), which is a proxy of energy expenditure (Alfonso et al., 2021). Five environmental sensors were installed in the cage (Figure 1). Among the five sensors, we installed three dissolved oxygen (DO) sensors, one turbidity sensor and one salinity sensor. In this trial, we unfortunately faced issues with the outputs of both turbidity and salinity sensors. Thus, being not satisfied of the outputs of these parameters, they have been not included in any of the model presented below. Receiver linked to a HUB continuously received acoustic signals from both environmental and physiological sensors acoustically at frequency of 69kHz. The Hub was installed at a radius of ~200 m of the cage and was connected to WIFI of the farm to send data to the cloud and to display data in real time to live dashboard.
Growth performance and mortality data gathered over the trial were modelled using generalized linear models depending on environmental and physiological parameters recorded by different sensors. Selection of parameters to fit the best models has been carried out and only the best model for each variable is presented below.
Results and discussion
Growth performance (i.e., weight gained by fish per day) was found to be significantly affected by both temperature and DO; the fish grew better with higher temperature and lower DO. It could appear surprising that fish grew better at lower DO but this can be explained first by the fact that the low DO values are overall associated to higher temperature occurring at the beginning of the trial (i.e. September), and so fish grew better. Also, even is the DO is overall low in this period (lowest measured at 40 % in some moments) and known to affect swimming and metabolism, this value is not significantly affecting critically the sea bass physiology. We also found a tendency for significant effects of both swimming activity and fish depth (p=0.054 and p=0.064 respectively). In more details, the growth performance tends to be enhanced when swimming activity recorded is greater, and fish is located near the water surface. This appears relevant because greater fish swimming activities are generally recorded when temperature is high. Overall, greater swimming activity is related to greater energy expenditure (Alfonso et al., 2021). However, greater energy consumed at higher temperature may be compensated by greater environmental conditions, overall favorizing better growth performances.
Also, we found that mortality is significantly affected by both temperature and the DO recorded. In more details, higher mortality is linked to low temperature and high DO levels. Oppositely to what has been observed for growth performance, the mortality did occur when the water temperature was cold. This is not very surprising but it is important to empathize that high temperature are also known to lead to such mortality events (Genin et al., 2020), but since we started the trial at the end of September, we did not face any huge warm vague triggering such mortality events, generally occurring during summer. In addition, the swimming activity and fish depth recorded by the physiological sensors implanted in fish were interestingly found to predict the mortality. High mortality over the experimental duration has been linked to greater swimming activity and fish located deeper in the cage. Therefore, here, contrary to what has been observed for growth performance, greater energy consumed could be linked to a response to acute stress event occurred in the sea cage which resulted in mortality events. Interestingly, location of fish at the bottom is known as thigmotaxis behaviour and indicative of stress in fish, including in sea bass (Sadoul et al., 2021), and could explain why mortality event are occurring in these moments.
In conclusion, predicting growth performance and mortality of sea bass using physiological and environment parameters of interest gathered by wireless sensors is very promising for the future, especially in the framework of precision livestock farming. It is, however, important to note that models and predictions have to be enhanced by including more parameters such as, species, salinity, turbidity, water current, stocking density, mass.
Acknowledgments
This research was supported by H2020 FutureEUAqua, under grant No. 817737.
References
Alfonso, S., Zupa, W., Spedicato, M. T., Lembo, G., and Carbonara, P. (2021). Mapping the Energetic Costs of Free-Swimming Gilthead Sea Marine Aquaculture. Biology (Basel). 10, 1357. doi:https://doi.org/10.3390/biology10121357.
Brijs, J., Føre, M., Gräns, A., Clark, T. D., Axelsson, M., and Johansen, J. L. (2021). Bio-sensing technologies in aquaculture: how remote monitoring can bring us closer to our farm animals. Philos. Trans. R. Soc. B 376, 20200218. doi:https://doi.org/10.1098/rstb.2020.0218.
Genin, A., Levy, L., Sharon, G., Raitsos, D. E., and Diamant, A. (2020). Rapid onsets of warming events trigger mass mortality of coral reef fish. Proc. Natl. Acad. Sci. U. S. A. 117, 25378–25385. doi:10.1073/pnas.2009748117.
Halachmi, I., Guarino, M., Bewley, J., and Pastell, M. (2019). Smart Animal Agriculture: Application of Real-Time Sensors to Improve Animal Well-Being and Production. Annu. Rev. Anim. Biosci. 7, 403–425. doi:10.1146/annurev-animal-020518-114851.
Sadoul, B., Alfonso, S., Cousin, X., Prunet, P., Bégout, M. L., and Leguen, I. (2021). Global assessment of the response to chronic stress in European sea bass. Aquaculture 544, 737072. doi:https://doi.org/10.1016/j.aquaculture.2021.737072.