Downgrading traits are economically important for the Atlantic salmon industry because they are directly related to the price of the products. Their prevalence is typically low , and it is therefore difficult to obtain data in a normal sib trait testing regime for genetic studies and thus obtain breeding values for selection as part of a breeding programme. In this study, we focus on melanin spots in the fillets , which is the most common fillet quality problem (Nordberg, 2018). Melanin spots is estimated to be the cause of 9–67% price losses, depending on the intensity and size of the discoloration (Färber, 2017).
There are only a few reported estimates of heritability of melanin spots in salmon fillets; they typically indicate only low genetic variation for melanin, and the heritability estimates were coupled with high standard errors (h 2 = 0.025±0.019 in a study where 32 % of fish had melanin on at least one of their fillets; Kettunen et al., 2020) and h2 = 0.01– 0.02 in a study where ca 15% of fish had melanin spots ( Mørkøre et al., 2015). We hypothesise that these low heritability estimates were a result of the poor data quality, in particular the low overall frequency and unbalanced data with regard to family origin. In this task we set up a system where we sampled production fish at a processing site of MOWI and used this fish for a genetic study, once we had obtained enough data. These fish are related to the nucleus fish and each batch has a limited number of families. The analysis included estimation of heritability, and a genome-wide association study. We compared the parameters obtained from a quantitative analysis assuming a normal distribution of the trait to a probit analysis, to take account of the binary nature of the trait.
Materials and Methods
Data were collected at the processing site of Mowi at Eggesbønes, Norway. At the slaughter line, filleted Atlantic salmon were sorted into Superior (no melanin spots) or Melanin (presence of melanin spot in the front part of the fillet under the spine). Four random batches of fish were sampled from cages with known geographical origin, but with unknown sires and dams . There were 1643 fish in total with 45% Superior and 55% Melanin fish. All fish were genotyped with the in-house custom made SNP-chip of MOWI. 63,826 SNPs passed filters and quality control and were used to set up a genomic relationship matrix using the GCTA software of Yang et al. (2011). G enomic estimates of variance components were estimated from a univariate threshold (with, !BIN and !PROBIT functions) and linear models implemented in ASREML 4.2 (Gilmour et al., 2015), using sex (2 levels) and batch (4 levels) as fixed effects . A genome wide association analysis was performed using the following linear mixed animal model implemented in GCTA program with the “–mlma-loco” function (Yang et al. 2011) usi ng sex, batch and five principal components showing stratification in the genetic structures.
Results and Discussion
The use of data from non-family nucleus fish from a processing site proved successful for variance component estimation of Melanin spots in Atlantic salmon fillets. The genetic variation for melanin accumulation in fillet obtained using genomic information were low to moderate with estimates of 0.202±0.040 and 0.326±0.035 with linear and threshold models, respectively. It should be noted that heritability of case/control traits often are overestimated.
In the genome-wide association study , chromosome ssa03 had 7 SNPs located in the region between bp 88.3Mbp - 92.4Mbp that surpassed the chromosome-wide significance threshold. On ssa07 , 9 SNPs located in the region between 37.1Mbp - 45.7Mbp surpassed the chromosome-wide significance threshold. This is the same genomic region where QTL linked to resistance to Pancreas Disease (PD) have also been detected (Hillestad et al., 2020; Aslam et al., 2020). A random sample of 10 fish were analyzed for the detection of PD virus which showed that 7 out of 10 had PD virus. These results indicate a possible genetic link between melanin spots and susceptibility to PD. They also agree with previous suppositions that the prevalence of melanin spots is linked to inflammatory diseases such as PD (Mørkøre et al, 2015).
Yang Am J Hum Genet. 88: 76-82.
Acknowledgements . AquaIMPACT is funded by the H2020 program Sustainable European aquaculture 4.0: nutrition and breeding (GA nr. 818367).