Model to Predict Outbreaks of Pancreas Disease in Salmon

The pancreas disease (PD), caused by salmonid alphavirus (SAV), wreaks havoc in Atlantic salmon farms in Norway, Scotland, and Ireland. This single-stranded RNA virus, primarily affecting farmed fish, devastates their health and profitability. Its impact is undeniable: reduced growth, increased mortality, poor feed conversion, and degraded products.

But what if we could predict PD outbreaks before they occur? That’s exactly what a team of researchers from Zoetis and Pharmaq Analytiq has achieved.

Harnessing cutting-edge machine learning and easily accessible data from BarentsWatch, they have developed a predictive model with an astonishing accuracy of 94.4%. This means salmon producers can now anticipate SAV transmissions up to 8 weeks in advance, allowing them to take proactive measures and safeguard their precious fish.

Unveiling the culprit and its transmission pathways

While SAV3 was long considered the dominant subtype, recent data reveals that SAV2 and SAV3 share the same prominence. This viral threat is primarily transmitted horizontally (87.8% of cases), spreading through water currents among fish.

Researchers have tirelessly investigated the risk factors and pathways of this alarming spread.

Infection and risks

Factors such as the number of fish in neighboring farms, the density of fish farms, the presence of a history of pancreas disease, and shared ownership with infected fish farms increase the risk of disease spread.

Additionally, water currents play a crucial role, with the distance from infected sites being a crucial measure. Even boat movements contribute to the problem, highlighting the interconnectedness of the aquaculture industry.

The presence of salmon lice (Lepeophtheirus salmonis) adds another layer of complexity. These troublesome parasites can act as virus vectors, facilitating SAV transmission and spread. Their immunosuppressive effects on salmon worsen the situation, making fish more susceptible to infections.

Vaccines, feeds, and biosecurity

The good news is that the industry is fighting back. Several vaccines offer varying degrees of protection against PD. Functional feeds, enhanced genetics, and strict biosecurity measures are employed to mitigate the impact of the disease.

Moreover, the Norwegian government has implemented a mandatory detection program, requiring fish farmers to conduct periodic tests for the presence of SAV. Data on salmon lice are also meticulously collected and monitored. All this information is available to the public through BarentsWatch, a valuable platform for sharing knowledge and ideas.

Predictive models as early warning systems

The researchers aimed to go beyond existing models by investigating the impact of various aquaculture parameters. The goal was to develop a high-precision model that predicts the likelihood of detecting SAV at specific Norwegian sites.

The secret weapon?

A powerful ensemble model trained with a unique combination of data. Publicly available resources like BarentsWatch, such as distance between sites, fish population/density, along with the influence of sea lice and boat movement, were combined with private PCR analyses from individual fish farms.

This rich dataset allowed researchers to create the predictive model and identify 11 key features influencing SAV transmission, including:

• Movements of wellboats: These crucial vessels, used to transport fish, can unknowingly spread the virus, making their movements vital indicators.

• Environmental factors: Water temperature, currents, and other environmental factors play a crucial role in viral spread and shape the risk landscape.

• Proximity to infected sites: The closer a fish farm is to a PD outbreak, the higher the level of threat, emphasizing the importance of geographical awareness.

• Seasonality: Like clockwork, PD outbreaks exhibit seasonal patterns, providing valuable predictive clues.

Application in the salmon industry

The new predictive model will enable salmon farmers to be proactive, intervening before pancreas disease takes hold and safeguarding their precious fish. Based on the information generated by the model, salmon producers can:

• Optimize wellboat routes: By adapting routes based on anticipated risk areas, the possibility of viral spread through transportation can be minimized.

• Adjust biosecurity measures: Knowing environmental parameters that increase risk allows specific biosecurity interventions, effectively strengthening defenses for salmon farms.

• Plan for seasonal threats: By anticipating peak periods of pancreas disease, aquaculturists can proactively increase fish resilience and implement preventive measures before danger strikes.

This innovative research is a testament to the power of collaboration. By combining publicly available data, analyses from private salmon farms, and cutting-edge machine learning, researchers have created a weapon in the fight against pancreas disease.

Conclusion

Pancreas disease poses a significant threat to the aquaculture industry, but the battle is far from over. By understanding the virus, its transmission pathways, and key risk factors, we can develop innovative solutions and predictive models. This study’s novel approach has the potential to transform the fight against PD, ensuring a healthier future for Atlantic salmon and the industry that depends on it.

The study presents a predictive model capable of forecasting SAV presence on any salmon farm in Norway using public and private data.

“It has been demonstrated that it is possible to predict the probability of ‘SAV’ detection 8 weeks before a positive ‘PD record’ using 9 parameters from the BarentsWatch database to develop 11 well-designed functions,” the researchers concluded.

With this early warning system at their disposal, salmon producers can breathe a sigh of relief. Armed with the power of prediction, they can safeguard their fish, their profits, and the future of sustainable salmon farming.

The study was funded by Zoetis and Pharmaq.

Reference
Öhlschuster, M., Comiskey, D., Kavanagh, M., Kickinger, F., Scaldaferri, C., Sigler, M., & Nilsen, P. (2023). On the prediction of SAV transmission among Norwegian aquaculture sites. Preventive Veterinary Medicine, 106095.