AquaGen was the first fish breeding company to implement marker-assisted selection directly in salmon egg production (2009), enabling rapid transfer of improved genetic material to salmon producers. The company is still a leader in the implementation of genomic information. In 2014 the reference sequence of the Atlantic salmon genome will be finalized, opening up for further progress in salmon breeding and genomics. This article reviews accomplishments up until this important milestone, and provides a brief look at the future prospect of salmon breeding.
It all began at the Agricultural University of Norway (the current Norwegian University of Life Sciences), where Professor Harald Skjervold, an expert in selective breeding of livestock, started using rainbow trout as a model organism for genetic research in the early 1960s. Contrary to conventional wisdom at that time, he and his colleagues were convinced that selective breeding had a great potential also in aquaculture. . Incidentally, this coincided with the infancy of the aquaculture industry in Norway. The first smolts of another salmonid species, the Atlantic salmon, were transferred to net sea cages in 1970. In 1971, the Norwegian aquaculture production totaled only 100 metric tons of Atlantic salmon and 540 tons of rainbow trout. In 1970, professor Skjervold asked his research fellow, Trygve Gjedrem, to take responsibility for research in aquaculture and the building of necessary facilities (Gjedrem, 2012). During the years 1971–74 Gjedrem and his staff at the Institute of Aquaculture Research (AKVAFORSK) sampled eggs and milt from brood fish in 41 rivers. The breeding nucleus was created during the first four years (with four parallel year classes). In the initial generations, selection was for high growth rate and later sexual maturation. From the fifth generation in the early 1990s other selection traits were also included; meat quality (fillet fat and fillet colour) and resistance to specific infectious diseases. Concurrently, AKVAFORSK also established a similar breeding program for rainbow trout by collecting existing genetic material from farmed fish in Norway and Sweden.
AKVAFORSK wanted the emerging industry to take responsibility for the genetic improvement of farmed salmon, and AquaGen was therefore established as a fish farmers’ cooperative in 1985. Minimizing risk of disease and accidents was also important, and parallel populations of the breeding nuclei of both Atlantic salmon and rainbow trout were handled over to AquaGen. During the first years, both breeding programs were run in parallel at AKVAFORSK and AquaGen, but in 1992 AquaGen became a limited company, and was from then on responsible for the commercial continuation of the Atlantic salmon and rainbow trout breeding programs.
From the beginning, the breeding program has been based on combined family (sib) and within family selection for improvement of growth rate and reduced early maturation, and later family selection for improvement of disease resistance and quality traits. The achievements obtained in improving the performance of Atlantic salmon through classical breeding have been considerable. The time to produce market-sized 4 kg fish has been more than halved and feed conversion ratio has been reduced from 3 in the 1970s to 1.2 today.
AquaGen was early aware that molecular genetic methods could provide new opportunities as a supplement, and perhaps also a substitute, to traditional quantitative genetics in fish breeding. From 1997, AquaGen participated in an EU-funded project on marker-assisted selection for resistance to infectious diseases. This was followed up with two user-driven research projects, the last one focusing on marker-assisted selection for resistance to the viral disease IPN (infectious pancreatic necrosis), one of the major diseases in Atlantic salmon farming. In 2007 a major gene, a so-called QTL, was identified through this project, explaining as much as 80% of the genetic variation in IPN resistance. This was immediately implemented into selective breeding, directly on the egg-producing broodstock in a manner that has contributed to the biggest commercial success in the history of AquaGen. In parallel with the implementation, the effect of the QTL-based selection has been documented both experimentally and by field-based data. The strength of the IPN QTL is so far unparalleled in animal breeding and is representing a paradigm shift in salmon breeding.
Genomics is, like so many other branches of science, reliant on the availability of different kinds of tools and resources. Researchers applying genomics in animal breeding use DNA-markers, which are sites or regions within the genome displaying variation between animals. The genotypes of individual animals can be assayed using such markers. In order to form and test biological hypotheses on the basis of DNA marker data, a researcher will also need information about the location of markers, genes and functional elements in the genome. The ultimate resource for genomics research is the entire DNA sequence of the genome of subject species. The genome makes it possible to uncover variable sites within the genome, to determine the position of genes, and to draw from the vast body of biological literature generated across species. Since the publication of the first human genome draft in 2001, an increasing number of species have been sequenced, starting with the major model species (mouse, puffer fish etc.) and production species (chicken, cattle, etc.).
The Atlantic salmon is not the typical "model species" for basic research, nor can it be compared to the major meat-producing species (cattle, chicken, pigs) in terms of funding available for genomic research. However, dedicated contributions from scientists and funding agencies throughout the past two decades have decreased the gap between the Atlantic salmon and the major model/production species as far as resources for genetic research is concerned. At around the turn of the millennium, the Genome Canada-funded GRASP project and the EU-funded SALMAP project provided large numbers of microsatellite markers, Expression Sequence Tag (EST) sequences, and sequences of genomic fragments. A little later, the Centre of Integrative Genetics (CIGENE) in Norway started focusing on the detection and utilization of single nucleotide polymorphisms (SNPs).
The sequencing of the Atlantic salmon genome was initiated in 2009, funded by Norwegian, Chilean and Canadian funding agencies as well as by AquaGen, Marine Harvest, Cermaq and Salmobreed. The Atlantic salmon turned out to be a difficult species to sequence, due to a high content of repeated sequences and by the very high prevalence of loci that are duplicated across chromosomes due to a recent 'whole-genome duplication event' in the previous history of the salmonids. In spite of these difficulties the project is expected to release a genome assembly of high quality in 2014. The availability of a draft genome sequence, and the opportunity to re-sequence the entire genomes of individual animals, has facilitated the detection of large number of SNPs in the Atlantic salmon. AquaGen and CIGENE recently identified millions of putative SNPs on the basis of Illumina-sequencing of 30 individual AquaGen animals, and made a chip with 930 000 (930k) SNPs in collaboration with Affymetrix in 2013. This SNP-chip and a more compact 220k version of the chip have been put to use for QTL detection for disease resistance and quality traits. These SNP-chips are very powerful tools for finding markers that can be used in the selection for traits of high value for the aquaculture industry. Altogether, the availability of a high-quality, annotated, genome sequence, high-density SNP-chips etc. has now brought salmon genomics into another 'league', and opened up entirely new possibilities with regard to that genomics can contribute to salmon breeding.
Infectious pancreatic necrosis (IPN) is a viral disease which is endemic in most Atlantic salmon-producing countries, causing frequent outbreaks, often with large mortalities. AquaGen have selected broodstock for increased IPN-resistance using classical selection based on family data from challenge trials since the turn of the millennium.
In 2007, a breakthrough in the fight against IPN came when two research groups, one in Norway and one in Scotland, independently identified a major quantitative trait locus (QTL) for resistance to IPN (Houston et al., 2008; Moen et. al., 2009). A QTL is a region of the genome harboring a mutation that causes differences between animals with regard to a phenotypic trait such as IPN resistance. The identity and exact position of the causative mutation itself is usually unknown, but other DNA markers in the vicinity of the QTL can be used as proxies in the selection of good or bad animals. The QTL for IPN resistance was exceptional as it was found to be responsible for almost all genetic variation of the trait. The Norwegian researchers from AquaGen, CIGENE, and Nofima, discovered that three microsatellite markers within the QTL region could be used jointly in a test to predict if a random fish originating from the AquaGen breeding program was IPN resistant or not. This test would prove to be a formidable tool in the fight against IPN.
AquaGen quickly put the DNA-test to work in selection for increased IPN-resistance. Initially, the test was used for improvement of IPN resistance in the breeding nucleus, the "core" breeding population and normally the primary target for genetic improvement. However, the DNA markers opened up for the possibility of selecting IPN-resistant parents directly in egg production, bringing the genetic improvements to the industry without delay. Male QQ salmon, i.e., male salmon harboring two copies of the high-resistance allele at the QTL, were mated to random females in order to create eggs each having at least one copy of the high-resistance allele (Q). These salmon eggs were offered to costumers with at a premium price, and the product was given a trademark name - QTL-innOva IPN.
The product was well received by the industry from the very start; one quarter of the costumers chose this product in the release season (2009). After that, the interest took off, and 3 years later, practically all costumers chose a product incorporating the DNA-test for IPN resistance. It turned out that the QTL-innOva IPN fish were far more resistant to IPN than randomly selected fish, and the disease vanished from hatcheries that used to have high mortality rates due to IPN in the past. Reduced IPN problems also lead to reduced reduction of losses during the seawater phase of the production. A number of experimental challenges have further demonstrated the strength of the IPN QTL, but the power of the marker assisted selection is best illustrated by the development in the fish disease statistics in Norway. In 2009, when the IPN QTL was introduced in egg production, a total of 223 IPN outbreaks were registered in Norway. From 2009 to 2013 the total production of Atlantic salmon increased by more than 20%, but despite this, the number of IPN outbreaks in Norway was reduced to 50 (Figure 1).
As noted above, a QTL represents an area or a region of the genome associated with the trait, and points towards markers that can be useful in selection. The QTL region also harbors a mutation or polymorphism responsible for creating the phenotypic differences between animals. The identification of this hidden causative mutation is, generally speaking, very challenging, but of high scientific value, as it can explain the functional mechanism behind the favorable trait. When searching for the causative mutation one is also likely to discover new DNA markers that are more specific and better diagnostic tools than the DNA markers already known. AquaGen and collaborators (CIGENE, Nofima, Simon Fraser University (Burnaby, Canada)) set out to identify the causative mutation behind the IPN QTL in 2007. The pivotal strategy of this project was to sequence the entire genomes of animals known to have the QQ and qq genotypes, and to search for variable sites displaying a large difference in allele frequency between these two groups of animals. Without a salmon reference genome sequence, this involved a lot of basic genomic work, as the research group had to make its own genome reference sequences with gene annotations. This work led to an improvement in the diagnostic test for IPN resistance implemented in 2011. However, no DNA variations were found to perfectly match the genotype pattern of the QTL or have a function associated with IPN resistance until early 2013. A breakthrough came when the group discovered an amino-acid shifting SNP in a gene that could potentially be involved in virus-host interaction. Functional studies have indicated strongly that the protein product of the gene in question is indeed part of the machinery used for cell infection by the virus. Thus, the seven-year search for the mutations causing IPN-resistance in Atlantic salmon now seems to be reaching its conclusion.
In parallel with the implementation of the IPN QTL in egg production and the development of SNP panels for high density, high throughput genotyping, AquaGen has invested in the collection of phenotype data from a number of important traits in Atlantic salmon and rainbow trout. The aim is to make a thorough mapping of traits that could be targeted by marker-assisted selection. So far, genome wide association studies have been performed for example with successful detection of DNA markers for two more viral diseases, pancreas disease (PD) and cardiomyopathy syndrome (CMS) in salmon. Selection for PD resistance was implemented in 2011, and we believe that this has been a contributing factor to the positive development in the western part of Norway, where the number of PD outbreaks is in decline, from 93 in 2012 to 48 in 2013. CMS is a disease that occurs late in the seawater phase of the production, and producers can experience large losses from this disease on their way to the slaughter house. CMS-selected eggs will be available for the salmon farmers in 2014.
AquaGen and CIGENE have also investigated if meat quality parameters could be improved by marker assisted selection. A firm textured and well colored fillet is important for the quality perception of the consumer and pale, soft textured fillets is a common cause for complaints from the value adding industry. Several QTLs for filet color and fillet texture have been detected by QTL-mapping in three consecutive generations of fish from the breeding nucleus. These DNA markers are used for general improvement of the AquaGen breeding populations, but are also offered as a QTL product directly to costumers. Improvement in fillet quality is of particular importance for salmon farmers selling premium products to differentiated markets. One example is Chilean producers selling to Japan, a market where the redness of the fillet is highly appreciated.
The Blue Genomics initiative, lead by the EW Group companies AquaGen (Chile and Norway) and Vaxxinova, is targeting some of the unique challenges of the Chilean aquaculture industry. In Chile, the most important cause of losses is the Salmon Rickettsial Syndrome (SRS), caused by the intracellular bacteria Piscirickettsia salmonis. SRS is a major concern as it leads to extensive use of antibiotics in order to prevent losses. A QTL search for SRS resistance has given encouraging results and we hope that marker assisted selection for improved SRS resistance can be implemented in Chilean salmon egg production very soon. Blue Genomics aspires to find markers for SRS resistance in rainbow trout and Coho salmon as well. Caligus resistance is another target, and challenge studies performed in Atlantic salmon demonstrate promising heritability for the trait.
Marker assisted selection for sea lice resistance would be a very important contribution to the profitability and future sustainability of the salmon industry. AquaGen have performed two major sea lice challenges of fish that were subsequently genotyped on the 220k SNP chip. The data entered a genome-wide association study, revealing a few markers that may be useful, but also demonstrating that a real increase in salmon sea lice resistance will have to rely on traditional selection combined with even more DNA marker information through the implementation of genomic selection.
The IPN QTL in Atlantic salmon has become a famous examples of successful implementation of QTL selection across species. Successful QTL selection is, however, based on several requirements:
1) One or more major QTL must be present
2) The QTL must be identified
3) The favorable QTL allele should not be close to fixation (i.e., there is room for considerable improvement).
Many of the important traits in animal breeding are typically polygenic, which imply that they are controlled by numerous QTLs of small individual effects. Examples of such traits are growth-related traits and milk production in dairy cattle. Even though the individual QTL effects may be small, genetic variation and heritability can be high, giving a substantial potential for selective breeding. A well-studied example of a polygenic trait is height in humans, which is highly heritable (~80%), and where 180 identified loci have been found to have an effect on the trait. Still, the combined effect of all these loci explains only ~10% of the variance in height (Lango Allen et al. 2010). For highly polygenic traits, QTL selection is therefore expected to be unsuccessful, due to detection problems combined with complicated and inefficient joint selection for numerous QTL of small effect.
Still, even for highly polygenic traits, information on dense genome-wide SNP-markers can be valuable for selective breeding through so-called genomic selection (GS). In GS, the focus is not to estimate the effect of one or more specific QTL, but to use the combined effect of thousands of genome-wide SNP markers to estimate breeding values of the animal (Meuwissen et al. 2001), i.e., the summed effect over all QTL genotypes in the genome of an animal. One of the main advantages of the GS methodology is that it provides a method for calculation of individual breeding values, even for fish that has not been tested for the trait. The breeding values are based on associations between markers and the trait among close fish that are both phenotyped and genotyped. Compared with classical selection, GS methodology in aquaculture has its highest potential for traits that are typically measured in sibs of the selection candidate (Ødegård et al. 2009). Traits recorded on sibs involve traits that are difficult or impossible to record on live animals (e.g., meat quality or disease resistance).
There are several methods for implementation of GS; The simplest assume that all SNPs across the genome explain equally small proportions of the genetic variation (GBLUP), while more advanced methods assume that some SNP loci have a relatively higher importance (Bayesian methods) (Meuwissen et al. 2001). Alternatively, the marker may be used to trace inheritance of genome segments within the known pedigree using linkage analysis (Villanueva et al. 2005; Luan et al. 2012; Ødegård and Meuwissen 2014).
AquaGen has recently tested GS methodologies on Atlantic salmon for the traits sea lice resistance and fillet color. A total of 157 full-sib families were included, and 1444 and 1869 fish were phenotyped and genotyped (220,000 SNPs) for the two traits, respectively. The traits had rather different heritabilities; 0.14 for lice resistance and 0.43 for fillet color. Different GS models were evaluated using cross validation. Compared with classical pedigree-based models, GS increased reliability of breeding values for genotyped animals without own phenotypes by 52% and 22% for lice resistance and fillet color, demonstrating the improved power of selection by GS methods.
Developments in gene technology have resulted in improved tools at an increasing pace, and it has become challenging for the breeding companies to keep up, and choose how and when to implement these tools in selection. AquaGen is leading on in aquaculture breeding; being the first company to implement marker-assisted and genomic selection in egg production and breeding. This work will be continued and refined in the years to come, improving selection accuracy and resulting in better performing fish in the net cages.
Genomics is a useful tool for enhancing performance, but can also be applied in the support of a responsible development of the aquaculture industry. A great concern both in Norway, Scotland and Canada is the possible negative effect of genetic interaction between wild salmon and escaped farmed salmon. DNA can be used as a tool to trace escaped fish back to the farm of origin and enabling identification of the responsible company. It can also be used to monitor of the genetic integrity of wild salmon populations.
The release of the Atlantic salmon reference genome sequence in June 2014 will bring salmon aquaculture into the post-genomic era. Lack of genomic information will no longer be the bottle-neck in the investigation of complex traits and functional pathways. Sequence information can be obtained at a lower cost and analyzed with greater ease, and each breeding candidate will enter the final rounds of selection with ever increasing amount of genomic information complementing the traditional breeding values. Eventually, we expect to make use of the full genome information sequence and epigenetic profile of our breeding candidates in selection, but this goal is still some distance ahead of us. For the moment, we work with the research partners CIGENE, CEES and Nofima towards a better understanding of the full genetic variation in farmed and wild salmon, in a project aiming at sequencing 1000 individual Atlantic salmon (The Aquagenome project, funded by the Norwegian Research Council).
So far, salmon aquaculture have actively avoided the use of GMO-technology. A GMO salmon made by first generation GMO technology has been available for more than a decade, but have moved closer to the market during the last couple of years (www.aquabounty.com). In the meantime, the growth rate of non-GMO salmon has improved through traditional breeding and is closing up on the genetically engineered salmon. We believe that the production time in salmon aquaculture will be further reduced by improved growth from traditional breeding also in the future, reducing the comparative advantage of the GMO salmon. There may however, be applications for gene editing technology in situations if it can provide solutions to important sustainability challenges in the industry. This can only come into play once we have investigated all possible applications of classical selection supported by genomic information, without the use of GMO. Recent progress have demonstrated the vast opportunities and potential in salmon genomics, giving exciting prospects for the aquaculture industry.
Gjedrem, T (2012) Genetic improvement for the development of efficient global aquaculture: a personal opinion review. Aquaculture 344–349:12–22.
Houston, Ross D., C.S. Haley, A. Hamilton, D. R. Guy, A.E. Tinch, J.B. Taggart and S. C. Bishop (2008) Major quantitative trait loci affect resistance to infectious pancreatic necrosis in Atlantic salmon (Salmo salar). Genetics, 178(2): 1109–1115.
Lango Allen, H., K. Estrada, G. Lettre, S. I. Berndt, M. N. Weedon et al.( 2010) Hundreds of variants clustered in genomic loci and biological pathways affect human height. Nature 467: 832-838.
Luan, T., J. A. Wooliams, J. Ødegård, M. Dolezal, S. I. Roman-Ponze et al. (2012) The importance of identity-by-state information for the accuracy of genomic selection. Gen. Sel Evol. 44: 28.
Meuwissen, T.H.E., B.J. Hayes and M.E. Goddard (2001) Prediction of total genetic value using genome-wide dense marker maps. Genetics 157: 1819-1829.
Moen, T., M.Baranski, A.K. Sonesson and S. Kjøglum (2009) Confirmation and fine-mapping of a major QTL for resistance to infectious pancreatic necrosis in Atlantic salmon (Salmo salar): population-level associations between markers and trait. BMC genomics, 10: 368-.
Villanueva, B., R. Pong-Wong, J. Fernández and M. A. Toro (2005) Benefits from marker-assisted selection under an additive polygenic genetic model. Journal of Animal Science 83: 1747-1752.
Ødegård, J. and T. Meuwissen( 2014) Identity-by-descent genomic selection using selective and sparse genotyping. Gen. Sel. Evol. 46: 3.
Ødegård, J., M. H. Yazdi, A. K. Sonesson and T. H. E. Meuwissen (2009) Incorporating Desirable Genetic Characteristics From an Inferior Into a Superior Population Using Genomic Selection. Genetics 181: 737-745.
Figure 1 text:
Figure 1: Number of IPN diagnoses for Atlantic salmon in fresh- and sea water in Norway from 2009 to 2013 compared to the number of QTL-eggs delivered by AquaGen (QTL-innOva) and the total number of eggs sold in the same period. Source: Norwegian Veterinary Institute and Norwegian Seafood Federation.
LOHMANN TIERZUCHT GmbH
Am Seedeich 9–11
BREEDING FOR SUCCESS ... TOGETHER