Understanding and monitoring mortality in farmed fish towards sustainable growth in aquaculture

Limiting mortality must be a top priority in an ethical, sustainable production of fish. Low mortality is one of several indicators of good animal welfare, and it is important to highlight the possibilities the industry has to limit the mortality in its production. The aim of this project was therefore to help the industry, the authorities and others to understand what causes mortality in the aquaculture industry. The first objective of the project was to provide an in-depth description of mortality and how it varies with season and year and between farms and geographical areas. Analyzes of data reported to Altinn, showed that mortality rates have increased for cohorts slaughtered in 2016-2018 compared to those slaughtered in 2014-2015. Further, mortality was shown to be at its greatest during the first two months after sea-transfer, with large variations between farms. Some farms report mortality up to 50% during this period. Mortality is lowest in months 3-9 after sea-transfer, with little variation between farms, but in months 12-20 there is increased mortality with some variation. It also appears that mortality is greatest in production zones with high density of farms. The results have been published in the article: Spatio-temporal variations in mortality during the seawater production phase of Atlantic salmon (Salmo salar) in Norway (https://doi.org/10.1111/jfd.13142) To give stakeholders a better opportunity to observe and compare mortality in time and space, an application has been developed that shows mortality at various levels from 2015-2019. In this app, the user can to see mortality over years, months, counties or production areas: http://apps.vetinst.no/Laksetap/ Subsequently, a model has been developed based on the same data, and with the inclusion of environmental data and treatments against salmon lice. The model showed that the number of dead fish per 1000 fish per month depends on surface temperature and salinity, production zone, weight at sea-transfer, month of sea-transfer, average weight and treatments against salmon lice. Of these, non-medicinal treatment against salmon lice is the single factor with the greatest significance for mortality, but medicinal treatment is also a significant factor. The results have been published in the article: Factors associated with baseline mortality in Norwegian Atlantic salmon farming (https://doi.org/10.1038/s41598-021-93874-6) During the project period, we gained access to data on mortality and production also in the freshwater phase. These data contain less information than those for the marine phase, but it was an important goal to try to find out if the data could be used to say something about survival and welfare in hatchery production. Therefore, we used a similar approach as for the first article on this data, and ended up with a model that describes the mortality in juvenile fish production. This is the first time this has been done in Norway, and the work has been published in the article: Mortality patterns during the freshwater production phase of salmonids in Norway (https://doi.org/10.1111/jfd.13522) Further, we analyzed production data to describe how various factors influence the daily mortality. The study included information on complete cycles from stocking to harvest, covering altogether 134 cohorts at 82 sites, supplied by five salmon producing companies. We modelled the number of dead fish each day as a function of days since stocking, weight at stocking, temperature, salinity, various lice treatments and information on pancreas disease outbreaks. We estimated that the total mortality could be reduced by about 21% if all lice treatments could be avoided, and that the total mortality could be reduced by about 21% if pancreas disease (PD) outbreaks could be avoided. As in the first study we found a high mortality right after time of stocking, decreasing over the first three months and thereafter increasing. An article is submitted to Preventive Veterinary Medicine. We have also investigated using mortality data to predict outbreaks of infectious diseases in farmed fish. Data from Production Area 3 (PO 3) from 2014 onwards were used to develop a model for expected mortality. An algorithm has been developed to to alert if mortality higher than 2% per month, or mortality higher than expected by the model was observed. PD was used to test how good the method is at detecting outbreaks, and it has been shown to detect PD outbreaks in 86 % of cases, when it was tested against the occurrence of PD in 235 fish groups. The method thus has potential for use in monitoring emerging diseases and non-notifiable diseases. We have also developed an interactive dashboard that shows mortality and with an indication of increased mortality. In collaboration with the EU project DECIDE (Decideproject.org), focus group interviews have been conducted to investigate the use of data in the management of fish health and the usefulness of this.

Gjennom prosjektet er beregning av dødelighet ved bruk av dødelighetsrater blitt introdusert og brukes til å lage de årlige oversiktene over dødelighet som publiseres fra Veterinærinstituttet gjennom for eksempel Fiskehelserapporten. Det er disse tallene det gjerne refereres til i media og som brukes som utgangspunkt for diskusjon rundt dødeligheten i norsk havbruk, både nasjonalt og internasjonalt. Metoden utmerker seg ved å gi robuste og konsistente beregninger, slik at utviklingen i dødelighet over år kan følges. Et meget viktig poeng har også vært kommunikasjon rundt den store variasjonen i dødelighet mellom anlegg, og dermed det store potensialet som ligger i å begrense dødeligheten. Med våre oversikter og analyser har det blitt satt i gang diskusjon i næringen og forvaltningen om benchmarking for å kunne redusere dødeligheten. En aktiv politikk på dette området vil naturlig medføre lavere dødelighet og bedre produksjon og fiskevelferd. Lanseringen av en app der interessenter kan se og sammenligne dødelighet over tid og rom er en viktig del av kommunikasjonen og gjennomsiktigheten rundt dødeligheten i akvakulturnæringen. Prosjektet har også for første gang tallfestet dødeligheten for den minste fisken –fisken i settefiskanleggene. Vi har belyst noen mangler ved dagens rapportering av dødelighet og produksjon for denne delen av produksjonskjeden. Prosjektet har bidratt til å sette i gang et doktorgradsarbeid som skal studere nærmere årsaker til dødelighet i settefiskfasen. I prosjektet ble ulike faktorers betydning for dødeligheten modellert både på nasjonalt og månedlig nivå med data rapportert via Altinn, og på lokalitetsnivå via daglige produksjonsdata fra produsenter. I begge modellene studerte vi effekter av ulike lusebehandlinger, og resultatene i de to modellene stemte godt overens. Det ble estimert at behandlinger mot lakselus var årsak til ca 21% av den totale dødeligheten, og at utbrudd av pankreassykdom (PD) gav en tilsvarende negativ effekt på dødeligheten. Begge modeller viste også, at fiskens overlevelse i høy grad er avhengig av sjøtemperatur og salinitet. Vi fant også at dødeligheten varierte med behandlingsmetode mot lakselus. Dette er viktig kunnskap når man skal velge type behandling og samtidig minimere tap. I prosjektet er utviklet en algoritme som gir signal dersom dødeligheten i et anlegg overstiger 2% eller går over forventet dødelighet en måned. Hensikten er å kunne bruke dette systemet til å varsle dersom det skjer plutselige endringer i dødelighetsmønstret. Et slikt system vil kunne brukes for eksempel av forvaltningen i deres arbeid med overvåking av helse i akvakultur. I tillegg til de oppnådde resultatene, har prosjektet også fungert som et springbrett for Veterinærinstituttets deltakelse i EU-prosjektet DECIDE (decide.org), der målet er bedre integrasjon av data for å kunne lage nyttige og brukbare beslutningsstøtteverktøy.

The ultimate goal of this project is to help producers, competent authorities and other stakeholders to understand drivers of mortality in aquaculture and provide them with tools to monitor and mitigate mortality in farmed fish. Low mortality can generally be regarded as a first indicator of good welfare, and minimizing mortality is a top-priority in an ethical, sustainable production of fish. The project will build on experience from producers, who have an extensive knowledge on individual factors driving mortality, but who have to adhere to regulations against specific diseases or environmental challenges that might be detrimental to the control of other adverse effects. The project will utilize extensive time series of 16 years of monthly data reported from all active Norwegian aquaculture farms for a model and a description of spatio-temporal differences in baseline mortality. The project will further develop a system that uses monthly reporting of production data to identify events of unusual mortality patterns in a syndrome-based surveillance system. Such events can be indicative of disease outbreaks or failures in management routines, and thus help stakeholders to activate prompt investigations and suitable responses. In order to reach a goal stated by the politicians of quintupling the aquaculture production by 2050 the producers are implementing new technologies, and changing their ways of producing. Measuring mortality and comparing to a baseline mortality will provide a robust tool to evaluate the impact on fish welfare due to these changes. High-resolution data from a subset of the producers will be used in a refined model that can calculate the effects of individual managemental factors on disease-driven mortality dependent on time and place of the application of these factors. Thus, the primary objective of the project is to disentangle the drivers for mortality and provide inputs for discussion on regulation of diseases.