Search

Mortality represents a major challenge in the salmon farming industry. In Norway the mortality in the marine phase of the salmon production cycle has been 15-20% annually the recent years. This situation is not sustainable for the fish, the farmers or society. To prevent mortality throughout the production cycle, fish-groups at risk of deteriorating health must be identified early, regarding both disease development and stage in the production cycle. Hence, robust population health monitoring methods are needed, which will provide farmers with information to make knowledge-based decisions to improve fish health. The aim of the thesis was to investigate monitoring methods that can be used to improve fish health management in salmon farming. This was approached through four studies investigating mortality patterns, cause-specific mortality classification, gill mucous cell count and gill mucus microbial composition, respectively, as potential factors relevant to monitoring fish health status in salmon farming. Mortality patterns in salmon farming were investigated through a retrospective study of fish-groups from two commercial fish farming companies, describing the cause-specific mortality during the production cycle and investigating sources of variation in mortality between the hatchery, marine farm, and the fish-group. Most of the variation in mortality was attributed to the fish-group. Based on the information from the cause-specific mortality registrations, smolt-related mortality was found to be the major cause of death during the first six months of production, while handling and treatment was the overall dominating cause of death in the full production cycle. However, this varied extensively between the fish-groups. The second study further explored the system of cause-specific mortality and suggested a unique classification code. A system for classifying causes of mortality based on underlying cause was created for salmon farming inspired by the human mortality classification system. The proposed standard has a three-level hierarchical structure of mortality causes. This accommodates different levels of details when mortality causes are registered and enhance the information possible to retrieve from the system. To target early responses of the immune system, variation in gill mucous cell count was studied as a potential method of monitoring health status. Salmon presmolt from six commercial hatcheries was sampled and mucous cells were histologically quantified based on certain criteria. The counts varied among both fish and farms. When “farm” was included as an independent variable in a regression model the proportion of variation in mucous cell counts explained by the model was twice as high compared to when only fish size was included. This indicates that the variation depends on farm-related factors. A subset of the fish (from four recirculating aquaculture system (RAS) facilities) included in the study of mucous cells was also sampled for gill mucus for microbial composition analyses using 16S rRNA gene sequencing. Quantification of extracted bacterial DNA in the gill mucus samples showed low levels in general, but fish from one farm had considerably more bacterial DNA compared to the others. Samples from the same farm were also compositionally different based on beta diversity metrics, compared to the others as one group. Assessed by the Shannon index as an outcome in a regression model, sources of variation attributed most variation to the individual fish, suggesting gill microbial structure was linked to the individual fish. No associations between gill microbial diversity and specific production parameters were detected. In conclusion, cause-specific mortality classification was found to identify important causes of mortality in the production cycle, thus providing farmers with substantial health information with limited effort and at little cost. The tool can further be applied at desired level of detail and adapted to the needs of each farmer. Novel methods of gill mucous cell count and microbial composition identifies variation among farms and individual fish for the different measured parameters. This variation can potentially be exploited to detect early signs of reduced health in the fish. However, further studies are necessary to establish the causal association between these factors and fish health before they can be used in systematic monitoring of health in salmon farming. Dødelighet i oppdrettsnæringen er en stor utfordring i dag. Norsk lakseoppdrett har de siste årene hatt mellom 15 og 20% dødelighet i sjøfasen. Dette er ikke bærekraftig, hverken for fisken, oppdretterne eller samfunnet. For å forebygge dødelighet gjennom produksjonssyklusen må fiskegrupper som har økt risiko for dårligere helse identifiseres tidlig. Derfor er robuste overvåkingsmetoder på populasjonsnivå nødvendige, det vil gi oppdretterne nyttig informasjon for å kunne ta kunnskapsbaserte avgjørelser i arbeidet med å forbedre helsesituasjonen til fisken. Målet med doktorgraden var å undersøke overvåkingsmetoder som kan brukes til å forbedre helsestyringen i oppdrettsnæringen. Dette ble gjort ved å undersøke mulige faktorer som kan gi informasjon om helsestatus, henholdsvis dødelighetsmønster, årsaks-spesifikk dødelighetsklassifisering, slimceller på gjellene og mikrobiell sammensetting på gjellenes slimlag.

Lakselus (Lepeophtheirus salmonis), er ein parasitt som er ei stor utfordring for vill laksefisk og for oppdrettsnæringa. Denne oppgåva er skriven i samarbeid med "Taskforce Lakselus", som er eit FoU-prosjekt, der målet er å "etablere grunnleggjande kunnskapar om korleis lakselus angrip oppdrettslaks og mekanismane om korleis parasitten spreier seg innad og mellom oppdrettslaks og ville bestandar av laks. Teljedata om lakselus vart lasta ned frå Barentswatch, Fiskeridirektoratet og eKlima. Barentswatch har informasjon om aktivitet i kyst- og sjøområde, og baserer seg på samarbeid mellom norske, statlege etatar og forskningsinstitusjonar. Fiskeridirektoratet er ansvarlege for fiskeri- og havbruksforvaltninga i Noreg. eKlima er ei nettside som gir tilgang til databasen for klimadata frå Meterologisk Institutt. For å analysere dataane har generaliserte lineære modellar (GLMar) som ein Poisson regresjonsmodell, ein quasi-Poisson regresjonsmodell, ein negativ-binomisk regresjonsmodell og ein null-inflatert negativ binomisk regresjonsmodell blitt tilpassa. Eit regresjonstre og ein random forest-analyse vart deretter laga. Til slutt vart det tilpassa ein generalisert additiv modell (GAM). Studien antyda at GLM-ane som vart brukt, ikkje gav ei god tilpasning til dataane. The salmon louse (Lepeophtheirus salmonis), is a parasite that represents direct threats to wild salmonids and indirectly threats to salmon farmers and farmed salmon. This thesis is written in cooperation with ”Taskforce Salmon lice”, a R & D project, where the objectiveis to ”establish fundamental knowledge on how sea lice infest farmed salmon and the mechanisms of how the parasites spread within and between farmed and wild populations of salmonids”. Salmon lice count data were downloaded from the sources Barentswatch, the Nor-wegian Directorate of Fisheries and eKlima. Barentswatch gives information on activityin sea and coastal areas, and is based on cooperation between different Norwegian state agencies and research institutes. The Norwegian Directorate of Fisheries is responsible for fisheries and aquaculture management. Eklima is a web site which gives access to the database of climate data of the Norwegian Meteorological Institute. To analyse the data, generalized linear models (GLMs) such as a Poisson regression model, a quasi-Poisson regression model, a negative binomial regression model and a zero-inflated negative binomial regression model were fitted. A regression tree and a random forest analysis were applied. Finally, a generalized additive model (GAM) was fitted. The study suggested that the fitted GLMs did not give a good fit to the data.