Summary and status of deep lights and deep feed use in commercial settings: welfare, behaviour and infestation at three case study sites

The national growth of the Atlantic salmon (Salmo salar) aquaculture industry has cemented Norway as the leading producer in the world, boasting 53% of global production in 2015. Expansion has been facilitated by the industry’s ability to refine and optimise production methods whilst responding rapidly to issues that arise, with innovative solutions that stem from both research and commercial resourcefulness. One of the major challenges that salmon aquaculture industries face is the issue of salmon lice (Lepeophtheirus salmonis) infestations, which is restricting the potential expansion of the Norwegian industry (Olaussen 2018). Salmon lice are an ectoparasite that have dramatically proliferated in parallel to the increasing abundance of farmed Atlantic salmon, prompting action by the Norwegian government to strongly regulate production limits. Farmers will be allowed to increase their producible biomass depending on the infection ‘status’ of their regions (see Myksvoll et al. 2018). Therefore, there is strong pressure on farming companies to control and minimise the salmon lice infestation levels on their sites.

Strategies to combat lice are critical for disease management, with preventative measures being of high priority. This becomes increasingly important given the fact that the majority chemical compounds typically used to treat against lice have lost their effect completely or lice have become increasingly less sensitive to the active substance (Aaen et al. 2015). One strategy to mediate salmon lice infestations is depth-related prevention methods, where the distribution of the salmon in the water column is manipulated to mismatch the distribution of infective lice. Passive guidance of salmon swimming depth has been achieved using attractant lights and the provision of feed away from the surface (e.g. Oppedal et al 2011).

This report aims to summarise the current status of knowledge for the use of submerged lights and feed zones as a means for lice prevention, and synergise recent studies that have tested this method at a commercial scale.

1.1 - Lice larvae distribution with depth

The salmon louse has 8 life-history stages, where the youngest stage, nauplius 1 are hatched from eggs carried in egg strings by the adult female; they are lecithotrophic and drift in this planktonic stage, undergoing two moults until they reach the infective copepodid stage (Hamre et al. 2013). It is this stage that starts exhibiting host-searching behaviours, whereby copepodids will actively disperse themselves near the surface of the water column. They are positively phototactic, but avoid brackish salinities that often are present in coastal or fjord waters (Heuch 1995, Genna et al. 2005). This strategy enables the largest possibility of encounter with salmonid hosts, who swim relatively shallow during their outmigration (Thorstad et al. 2012, Eldøy et al. 2017). Once a potential host is located, copepodids attach to the host and moult through two chalimus stages at that fixed position (termed sessile stages). When they reach the pre-adult 1 stage, they become mobile and begin to move around the host; after an additional pre-adult stage, lice moult into adults and females become reproductively active.

The concept of depth-related lice prevention technologies in aquaculture is to draw the salmon school away from the most parasite-risky surface waters, either through cage manipulations (e.g. snorkel structures, or submerged cages) or passive attraction using positive stimuli. Here, we focus on the latter method, with the use of submerged lights and provision of a deep feeding point as attractants.

1.2 - Use of light for influencing salmon behaviour and controlling sexual maturation

The observation that artificial lights, long used in salmon production to reduce early maturation, are highly attractive to salmon (Juell and Fosseidengen 2004, Oppedal et al. 2011, Wright et al. 2015) prompted researchers to explore the application of lights as a passive means of influencing salmon swimming depth. However, the use of artificial lights during a full production cycle has a potential negative side effect. Photoperiod, the alternation of light and dark in a 24-hour cycle, is a key environmental cue which modulates growth, reproduction and behaviour in fish (Falcón et al. 2007, Iigo et al. 2007). Because salmonids appear to lack any endogenous clock, production of the biological time keeping hormone, melatonin, is controlled entirely by photoreception in the retina and pineal gland (Falcón et al. 2007, Iigo et al. 2007). As a result, photoperiod modification with artificial lights has profound implications for the growth and development of farmed salmon. Currently, artificial lighting is widely used in salmon aquaculture to inhibit maturation in spring (Leclercq et al. 2011, Hansen et al., 2017). By super-imposing artificial lights at night on the natural photoperiod from January to June, fish skip directly from the short daylengths of winter to the long daylengths of summer, and thus are signalled to have missed the spring window of opportunity for maturation (Migaud et al. 2010). However, while photoperiod advancement in spring inhibits maturation, extending summer light conditions into autumn can do the opposite and stimulate sexual development (Fjelldal et al. 2011, Fjelldal et al. 2018). Thus, behavioural modification using submerged lights is only a viable lice prevention measure if the photic signals which initiate sexual maturation in salmon are understood and accounted for.

Powerful, full spectrum (white) metal-halide lights from 400 to 1000W have traditionally been used to prevent maturation in the spring, but narrow-spectrum violet LED lights (peak at 400 nm) have been developed for use as an attractant the rest of the year. These lights are built to emit wavelengths with the least physiological effect on salmon, and light intensity from these 100W lights can be decreased as low as 0.1 μE cm^-2 sec^-1 (irradiance measured at 1 m) as compared to regular anti-maturation lights with an irradiance of 80-140 at the same distance (Migaud et al., 2006, Leqlercq et al., 2011; Hansen et al., 2017). Studies on S1 smolts have shown that prolonged exposure to these lights did not induce increased maturation (Hansen et al., in prep).

Thus, if lights are deployed and kept lights at 10 m depth during a complete production cycle, we may be able to continuously keep the fish deeper in the pen during night-time. Proposed regimes include using violet UV-LED from first autumn to midwinter, anti-maturation lights at the same depth from mid-winter to mid-summer, and then reverted back to violet UV-LED from mid-summer and until harvest. By implementing lights over an entire production cycle, we can also determine whether or not light attraction can override temperature (vertical gradients) which in the literature have been recognized as the main driver of vertical positioning in farmed salmon (Oppedal et al. 2011).

1.3 - The influence of feed on salmon swimming depth and behaviour

The provision of feed is a strong stimulus for salmon, whereby many of their behaviours can be anticipatory or related to feeding. As appetite is a strong driver of behaviour, salmon can be conditioned to new environments or negative stimuli (such as aversive flashing lights) through food reward (Bratland et al. 2010). Thus, although lights may keep the fish deep during the night, surface-feeding during day-time will still attract the fish towards the surface during feeding (Bjordal et al. 1993, Juell et al. 1994), and hungry fish will in addition be attracted to the surface at times with no feeding (Juell et al. 1994, Frenzl et al. 2014). Contrary to trickle-feeding where fish will spend more time in the surface-layers where infective lice-stages are most abundant, shorter and more intensively distributed meals will leave the fish more time to position itself according to other preferences than feeding / hunger level (Bjordal et al. 1993, Juell et al. 1994), i.e. feeding motivation declines below some threshold level where other parameters are prioritized. Both empirical knowledge, also from several Marine Harvest sites, and a study done by Frenzl et al. (2014) show that salmon may be attracted towards greater depths when the feeding-area is re-situated from the surface and down to 5-8 m depth. The same study indicated that underfed fish are still attracted to the surface even if feed is distributed under water. This illustrates the importance of avoiding near continuous trickle-feeding and give priority to more intense feeding that may rapidly lower feeding motivation and therefore also tendencies to move towards the surface.

1.4 - Use of deep lights and feed as a lice prevention method

The negative correlation between swimming depth of Atlantic salmon and their experienced lice infection pressure or lice levels in general is well described (Huse and Holm 1993, Osland et al. 2001, Hevrøy et al. 2003, Stien et al 2016; Oppedal et al. 2017, Wright et al. 2017). The principle is shown in Figure 1. Similar results have been found when using skirts around farming cages to shield salmon from the upper water layers (Næs et al. 2012, Lien et al. 2014; Stien et al 2018) where infective stages of L. salmonis are most abundant (Heuch 1995, Heuch et al. 1995).

The approach of using lights and feed to attract salmon deeper in to the cage has a promising potential as a prevention strategy. Salmon can be attracted to swim deeper in the sea cage by use of lights (Juell et al. 2003, Juell and Fosseidengen 2004, Frenzl et al. 2014, Stien et al. 2014) and by moving the feeding zone to match the lights’ depth (Frenzl et al. 2014, Nilsson et al. 2017). If successful, salmon will thereby be avoiding the shallow waters except for infrequent forays to the surface to refill their swim bladders (Korsøen et al. 2012). Positioning salmon deeper in the water column, whether with cage structures or otherwise, has been proven as an efficient way to prevent infestation (see above). However, the effectiveness of reducing infestation may be highly dependent on seasonal variation in salinity stratification and water temperature. If a brackish layer is present at the surface, sea lice larvae may actively avoid low-salinity waters and aggregate below the halocline (Heuch 1995, Wright et al 2018) and the infective zone can be brought deeper in the water column. Alternatively, constant or periodic hydrodynamic mixing can also transport larvae to meet the deeper-swimming salmon. In contrast, warmer temperatures can also override phototactic preferences in the fish, resulting in shallow swimming if optimal temperatures are towards the surface (Oppedal et al. 2007, Oppedal et al. 2011, Stien et al. 2016). As deep lights can attract fish to greater depth and stimulate schooling during night-time, and granted that deep feeding (removing the surface as a feeding area) stimulate fish to descend deeper in the cage during day-time, the combination of the two should in theory imply that the fish will stay deeper in the pen for significantly longer periods than will fish in control pens, and thus lice infection rates and levels in general should be reduced.

1.5 - Case studies

Three commercial sites provided case studies for the case of using submerged lights and feeding systems in a commercial production setting. Two sites were located in the Rogaland region of south-western Norway; Kobbavika was located at a relatively open site towards the entrance of Boknafjord (Nilsson et al 2017), whereas Vindsvik was situated in a narrow, more inland fjord that branched off Boknafjord (Bui et al 2018). Skrubbholmen is located further north, in Nord-Trøndelag region of Norway, at an open-water site with little salinity stratification (Nilsson et al 2018).

2.1 - Experimental setup

On all sites the principal setup was to use submerged lights positioned deep in the cages and deep feeding of which were to attract the salmon night and day to swim deeper, preferably under the so called “lice belt”, and thereby be infested with less lice (Figure 2). Control cages were originally to have surface feeding and no lights.

Kobbavika: The site is positioned in Rogaland, Fognafjorden, on the south-east side of the island Fogn. The site consists of 8 circular cages, of which only 4 were used in the trial (2 surface fed control cages, 2 deep fed cages). Cages were 200 m (cone-nets) in circumference, with a depth of 35m (and later 45m), initially with a pen-wise volume of 47 745 m³ stocked with 196-198 000 smolts from transfer. All cages had cleaner fish: around 4% at the start of the trial and between 13 and 18% at the end of the trial when the number of salmon had been reduced due to harvest. The study period was from December 2015 until December 2016.

In the two treatment cages, systems for deep feeding at 7 m depth (AKVA SubFeeder; AKVA Group, Norway) was installed for the entire production cycle, with the exception of shorter periods when they were removed for technical reasons. The control cages were fed at the surface. In addition to deep feeding, nine low-intensity violet LED-UV lights were installed at 10 m depth in treatment cages from trial start to 25th January, and from 15th July to end of trial, with the exception of shorter periods when they were removed for technical reasons. From 25th January to June, standard anti-maturation lights with higher intensities were used in all 4 cages, placed at the same depth as LED-UV lights (10 m).

Vindsvik: The site was located in the same region as Kobbavika, but further inland in the long, narrow fjord of Jøsenfjorden. The experimental site had 12 circular sea cages (circumference = 120 m, 35 m deep) in two parallel rows. In September 2016, approximately 65 000 smolt (~100 g) were transferred into each cage and raised with standard production procedure throughout the study. All cages were stocked with cleaner fish throughout the production cycle, using whichever species were available at the time. Cleaner fish species used included the ballan wrasse (Labrus bergylta), lumpfish (Cyclopterus lumpus), corkwing wrasse (Symphodus melops) and goldsinny wrasse (Ctenolabrus rupestris). Intended stocking density of cleaner fish was to be 5% of salmon number per cage, with restocking multiple times during the experimental period. The study period was from October 2016 until December 2017.

Cages were assigned to one of four treatment groups (3 replicate cages per treatment), however in this report, only 3 groups will be represented: surface-fed cages (cleaner fish and functional feed provided), deep lights + feed cages (cleaner fish, functional feed, and deep lights + feed), and deep lights + feed + skirt cages (cleaner fish, functional feed, deep lights + feed, and a lice skirt).

Functional feed (Skretting Shield, Skretting, Norway) was provided from trial initiation until trial completion. In December 2016, relevant treatment cages had a structure installed in the centre that supplied feed at 7 m depth (AKVA SubFeeder; AKVA Group, Norway). In addition, five UV LED lights (Aurora SubLED Combi, AKVA Group, Norway) were suspended between 7 to 10 m depth. The lights emitted a deep violet (120 W) colour from end of Dec 2016 – mid-Jan 2017, a green-blue anti-maturation (600 W) colour from mid-Jan – mid-June 2017, then returned to deep violet colour thereafter. At the same time, a semi-permeable canvas lice skirt (Norwegian Weather Protection, Norway) was installed outside of skirt treatment cages that extended from the surface down to 6 m.

Skrubbholmen: The site was located in a relatively exposed area south of Nærøysund in Nord-Trøndelag. Four cages (circumference = 157 m, 15 m deep + cone) were used in the study. The cages were stocked with 162 000 salmon in each, transferred on 22 May 2017. The fish were around 2.7 kg on average at trial start and 3.3 kg at trial termination. All cages had 5 m deep lice skirts and 6 LED lights at 9 m depth. The lights were of the same type as used at Vindsvik (AKVA Aurora SubLED Combi), with the green-blue anti-maturation colour switch on continuously throughout the study period. Two of the cages were fed at 7 m depth by the AKVA SubFeeder (AKVA Group, Norway). The remaining two control cages were fed at the surface. Prior to the trial all fish had been surface fed. The study period was from 24 January until 22 March 2018.

2.2 - Welfare and infestation assessments

Lice have a temperature-dependent growth rate, whereby the time taken to reach the mobile pre-adult stage is between 6 days (at 21°C) to 69 days (at 3°C) post-infestation for males (S. Dalvin, unpublished data). For evaluating the effectiveness of deep lights and feeding on preventing infestation, we only use data on abundances of sessile lice stages (i.e. copepodid, chalimus 1 and chalimus 2 stages) found on salmon. This is due to the uncertainty of the grazing effect cleaner fish had on mobile stages, and the occurrence of delousing treatments that often only target mobile stages.

Procedures for assessing welfare and recording infestation status was the same for all sites. Every 3 to 4 weeks during the study period (except during summer at Kobbavika), fish were sampled and assessed for lice infestation levels and welfare status. Fish were captured using a hand net at the surface, a seine net from the surface and 10 m deep, or small ring-net pulled from 5 or 10 m depth to the surface by a boat crane; cages were sampled using the same method at each separate sample event. At Vindsvik, sampling was conducted a minimum of 3 weeks after any delousing treatment, so that new infestations were certain to be unaffected by previous treatments. At Kobbavika, there were delousing events approximately two weeks before Sample 3 (Cage 4 only) and Sample 4 (Cage 3 only); these has not been excluded from analysis. At each sampling point, 20 salmon (occasionally more) from each cage were euthanised and any lice attached were quantified and staged. Also lice in the anesthetization tub was counted, and all lice were counted and divided by the total number of fish sampled. Each fish was also evaluated using the standardised SWIM 1 and 1.1 model (Stien et al. 2013), which involves the scoring of 14 indicators of welfare ranging from undamaged ormal to severely damaged/abnormal (Table 2). Welfare indicators are weighted in the model and are used to calculate an overall welfare index, a value bounded from 0 (worst) to 1 (best) that represents an individual’s welfare status. Gill diseases were also recorded, specially presence of PGI and AGD, however the prevalence of these diseases were negligible throughout the study.

2.3 - Monitoring of swimming depth and environment

To monitor group swimming behaviour, swimming depth distributions were continuously recorded using a PC-based echo integration system (Lindem Data Acquisition, Oslo, Norway; Bjordal et al. 1993, now CageEye AS, cageeye.no). The system includes a transducer submerged at ~30 m deep inside every cage, positioned to face upwards with a 42° acoustic beam. The strength of the returned echo signal indicates the presence of fish, with higher signal strengths indicating more dense groups of individuals. Environmental conditions of the water column were profiled daily using a CTD sensor (SD204, SAIV AS, Bergen, Norway). Profiles of temperature and salinity were taken from the surface down to 40 m depth, dependent on depth at the site, at a reference point outside of the experimental cages.

2.4 - Data handling

Skrubbholmen lacks comprehensive data for swimming depth and therefore we only represent and discuss lice data for this short experimental period.

Lice stages were categorised by whether they were new infestations since the previous sampling (copepodid, chalimus 1 and 2 stages), or could possibly have been present at the previous sampling (pre-adult 1 and 2, and adult stages). All lice considered a new infestation were summed and averaged by the number of individuals sampled per cage. This value was used to compare the efficiency of treatments on prevention of new infestation, using a generalised linear model (GLM). The model included treatment, site, and sample date as predictor variables, with cage number as random effect and used a Poisson distribution (package ‘glmmTMB’, function glmmTMB in R). Date of lice attachment was back-calculated for new infestations stages using their temperature-dependent development rate (S. Dalvin, unpublished data) based on the average sea temperature 2-3 weeks prior to the sample event.

Existing infestations (mobile lice stages) were not analysed as many confounding factors could affect the value (such as actual cleaner fish stocking density or recent delousing treatments), which could vary between sample points.

Echosounder data was frequently unavailable due to equipment damage or incorrect placement, particularly at the Skrubbholmen site. Thus, analyses of salmon swimming depth were only conducted on the Kobbavik and Vindsvik sites. The data provided information on the vertical dispersion of the salmon; occasionally the school exhibited a bimodal distribution where there were 2 (and rarely, 3) main groups within the school. The swimming depth of salmon was estimated using the relative vertical distribution seen from the echosounder data, at 1-m depth intervals whereby more than 4% of fish were present. For analysis of school vertical distribution in relation to halocline depth, values for daily swimming depth were calculated as the median depth of the upper (shallower) school from the mean of their upper and lower depth limits, pooled across all hours within the day.

As copepodids gradually avoid brackish waters, particularly of 28 ppt or fresher (Wright et al 2018), a stratification with salinity value of 28 ppt was considered the halocline for this study. The estimated depth of the halocline was determined by the deepest depth with a salinity of 28 ppt or lower, calculated daily. The relationship between the distance of median salmon swimming depth from the halocline depth (and therefore assumed depth of infective copepodids), and actual lice infestation (back-calculated from stages) was assessed for the study period, whereby only days when all three parameters (environment, salmon swim depth, and lice attachment rate) were available for a replicate cage were used. Intensity of lice acquired per fish was analysed with a GLM using distance from halocline (as an absolute value) and site as predictor variables, with cage as a random factor. Data were pooled across treatments, as distance from halocline was the most interesting factor and is intrinsically linked with the influence of deep lights and feed on school swimming depth.

Overall Welfare Indexes (OWI) were averaged across individuals in a cage and compared among groups using a GLM. Similar to the lice model, treatment, site and sample date were included as predictor variables in the model.

3.1 - Salmon welfare

Overall welfare, as calculated using the SWIM model, was similar between standard cages and those with deep lights and feed; the null model was not different from the full model that included treatment (χ² = 0.277). The full model therefore did not show a significant effect of deep lights and feed on welfare compared to control cages, but OWI was different between sites (Vindsvik from the other two; z = -7.61, p < 0.001) and among sample points (z = -4.70, p < 0.001). For Kobbavika, OWI was similar among all cages for the 12-month study period, with a small non-significant drift of treatment cages in autumn months (Figure 3). At Skrubbholmen, the shorter study period showed almost identical welfare statuses among cages. Vindsvik was more variable, with the largest discrepancy between groups also occurring in autumn; however, it was the surface-fed cages that suffered lower welfare status in August to October, reaching the lowest OWI score of 0.56 (Figure 3). The addition of a lice skirt did not affect welfare, however scores were lower than deep feed and lights cages in October to December.

3.2 - Environment and swimming depth

The exposed site Skrubbholmen had relatively homogenous water with temperature and salinity changing little with depth, and no halocline was present at this site. As this site was studied during winter-spring, the temperatures were low (< 4 °C) and falling throughout the study period (Figure 4). Kobbavika was studied over almost a year and temperature changed both with time and depth. Kobbavika is located in a fjord and therefore affected by freshwater, and a halocline (salinity < 28 ppt) was present for long periods, usually between 5 and 10 m depth (Figure 4). Vindsvik was also studied over a year, with temperature changing with time and depth. Vindsvik is located in a more inland fjord site, and is more affected by freshwater than Kobbavika, with a more pronounced and partly deeper halocline (Figure 4).

Brackish layers were present almost throughout the study period, sometimes extending below the depth of the deep lights and feeders (Figure 4).

At Kobbavika, salmon in all cages swam relatively shallow in the first months of the study, when there was little temperature difference in the water column, and generally swam between 5 to 15 m over the study period. For Kobbavika, there was little difference in the swimming depths of treatment and control cages over time. At Vindsvik, fish exposed to deep lights and feed began to swim deeper than control fish almost straight away, but then fluctuated between matching swimming depths and swimming deeper between February to April. During this period, fish were more likely to swim at the same depth if there was no thermal stratification. From June to August, swimming depths were very similar between groups, as the preference for the thermocline was higher than any attraction to lights. Larger differences in swimming depths were then observed for both sites in the late autumn to early winter months, but with contrasting patterns: treatment cages swam shallower than control fish during this period at Kobbavika, but deeper at the Vindsvik site (Figure 4).

3.3 - Lice infestation status

Across the study period and case study sites, the provision of deep lights and feed did not influence the number of lice salmon acquired. The full model was significantly different from the null model which excluded treatment (χ² = 0.001), however the variation that treatment contributed to lice was largely driven by the lice skirts in the full model (deep lights and feed and skirt cages compared to deep lights and feed: z = -2.89, p = 0.004; control cages compared to deep lights and feed only treatment: z = 1.71, p = 0.087). Control cages Site and sample date were significant factors in the model (Skrubbholmen compared to Kobbavika: z = -3.30, p < 0.001, and Sample: z = 4.60, p < 0.001, respectively).

The rate of lice acquisition over the study period varied, with relative peaks occurring in July onwards (Figure 5) at both Kobbavika and Vindsvik sites. The period of study for Skrubbholmen was in winter months, and therefore overall lice levels were low throughout. The exposed and halocline conditions of Kobbavika resulted in higher lice levels of almost 8 lice per fish, compared to the highest level of nearly 3 lice per fish in Vindsvik (Figure 5).

The largest spikes in infestation rates occurred in the latter part of the year for both Kobbavika and Vindsvik, but did not necessarily correlate to periods with particular halocline characteristics (Figure 6). Although the salmon responded to multiple environmental variables that fluctuated with time, there was a consistent difference in swimming depth at Vindsvik when time-pooled means were estimated. The average median swimming depth of the shallowest school, over the entire study period, was 5.55 m for cages without deep lights and feed, compared to 11.05 m for deep lights/feed and 11.49 m for deep lights/feed + skirt cages. This translated to an average of ~ 0.16 m above the halocline for control cages over the period, whereas both deep lights and feed groups were ~5.76 and 6.79 m below the halocline, but with larger variation between days (Figure 6). However, at Kobbavika, average median swimming depth was very similar between control and treatment cages (5 and 6 m, respectively), which meant a similar distance from halocline over time (~1.8 and 2.8 m, respectively; Figure 6).

Numerous assumptions were made when calculating the daily infection success (i.e. date of lice attachment) and swimming depth of the school, however an interesting pattern arises when looking at the relationship between the schools’ relative distribution around the halocline and the number of lice acquired (Figure 7). The correlation between distance from halocline and lice acquired was weak but non-significant for the model that included all data (z = -1.83, p = 0.067). However, salmon in cages that also had a lice skirt are theoretically not interacting with the halocline in the same way if it were within the skirt depth, so these data were removed for a subsequent analysis. When pooling data only from control cages and those with deep lights and feed, the model thereby demonstrates a significant relationship between larger distances away from the halocline and lice acquired (z = -2.40, p = 0.017), with site also as a significant factor (z = -3.80, p < 0.001). To investigate whether swimming depth, irrespective of halocline, influenced the rate of infestation, a GLM was used with the same variables included as the distance from halocline model but inserting swim depth data instead. This model did not reveal a difference in lice acquired due to swimming depth but suggested that a trend was present with (z = -1.66, p = 0.096) and without the skirt cages included (z = -1.85, p = 0.064).

4.1 - Impact of deep lights and feeders on salmon welfare

The implementation of deep lights and a deep feeding zone was not anticipated to affect the physical condition of salmon, but welfare indicators related to growth could possibly be influenced by a deeper feeding zone. The results from these case studies using deep lights and feed for 14 months showed that there was no long-term effect of this technology on salmon welfare, and in the case of Vindsvik, could actually provide an advantage over standard cages. Short-term declines in welfare status at Vindsvik could be driven by higher lice infestation levels in standard cages, or indicators such as eye status (Bui et al 2018). At Kobbavika, the largest difference in Overall Welfare Index (OWI) was 0.05, which is a negligible variation. Welfare status at Skrubbholmen was exceedingly similar over the 2 sample points during the experimental period, demonstrating no short-term effect on applying deep lights and feed at later stages of the production cycle.

Using the Salmon Welfare Index Model (SWIM) to assess overall welfare status is an invaluable tool to compare large groups of salmon that are exposed to different experimental conditions; the OWI over time for these sites were ~ 0.76 (Vindsvik), ~ 0.84 (Kobbavika), or ~ 0.83 (Skrubbholmen) which is comparable to standard commercial production cages (Folkedal et al 2016). Kobbavika and Vindsvik sites showed a temporal decline in OWI, with a maximum reduction of 0.16 or 0.17 in OWI, respectively; this is not uncommon for salmon held in production 14 months after sea-transfer.

In terms of the feed uptake of salmon with the deep feeding zone, the indirect measure of coping could be growth of these groups compared to those with standard surface feed. At Vindsvik, cages with deep feed grew to ~ 8.5 times their original sampled body weight (i.e. from 322 g to 2701 g), compared to ~ 10.5 in standard-fed cages. This discrepancy could be due to reduced feed intake either because of the feeding method, or because they experienced alternative temperatures, or a combination of both. Alternatively, the recorded weights could be confounded by unrepresentative sampling, whereby emaciated or smaller-class fish were captured instead of individuals from the deeper-swimming main school (see discussion below). In contrast, the deep fed fish at Kobbavika grew more (~ 7 times heavier at sampling 12 than at sampling 1, estimated on sampled fish) than the surface fed controls (~ 6 times heavier). At both sites, growth predicted by the managing program estimated an increase of 7.4 and 15.9 times for the deep fed fish (at Kobbavika and Vindsvik, respectively) and 7.1 and 16 times for the surface fed. This may indicate some sampling bias and suggests a smaller deviation in growth between the groups that were fed deep and those with standard feed regimes.

4.2 - Swimming depth of salmon in relation to light and feed system

The difference in swimming depth between deep light and feed attracted fish was small and inconsistent at Kobbavika, even if there were tendencies of deep attracted fish swimming deeper during periods with little temperature stratification. In contrast, the deep attracted fish at Vindsvik swam on average 5.5 m deeper than the control fish, but with small difference when there was a clear temperature stratification with warmer water near the surface (e.g. July). Manipulation of swimming depth with the use of lights has been demonstrated in several studies (Juell et al. 2003, Juell and Fosseidengen 2004, Oppedal et al. 2007, Frenzl et al. 2014), and also with weak violet lights as used when the anti-maturation lights were switched off in the present study (Stien et al. 2014; Wright et al. 2015). Salmon do however trade-off between illumination and temperature. If temperature is slightly lower at the illuminated depth than at other depths, salmon prefer the higher temperature during the day and illumination during night, while at larger differences in temperature the higher temperature is preferred also at night (Oppedal et al. 2007). This is in accordance with the observations at Vindsvik with all groups swimming in the warmer water above the illuminated zone during the summer while salmon with deep light swam deeper when the temperature stratification was smaller. Also at Kobbavika all fish were swimming above the illuminated depth when the temperature was highest near the surface during the spring, and there were tendencies of fish with deep lights swimming deeper during late summer when the warm water was less stratified. Generally, the differences in swimming depth between the Kobbavika groups was small with all groups swimming relatively near the illumination depth most of the study period, which may explain the minor effect of the light and feed attractants.

During the spring, also the surface fed control groups had lights at the same depth as the treatment groups, in all sites. Still, deep fed fish at Vindsvik swam deeper than control fish during much of this period, suggesting that deep feeding attracted salmon to swim deeper at this site. In the period with anti-maturation lights when the only difference between groups was feeding depth, all groups in Kobbavika swam at depths similar to the feeding depth most of the period, except when warmer surface water attracted them to swim higher, and thus no clear effect of feeding depth was found. In the study Frenzl et al. (2014) there was no clear additional effect on swimming depth of deep lights and deep feeding compared to deep light only, but the dubious results could be because some of the deep fed fish that did not find the food stayed close to the surface as they became hungry (Juell et al. 1994). Trials with groups without any additional light that are either fed at the surface or deep fed would be more suited to study the isolated effect of deep feeding on salmon swimming depth.

4.3 - Lice infestation rate in relation to swimming depth and distance from the halocline

Infective lice larvae are positively phototactic and are therefore more abundant towards the surface (Heuch 1995), and previous findings have found that infestation is negatively affected by salmon swimming depth (Osland et al. 2001, Hevrøy et al. 2003, Oppedal et al. 2017, Wright et al. 2017). In the present study lice infestation rates was not clearly correlated by swimming depth. However, as lice larvae avoid brackish water (Heuch 1995) the maximum density of larvae and thus probability of being infested is not at the surface if a halocline with brackish water above is present, which was the case during large parts of the study period in both Kobbavika and Vindsvik sites. Lice would then be most abundant as near the sun light as possible but within the preferred salinity, i.e. just below the halocline. In accordance with this, distance from halocline explained the lice infestation rates better than swimming depth per se in the present study. One must keep in mind that in our model several assumptions and simplifications were made. For instance, halocline depth was defined as the deepest point with a salinity of 28 ppt or less, and lice assumed to prefer salinities > 28 ppt and avoid lower salinities. In reality the avoidance breakpoint is more gradual (Wright et al. 2018). Furthermore, swimming depth was boiled down to one depth value per day (daily median value of the mean depth of the shallowest school), while in reality the depth distribution of the school usually several meters and varied during the day, and there were sometimes more than one school. In addition, the possibility of some degree of sampling bias when collecting fish from the surface in large cages should not be ignored. Still, the infestation rates declined with the estimated distance from the halocline, and most of the higher infestation rates occurred when the distance to the halocline was low. We therefore consider distance from the halocline to be a better predictor of infestation rate than swimming depth per se. In order to avoid infestations salmon should therefore not necessarily be attracted to swim deeper, but further away from the halocline. In this study, salmon were attempted to be attracted to deep feeding at 7 m and to lights located at 9-10 m depth, and these depths were often similar to the halocline depth. In cases with a deep halocline it would possibly be more efficient to attract the fish towards the surface, while when the halocline is shallow or non-existing deep attraction is a better option. Dynamic control of the swimming depth attractors then requires monitoring of fish vertical distribution and the vertical profile of the environment, and operational depth adjustment of the attractors.

4.4 - Sampling method effect

True estimates of lice loads and welfare status are vulnerable to sampling error, through either capture methods or processing techniques (Heuch et al. 2011). For lice assessments, euthanising fish in a water bath and conducting lice counting inspections directly afterwards is a more accurate method for recording lice abundances, compared to a blow to the head and individual bagging (Copley et al. 2005). Therefore, capture methods are the likely source of any unrepresentative sampling for both welfare and lice assessments. In a sea cage, differences in spatial location of individual salmon within the school is driven by size (Folkedal et al. 2012), hunger (Juell et al. 1994), physiological state (i.e. emaciation status, Vindas et al. 2016), and infection status (Bui et al. 2016), all of which are likely to interact with the deep lights and feed; if individuals caught for sampling are only from the upper 5 or 10 m depth, it is possible that the sampled individuals are unrepresentative due to these factors. In fact, the median swimming depth of the shallowest school for deep lights and feed groups at Vindsvik was ~ 11 m (Figure 4), and therefore collections from shallower than 10 m is likely to capture individuals that are not in the larger school. This was attempted to be mitigated by the restriction of feed in the sample cage prior to netting fish: as these production fish are fed well throughout the day, withholding feed and then hand-throwing pellets before capturing improves the chance for sampling ‘normal’ fish. On the other hand, Kobbavika is less likely to have biased sampling as schools swam at similar depths in both treatment and control cages (Figure 4), on average ~6.5 m deep over the study period. Thus, the discrepancy in weight between groups at Vindsvik but not Kobbavik could be a by-product of unrepresentative sampling in relation to the main school’s swimming depth.

In the cages with deep lights and deep feed system, there is likely to be a complex interaction between the temperature profile inside the cage, the thermo-regulatory swimming depth preference of the school, and the distribution of infective lice in response to salinity gradients. Preliminary evidence from two case studies showed that the distance of the school from the halocline could be a greater driver of infestation prevention over a long period than the swimming depth, and that attracting fish with lights and feed should take this into account. Although data points for the analysis of distance from halocline and subsequent lice acquisition was limited and carried many assumptions, this model also showed that cages with lice skirts masked the preventive effect of distance from halocline. This suggests that dealing with a dataset of this nature using the ‘actual swim depth-halocline depth-lice acquired’ approach could be a promising method for understanding the host-parasite interaction at a finer scale. These case studies showed no difference in lice infestation between control or treatment cages, however time-specific preventive successes were observed, suggesting that calculating distance of school swimming depth from halocline could be an explanatory variable during these periods.

These case studies and other previous works with depth-related prevention approaches (Stien et al. 2016, Oppedal et al. 2017, Stien et al. 2018) demonstrate that the concept of host-parasite mismatching can be successful, but under certain conditions. With constant changes in environmental conditions that drive the behaviour of the salmon and their interaction with cage prevention technologies, an approach to maximise efficiency is to utilise these tools in response to specific environments. By understanding under what conditions a technology successfully functions (through both host and parasite behaviours), and having flexible responses to the current temperature and salinity profile, farmers could maximise lice prevention potential. For instance, submerged lights and feed could encourage the school’s depth preferences towards areas with lower infective risk, as predicted by the salinity profile and halocline depth, if the expected temperature preference matches the intended depth. The use of lice skirts could facilitate prevention during periods when shallow temperatures are likely to override light or feed attraction, and skirts could also be lowered deeper if there is a brackish layer at the surface.

5.1 - Main findings

• Welfare is not affected by the use of deep light and deep feeding./ Dype lys og dyp fôring påvirker ikke fiskens velferd.

• Deep light and feeding encouraged the fish to swim deeper in periods./ Laksen svømte periodevis dypere ved bruk av dype lys og fôring.

• Reduction in new lice infestation was not consistent with deep light and deep feeding, but a pattern for prevention efficiency was present./ Reduksjon av lusepåslag med dype lys og dyp fôring var ikke konstant, men et forebyggende mønster ble observert.

• Increased distance from halocline (brackish surface waters) seemed to reduce susceptibility to lice infestation./ Økende svømmedyp fra haloklinen (overgang brakkvann-sjøvann) reduserte lusepåslag.

Aaen S.M., Helgesen K.O., Bakke M.J., Kaur K. & Horsberg T.E. (2015) Drug resistance in sea lice: a threat to salmonid aquaculture. Trends in Parasitology 31, 72-81. Doi: 10.1016/j.pt.2014.12.006

Bjordal Å., Juell J.-E., Lindem T. & Fernö A. (1993) Hydroacoustic monitoring and feeding control in cage rearing of Atlantic salmon (Salmo salar L.). In: Fish Farming Technology, pp. 203-208. Balkema, Rotterdam.

Bratland S., Stien L.H., Braithwaite V.A., Juell J.-E., Folkedal O., Nilsson J., Oppedal F., Fosseidengen J.E. & Kristiansen T.S. (2010) From fright to anticipation: using aversive light stimuli to investigate reward conditioning in large groups of Atlantic salmon (Salmo salar). Aquaculture International 18, 991-1001. Doi: 10.1007/s10499-009-9317-8

Bui S., Oppedal F., Stien L. & Dempster T. (2016) Sea lice infestation level alters salmon swimming depth in sea-cages. Aquaculture Environment Interactions 8, 429-435. Doi: 10.3354/aei00188

Bui, S., Stien L.H., Nilsson J., & Oppedal F. (2018) Assessment of long-term implementation of sea lice prevention technologies: efficiency in reducing infestations and impact on fish welfare. In: Rapport fra Havforskningen 45. Institute of Marine Research, Bergen. www.hi.no/filarkiv/2018/12/45-2018_assessment_of_long-term_implementation.pdf b-no

Copley L., O’Donohoe P., McGrath D. & Jackson D. (2005) Sampling errors in examining salmon (Salmo salar L.) for sea lice (Lepeophtheirus salmonis (Krøyer) and Caligus elongatus Nordmann) burden: a comparison of methods. Fish Veterinary Journal 8, 32-44. Doi:

Eldøy S.H., Davidsen J.G., Thorstad E.B., Whoriskey F.G., Aarestrup K., Naesje T.F., Ronning L., Sjursen A.D., Rikardsen A.H. & Arnekleiv J.V. (2017) Marine depth use of sea trout Salmo trutta in fjord areas of central Norway. J Fish Biol. Doi: 10.1111/jfb.13463

Falcón J., Besseau L., Sauzet S. & Boeuf G. (2007) Melatonin effects on the hypothalamo–pituitary axis in fish. Trends in Endocrinology & Metabolism 18, 81-88. Doi: https://doi.org/10.1016/j.tem.2007.01.002

Fjelldal P.G., Hansen T. & Huang T.-s. (2011) Continuous light and elevated temperature can trigger maturation both during and immediately after smoltification in male Atlantic salmon (Salmo salar). Aquaculture 321, 93-100. Doi: https://doi.org/10.1016/j.aquaculture.2011.08.017

Fjelldal P.G., Schulz R., Nilsen T.O., Andersson E., Norberg B. & Hansen T.J. (2018) Sexual maturation and smoltification in domesticated Atlantic salmon (Salmo salar L.) – is there a developmental conflict? Physiological Reports 6, e13809. Doi: 10.14814/phy2.13809

Folkedal O., Stien L.H., Nilsson J., Torgersen T., Fosseidengen J.E. & Oppedal F. (2012) Sea caged Atlantic salmon display size-dependent swimming depth. Aquatic Living Resources 25, 143-149. Doi: 10.1051/alr/2012007

Folkedal O., Pettersen J.M., Bracke M.B.M., Stien L.H., Nilsson J., Martins C., Breck O., Midtlyng P.J. & Kristiansen T. (2016) On-farm evaluation of the Salmon Welfare Index Model (SWIM 1.0): theoretical and practical considerations. Animal Welfare 25, 135-149. Doi: 10.7120/09627286.25.1.135

Frenzl B., Stien L.H., Cockerill D., Oppedal F., Richards R.H., Shinn A.P., Bron J.E. & Migaud H. (2014) Manipulation of farmed Atlantic salmon swimming behaviour through the adjustment of lighting and feeding regimes as a tool for salmon lice control. Aquaculture 424-425, 183-188. Doi: 10.1016/j.aquaculture.2013.12.012

Genna R.L., Mordue W., Pike A.W. & Mordue A.J. (2005) Light intensity, salinity, and host velocity influence presettlement intensity and distribution on hosts by copepodids of sea lice, Lepeophtheirus salmonis. Canadian Journal of Fisheries and Aquatic Sciences 62, 2675-2682. Doi: 10.1139/f05-163

Hamre L.A., Eichner C., Caipang C.M., Dalvin S.T., Bron J.E., Nilsen F., Boxshall G. & Skern-Mauritzen R. (2013) The Salmon Louse Lepeophtheirus salmonis (Copepoda: Caligidae) life cycle has only two Chalimus stages. PLoS One 8, e73539. Doi: 10.1371/journal.pone.0073539

Hansen T.J., Fjelldal P.G., Folkedal O., Vågseth T. & Oppedal F. (2017) Effects of light source and intensity on sexual maturation, growth and swimming behaviour of Atlantic salmon in sea cages. Aquaculture Environment Interactions 9, 193-204. Doi:

Heuch P.A. (1995) Experimental evidence for aggregation of salmon louse copepodids (Lepeophtheirus Salmonis) in step salinity gradients. Journal of the Marine Biological Association of the UK 75, 927-939. Doi:

Heuch P.A., Parsons A. & Boxaspen K. (1995) Diel vertical migration: a possible host-finding mechanism in salmon louse (Lepeophtheirus salmonis) copepodids? Canadian Journal of Fisheries and Aquatic Sciences 52, 681-689. Doi:

Heuch P.A., Gettinby G. & Revie C.W. (2011) Counting sea lice on Atlantic salmon farms ? Empirical and theoretical observations. Aquaculture 320, 149-153. Doi: 10.1016/j.aquaculture.2011.05.002

Hevrøy E.M., Boxaspen K., Oppedal F., Taranger G.L. & Holm J.C. (2003) The effect of artificial light treatment and depth on the infestation of the sea louse Lepeophtheirus salmonis on Atlantic salmon (Salmo salar L.) culture. Aquaculture 220, 1-14. Doi: 10.1016/s0044-8486(02)00189-8

Huse I. & Holm J.C. (1993) Vertical distribution of Atlantic salmon (Salmo salar) as a function of illumination. Journal of Fish Biology 43, 147-156. Doi: 10.1111/j.1095-8649.1993.tb01184.x

Iigo M., Abe T., Kambayashi S., Oikawa K., Masuda T., Mizusawa K., Kitamura S., Azuma T., Takagi Y., Aida K. & Yanagisawa T. (2007) Lack of circadian regulation of in vitro melatonin release from the pineal organ of salmonid teleosts. Gen Comp Endocrinol 154, 91-97. Doi: 10.1016/j.ygcen.2007.06.013

Juell J.-E. & Fosseidengen J.E. (2004) Use of artificial light to control swimming depth and fish density of Atlantic salmon (Salmo salar) in production cages. Aquaculture 233, 269-282. Doi: http://dx.doi.org/10.1016/j.aquaculture.2003.10.026

Juell J.-E., Ferno A., Furevik D. & Huse I. (1994) Influence of hunger level and food availability on the spatial distribution of Atlantic salmon, Salmo salar L., in sea cages. Aquatic Fish Management 25, 439/451. Doi:

Juell J.-E., Oppedal F., Boxaspen K. & Taranger G.L. (2003) Submerged light increases swimming depth and reduces fish density of atlantic salmon in production cages. Aquaculture Research 34, 469-477. Doi:

Korsøen Ø.J., Fosseidengen J.E., Kristiansen T.S., Oppedal F., Bui S. & Dempster T. (2012) Atlantic salmon (Salmo salar L.) in a submerged sea-cage adapt rapidly to re-fill their swim bladders in an underwater air filled dome. Aquacultural Engineering 51, 1-6. Doi: 10.1016/j.aquaeng.2012.04.001

Leclercq E., Taylor J.F., Sprague M. & Migaud H. (2011) The potential of alternative lighting-systems to suppress pre-harvest sexual maturation of 1+ Atlantic salmon (Salmo salar) post-smolts reared in commercial sea-cages. Aquacultural Engineering 44, 35-47. Doi: https://doi.org/10.1016/j.aquaeng.2010.12.001

Lien A.M., Volent Z., Jensen Ø., Lader P. & Sunde L.M. (2014) Shielding skirt for prevention of salmon lice (Lepeophtheirus salmonis) infestation on Atlantic salmon (Salmo salar L.) in cages – A scaled model experimental study on net and skirt deformation, total mooring load, and currents. Aquacultural Engineering 58, 1-10. Doi: 10.1016/j.aquaeng.2013.11.003

Migaud H., Davie A. & Taylor J.F. (2010) Current knowledge on the photoneuroendocrine regulation of reproduction in temperate fish species. Journal of Fish Biology 76, 27-68. Doi: 10.1111/j.1095-8649.2009.02500.x

Migaud H., Taylor J.F., Taranger G.L., Davie A., Cerdá-Reverter J.M., Carrillo M., Hansen T. & Bromage N.R. (2006) A comparative ex vivo and in vivo study of day and night perception in teleosts species using the melatonin rhythm. Journal of Pineal Research 41, 42-52. Doi: 10.1111/j.1600-079X.2006.00330.x

Myksvoll M.S., Sandvik A.D., Albretsen J., Asplin L., Johnsen I.A., Karlsen O., Kristensen N.M., Melsom A., Skardhamar J. & Adlandsvik B. (2018) Evaluation of a national operational salmon lice monitoring system-From physics to fish. PLoS One 13, e0201338. Doi: 10.1371/journal.pone.0201338

Næs M., Heuch P.A. & Mathisen R. (2012) Bruk av ↔luseskjørt≈ for å redusere påslag av lakselus Lepeophtheirus salmonis (Krøyer) på oppdrettslaks. In: Vesterålen fiskehelsetjeneste.

Nilsson J., Stien L.H. & Oppedal F. (2018) Lusepåslag og velferd hos dypfôret og overflatefôret laks på lokalitet Skrubbholmen midtvinters. Rapport fra Havforskningen 35. Institute of Marine Research, Bergen. https://www.hi.no/filarkiv/2018/09/lusepaslag_og_velferd_hos_dypforet_og_overflateforet_laks_pa_
lokalitet_skrubbholmen_midtvinter.pdf b-no

Nilsson J., Stien L.H. & Oppedal F. (2017) Environment, lice levels, welfare and salmon swim depth at Kobbavika site with surface or deep feeding combined with artificial light. In: Rapport fra Havforskningen 35. Institute of Marine Research, Bergen. https://www.hi.no/filarkiv/2017/12/kobbavika2015g_imr_rapport_fra_havforskningen_en_21122017.pdf b-no

Olaussen J.O. (2018) Environmental problems and regulation in the aquaculture industry. Insights from Norway. Marine Policy In press. Doi: 10.1016/j.marpol.2018.08.005

Oppedal F., Juell J.-E. & Johansson D. (2007) Thermo- and photoregulatory swimming behaviour of caged Atlantic salmon: Implications for photoperiod management and fish welfare. Aquaculture 265, 70-81. Doi: http://dx.doi.org/10.1016/j.aquaculture.2007.01.050

Oppedal F., Dempster T. & Stien L.H. (2011) Environmental drivers of Atlantic salmon behaviour in sea-cages: A review. Aquaculture 311, 1-18. Doi: 10.1016/j.aquaculture.2010.11.020

Oppedal F., Samsing F., Dempster T., Wright D.W., Bui S. & Stien L.H. (2017) Sea lice infestation levels decrease with deeper 'snorkel' barriers in Atlantic salmon sea-cages. Pest Manag Sci. Doi: 10.1002/ps.4560

Osland H., Sandvik J.I., Holm J.C., Heuch P.A. & Bakke S. (2001) Studie av lakseluspåslag og tilvekst hos Atlantisk laks (Salmo salar) i nedsenkede merder. In: HSF Report, p. 22 (in Norwegian).

Stien L.H., Lind M.B., Oppedal F., Wright D.W. & Seternes T. (2018) Skirts on salmon production cages reduced salmon lice infestations without affecting fish welfare. Aquaculture. Doi: https://doi.org/10.1016/j.aquaculture.2018.02.045

Stien L.H., Fosseidengen J.E., Malm M.E., Sveier H., Torgersen T., Wright D.W. & Oppedal F. (2014) Low intensity light of different colours modifies Atlantic salmon depth use. Aquacultural Engineering 62, 42-48. Doi: 10.1016/j.aquaeng.2014.05.001

Stien L.H., Dempster T., Bui S., Glaropoulos A., Fosseidengen J.E., Wright D.W. & Oppedal F. (2016) ‘Snorkel’ sea lice barrier technology reduces sea lice loads on harvest-sized Atlantic salmon with minimal welfare impacts. Aquaculture 458, 29-37. Doi: 10.1016/j.aquaculture.2016.02.014

Stien L.H., Bracke M.B.M., Folkedal O., Nilsson J., Oppedal F., Torgersen T., Kittilsen S., Midtlyng P.J., Vindas M.A., Øverli Ø. & Kristiansen T.S. (2013) Salmon Welfare Index Model (SWIM 1.0): a semantic model for overall welfare assessment of caged Atlantic salmon: review of the selected welfare indicators and model presentation. Reviews in Aquaculture 5, 33-57. Doi: 10.1111/j.1753-5131.2012.01083.x

Thorstad E.B., Whoriskey F., Uglem I., Moore A., Rikardsen A.H. & Finstad B. (2012) A critical life stage of the Atlantic salmon Salmo salar: behaviour and survival during the smolt and initial post-smolt migration. Journal of Fish Biology 81, 500-542. Doi: 10.1111/j.1095-8649.2012.03370.x

Vindas M.A., Johansen I.B., Folkedal O., Hoglund E., Gorissen M., Flik G., Kristiansen T.S. & Overli O. (2016) Brain serotonergic activation in growth-stunted farmed salmon: adaption versus pathology. R Soc Open Sci 3, 160030. Doi: 10.1098/rsos.160030

Wright D.W., Glaropoulos A., Solstorm D., Stien L.H. & Oppedal F. (2015) Atlantic salmon Salmo salar instantaneously follow vertical light movements in sea cages. Aquaculture Environment Interactions 7, 61-65. Doi: 10.3354/aei00136

Wright D.W., Stien L.H., Dempster T., Vagseth T., Nola V., Fosseidengen J.E. & Oppedal F. (2017) 'Snorkel' lice barrier technology reduced two co- occurring parasites, the salmon louse (Lepeophtheirus salmonis) and the amoebic gill disease causing agent (Neoparamoeba perurans), in commercial salmon sea-cages. Prev Vet Med 140, 97-105. Doi: 10.1016/j.prevetmed.2017.03.002

Wright, D., Crosbie, T., Dalvin, S., Oppedal, F., Dempster, T., (2018). Vertikal fordeling av luselarver i brakkvannsgradienter. Norsk Fiskeoppdrett 3/2018, 40-43

Nilsson, J., Stien, L. and Oppedal, F., 2017. Environment, lice levels, welfare and salmon swim depth at Kobbavika site with surface or deep feeding combined with artificial light/ Miljø, lusenivå, velferd og laksens svømmedyp ved lokalitet Kobbavika ved bruk av utfôring i overflate eller under vann kombinert med kunstig lys. Rapport fra Havforskningen 35-2017. 31p

Nilsson, J., Stien, L. og Oppedal, F., 2018. Lusepåslag og velferd hos dypfôret og overflatefôret laks på lokalitet Skrubbholmen midtvinters. Rapport fra Havforskningen 35-2018. 19s

Bui, S., Stien, L. Nilsson, J., and Frode Oppedal, 2018. Assessment of long-term implementation of sea lice prevention technologies: efficiency in reducing infestations and impact on fish welfare. Rapport fra Havforskningen 45-2018. 38p

Nilsson, J., Bui, S., Stien, L. and Oppedal, F., 2019. Use of deep lights and deep feed in commercial settings: effect on welfare, behaviour and infestation, summary and status at three case study sites/ Bruk av dype lys og dyp fôring i kommersiell skala; effekter på velferd, atferd og lusepåslag ved 3 oppdrettsanlegg: sluttrapport FHF 901154. Rapport fra Havforskningen denne rapport-2019. 28p

Bui, S., Trengereid, H., Stien, L., Nilsson J., Oppedal, F. Efficiency and welfare impact of multiple-prevention methods for salmon louse reduction in commercial sea cages. In prep

Presentations

Oppedal, F., m.fl. 2019. Lakselusa – når og hvor? Hvordan utnytte kunnskapen til forebygging. FHFs Lusekonferanse 2019 Trondheim 21.-23. januar

Bui, S., Trengereid, H., Stien, L., and Oppedal, F., 2019. Use of deep lights and feeding, and skirts for lice infestation prevention. FHFs Lusekonferanse 2019 Trondheim 21.-23.januar

Oppedal et al 2018. Examples of alternative preventative methods on sealice control. Workshop Non-Medicinal control, SEALICE 2018, Dream hotel, Punta Arenas, 5-9 November 2018

Bui, S., Trengereid, H., Stien, L., Oppedal, F., 2018. Effect of lice prevention technologies on salmon welfare and infection status. International Symposium on Aquatic Animal Health, Charlottetown, Canada, 2-6 September

Bui, S., Trengereid, H., Stien, L., Oppedal, F., 2017. Efficiency of sea lice prevention technologies in a commercial setting: salmon welfare, infestation levels, and treatment frequencies. European Aquaculture Society, Dubrovnik, Croatia. 16-20 October