Fish are capable of excellent vision and can be profoundly influenced by the visual properties of their environment. Ambient colours have been found to affect growth, survival, aggression and reproduction, but the effect of background darkness (i.e., the darkness vs. lightness of the background) on preference and aggression has not been evaluated systematically. One-hundred Coho salmon (Oncorhynchus kisutch), a species that is increasing in popularity in aquaculture, were randomly assigned to 10 tanks. Using a Latin-square design, every tank was bisected to allow fish in each tank to choose between all the following colour choices (8 choices in total): black vs. white, light grey, dark grey, and a mixed dark grey/black pattern, as well as industry-standard blue vs. white, light grey, dark grey, and black. Fish showed a strong preference for black backgrounds over all other options (p < 0.01). Across tests, preference strength increased with background darkness (p < 0.0001). Moreover, having darker backgrounds in the environment resulted in less aggressive behaviour throughout the tank (p < 0.0001). These results provide the first evidence that darker tanks are preferred by and decrease aggression in salmonids, which points to the welfare benefits of housing farmed salmon in enclosures containing dark backgrounds.
Citation: Gaffney LP, Franks B, Weary DM, von Keyserlingk MAG (2016) Coho Salmon (Oncorhynchus kisutch) Prefer and Are Less Aggressive in Darker Environments. PLoS ONE 11(3): e0151325. https://doi.org/10.1371/journal.pone.0151325
Editor: Carlos Garcia de Leaniz, Swansea University, UNITED KINGDOM
Received: September 2, 2015; Accepted: February 27, 2016; Published: March 30, 2016
Copyright: © 2016 Gaffney et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data for the experiment are available from the figshare database. The DOI for our data is: (https://dx.doi.org/10.6084/m9.figshare.3083704.v1).
Funding: This work was funded through the generous donations to the UBC Animal Welfare Program. B.F. is supported through a Killiam Postdoctoral Scholarship.
Competing interests: The authors have declared that no competing interests exist.
In general, fish rely on vision to detect prey, predators, and mates [1–3] and exhibit one of the greatest visual sensitivities among vertebrates . A large body of literature has shown that fish are strongly affected by the visual properties of their environment: light conditions, substrate colour, and tank wall colour can impact fish stress responses, growth, and aggression [5–10]. Across these studies, there is some evidence that darkness—how much light the environment absorbs rather than reflects —may be a critical dimension. For example, Nile tilapia (Oreochromis niloticus) and summer flounder (Paralichthys dentatus) show lower plasma cortisol levels in darker versus lighter environments . Höglund et al. , found that pairs of Arctic charr (Salvelinus alpinus) interacting on a black background displayed darker colouration and less aggression than fish interacting on a white background. Collectively, these studies suggest that fish may prefer dark environments and that having access to dark environments may also reduce aggression.
With the demand for fish projected to increase by almost 50 million tonnes and aquaculture supplying approximately 73% of this growth , there is increasing interest in developing effective and humane land based captive rearing systems. Surprisingly little attention has been paid to the environmental parameters that would best promote fish welfare. Along with the ethical implications, improved fish welfare has the potential to improve industry production quality and quantity as well as public perception and product acceptance [14–16]. Previous research on the importance of the visual properties of the environment suggests that even simple choices such as tank colour could dramatically affect fish health [9,10,12]. Though tanks could easily be fabricated in any colour, the prevailing industry practice has become to house farmed fish in light blue tanks.
The aim of this study was to assess tank colour preferences of Coho salmon (Oncorhynchus kisutch) and quantify how background colour affects aggression. Extending the previous research, we focused specifically on background darkness—testing several shades of gray including black and white—against the industry standard light blue. As an initial exploration into the possibility that fish might prefer certain backgrounds because they allow for better concealment, we also tested a black background (absolute darkness) against a dark, dappled background that was lighter but thought to provide better camouflaging properties.
This protocol was approved by the University of British Columbia Animal Care Committee (application number: A13-0210). All experimental procedures were performed in accordance with the Canadian Council on Animal Care guidelines on care and use of fish in research.
Animals and Housing
One hundred, 2-year-old coho salmon (mixed sex, mean ± s.d. weight 100 ± 5 g) were obtained from a collaborating research laboratory, originally purchased from Target Marine Hatcheries Ltd. (Sechelt, BC, Canada). Fish were reared during this time in circular, 244 cm diameter (5.6m3), light blue, fibreglass tanks (D&T Fiberglass Inc., Sacramento, CA, USA).
Fish were randomly assigned to groups of ten, each housed in a circular 600 L, 107 cm diameter (0.7m3), light blue, fibreglass tank (D&T Fiberglass Inc., Sacramento, CA, USA). The 10 tanks were serviced by a freshwater recirculating system providing mechanical and biological filtration, UV-sterilization, and oxygen injection (components were supplied by Integrated Aqua Systems, Inc., Escondido, CA, USA). During the experiment, mortality, water temperature, dissolved oxygen content, and pH were measured daily and ammonia-N was measured weekly. No mortalities occurred. Water temperature was maintained between 8 and 12°C. The dissolved oxygen content of the water was kept between 70–90% air saturation, pH between 6.0 and 6.8 and ammonia-N was less than 0.5 ppt. The average water flow rate was 360 L/min (± 1 L/min). These water quality parameters are well within the range of what salmon would experience in the wild  and adhere to industry standard ranges . Tanks were cleaned once a week.
Fish were hand fed to satiety with a commercially available pelleted diet (Transfer Plus 3.0MM and 4.0MM, Skretting Canada, Vancouver, BC, Canada) once daily from Monday to Saturday at 1000h; no food was given on Sunday. All experimental populations were housed under a 10-h light; 14-h dark cycle (lights on and off at 0730h and 1700h, respectively). To ensure even light distribution in each tank and to avoid complications due to room lighting, a full spectrum daylight fluorescent lamp (CFL T2 23W Daylight Bulb, Globe Electric, Montreal, QC, Canada) was suspended 1 m above the center of each tank and light intensity was adjusted to 410 lx at water surface.
The interior sides and base of each of the experimental tanks were lined with laminated paper (InPrint Graphics & Copying Ltd., Vancouver, BC, Canada) varying in pigmentation: black (0°, 0%, 0%; hue, saturation, value, respectively), white (0°, 0%, 100%), light grey (0°, 0%, 85%), dark grey (0°, 0%, 45%), and a mixed dark grey/black dappled/sponge pattern. The industry-standard, original blue tank background had a hue of 200.4°, a saturation of 22.8%, and a value of 85% (the same % value as the light grey laminated sheets). Sixteen 1.27cm x 0.32cm magnets (Rare-Earth Circular Magnets, Lee Valley Tools Ltd. and Veritas Tools Inc., Vancouver, BC, Canada) were used to hold the laminated pieces of paper in place in each tank. To habituate fish to each of the conditions, tanks were lined with each background option for a total of 3 days before preference testing began. After each individual preference test, the tanks were lined with the subsequent preference test background options and preference testing began 3 days afterwards.
Preference tests were conducted between (a) black versus white, light grey, dark grey and patterned and (b) blue versus white, light grey, dark grey, and black. Every tank experienced every preference test. To avoid order effects, exposure was set by a replicated 5x5 Latin square such that on any given day, every tank experienced a different colour choice option in a randomized order. All tanks were split into two to provide fish a choice of background colour. To avoid an effect of side on preferences, colour placement was alternated across the 10 tanks.
The positions of the fish in each tank were video recorded from above using a Microsoft LifeCam StudioTM webcam (USA). Experimental trials began 15 min after the camera was placed above the tank to control for disturbances. The fish in each tank were recorded for a total of 10 min, starting at 1100h. Videos were scored using scan sampling at 30-s intervals to determine the number of fish on each side of the tank, yielding 20 observations/trial.
Videos were scored continuously for aggressive acts (charging, a rapid dart towards another fish with attempts to bite the other fish, and chasing, rapid pursuit of another fish) throughout each 10-min trial.
To determine background darkness preference, we ran one-sample t-tests on the average number of fish per side using 5 (half the number of fish per tank) as the critical value. Across trials, we modelled the effect of background darkness on preference strength (the degree to which an animal prefers a choice in a preference test) using a multilevel model with fixed effects of background darkness (continuous from 0 [white] to 1 [black]), blue (coded as an indicator/dummy variable of black vs. blue trial) and pattern (indicator variable of a pattern vs. regular trial) and a random effect of tank to control for repeated testing of tanks and to avoid pseudoreplication [19–21].
To model aggression on each side of the tank, we log-transformed aggressive-behaviour-per-fish (side-aggression = log[aggressive acts per side/average number of fish per side across the trial]) as a function of side-darkness (continuous) and blue and pattern indicator variables with a multilevel model that included tank and trial as nested random effects. For total number of aggressive acts per trial, we applied generalized multilevel models (Poisson error structure and log link) with a random effect of tank and fixed effects for mean tank darkness (the average darkness of the two sides), blue and pattern indicator variables, and whether or not the trial contained a side with greater than 0.55 darkness (a refuge).
Tank Colour Preference
Pair-wise comparisons for the black-trials revealed a strong preference for black background over all other options (black vs. white: t(9) = 41.24, p < .0001; black vs. light grey: t(9) = 11.88, p < 0.0001; black vs. dark grey: t(9) = 12.11, p < 0.0001; black vs. pattern: t(9) = 3.66, p < .01; Fig 1a). In the blue-trials, fish showed no preference for white vs. blue (t(9) = 0.47, p > 0.6), a tendency to prefer light grey over blue (t(9) = 2.22, p < 0.1), and a preference for both dark grey and black over blue (t(9) = 5.63, p < 0.001, and t(9) = 13.06, p < 0.0001, respectively; Fig 1b). Across all trials, we found that preference strength increased with background darkness (t(67) = 8.81, p < 0.0001, Fig 1a and 1b). We found limited evidence that the patterned background was preferred beyond what would be expected based on its darkness value alone (t(67) = 1.92, p = 0.06, Fig 1b).
Preference strength increased with increasing background darkness (p < 0.0001). For the (a) black-trials and (b) blue-trials, dots represent each tank’s average number of Coho salmon (Oncorhynchus kisutch) located on the comparison side. Background darkness ranges from 0 (white) to 1 (black). Large symbols represent average values across all tanks (n = 10) with the fill colour corresponding to the background darkness of the comparison side (white, light grey, dark grey, or black) and the triangle (▿) indicating the patterned background trial.
Background darkness dramatically decreased aggression (df = 157, t = 11.06, p < 0.0001) Fig 2): aggression was nearly four times greater on blue backgrounds (the industry standard) than on black backgrounds. The effect of blue and patterned background on aggression was entirely explained by their darkness value or, in other words, aggression on the blue side was no different than aggression on the light grey side (which had the same darkness value) and aggression on the patterned side was consistent with what would have been expected given the patterned background’s average darkness value. There was virtually no variation in aggression per tank—the ICC (intraclass correlation coefficient) was 0. Examining aggression at the tank level, we found that increasing the average darkness of the environment decreased the number of overall aggressive acts (z = 2.79, p < 0.01; Fig 3). Interestingly, however, the decrease in aggression with background darkness was not linear even on a log scale. When the tank contained a dark ‘refuge’—a side of the tank with a darkness value of 0.55 or higher—overall aggression was reduced beyond what would be expected by the effect of mean tank darkness alone. Fish engaged in only one third as many aggressive acts in trials with a dark ‘refuge’ compared to trials without a dark ‘refuge’ (Fig 3; z = 9.41, p < 0.0001).
Average aggression (the total number of aggressive acts displayed divided by the average number of fish across a 10-min trial) on a side (log scale) decreased with background darkness (p < 0.0001). Background darkness ranges from 0 (white) to 1 (black). Filled-in circle (◯) colour corresponds to the background colour (white, blue, light grey, dark grey, or black) and the triangle (▿) indicates the patterned background.
Overall aggression (total number of aggressive acts occurring per 10-min trial) throughout the tank decreased with average tank darkness (p < 0.01), but especially when the tank contained an area available with darkness of .55 or greater (p < 0.0001). Dots represent aggression counts in a trial for each tank. Background darkness ranges from 0 (white) to 1 (black). Filled-in circle (◯) colour corresponds to the average tank darkness and the triangle (▿) indicates the average tank darkness of the pattern vs. black trial. The bi-colour panel at the bottom of the plot shows the two colours tested in the trial.
The fish in this study showed a clear preference for black backgrounds over all other colour options and for darker backgrounds in general. Dark backgrounds also substantially decreased aggression, both in terms of the aggression observed on the background side itself and in terms of aggression observed throughout the tank. We found only limited evidence that patterned backgrounds may be preferred beyond what would be expected given the background darkness value alone and found no evidence that the patterned backgrounds uniquely reduced aggression. A wide range of studies have shown that prey animals prefer parts of the habitat that match their own visual appearance to reduce the risk of predator detection (e.g. [27–29]). Similarly to Kjernsmo and Merilaita , we speculate that the presence of a predator or the use of a patterned background that more accurately matches the colouration of coho salmon may have resulted in a higher preference for the dark, dappled background used in our study.
With the rapid expansion of fish aquaculture, the welfare of farmed fish has received increasing attention . However, the majority of studies on aquaculture fish welfare have focused on the biological functioning aspect of welfare, addressing issues such as growth rates, flesh quality, injury, disease, and reproductive problems (e.g. [14–16, 31]). However, good animal welfare goes beyond physical health. It also involves the ability of an animal to express natural behaviours and a lack of mental suffering from pain, fear, etc. . The present study focuses on fish preferences and behaviour as indicators of fish welfare. There is a growing body of physiological, behavioural, and anatomical evidence indicating that fish are able to experience pain and fear (e.g. [33–47]). Thus, affective states, revealed through behaviour and potentially impacted by preferences, should be considered when assessing fish welfare in aquaculture systems.
Preference tests allow animals to express their priorities, giving us insight into what may be important to their health and welfare ; they are widely used in the welfare literature, for example, to assess flooring temperatures for sows , lighting type and intensity for poultry [50, 51], bedding for dairy cows , and nesting material for laboratory mice . However, the results of preference tests vary with context  including the animal’s previous experiences . Previous research suggests that animals may initially prefer environments they are familiar with (Dawkins, 1977). In the current study, for example, the fish may have shown a preference for blue tanks over other options because they were previously reared in blue tanks. Our results, however, show that the fish had a clear preference for black backgrounds over all other colour options and for darker backgrounds in general. Moreover, animals do not always make choices that are best for their long-term health and welfare . The implications for welfare of preference tests must therefore be interpreted for with care. For example, in the current study the fish clearly preferred the darker sides, but, in absolute terms, were also observed to engage in more aggressive acts on those sides. However, the higher frequency of aggressive interactions observed on the black sides was explained entirely by the greater fish numbers and higher densities on those sides. In fact, accounting for the fish densities revealed the reverse pattern to be the case: darker colours resulted in lower aggression on a per fish basis.
The preference for darker colours may be explained, in part, by the feeding habits of Coho salmon. Salmon rely almost entirely on vision to detect their prey, which are mostly drifting, light-coloured invertebrates [1, 2]. It is likely that these prey are easier to detect against a dark background, thus leading salmon to have a potentially innate preference for darker environments . Additionally, bright environments, especially those without cover, have been shown to elicit a physiological stress response such as increased plasma cortisol levels for fish . Possibly as a form of anti-predator behaviour , fish tend to seek dark environments when anxious [59, 60]. Thus, fish preference for darker background may also be related to protective, escape behaviours.
Salmonids are naturally territorial and aggressive behaviours (charging and chasing) are known to increase with increasing density [31, 61, 62]. Aggression is thus of particular concern in closed containment aquaculture systems . In the present study, darker backgrounds were also shown to reduce overall fish aggression levels within the tank, despite the higher fish densities on the darker sides. Salmonids, like other teleost fish, have the ability to adjust the colour of their skin in response to changes in background colour [64, 65]. In salmon, darkening of the skin and eyes signals social subordination, while lightening of the skin signals dominance [65–67]. Thus, a darker fish may represent less of a threat and elicit less aggression than a conspecific displaying paler body colouration. Our results show that the effect of background darkness on aggression extends beyond the immediate environment: having a discrete area of darkness in the environment reduced aggression throughout the entire environment. Follow up studies should examine this result further by directly comparing environments with equivalent levels of average darkness, but varying whether they contain a dark refuge. Determining the visual parameters that minimize aggression could have important implications for the welfare of fish in captivity.
Salmonids are the most produced species of fish in aquaculture  and there is growing interest rearing coho salmon in land-based, closed containment aquaculture systems [18, 68]. Aggression is a major concern in closed containment aquaculture and is often the primary cause of fin and skin damage in fish [31, 61, 62]. Fin and skin damage have the potential to cause inflammation and pain, and may lead to infection, reduced survival in salmon, decreased production and decreased product quality [31, 62, 66, 69, 70]. Furthermore, salmon smolts have a long-history of being cultured in hatcheries . Many studies have investigated how rearing salmonids in semi-naturalistic conditions has the potential to improve the post-release survival and performance of hatchery-reared fish. For example, adding structure to the rearing environment decreases stress levels  and promotes survival skills . However, no study to date has investigated the effect of tank colour on survival and success of hatchery-reared fish. Thus management practices such as selecting appropriate tank colours have the potential to improve the welfare of many fish. The present research indicates that tank darkness matters to fish and can alter their social behaviour.
Limitations and Future Research
One limitation of the current study was that the fish were kept under a relatively narrow range of environmental conditions. Preference tests depend on context, and external and internal variables such as, temperature, age of the animal, or time of the year, must be considered . Preferences may also fluctuate in response to changing variables; for example, pigs prefer to rest on straw in the morning when it is cold, but not in the evening when it is warmer . Future research should determine temporal, spatial and environmental influences on tank darkness preferences.
In the case of wild coho salmon, the majority of their first year of life is spent living in freshwater streams, after which they migrate to the ocean, undergoing a metamorphic process called smoltification . Thus, the salinity of water ranges from 0ppt (freshwater) to 35ppt (saltwater) and temperature ranges from 0°C to 25°C during the lifecycle of a wild coho salmon . Considering the visual pigments of coho salmon are maximally sensitive to a specific region of the spectrum and can be altered as fish develop and undergo migration [4, 75, 76], changes in water salinity or temperature may affect how coho salmon react to different tank colours. Thus future research should examine how colour affects aggression and preference in salmon held in differing water salinity and temperature levels.
In addition, fish age may influence the effect that colour has on aggressive behaviour and colour preference in fish. For example, as juvenile coho salmon mature in the wild, individuals acquire their own territories for feeding and begin to engage in intraspecific aggressive behaviours in the form of territory defense and hierarchy establishment . Thus, coho salmon aggression varies with life stage. Although no research to date has looked at how colour interacts with fish aggression at differing ages, it is possible that as salmon grow and their aggression levels change, colour preferences may also change. The results presented in this study should be considered specific to the age of fish examined.
The stocking density of the tanks in this study, measured using the weight of fish per unit volume of the tanks (expressed as kg/m3), was approximately 14 kg/m3 (compared to approximately 3 kg/m3 used in the other studies), which is considerably lower than what is normally used in commercial salmonid aquaculture farms, where stocking densities in excess of 80 kg/m3 are common . It is necessary to determine, therefore, how the effects of tank colour may interact with stocking density. One possibility is that tank colour would have an even greater impact on aggression levels in these larger, more densely stocked tanks. Alternatively, at such high stocking densities, the visual properties of the environment may matter less. More research needs to be conducted at higher densities more similar to those found on commercial farms.
We are grateful to Dr. Jeff Richard for his comments on a previous version of this manuscript and to Josh Emerman, Victor Chan and Betsy Robertson for their assistance with feeding the fish. This work was funded through the generous donations to the UBC Animal Welfare Program. B.F. is supported through a Killiam Postdoctoral Scholarship.
Conceived and designed the experiments: LPG DMW MAGvK. Performed the experiments: LPG. Analyzed the data: LPG BF. Contributed reagents/materials/analysis tools: MAGvK DMW. Wrote the paper: LPG BF DMW MAGvK.
- 1. Keenleysidme M.H.A. (1962) Skin-diving observations of Atlantic salmon and brook trout in the Miramichi River, New Brunswick. Journal of the Fisheries Research Board of Canada. 19:625–634.
- 2. Stradmeyer L. and Thorpe J.E. (1987) Feeding behaviour of wild Atlantic salmon, Salmo salar, L., parr in mid- to later summer in a Scottish stream. Aquaculture and Fisheries Management. 18:33–49.
- 3. Levine J.S. and MacNichol E.F. Jr. (1982) Color vision in fishes. Scientific America. 246:140–149.
- 4. Temple S. E., Ramsden S.D., Haimberger T.J., Veldhoen K.M., Veldhoen N.J., Carter N.L., W. M. et al. (2008). Effects of exogenous thyroid hormones on visual pigment composition in coho salmon (Oncorhynchus kisutch). Journal of Experimental Biology. 211:2134–2143. pmid:18552303
- 5. Volpato G.L. and Barreto R.E. (2001) Environmental blue light prevents stress in the fish Nile tilapia. Brazilian Journal of Medical and Biological Research. 34:1041–1045. pmid:11471044
- 6. Karakatsouli N., Papoutsoglou E.S., Sotiropoulos N., Mourtikas D., Stigen Martinsen T. and Papoutsoglou S.E. (2010) Effects of light spectrum, rearing density and light intensity on growth performance of scaled and mirror common carp Cyprinus carpio reared under recirculating system conditions, Aquacultural Engineering. 42:121–127.
- 7. Luchiari A.C. and Freire F.A.M. (2009) Effects of environmental colour on growth of Nile tilapia, Oreochromis niloticus (Linnaeus, 1758), maintained individually or in groups. Journal of Applied Ichthyology. 25:162–167.
- 8. Batzina A. and Karakatsouli N. (2012) The presence of substrate as a means of environmental enrichment in intensively reared gilthead seabream Sparus aurata: growth and behavioural effects. Aquaculture. 370–371:54–60.
- 9. McLean E., Cotter P., Thain C. and King N. (2008) Tank colour impacts performance of cultured fish. Ribarstvo. 2:43–54.
- 10. Ullmann J.F.P., Gallagher T., Hart N.S., Barnes A., Smullen R.P., Collin S.P. et al. (2011) Tank colour increases growth, and alters colour preference and spectral sensitivity, in barramundi (Lates calcarifer). Aquaculture. 322–323:235–240.
- 11. Smith A. R. (1978) Color Gamut Transform Pairs. Computer Graphics. 12:12–19.
- 12. Höglund E., Balm P.H.M. and Winberg S. (2002) Behavioural and neuroendocrine effects of environmental background colour and social interaction in Arctic charr (Salvelinus alpinus). Journal of Experimental Biology. 205:2535–2543. pmid:12124377
- 13. Food and Agricultural Organization of the United Nations (2014) The State of World Fisheries and Aquaculture: Opportunities and Challenges. [WWW Document] URL: http://www.fao.org/3/a-i3740t.pdf (accessed 10.11.14).
- 14. Broom, D. (1998) Fish welfare and the public perception of farmed fish. pp 89–91. In: C. Nash and V. Julien, (eds.), Report Aquavision ‘98. The Second Nutreco Aquaculture Business Conference Stavanger Forum, Norway, 13–15 May 1998.
- 15. Southgate P. and Wall T. (2001) Welfare of farmed fish at slaughter. In Practice. 23:277–284.
- 16. FSBI (2002) Fish Welfare. Briefing Paper 2, Fisheries Society of the British Isles, Granta Information Systems. [WWW Document] URL: http://www.fsbi.org.uk/assets/brief-welfare-refs.pdf (accessed 5.10.14).
- 17. Sandercock F.K. (1991) Life history of coho salmon Oncorhynchus kisutch. pp 397–445. In: Groot C. and Margolis I. (eds.), Pacific salmon life histories. UBC Press, Vancouver, BC (1991), 567 pp.
- 18. Thorarensen H. and Farrell A.P. (2011) The biological requirements for post-smolt Atlantic salmon in closed-containment systems. Aquaculture. 312:1–14.
- 19. Gelman A., and Hill J. (2006) Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge: Cambridge University Press.
- 20. Rabe-Hesketh S., and Skrondal A. (2008) Multilevel and Londgitudinal Modeling Using Stata (2nd ed.). College Station, TX, Stata Press.
- 21. Snijders T.A.B., and Bosker R.J. (2012) Multilevel Analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA, SAGE Publications. 2:354.
- 22. R Core Team (2014). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. [WWW Document] URL: http://www.R-project.org/ (accessed 9.10.14).
- 23. Wickham H. (2011) The Split-Apply-Combine Strategy for Data Analysis. Journal of Statistical Software. 40:1–29. [WWW Document] URL: http://www.jstatsoft.org/v40/i01/ (accessed 5.10.14).
- 24. Wickham H. (2007) Reshaping data with the reshape package. Journal of Statistical Software. 21(12).
- 25. Bates, D., M. Maechler, B. Bolker and S. Walker (2014) _lme4: Linear mixed-effects models using Eigen and S4_. R package version 1.1–7. [WWW Document] URL: http://CRAN.R.project.org/package=lme4 (accessed 4.10.14).
- 26. Kuznetsova, A., P.B. Brockhoff and R.H.B. Christensen (2015). lmerTest: Tests in Linear Mixed Effects Models. R package version 2.0–25. http://CRAN.R-project.org/package=lmerTest (accessed 25.02.15).
- 27. Cott H.B. (1940) Adaptive coloration in animals. London, UK: Methuen & Co., Ltd.
- 28. Godin J.-G.J. (1997) Behavioural ecology of teleost fishes. Oxford, UK: Oxford University Press.
- 29. Ruxton G.D., Sherrat T.N. and Speed M.P. (2004) Avoiding attack: the evolutionary ecology of crypsis, warning signals and mimicry. Oxford, UK: Oxford University Press.
- 30. Kjernsmo K. and Merilaita S. (2012) Background choice as an anti-predator strategy: the roles of background matching and visual complexity in the habitat choice of the least killifish. Proceedings of the Royal Society B. 279:4192–4198. pmid:22915675
- 31. Ashley P.J. (2007) Fish welfare: Current issues in aquaculture. Applied Animal Behaviour Science. 104:199–235.
- 32. Fraser D. (2008) Understanding Animal Welfare: The Science in its Cultural Context. Wiley-Blackwell, Oxford.
- 33. Braithwaite V.A. and Huntingford F.A. (2004) Fish and welfare: do fish have the capacity for pain perception and suffering? Animal Welfare. 13: 87–92.
- 34. Chandroo K.P., Duncan I.J.H., and Moccia R.D. (2004a) Can fish suffer? Perspectives on sentience, pain, fear and stress. Applied Animal Behaviour Science. 86:225–250.
- 35. Chandroo K.P., Yue S., and Moccia R.D. (2004b) An evaluation of current perspectives on consciousness and pain in fishes. Fish and Fisheries. 5:281–295.
- 36. Portavella M., Salas C., Vargas J.P., and Papini M.R. (2003) Involvement of the telencephalon in spaced-trial avoidance learning in the goldfish (Carassius auratus). Physiological Behaviour. 80: 49–56.
- 37. Portavella M., Torres B., and Salas C. (2004a) Avoidance response in goldfish: Emotional and temporal involvement of medial and lateral telencephalic pallium. Journal of Neuroscience. 24: 2335–2342.
- 38. Portavella M., Torres B., Salas C., Papini M.R. (2004b) Lesions of the medial pallium, but not of the lateral pallium, disrupt spaced-trial avoidance learning in goldfish (Carassius auratus). Neuroscience Letters. 362:75–78.
- 39. Sneddon L.U. (2002) Anatomical and electrophysiological analysis of the trigeminal nerve in a teleost fish, Oncorhynchus mykiss. Neuroscience Letters. 319:167–171. pmid:11834319
- 40. Sneddon L.U., Braithwaite V.A. and Gentle M.J. (2003a) Do fish have nociceptors? Evidence for the evolution of a vertebrate sensory system. Proceedings of the Royal Society of London, Series B. Biological Sciences. 270:1115–1121.
- 41. Sneddon L.U., Braithwaite V.A. and Gentle M.J. (2003b) Novel object test: examining pain and fear in the rainbow trout. Journal of Pain. 4:431–440.
- 42. Dunlop R. and Laming P. (2005) Mechanoreceptive and nociceptive responses in the central nervous system of goldfish (Carassius auratus) and trout (Oncorhynchus mykiss). Journal of Pain. 6:561–568. pmid:16139775
- 43. Rodríguez F., Durán E., Vargas J.P., Torres B. and Salas C. (1994) Performance of goldfish trained in allocentric and egocentric maze procedures suggest the presence of a cognitive mapping system in fishes. Animal Learning and Behaviour. 10:108–114.
- 44. Braithwaite V.A. (1998) Spatial memory, landmark use and orientation in fish. pp 86–102. In: Healy S.D. (ed.) Spatial Representations in Animals. Oxford University Press: Oxford, UK.
- 45. Odling-Smee L. and Braithwaite V.A. (2003b) The role of learning in fish orientation. Fish and Fisheries. 4:235–246.
- 46. Laland K.N., Brown C. and Krause J. (2003) Learning in fishes: from three-second memory to culture. Fish and Fisheries. 4:199–202.
- 47. Kelley J.L. and Magurran A.E. (2003) Learned predator recognition and antipredator responses in fishes. Fish and Fisheries. 4:216–226.
- 48. Fraser D. and Matthews L.R. (1997) Preference and motivational testing. pp 159–173. In: Appleby M.C. and Hughes B.O. (eds.), Animal Welfare. CAB International, Wallingford.
- 49. Phillips P.A., Fraser D. and Pawluczuk B. (2000) Floor temperature preference of sows at farrowing. Applied Animal Behaviour Science. 67:59–65. pmid:10719189
- 50. Widowski T.M. (1992) The preferences of hens for compact fluorescent over incandescent lighting. Journal of Animal Science. 72:203–211.
- 51. Davis N., Prescott N., Savory C. and Wathes C. (1999) Preferences of growing fowls for different light intensities in relation to age, strain and behaviour. Animal Welfare. 8:193–203.
- 52. Tucker C.B., Weary D.M. and Fraser D. (2003) Effects of three types of free-stall surfaces on preferences and stall usage by dairy cows. Journal of Dairy Science. 86:521–529. pmid:12647958
- 53. Van de Weerd H.A., van Loo P.L.P., van Zutphen L.F.M., Koolhaas J.M. and Baumans V. (1997) Preferences for nesting material as environmental enrichment for laboratory mice. Laboratory Animals. 31:133–143. pmid:9175010
- 54. Dawkins M.S. (1977) Do hens suffer in battery cages? Environmental preferences and welfare. Animal Behaviour. 225:1034–1046.
- 55. Duncan I.J.H. (1978) The interpretation of preference tests in animal behaviour (letter to the Editor). Applied Animal Ethology. 4:197–200.
- 56. Naas K., Huse I. and Iglesias J. (1996) Illumination in first feeding tanks for marine fish larvae. Aquacultural Engineering. 15:291–300.
- 57. Volpato G.L., Gonçalves-de-Freitas E., and Fernandes-de-Castilho M. (2007) Insight into the concept of fish welfare. Diseases of Aquatic Organisms. 75:165–171. pmid:17578256
- 58. Fraser N.H.C., Metcalfe N.B. and Thorpe J.E. (1993) Temperature-dependent switch between diurnal and nocturnal foraging in salmon. Proceedings of the Royal Society B. 252:135–139.
- 59. Chaouloff F., Durand M. and Mormède P. (1997) Anxiety- and activity-related effects of diazepam and chlordiazepoxide in the rat light/dark and dark/light tests. Behavoural Brain Research. 85:27–35.
- 60. Steenbergen P.J., Richardson M.K. and Champagne M.K. (2011) Patterns of avoidance behaviours in the light/dark preference test in young juvenile zebrafish: a pharmacological study. Behavioural Brain Research. 222:15–25. pmid:21421013
- 61. Ellis T., North B., Scott A.P., Bromage N.R., Porter M. and Gadd D. (2002) The relationships between stocking density and welfare in farmed rainbow trout. Journal of Fish Biology. 61:493–531.
- 62. Turnbull J., Bell A., Adams C., Bron J. and Huntingford F. (2005) Stocking density and welfare of cage farmed Atlantic salmon: application of a multivariate analysis. Aquaculture. 243: 121–132.
- 63. Marchesan M., Spoto M., Verginella L. and Ferrero E.A. (2005) Behavioural effects of artificial lights on fish species of commercial interest. Fisheries Research. 73:171–185.
- 64. Fujimoto M., Arimoto T., Mosichita F. and Naitoh T. (1991) The background adaptation of the flatfish, Paralichthys olivaceus. Physiology and Behaviour. 50:185–188.
- 65. Keenleyside M.H.A., and Yamamoto F.T. (1962) Territorial behaviour of juvenile Atlantic salmon (Salmo salar L.). Behaviour, 19:139–169.
- 66. Abbott J.C. and Dill L.M. (1985) Patterns of aggressive attack in juvenile steelhead trout (Salmo-Gairdneri). Canadian Journal of Fisheries and Aquatic Sciences. 42:1702–1706.
- 67. O’Connor K.I., Metcacalfe N.B. and Taylor A.C. (1999). Does darkening signal submission in territorial contests between juvenile Atlantic salmon, Salmo salar?. Animal Behaviour. 58:1269–1276. pmid:10600149
- 68. Canada House of Commons (2013) Closed Containment Salmon Aquaculture. Report of the Standing Committee on Fisheries and Oceans, First Session, 41st Parliament, March 2013. [WWW Document] URL: http://www.fao.org/3/a-i3720e.pdf (accessed 21.10.14).
- 69. Schneider R. and Nicholson B. (1980) Bacteria associated with fin rot disease in hatchery-reared Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences. 37:1505–1513.
- 70. Turnbull J.F., Adams C.E., Richards R.H., R. H. and Robertson D.A. (1998) Attack site and resultant damage during aggressive encounters in Atlantic salmon (Salmo salar L.) parr. Aquaculture. 159:345–353.
- 71. Hyvärinen P. and Rodewald P. (2013) Enriched rearing improves survival of hatchery-reared Atlantic salmon smolts during migration in the River Tornionjoki. Canadian Journal of Fisheries and Aquatic Sciences. 70:1386–1395.
- 72. Näslund J., Rosengren M., Del Villar D., Gansel L., Norrgård J.R., Persson L. et al. (2013) Hatchery tank enrichment affects cortisol levels and shelter-seeking in Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences. 70:585–590.
- 73. Braithwaite V.A. and Salvanes A.G.V. (2005) The importance of environmental variability for coastal cod r eared for the rehabilitation of wild populations. Proceedings of the Royal Society of London, Series B. 272:1107–1113.
- 74. Steiger A., Tschanz B., Jakob P., and Scholl E. (1979) Behavioral studies of fattening pigs on different floor coverings and with a varying rate of stocking. Schweizer Archiv Fur Tierheilkunde No. 121:109–126.
- 75. Cheng C.L. and Flamerique I.N. (2004) New mechanism for modulating colour vision; single cones start making a different opsin as young salmon move to deeper waters. Nature. 428:279.
- 76. Temple S.E., Ramsden E.M.S., Haimberger T.J., Roth W.M. and Hawryshyn C.W. (2006). Seasonal cycle in vitamin A1/A2-based visual pigment composition during the life history of coho salmon (Oncorhynchus kisutch). Journal of Comparative Physiology. A, Neuroethology, Sensory, Neural, and Behavioral Physiology. 192:301–313. pmid:16292551
- 77. Shapovalov L. and Taft A.C. (1954) The life histories of the Steelhead Trout (Salmo gairdneri) and Silver Salmon (Oncorhynchus kisutch) with special reference to Waddell Creek, California, and recommendations regarding their management. California Department of Fish and Game. Fisheries Bulletin. 98:1–375.