DIGITAL PHENOTYPING IN FISH TELEMETRY

Digital phenotyping is increasingly used in aquaculture to extract biologically and operationally relevant information from large volumes of sensors data. To date, most applications of digital phenotyping in farmed fish are based on image and video monitoring systems combined with artificial intelligence. Such approaches were successfully applied on feeding behavior, swimming patterns, biomass estimation and external welfare indicators. However, despite the advances, image based systems are also inherently limited to externally observable behavior and are sensitive to visibility, illumination and camera positioning. Behavior occurring at depth, during night time, or under poor visual conditions remains largely inaccessible. In addition, internal physiological responses to environmental or operational challenges are difficult to analyze using visual data alone.

Telemetry as under-used source for digital phenotyping

At the same time, fish telemetry and biologging technologies are already well established. Acoustic telemetry of various sensors is widely used to study behavior, energetics and stress responses. They are typically analyzed using isolated metrics or descriptive summaries and only rarely framed within explicit digital phenotyping paradigm. The potential of telemetry to provide integrated, interpretable and operational phenotypes remains under utilized.

Here, digital phenotyping provides a unifying conceptual framework for telemetry based aquaculture monitoring by explicitly linking raw sensor data to biologically meaningful response profiles. Instead of treating telemetry outputs as standalone metrics, digital phenotyping emphasizes systematic data quality control, feature aggregation across temporal scales, and the synthesis of multivariate patterns into individual level digital phenotypes. These phenotypes can represent functional traits such as activity regulation, stress sensitivity, resilience, or recovery dynamics, bridging telemetry, physiology, and welfare together. Digital phenotyping does not replace existing telemetry methodologies, but reframes them as a phenotype centric pipeline with direct relevance for welfare monitoring, early warning systems, and adaptive aquaculture management.

Applications for welfare and management monitoring

Key application is continuous welfare monitoring in aquaculture systems. Digital welfare phenotypes quantify deviations from baseline behavior and physiology following routine husbandry procedures such as crowding, delousing or transfer. By focusing on recovery trajectories rather than single stress indicators, this approach enables objective comparison of welfare impacts across procedures, seasons and production systems.

It may also support early warning and anomaly detection. Gradual shifts in activity patterns, ventilation dynamics or depth use may precede visible signs of welfare compromise or production loss. Monitoring deviations from established phenotypic envelopes provides a basis for identifying emerging risks earlier than traditional threshold or mortality based indicators.

It is not proposed as replacement of existing image based systems. Both approaches are actually complementary. Image based phenotyping provides valuable information on external behavior at specific locations and times, while telemetry based phenotyping offers continuous insight into internal state and behavioral dynamics independent of visibility or daylight. Combined, these approaches have the potential to provide more complete representation of fish phenotypes in aquaculture. Future work should focus on validation across species and production systems, integration with feeding and bioenergetic models, and alignment with farm level decision processes.

Figure 1. Conceptual diagram of telemetry based digital phenotyping in aquaculture, integrating multi modal sensor signals to characterise internal behavioural and physiological dynamics as continuous phenotypic trajectories.

References

D'Agaro, E. (2025). Fish Farming 5.0: Advanced Tools for a Smart Aquaculture Management. Applied Sciences, 15(23), 12638. https://doi.org/10.3390/app152312638

Neethirajan, S.; Kemp, B. (2021). Digital Phenotyping in Livestock Farming. Animals, 11(7), 2009. https://doi.org/10.3390/ani11072009