Portal de Periódicos da CAPES

Does waste feed from salmon farming affect the quality of saithe ( Pollachius virens L.) attracted to fish farms?

2020; Wiley; Volume: 51; Issue: 4 Linguagem: Inglês

10.1111/are.14519

1365-2109

Ingebrigt Uglem, Kilian Toledo‐Guedes, Pablo Sánchez-Jérez, Eva Marita Ulvan, Tor Hatten Evensen, Bjørn Steinar Sæther,

Marine and fisheries research

Aquaculture ResearchVolume 51, Issue 4 p. 1720-1730 ORIGINAL ARTICLEOpen Access Does waste feed from salmon farming affect the quality of saithe (Pollachius virens L.) attracted to fish farms? Ingebrigt Uglem, Ingebrigt Uglem Norwegian Institute of Nature Research, Trondheim, NorwaySearch for more papers by this authorKilian Toledo-Guedes, Corresponding Author Kilian Toledo-Guedes ktoledo@ua.es orcid.org/0000-0002-6819-1902 Department of Marine Sciences and Applied Biology, University of Alicante, Alicante, Spain Correspondence Kilian Toledo-Guedes, Department of Marine Sciences and Applied Biology, University of Alicante, P.O. Box 99, Alicante 03080, Spain. Email: ktoledo@ua.esSearch for more papers by this authorPablo Sanchez-Jerez, Pablo Sanchez-Jerez Department of Marine Sciences and Applied Biology, University of Alicante, Alicante, SpainSearch for more papers by this authorEva Marita Ulvan, Eva Marita Ulvan Norwegian Institute of Nature Research, Trondheim, NorwaySearch for more papers by this authorTor Evensen, Tor Evensen NOFIMA Marin, Tromsø, NorwaySearch for more papers by this authorBjørn Steinar Sæther, Bjørn Steinar Sæther NOFIMA Marin, Tromsø, NorwaySearch for more papers by this author Ingebrigt Uglem, Ingebrigt Uglem Norwegian Institute of Nature Research, Trondheim, NorwaySearch for more papers by this authorKilian Toledo-Guedes, Corresponding Author Kilian Toledo-Guedes ktoledo@ua.es orcid.org/0000-0002-6819-1902 Department of Marine Sciences and Applied Biology, University of Alicante, Alicante, Spain Correspondence Kilian Toledo-Guedes, Department of Marine Sciences and Applied Biology, University of Alicante, P.O. Box 99, Alicante 03080, Spain. Email: ktoledo@ua.esSearch for more papers by this authorPablo Sanchez-Jerez, Pablo Sanchez-Jerez Department of Marine Sciences and Applied Biology, University of Alicante, Alicante, SpainSearch for more papers by this authorEva Marita Ulvan, Eva Marita Ulvan Norwegian Institute of Nature Research, Trondheim, NorwaySearch for more papers by this authorTor Evensen, Tor Evensen NOFIMA Marin, Tromsø, NorwaySearch for more papers by this authorBjørn Steinar Sæther, Bjørn Steinar Sæther NOFIMA Marin, Tromsø, NorwaySearch for more papers by this author First published: 16 January 2020 https://doi.org/10.1111/are.14519Citations: 3AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Salmon farms attract large amounts of wild fish, which prey on uneaten feed pellets. The modified diet of the wild fish aggregating at salmon farms may reduce the flesh quality of the fish, influencing the local fisheries. We compared the quality of saithe (Pollachius virens) captured near (farm associated—FA) or more than 5 km away (un-associated—UA) from salmon farms in Norway. The fish were captured during summer, autumn and spring using two commercial fishing methods (jigging and bottom nets). Overall, the fillet quality of FA saithe was good, although it was clearly reduced for almost 10% of the catch. Moreover, the quality of the FA saithe was significantly reduced compared with UA saithe, but the differences were small. Our results also showed that fish caught with jigging had better quality than fish caught with nets, and that fish that died in the nets were of lower quality than fish that were alive after hauling. There was no clear variation among seasons in fillet quality. Although no major and overall differences in quality were found between FA and UA saithe, reduced quality for even a modest proportion of the fish may influence the value of the total catch. 1 INTRODUCTION Salmon farming can modify ecosystems in coastal waters (Maurstad, Dale, & Bjørn, 2007; Wiber, Young, & Wilson, 2012). The large fish feed requirement results in considerable amounts of organic by-products, in terms of uneaten feed pellets falling through the cages, fragmentation of pellets during feeding, and dissolved and particulate nutrients originating from faeces (Aas et al., 2011; Dempster et al., 2011; Holmer, 2010). The amount of uneaten feed (waste feed) during commercial operation of salmon farms has to our knowledge not been measured, but it has been assumed that the loss lies in the region of 3%–5% (Otterå, Karlsen, Slinde, & Olsen, 2009; Svåsand et al., 2015). In 2015, the salmon farming industry in Norway used more than 1.7 million tons of food (Norwegian Directorate of Fisheries, 2015), suggesting that thousands of tons of waste feed are available to wild fish each year. Attraction of wild fish to open cage fish farms, that is farms consisting of floating net cages, is a global phenomenon (Barrett, Swearer, & Dempster, 2019; Callier et al., 2018; Uglem, Karlsen, Sanchez-Jerez, & Sæther, 2014). More than 160 fish species, belonging to about 60 families, have been detected in the near vicinity of open cage farms (Sanchez-Jerez et al., 2011). In Norway, 15 fish species belonging to nine families have been observed underneath salmon farms, with the most common species being saithe (Pollachius virens), Atlantic cod (Gadus morhua), haddock (Melanogrammus aeglefinus) and mackerel (Scomber scomber) (Dempster et al., 2009). Waste feed from the farms is believed to be the primary cause for aggregation of wild fish at open cage farms (Dempster et al., 2011; Fernandez-Jover, Sanchez-Jerez, Bayle-Sempere, Valle, & Dempster, 2008; Sanz-Lazaro, Belando, Marin-Guirao, Navarrete-Mier, & Marin, 2011). The occurrence of waste feed in stomach samples from wild fish caught at salmon farms has been quantified for saithe and cod only (Dempster et al., 2011), but waste feed has also been observed in stomach samples from haddock, mackerel and pollack (Pollachius pollachius) I. Uglem (unpublished data). Waste feed has been found in stomach samples of 14%–92% and 11%–32% of farm associated (hereafter FA) saithe and cod respectively (reviewed in Uglem et al., 2014). In general, FA gadoids are significantly fatter and have larger energy stores than un-associated (hereafter UA) fish (Dempster et al., 2011), and have higher concentrations of terrestrial derived fatty acids and lower concentrations of docosahexaenoic acid (DHA) in muscle and liver compared with UA fish (Fernandez-Jover et al., 2011). The metabolic status of FA and UA saithe is also different and FA fish have higher levels of, for example lactate and lower levels of creatine than UA fish, both relationships indicating a potential for reduced quality of the FA fish (Maruhenda Egea, Toledo-Guedes, Sanchez-Jerez, & Uglem, 2015). Salmon feed is designed for promoting optimal quality and growth of salmon, and may not be an optimal food for other fish species. A diet change from natural prey to salmon feed may therefore affect the flesh quality of FA gadoids (Uglem et al., 2014). Norwegian coastal fishermen and fish buyers have raised concerns on the quality of wild gadoids feeding on waste feed (Otterå & Skilbrei, 2014; Skog, Hylland, Tortensen, & Berntssen, 2003). Fillets from FA fish are claimed to be soft and with a high degree of gaping. Abnormal coloration and unappealing smell have also been reported. It can be hypothesized that the reduction in quality, in terms of soft texture, is related to both ante-mortem and post-mortem glycolysis. The former is probably related to stress and/or activity before death (Kiessling, Espe, Ruohonen, & Mørkøre, 2004). In both cases, the glycogen is broken down to lactic acid, thus making the fish muscle slightly acidic, which in turn may increase flesh softness and gaping (Bremner, 1999; Kristoffersen, Tobiassen, & Steisund, 2006). The quality may also depend on the nutritional state of the fish. Since well-fed and fat FA fish have large amounts of lipids and glycogen in liver and muscle, they will have a high glycolytic potential and possibly low ultimate post-mortem pH and reduced fillet quality (Kristoffersen et al., 2006). The quality of FA saithe has been evaluated in several ways, but with inconsistent results (Uglem et al., 2014). Results from sensory tests of saithe that have had a diet consisting of salmon feed compared with fish that have had a more natural diet indicate slight variations, but no consistent trend, in taste and appearance (Otterå et al., 2009; Sæther et al., 2012; Uglem et al., 2017). FA and UA saithe have also been compared with different quality index assessment methods, that is evaluation and combination of several fillet properties such as smell, splitting/gaping, colour, consistency and surface appearance into an index value (Akse, Tobiassen, Midling, & Aas, 2007; Otterå et al., 2009). In the same way as for the sensory analyses, variation in fillet quality index has been relatively small, without consistent trends between either FA saithe caught in the wild or saithe fed salmon/cod pellets in captivity compared with wild-caught FA fish (Bjørn et. al., 2007; Otterå et al., 2009; Sæther et al., 2012). However, the interpretation of the fillet index results for saithe may have been confounded by the different groups in the previous studies being caught with different fishing techniques (jigging, pots or bottom nets). In general, fish caught by jigging or in pots are expected to be of a better quality than fish caught by nets, as they are caught alive and rapidly killed and bled, while fish caught in nets might be considerably stressed or even dead at sea after having been kept trapped in the nets for many hours (Toledo-Guedes, Ulvan, & Uglem, 2016). In support of this, Sæther et al. (2012) found that the fillet quality index of saithe caught with nets (UA fish) was significantly lower compared with saithe caught alive in pots (FA fish). Other confounding factors may be that the quality depends on the amount of waste feed consumed over time and thus the condition of the fish, as well as time of capture since flesh quality in fish may vary throughout the reproduction cycle and possibly also with environmental variation. In this study, we have captured FA and UA saithe in two areas in Norway during summer, autumn and spring using two commercial fishing methods (jigging and bottom nets). This study design allowed us to compare quality directly among groups of saithe, as well as assess if the quality varied among seasons. Hence, by avoiding confounding factors, such as capture by different types of gear, the present study represents a novel approach to evaluate the influence of salmon farming on the quality of saithe. Texture measurements and two different index-based assessment methods were used to evaluate fillet quality in a way that is realistic for the first part of the distribution chain of saithe, that is fish landing facilities and wholesalers, since this would be the most relevant stage in the value chain for assessing and sorting fish with respect to quality. 2 MATERIAL AND METHODS 2.1 Study location and fish sampling Saithe were collected in Ryfylke (N59°11, E05°53) and close to the island Hitra (N63°62, E08°99) in southwestern and mid-western Norway at six different occasions during summer, autumn and spring in 2013–2015 (Figure 1, Table 1). FA fish were captured less than 500 m from farms, containing salmon above 1 kg, in each of these two areas (Figure 1). Since Norwegian salmon farms are fallowed between each production cycle and the size of the farmed fish vary throughout the cycle, sampling of saithe was for practical reasons carried out at different farms within the same area throughout the study period to collect standardized samples at farms that contained larger salmon. This is the part of the production cycle when the feeding rate, and therefore the amount of waste feed available to the wild fish, is at its highest. UA fish were sampled from locations 5–10 km distant from the nearest farm (Figure 1). Nonetheless, the UA fish from Ryfylke were clearly influenced by salmon farms; between 7% and 19% of pellets, prevalence in their diet was found, while UA fish from Hitra had no pellets in their stomachs (Table 1). Thus, UA fish from Ryfylke were excluded from further analyses. Figure 1Open in figure viewerPowerPoint Map showing the areas in Hitra and Ryfylke where the saithe were captured. The boxes indicate where farm associated (FA) and un-associated (UA) saithe were captured Table 1. Morphometric data and quality indices (mean ± standard deviations) for farm associated (FA) and un-associated (UA) saithe from Hitra and Ryfylke, captured either by jigging or in nets Area Season Fishing gear FA/UA N Length (cm) Weight (kg) Condition factor (K) HSI GSI Pellets prevalence (%) Pellets weight (g) Pellet in diet (%) QIM IT Hitra Autumn Jigging UA 30 68 ± 4 3.05 ± 0.7 0.95 ± 0.13 6.9 ± 3.1 1.82 ± 1.52 0 — — 1.91 ± 0.75 0.67 ± 0.71 FA 31 60 ± 8 2.14 ± 0.97 0.96 ± 0.13 5.6 ± 2 0.93 ± 1.14 29 56 ± 77 100.0 2.73 ± 1.36 2.70 ± 1.88 Net UA 30 68 ± 3 2.96 ± 0.48 0.94 ± 0.1 6.8 ± 1.4 1.79 ± 1.63 0 — — 2.39 ± 1.40 1.10 ± 1.32 FA 30 66 ± 7 3.1 ± 0.97 1.05 ± 0.17 8.2 ± 2.4 1.27 ± 0.93 33 141 ± 163 88.8 2.96 ± 1.51 2.17 ± 2.09 Spring Jigging UA — — — — — — — — — — — FA 30 69 ± 4 3.09 ± 0.73 0.93 ± 0.16 6.1 ± 3.5 0.63 ± 1.39 60 166 ± 106 93.6 2.82 ± 1.29 2.40 ± 1.77 Net UA 31 66 ± 6 2.44 ± 0.63 0.84 ± 0.07 6.2 ± 4.4 0.63 ± 0.81 0 — — 1.72 ± 0.85 1.13 ± 1.18 FA 30 68 ± 6 3.19 ± 0.68 1.04 ± 0.18 10 ± 4.8 1.12 ± 3.09 50 89 ± 63 97.1 2.57 ± 1.10 2.03 ± 1.43 Summer Jigging UA 20 62 ± 6 2.09 ± 0.64 0.86 ± 0.07 6 ± 2.2 0.25 ± 0.28 0 — — 1.86 ± 0.57 2.42 ± 1.57 FA 25 68 ± 7 3.18 ± 0.89 0.98 ± 0.13 10 ± 3.2 0.36 ± 0.41 88 78 ± 87 92.0 2.41 ± 0.96 2.84 ± 1.80 Net UA 20 69 ± 4 2.99 ± 0.47 0.9 ± 0.08 6.8 ± 2.7 0.43 ± 0.3 0 — — 2.57 ± 1.34 2.60 ± 2.16 FA 48 66 ± 6 2.92 ± 0.74 1.01 ± 0.13 10.7 ± 3.8 0.24 ± 0.29 23 96 ± 112 95.3 2.91 ± 1.46 3.40 ± 2.29 Ryfylke Autumn Jigging UA — — — — — — — — — — — FA 25 45 ± 2 0.9 ± 0.1 1.02 ± 0.12 11.6 ± 2 0.12 ± 0.15 72 5 ± 3 100.0 6.85 ± 1.49 6.29 ± 1.76 Net UA 36 59 ± 6 2.05 ± 0.73 0.96 ± 0.1 8.9 ± 2.8 3.11 ± 3.02 19 94 ± 127 87.0 4.31 ± 1.88 3.53 ± 2.02 FA 25 57 ± 5 1.85 ± 0.7 0.93 ± 0.12 8 ± 3.8 3.44 ± 3.67 40 38 ± 75 85.5 4.70 ± 1.98 4.92 ± 1.53 Spring Jigging UA — — — — — — — — — — — FA 30 57 ± 5 2.09 ± 0.52 1.15 ± 0.14 7.2 ± 3.4 0.37 ± 0.28 70 54 ± 41 96.8 1.85 ± 0.59 1.80 ± 1.42 Net UA 30 62 ± 5 3.07 ± 0.85 1.26 ± 0.18 12 ± 5.1 3.52 ± 5.8 7 79 ± 67 100.0 2.25 ± 1.02 2.63 ± 2.16 FA 30 57 ± 7 2.29 ± 0.52 1.27 ± 0.21 9.1 ± 3.4 0.77 ± 1.34 70 156 ± 73 98.1 2.35 ± 0.65 2.97 ± 1.56 Summer Jigging UA — — — — — — — — — — — FA 30 58 ± 7 1.95 ± 0.67 1 ± 0.09 9.5 ± 2.5 0.38 ± 0.3 90 32 ± 38 78.5 3.16 ± 0.81 2.59 ± 1.08 Net UA — — — — — — — — — — — FA 23 65 ± 8 2.61 ± 0.94 0.9 ± 0.13 7.8 ± 3.6 0.42 ± 0.41 26 168 ± 208 100.0 2.96 ± 1.06 1.82 ± 1.11 Abbreviations: GSI, gonadososmatic index; HIS, hepatosomatic index; IT, industry test; N, sample size; Pellets in diet, percentage of diet that consisted of pellets; Pellets prevalence, percentage of sample with pellets in the stomach; pellets weight, average amount of pellets in stomach; QIM, quality index method. The fish were captured by commercial coastal fishermen using bottom gillnets which were soaked between 15 and 17 hr, and automatic jigging machines that hauled the fish on board in 2–3 min. After hauling, the fish that were alive were immediately euthanized with a blow to the head. The consciousness of the fish that were alive was evaluated according to Kestin, van de Vis, and Robb (2002), defined by physical responses towards handling and operculum movement prior to euthanization. Then, the fish was bled and transferred to a 500 L tank with running seawater at ambient seawater temperature where it was kept for approximately 1 hr before it was gutted and cleaned. Following cleaning and rinsing, the fish were transferred to Styrofoam boxes (BEWI. Dimensions: 22 × 39 × 79 cm) cooled with ice. The boxes with the fish were stored at low temperature (2–4°C) for 96 hr before the quality analyses. This study conforms to Directive 2010/63/EU on the protection of animals used for scientific purposes. 2.2 Size, diet and condition indices Upon capture and before bleeding the total length (TL) and the whole-body weight (W) of the fish were measured to the nearest 1 cm and 10 g respectively (Kern & Sohn, Scale model: HDB 10K 10N). Each fish was tagged with an individually coded T-bar tag (Hallprint Fish tags) to allow identification during subsequent processing. Immediately after gutting, liver (LW) and gonad (GW) weights were recorded for each fish to the nearest 1 g (Kern & Sohn, Scale model: FFN 6K 1IPN). Sex was determined by macroscopic examination of the gonads. Three morphometric indices were calculated on basis of the measurements. Fultons condition index (FCI) was calculated with the formula: FCI = (W/TL3) × 100. The hepatosomatic index (HSI) and gonadosomatic (GSI) indexes were calculated using the formulas: HIS = (LW/W) × 100 and GSI = (GW/W) × 100. Stomach contents from the foregut were examined and the prey identified into five broad categories (salmon feed pellets, fish, crustaceans, bivalves and other organic matter) and weighed to the nearest gram (Kern & Sohn, Scale model: FFN 6K 1IPN). 2.3 Samples Altogether, 554 saithe were captured during the project, 221 with jigging (50 at control sites and 171 at farm sites) and 333 with nets (147 at control sites and 186 at farm sites; Table 1). At four occasions, it was not possible to obtain control samples using jigging, partly due to extreme weather conditions (Hitra) or limited availability of saithe in areas away from the salmon farms (Ryfylke) (Table 1). At one occasion (Ryfylke, autumn), the fish caught by jigging close to farms were considerably smaller than the commercial size of saithe. This sample was therefore not included in the further analyses. Apart from these occasions between 20 and 48, saithe were collected for the different groups (Table 1). The fish caught at Hitra were larger than the fish caught in Ryfylke, most likely due to a general variation in the size composition of saithe in these two areas. In general, the FA saithe from Hitra had a higher condition factor and larger livers per body mass than the UA fish (HSI: t = 5.4, p < .001, K: t = 6.6, p < .001). The HSI was slightly higher for the UA fish than the FA fish from Ryfylke (t = −2.4, p = .03), while there was no difference between FA and UA fish from Ryfylke in condition factor (t = −1.5, p = .21). None of the UA fish from Hitra had pellets in their stomach, while pellets were found in the stomachs of 19 and 7% of the UA fish from Ryfylke in autumn and spring respectively (Table 1). The findings from Ryfylke may indicate that UA fish is difficult to obtain in this region due to high farm density and a high degree of inter-farms movements (Otterå & Skilbrei, 2014; Uglem, Dempster, Bjorn, & Sanchez-Jerez, 2009). The average GSI was low for all groups indicating that the majority of the fish were immature. Saithe should normally spawn during late winter or early spring, and it was thus expected that the GSI should be highest in the spring samples. Out of 181 fish caught during spring, only 15 (8.2%) had clearly mature or maturing gonads with a GSI above 2 (Table 1). The average prevalence and mass of pellets in stomachs from FA saithe from Hitra was 44% and 105 g (±104 g), while 61% of the saithe in Ryfylke had on average 65 g (±87 g) pellets in their stomach (Table 1). 2.4 Fillet quality assessment The quality of the saithe was assessed, 4 days after capture, using two quality tests: a modified quality index method (QIM) and a simple industry test (IT). Moreover, the texture of thawed loins was evaluated for a selection of the samples (stored for 4 months at −18°C). Quality index methods have been developed and used for evaluating the quality of a range of fish species (Bonilla, Sveinsdottir, & Martinsdottir, 2007 and references therein) and is recognized as a reference method in sensory research (Martinsdóttir, Luten, Schelvis-Smit, & Hylding, 2003; Olafsdóttir et al., 1997). A QIM approach has for instance been used for documenting reduced quality of bogue (Boops boops L.) captured in the proximity of fish farms (Bogdanovic, Šimat, & Marković, Šimat, Frka-Roić & Marković, 2012). In this study, our rationale for using a QIM approach was the need for a fast, reliable and simple method for assessing a potential alteration of quality as experienced by fishermen and wholesalers. As the purpose of this study was to detect potential differences between two groups and not to analyse fillet degradation over time or shelf life, the quality assessment was carried out 4 days after capture, that is after a sufficient period for avoiding confounding effects due to onset of rigor mortis. The QIM test (Table 2) was based on a method developed for assessment of quality of Atlantic cod fillets (QIM; Akse et al., 2007), while the industry test (Table 2) was based on simpler test designed for rapid detection of significant texture related quality defects for salmon fillets (IT; Erikson, Bye, & Oppedal, 2009). Both methods were slightly modified after pre-observations of saithe captured in a pilot study in 2012 as they originally were developed for other species. In the QIM test, five fillet attributes (odour, gaping, colour, texture and surface) were assessed and the maximum sum was 12, while in the industry test, three attributes (inelasticity, softness and gaping) were assessed with a maximum sum of 9. The total score from both tests indicate that the fillet quality decrease with increasing value. Three trained assessors carried out the quality evaluation, and the arithmetic mean of these three independent assessments was used in the subsequent analyses. Table 2. Individual quality parameters and their ranges for each quality index used Quality parameter Description Score QIM Odour Fresh smell of sea 0 Neutral 1 Fishy 2 Ammonium 3 Gaping No gaping 0 Initial gaping 1 Some, loose fillet 2 Major, disjoint fillet 3 Colour Normal—fresh colour 0 Abnormal—colour 1 Texture Firm, natural 0 Marginally soft 1 Soft 2 Very soft 3 Surface Dry and shiny 0 Partly dissolved 1 Very dissolved 2 IT Inelasticity Folded fillet, with skin side down, straightens out quickly 0 Somewhat elastic, the fillet straightens out slowly 1 Inelastic, the fillet remains folded over 2 Softness Firm, surface restored short time after approximately 1 kg pressure is applied with forefinger 0 Reduced firmness, finger pressure leaves lasting imprint 1 Soft fillet, the finger goes through the fillet 2 Gaping (loin, belly and tail) No gaping 0 Minor signs of gaping 1 Slight gaping 2 Moderate gaping 3 Considerable gaping 4 Major, disjoint fillet 5 Note Modified after Akse et al. (2007 and Erikson et al. (2009). Abbreviations: IT, industrial test; QIM, quality index method. The texture analyses followed established procedures used for evaluation of the quality of fish fillets (Larsson et al., 2012). Before analysis, the fillets were thawed at 4°C in a thermo-controlled cabinet. The analyses were performed instrumentally (TA-HDi Texture Analyser; Stable Micro Systems Ltd.) by pressing a flat-ended cylinder (12.5 mm diameter, type P/0.5) into the fillet perpendicular to the muscle fibres at 1 mm/s until the fillet thickness was reduced to 90% of original thickness. Texture was measured at three locations of each loin for fish collected at Hitra during spring 2015 and autumn 2014. The arithmetic mean of the three measurements was used in the subsequent analyses. 2.5 Data analyses Extreme weather conditions and low abundance of fish in some of the sampling periods, together with the presence of pellets in the diet of UA fish from Ryfylke (Section 2.1), hindered full factorial analyses. The quality of FA fish from Hitra (Farm 1) and Ryfylke (Farm 2) was therefore compared with UA fish from Hitra as a control. Since previous studies have shown that the metabolic status of UA fish from the same locations at Hitra differ significantly from FA fish (Maruhenda Egea et al., 2015), and no feed pellets were found in the stomach of these fish we assume that the UA fish from Hitra were unaffected by salmon farming. For analyses that involved comparisons of FA and UA fish between regions, the QIM and IT scores were standardized (Z-standardization) within each sampling season and region to control for unaccounted variation. Moreover, pairwise tests (i.e., pseudo-t statistic as given by PERMANOVA routine or chi-squared test; Anderson, Gorley, & Clarke, 2008) were carried out when needed to explore intergroup differences and interactions between fixed factors. To examine potential effects of farming or fishing gear on fillet quality, standardized scores for each of the fillet quality indices (QIM and IT) were analysed with permutational multivariate ANOVAs (PERMANOVA; Anderson et al., 2008) over a similarity matrix built using Euclidean distance and performing 4,999 permutations. The models included the fixed factors origin (Hitra UA, Hitra FA and Ryfylke FA) and fishing gear (Gillnet and Jigging). Furthermore, to analyse how the individual fillet quality parameters (Table 2) used to calculate the QIM and IT scores contributed to explain the observed variation in quality between origin and gear, SIMPER analyses were performed on standardized values of the different fillet quality parameters. Potential influence of region and season on fillet quality was examined by using the non-standardized QIM and IT scores of FA fish with permutational univariate ANOVAs in the same way as for the analyses of quality. In this case, the two fishing methods were tested separately as the quality to some extent varied between these two methods (see Section 3). These models included the fixed factors region (Hitra and Ryfylke) and season (spring, summer and autumn). The relationships between quality (standardized QIM or IT scores), condition factor (K) and hepatosomatic index (HIS) were explored with linear regressions, where fishing gear were added as an independent factor, since quality varied between different gear types. To assess whether presence of pellets in the stomach content of FA fish influenced fillet quality (standardized values of QIM or IT), permutational univariate ANOVAs were performed with the fixed factors pellets (presence and absence) and fishing gear (gillnet and jigging). Likewise, permutational univariate ANOVAs were used to determine whether the quality (standardized values of QIM or IT) differed between gillnetted fish that were alive or dead when hauled aboard (Fixed factors: dead vs. alive and origin). Fillet texture was measure for fish collected at Hitra during autumn 2014 and spring 2015 only. As UA saithe captured with nets were not collected during spring 2015, due to extreme weather conditions, the two samples were analysed separately. A one-way ANOVA with Tukey's post hoc tests was used to analyse the sample from spring 2015, while a univariate GLM with origin and fishing gear as fixed factors was used to analyse the autumn 2014 sample. 3 RESULTS The quality of the saithe was predominantly good with most of the total QIM scores being less than 3 (74.2%; Table 1). Only 25.8% of the saithe were given QIM scores above 3% and 11.6% above four. When split in different groups, 8.2% and 9.4% of the FA saithe from Hitra and Ryfylke had QIM scores at or above five, that is a sufficiently high score for the overall quality of the fish to be regarded as reduced. In comparison, 1.5% of the UA saithe from Hitra were give a QIM score at or above five. The proportion of the fish with a QIM score at or above five was higher for the FA fish compared with the UA fish (chi-square test, df = 2, χ2 = 7.95, p = .02). The origin of the fish (FA fish from Hitra and Ryfylke and UA fish from Hitra; pseudo-F = 13.19, p < .001) was significantly associated with variation in fillet quality as estimated by the QIM test. When FA fish from Hitra and Ryfylke were compared with UA fish from Hitra, the QIM scores show that the fillet quality of the FA fish was similar between sites (t = 2.02, p = .056), but significantly reduced compared with the UA fish (FA Hitra vs. UA fish: t = 5.17, p < .001; FA Ryfylke vs. UA fish: t = 3.04, p = .002) regardless of the fishing gear used (Figure 2a). The QIM scores also indicate that the quality of the fish captured with nets was significantly reduced compared with fish captured with jigging (Figure 2a, pseudo-F = 4.32, p = .035). Figure 2Open in figure viewerPowerPoint Fillet quality measured as (a) Quality index method and (b) Industry test for farm associated (FA) and un-associated (UA) saithe from Hitra and Ryfylke. The grey boxes represent saithe captured by gillnets, while white boxes indicate saithe captured by jigging. Horizontal solid lines are the median values, and the upper and lower edges on the boxes represent 75 and 25 percentiles. Vertical solid lines indicate extremal values (≤1.5 times the length of the box). Outliers are indicated with solid black circles Fillet quality assessments based on the IT method also differed between FA and UA fish (pseudo-F = 16.38, p < .001), but not between fis