SEARCH

SEARCH BY CITATION

Contents

  1. Top of page
  2. Contents
  3. Introduction
  4. Development and General Aspects of Genomic Selection
  5. Applications of Genomic Selection and Specific Challenges for Breeding
  6. Conclusions
  7. Conflict of interest
  8. Author contributions
  9. References

Technical advances and development in the market for genomic tools have facilitated access to whole-genome data across species. Building-up on the acquired knowledge of the genome sequences, large-scale genotyping has been optimized for broad use, so genotype information can be routinely used to predict genetic merit. Genomic selection (GS) refers to the use of aggregates of estimated marker effects as predictors which allow improved individual differentiation at young age. Realizable benefits of GS are influenced by several factors and vary in quantity and quality between species. General characteristics and challenges of GS in implementation and routine application are described, followed by an overview over the current status of its use, prospects and challenges in important animal species. Genetic gain for a particular trait can be enhanced by shortening of the generation interval, increased selection accuracy and increased selection intensity, with species- and breed-specific relevance of the determinants. Reliable predictions based on genetic marker effects require assembly of a reference for linking of phenotype and genotype data to allow estimation and regular re-estimation. Experiences from dairy breeding have shown that international collaboration can set the course for fast and successful implementation of innovative selection tools, so genomics may significantly impact the structures of future breeding and breeding programmes. Traits of great and increasing importance, which were difficult to improve in the conventional systems, could be emphasized, if continuous availability of high-quality phenotype data can be assured. Equally elaborate strategies for genotyping and phenotyping will allow tailored approaches to balance efficient animal production, sustainability, animal health and welfare in future.


Introduction

  1. Top of page
  2. Contents
  3. Introduction
  4. Development and General Aspects of Genomic Selection
  5. Applications of Genomic Selection and Specific Challenges for Breeding
  6. Conclusions
  7. Conflict of interest
  8. Author contributions
  9. References

In the last decade, parallel advances in molecular genetic technologies and bioinformatics have made it possible to establish genomic selection as a new tool to increase the genetic gain in animal breeding. Starting from research focused on improved understanding of structure and function of the human genome, revolutionary developments in animal breeding have been initiated. Whole-genome sequencing, which had long been extremely time-consuming and expensive, is today part of standard laboratory service, and knowledge of the DNA sequences of multiple species provided the basis for accelerated research progress and extension of the use of genome-wide information.

Using genomic data for improvement of selection in animal breeding was already suggested in the late 1960s (Smith 1967) and became possible with identification of major genes or genetic markers linked to quantitative trait loci (QTL). Particularly for traits, which were difficult to improve in the conventional breeding programmes because of low heritabilities or onerous collection of phenotype data, marker-assisted selection (MAS) promised enhancement of selection response. However, the cost benefit ratio hardly ever justified routine work with MAS: considerable research investments were needed for the QTL studies, QTL genotyping of selection candidates was expensive, and the benefits of MAS in commercial breeding programmes were clearly less than expected (Dekkers 2004).

The success of genomics in animal breeding set in with departure from the concept of using information on some genetic loci with large effect on the traits of interest to using whole-genome information (Meuwissen et al. 2001). Since then, genomic breeding values and their use for selection, genomic selection (GS), have been extensively studied worldwide. The aim of this work was to review some general aspects of GS and provide an overview over its current applications in animal breeding, with particular focus on the challenges of implementation and long-term use in different species, breeds and populations.

Development and General Aspects of Genomic Selection

  1. Top of page
  2. Contents
  3. Introduction
  4. Development and General Aspects of Genomic Selection
  5. Applications of Genomic Selection and Specific Challenges for Breeding
  6. Conclusions
  7. Conflict of interest
  8. Author contributions
  9. References

Accessibility of whole-genome data

Developments towards better exploitation of the potential of genomics for breeding purposes were triggered by the technical advances in the field of molecular genetics and the related drop of costs for genomic studies (Eggen 2012). Opportunities to work with whole-genome information have first stimulated research and later allowed developments towards routine use for selection. Building-up on the experiences from the human genome project, whole-genome sequencing and related use of genomic tools continued to become easier, faster and cheaper, so genome sequences of diverse domestic animal species can be assessed today (Fan et al. 2010). High-performance infrastructure and sophisticated bioinformatics tools have been developed around freely assessable data bases, strengthening the scientific community and facilitating further progress of genomic research. A summary of whole-genome sequence information provided by the National Center for Biotechnology Information (NCBI) and the Ensembl project is given in Table 1 for some important animal species.

Table 1. Summary of whole-genome sequence information on important animal species according to the public databases of the National Center for Biotechnology Information (NCBI) and the Ensembl project, with genome size (total sequence length) and number of single nucleotide polymorphisms (NSNP_NCBI, NSNP_Ensembl)
Species, year of full genome sequence completionGenome sequence information (most recent release with NCBI GenBank assembly ID and submission information)aGenome sizea, Total no. SNPsb,c,d
  1. Sources (assessed on 8 February 2013): ahttp://www.ncbi.nlm.nih.gov/genome; bhttp://www.ncbi.nlm.nih.gov/snp; chttp://www.ensembl.org; dFor Ensembl: total number of shorts variants, that is, SNPs, indels, somatic mutations; eAssembly level not chromosome yet, but contig (Salmo salar) or scaffold (Gadus morhua).

Cattle (Bos taurus), 2009

Bos_taurus_UMD_3.1 (GCA_000003055.3),

Center for Bioinformatics and Computational Biology (CBCB) at University of Maryland, Nov 2009

2 670 422 299;

NSNP_NCBI 13 146 622,

NSNP_Ensembl 13 393 280

Pig (Sus scrofa), 2009

Sscrofa10.2 (GCA_000003025.4),

The Swine Genome Sequencing Consortium, Sep 2011

2 808 525 991;

NSNP_NCBI 566 003,

NSNP_Ensembl 482 203

Sheep (Ovis aries), 2008

Oar_v3.1 (GCA_000298735.1),

International Sheep Genome Consortium, Sep 2012

2 619 054 388;

NSNP_NCBI 2 914 764

Goat (Capra hircus), 2012

CHIR_1.0 (GCA_000317765.1),

International Goat Genome Consortium, Jan 2013

2 635 832 257;

NSNP_NCBI 60 094

Horse (Equus caballus), 2007

EquCab2.0 (GCA_000002305.1),

The Genome Assembly Team, Oct 2007

2 474 929 062;

NSNP_NCBI 1 163 580,

NSNP_Ensembl 1 154 177

Rabbit (Oryctolagus cuniculus), 2009

OryCun2.0 (GCA_000003625.1),

The Genome Sequencing Platform, The Assembly Computation and Development Core Team at Broad Institute, Oct 2009

2 737 445 565
Dog (Canis lupus familiaris), 2003

CanFam3.1 (GCA_000002285.2),

Dog Genome Sequencing Consortium, Nov 2011

2 410 976 875;

NSNP_NCBI 3 328 578,

NSNP_Ensembl 3 225 735

Chicken (Gallus gallus), 2004

Gallus_gallus-4.0 (GCA_000002315.2),

International Chicken Genome Consortium, Nov 2011

1 046 932 099;

NSNP_NCBI 3 295 452,

NSNP_Ensembl 3 526 256

Turkey (Meleagris gallopavo), 2009

Turkey_2.01 (GCA_000146605.2),

Turkey Genome Consortium, Feb 2011

1 061 817 103;

NSNP_NCBI 9 282

Spotted gar (Lepisosteus oculatus), 2011

LepOcu1 (GCA_000242695.1),

Broad Institute, Jan 2012

945 861 706
Atlantic salmon (Salmo salar), 2011

ASM23337v1 (GCA_000233375.1),

International Cooperation to Sequence the Atlantic Salmon Genome, Oct 2011

2 435 040 521e;

NSNP_NCBI 8 635

Atlantic cod (Gadus morhua), 2010

GadMor_May2010 (GCA_000231765.1),

Genofisk, Aug 2011

824 311 139e;

NSNP_NCBI 2 141

Based on high-resolution sequence data and improved knowledge about the genome structure, dense marker maps could be compiled, providing the basis of whole-genome screening for various applications. Across species, single nucleotide polymorphisms (SNPs) are the most abundant type of variation in the DNA sequence, and their mostly biallelic nature makes them easy to assay and interpret (Brookes 1999). SNP libraries could therefore be used to develop genotyping arrays facilitating large-scale genotyping as needed in a breeding context (Kwok and Gu 1999). With good coverage of the genome and high density of the markers, the SNP genotypes capture the information on linkage disequilibrium (LD) and allow conclusions on adjacent genes or quantitative trait loci (QTL). With the intensification of molecular genetic research in the early 21st century, the foundation of international consortia had paved the way for whole-genome sequencing, compilation of SNP libraries and development of informative and affordable SNP panels for multiple animal species. Enrolment of partners from science and breeding industries allowed ever increasing speed of progress in the genomics sector, providing broadly applicable genomic tools soon after collaborations have been set up (e.g. International Chicken Genome Sequencing Consortium; Eggen 2012). High-throughput technologies as realized with the SNP chips and genotyping platforms made it possible to obtain SNP genotypes from large numbers of samples reliably, rapidly and cheaply. Although SNP chips are not yet commercially available for all livestock species (Table 2), opportunities of routine access to enormous amounts of genomic data now or in the near future exist across species. Technical flexibility with high multiplexing and various multisample array formats promise custom SNP chips for any species and study context (Illumina Inc., San Diego, CA, USA; http://www.illumina.com). Across species, available SNP chips are today optimized for use in multiple breeds and populations, ensuring strong performance in diverse applications (genome-wide association studies, LD studies, biodiversity research, routine genetic evaluations). Given the broadest use of whole-genome information in cattle, different formats of SNP chips with a range from approximately 3000 to almost 778 000 SNPs are marketed, covering the full spectrum of applications from screening over routine genomic evaluation to refined molecular genetic research in the bovine.

Table 2. Available whole-genome single nucleotide polymorphism (SNP) chips for important animal species, modified from Eggen (2012) according to provider information (assessed on 8 February 2013: http://www.illumina.com/, http://www.affymetrix.com/)
SpeciesClassification and consortium (for private chips)Identification (provider) and number of SNPs
Cattle (Bos taurus)Commercial

BovineHD (Illumina),

NSNP 777 962

Commercial

BOS 1 (Affymetrix),

NSNP 648 855

Commercial

BovineSNP50v2 (Illumina),

NSNP 54 609

Commercial

BovineLD (Illumina),

NSNP 6 909

Commercial

Bovine3k (Illumina),

NSNP 2 900

Pig (Sus scrofa)Commercial

PorcineSNP60v2 (Illumina),

NSNP 64 232

Sheep (Ovis aries)Commercial

OvineSNP50 (Illumina),

NSNP 54 241

Private (public sale),

AgResearch

Ovine (Illumina),

NSNP 5 409

Goat (Capra hircus)Private (consortium chip)

Goat (Illumina),

NSNP 53 347

Horse (Equus caballus)

Private (public sale),

Neogen (GeneSeek)

Equine (Illumina)

NSNP 65 157

Dog (Canis lupus familiaris)Commercial

CanineHD (Illumina)

NSNP 173 662

Private (public sale for research use only), Broad Institute

DogSNPs520431 (Affymetrix),

NSNP 127 132

Chicken (Gallus gallus)

Private (public sale),

Cobb Vantress-Hendrix-USDA

Chicken (Illumina)

NSNP 57 636

Commercial

Chicken (Affymetrix)

NSNP 580 961

Routine use of whole-genome data

Feasibility of large-scale genotyping for 1000s of SNPs in large numbers of animals allowed reconsidering the concept of breeding values and whole-genome information. The rationale of genomic selection is to use dense marker data to image the genome, link genotype and phenotype information for estimating marker effects, and finally use genomic breeding values (GEBV) as aggregates of all marker effects for selection (Meuwissen et al. 2001). Bioinformatics had to keep up with the improvement of the laboratory capacities to transfer the investments into increased genetic gain. To derive GEBV, effects of SNP genotypes need to be estimated, implying availability of reliable estimation procedures and thorough definition of the estimation basis. For routine applications of GS, effect estimation is usually based on large training sets or reference populations of genotyped animals with (highly) reliable information on trait phenotypes. Once prediction equations have been set up, the genetic merit of any individual can be predicted from its SNP genotypes alone, that is, as soon as DNA material becomes available, regardless of sex, age and availability of phenotype information. For reliable predictions, the reference animals need to be closely related to the selection candidates.

Particularly for smaller breeds and traits with low heritabilities where larger amounts of phenotype data are needed, assembly of a reference population of sufficient size and appropriate structure is challenging. Furthermore, marker effects decay under selection, so regular re-estimation of SNP effects must ensure persistence of prediction accuracy of GEBV. Accordingly, the need of a reference of animals with genotype and phenotype information remains in routine GS (Schefers and Weigel 2012). For the breeding companies and organizations, the long-term success of GS depends on finding efficient strategies which optimize investments for genotyping and continued collection of high-quality phenotype data, as the basis for reliable GEBV prediction, and balance the determinants of the genetic gain, that is, selection intensity, selection accuracy and age of selection and breeding use (Dürr and Philipsson 2012; Pryce and Daetwyler 2012).

Implementation of GS further stimulated international collaborations because of the need to assemble large reference populations for the genomic predictions models. Coincidence of genotype and accurate phenotype records and close relations to the selection candidates were the preconditions for the reference animals, the required number of which was shown to be proportional to the effective population size and inversely proportional to the heritability of the trait of interest (Goddard 2009). Links between populations of the same breed imply opportunities for synergistic initiatives to establish GS, because sharing of genotypes and comparable phenotypes lowers the logistic and financial burden for each of the partners (Dürr and Philipsson 2012; Pryce et al. 2011). Prospects of collaborations with mutual benefits differ considerably between species and breeds, being largest in systems with already intense exchange of genetic merit information and genetic material as in the dairy industry.

Effects of genomic selection on breeding programmes

The more difficult and expensive the collection of phenotype data and longer it takes for a breed until enough data is available for predicting breeding values reliably in the conventional systems, the larger is the potential of increasing the genetic gain by GS. In a population undergoing selection, genetic change (ΔG) is determined by the selection intensity (i), the accuracy of selection (r), the genetic variability of the trait (σa) and the generation interval (L) as ΔG = (× × σa)/L (Falconer 1989). Accordingly, enhancement of the genetic gain for a given trait may be achieved by an increase in selection intensity, more accurate prediction of the genetic merit of breeding animals and reduction in the generation interval. Substantial increases in selection intensity involve the risk of negative effects due to loss in genetic diversity at least in the long-term, so it may become necessary to use whole-genome data for diversity and inbreeding management (Cervantes and Meuwissen 2011; De Cara et al. 2011). Because of the long time needed to get enough information for accurate selection, the generation interval is far beyond the biological limits in several livestock species, implying considerable potential for improvements by earlier availability of predictions of equal or even higher accuracy via GEBV than in the conventional systems (e.g. dairy cattle). For breeds with already short general interval, enhancement of genetic gain by an increase in prediction accuracy is the major argument for using GEBV (e.g. pig).

Across species, breeding goals often include traits that have low heritabilities and are difficult or not at all measurable directly in the breeding animals, because they are sex limited (e.g. milk traits in dairy, egg traits in layer chickens, female fertility) or accessible only under production conditions (e.g. disease resistance), very late in life (e.g. longevity) or even after death (e.g. meat or slaughter traits, processing ability traits). Many of the traits which have become increasingly important over the years, because they relevantly impact the production conditions in the livestock sector, fall into the category of the traits where the success of the conventional selection systems is limited or requires high investments. For these functional traits, GS provides new opportunities because once reliable prediction formulae have been set up, higher weighting of functionality in the selection indices may become feasible (Muir 2007).

Population structures can be significantly impacted by GS, because of the opportunities of earlier and improved within-family selection. In the conventional systems, full-sibs have identical EBV until own or progeny phenotypes become available, so pre-selection based on pedigree indices resembles selection between families. Conversely, GEBV capture the Mendelian sampling term and allow refined differentiation between individuals early in life. GS may therefore represent a valuable tool to manage genetic diversity and inbreeding (Engelsma et al. 2011; Simianer et al. 2011). However, thorough strategies for screening of selection candidates may be needed to avoid genotyping bias with focus on high-merit families and increases in inbreeding due to intense breeding use of the animals with most promising GEBV (Dürr and Philipsson 2012). Higher accuracies of GEBV come along with more accurate estimation of the Mendelian sampling term and improved within-family selection (Pryce and Daetwyler 2012), and in this context, higher SNP densities or even replacement of SNP genotypes as markers by genome sequence data may facilitate maximizing breeding progress with constraint on coancestry (Kemper et al. 2012; Meuwissen et al. 2013). With the current methodology, genetic markers explain clearly less than 100% of the genetic variation, so denser marker data promise to capture more of the relevant loci for polygenic traits, and causative mutations could be directly considered via sequence data (De Roos et al. 2011; Pryce and Daetwyler 2012; Meuwissen et al. 2013). Methods have been developed to not only consider additive effects, but also account for non-additive effects like dominance and epistasis in routine analyses (Boysen et al. 2013), whereas estimation of epigenetic and imprinting effects may also in the near future confine to research applications (Sellner et al. 2007).

Applications of Genomic Selection and Specific Challenges for Breeding

  1. Top of page
  2. Contents
  3. Introduction
  4. Development and General Aspects of Genomic Selection
  5. Applications of Genomic Selection and Specific Challenges for Breeding
  6. Conclusions
  7. Conflict of interest
  8. Author contributions
  9. References

Dairy cattle

With long generation and sex-specific phenotype data, dairy breeding was expected to benefit enormously from implementation of GS (Meuwissen et al. 2001; Schaeffer 2006). In the traditional system of progeny testing, breeding organizations have high investments for collecting enough phenotype data and need to keep larger numbers of bulls for prolonged time periods, until reliable breeding values become available and allow selection decisions. When accuracies of GEBV for a bull calf at birth are equal to those of conventional breeding values after progeny test, cost reductions in the order of 90% appeared realizable (Schaeffer 2006). Accordingly, the international dairy industry was and is still leading with regard to use of genomic tools and the routine implementation of GS. After the release of the first whole-genome SNP panel for cattle in 2007 (Illumina BovineSNP50 BeadChip; Illumina Inc), the first official GEBV were published for US-Holstein in 2009. Since then, routine genomic evaluations have been established worldwide for the main Holstein populations and some populations of other dairy breeds, with collaborations and partnerships as key factors for success (Liu et al. 2010; Ibañez-Escriche and Gonzalez-Recio 2011, Dürr and Philipsson 2012; De Haas et al. 2012).

Challenges regarding assembly of reference populations were successfully met by regulations for sharing of genotype data and continuation of exchange of non-genomic data (Liu 2011). GS-related changes of the structures of dairy breeding are obvious and primarily driven by the significant reduction in the generation interval due to removal or at least reduction in progeny testing (Hayes et al. 2009; Chesnais 2010). GEBV are used for the identification of elite animals at very young age, with routine genotyping of calves and tests with running embryonic DNA on the SNP arrays. Early selection allows in the most offensive GS artificial insemination breeding programmes shortening of the generation interval in the sires-of-males path from approximately 63 months with progeny testing to approximately 21 months when breeding use as sires of sons starts with sexual maturity (Schefers and Weigel 2012). Similar reductions are possible in the other selection paths, and DNA-tested young bulls without progeny data (‘genomic bulls’) have been increasingly used worldwide in the last years (Dürr and Philipsson 2012; Schefers and Weigel 2012). The market share of ‘genomic bulls’ without milking daughters ranged internationally between 25% and 50% in 2011, with further increase being driven by success of GS and trust in GEBV (Pryce and Daetwyler 2012). Genotyping investments, that previously confined to male selection candidates (bull calves), have started to include females which will allow better exploitation of the potential of genomics in future (Schaeffer 2006; Pryce and Hayes 2012; Weigel et al. 2012). In nucleus breeding schemes, the selection intensities in the female pathways can be increased, and the generation intervals can be considerably reduced using reproductive technologies like multiple ovulation and embryo transfer (MOET) and juvenile in vitro embryo transfer (JIVET), but inbreeding rates may at the same time increase to inacceptable levels if not managed appropriately (Pryce and Daetwyler 2012; Pryce et al. 2012). Technical and methodical advances have been triggering these developments, enabling cost-efficient low-density genotyping and subsequent imputation to whole-genome data to be used for genomic evaluations (Weigel et al. 2010). Overall, possible doubling of the genetic gain was claimed (Schaeffer 2006; Schefers and Weigel 2012), with options for adjustments of breeding programmes in favour of low-heritability traits with not yet satisfactory genetic progress, especially functional traits (Miglior et al. 2012). Many countries are currently facing the challenges of transition to genome-assisted breeding, and a recent review of studied strategies of GS applications in dairy breeding indicates persisting need of research into improved genomic breeding programmes (Pryce and Daetwyler 2012).

For long-term success of GS in the dairy industry concepts need to be developed to monitor and efficiently manage genetic variation and cope with the changed structure of phenotype data with regard to data flow and distribution (Pryce and Daetwyler 2012). Statistical methods need to account for the intensified pre-selection and the resulting violation of normality assumptions in the prediction models to avoid significant bias (Ducrocq and Santus 2011; Patry and Ducrocq 2011). Low and probably further decreasing genotyping costs increase the value of phenotype data, which are essential for the estimation of marker effects for the traits of interest. Effects of abandoned incentives from the progeny testing are expected to be larger for functional traits than for the production traits, so improvement of breeding programmes by higher weighting of functionality and inclusion of new traits will require intense engagement in phenotype sampling (Dürr and Philipsson 2012; Pryce and Daetwyler 2012). Recent studies have shown how GS can successfully be applied for a trait, which is complicated and expensive to measure, restricting availability of phenotype data to a subset of the population (De Haas et al. 2012). Because of its links to profitability, sustainability and public demands, dry matter intake (feed intake) in dairy cattle underpins the huge and not yet fully exploited potential of GS for animal breeding.

For the farmer, the value of genomic technologies is likely to increase with profitability of partial or whole herd low-density genotyping. Optimization of mating plans, choice of replacement animals and selection of functional traits can be performed with distinct strategies in breeding and commercial herds and according to the individual farm needs (Pryce and Hayes 2012; Weigel et al. 2012).

Beef cattle

Beef cattle industry shares the long generation interval with dairy cattle and could principally benefit from the advances of bovine genomics, particularly the availability of different formats of commercial SNP chips. Difficulties of collecting phenotype data of sufficient quality and quantity, together with often incomplete pedigree data, hamper conventional genetic evaluations, so GS has the potential to substantially increase the genetic gain by increased selection accuracy at an early age (Montaldo et al. 2012). However, the heterogeneity of breeds, less developed infrastructures and breeding programmes, predominance of natural service (with lower value of breeding animals) and the population substructures with frequent crossbreeding in the commercial herds has interfered with the adoption of GS in beef cattle. Compilation of a reference population of large enough size would require extensive collaborations across breeding organizations, but GS methods for admixed and crossbred populations need to be developed further to allow appropriate performance in the field (Toosi et al. 2010). Concerning the choice of SNPs on the commercially available bovine SNP chips, multiple breeds were considered for the optimizations (Illumina Inc, San Diego, CA, USA; Affymetrix, Santa Clara, CA, USA), so markers should be informative for the major beef or dual-purpose breeds used for beef production. However, conditions for local breeds for beef production in specific environments and GS in the presence of relevant non-additive and epistatic effects may require further investigations. In the long-term, concentrated and improved phenotype recording could in combination with appropriate genotyping strategies allow routine genomic evaluations in beef cattle with better representation of breeding goals in the breeding programmes and increased and more targeted breeding progress.

Small ruminants

For small ruminants, opportunities to routinely access whole-genome information exist (Table 2), but genomic evaluations are much more likely to be established in sheep than in goats. When compared to dairy cattle, the organization levels of breeding are usually lower, systematic recording of phenotypes is often rudimentary, and breeding goals and programmes are rather heterogeneous, so conventional genetic evaluations are more likely to be performed by and within breeding organization than nationally. According to the importance of the sheep industry in Oceania, most advanced interflock genetic improvement programmes and genomic research activities have been reported from Australia (Brown et al. 2007). The expected benefits of GS for sheep and goat breeding may be smaller than for cattle breeding (15–40% for wool and meat sheep; Van der Werf 2009) and primarily relate to the opportunities of more accurate selection at an early age. The breeding goals include traits which need to be measured in the field, but for which the flow of phenotype data is difficult to establish. Provided feasibility of representative phenotype recording, GS promises considerably improved conditions for genetic improvement for traits like parasite resistance, slaughter traits, adult wools traits and female fertility (Van der Werf 2009). GS may accordingly allow adjustments of trait emphases in the breeding programmes towards economically important traits, hard to measure and difficult to improve in the conventional systems. However, the generally lower value of breeding animals and the continuing costs for both phenotyping and genotyping need to be taken into account in GS-related cost-benefits calculations. Practicable concepts may involve collection of relevant phenotypes in selected flocks (like the Australian Information Nucleus; Fogarty et al. 2007) and specific genotyping schemes. For goats, whole-genome screening has only recently become an option through international collaboration and development of a consortium SNP chip (Table 2) and with the given breeding structures routine application of GS cannot be expected in the near future. However, goats play an important role among the domesticated animals, and in the long-term GS may valuably assist the management of caprine genetic resources and population development with regard to sustainable food production in large parts of the world.

Pigs

Pig breeding is today dominated by worldwide acting suppliers of swine genetics, large organizations with distinct breeding programmes and nucleus farms, in which purebred lines are kept and breeding animals are produced. Because of the already short generation interval in pigs, main benefits of GS are conveyed by increased selection accuracy, particularly for the traits which require onerous measurement in the field and are difficult to improve in the conventional systems. The strong position of the leading pig breeding organizations is relevantly determined by their infrastructure for phenotype recording, implying considerable investments and substantial interest to maximize their use for the selection of elite breeding animals. Additional genetic gain was expected from GS for finisher and slaughter traits and in particular for several traits that can only be measured in females (reproduction traits like litter size, uniformity of litters, vitality, mothering abilities) or on the production level (feed conversion, leg health, longevity). At young age, conventional breeding values for these traits have low reliabilities and are equal among full-sibs, until further information becomes available. Accordingly, accurate predictions allowing within-family selection implied opportunities to optimize progeny testing under GS. Because competitiveness on the pig market is increasingly dependent on the ability to enable efficient and sustainable production, main emphasis in genomic research and promotion of GS was on reproduction and functionality. Additional genetic gain of 20–50% was expected in the maternal line where focus is on the reproduction traits through more accurate selection of replacement animals (Hypor 2012).

The porcine SNP chip was first offered in late 2008, with subsequent intensification of research on the use of whole-genome data for the improvement of commercial pig breeding. Given the recent routine implementation of GS in pigs, competitors may be forced to move towards broad use of whole-genome information to strengthen their market position. However, efficiency calculations of the breeding companies must take into account the initial and running costs and efforts of GS, so collaboration plans and strategic genotyping may be necessary for broader use of GS in pigs. Single-gene tests are likely to become dispensable with access to whole-genome information, but cost saving will not outweigh the genotyping costs, the prohibitive effect of which needs to be compensated by significantly tailored selection for traits of (increasing) economic importance. GS may facilitate parallel improvement in production efficiency and pig welfare, requested by the modern society.

Poultry

In poultry breeding, the dominance of large internationally acting breeding companies is even more pronounced than in the pig industry. Enormous efforts are made to collect phenotype and pedigree data, keep large databases and continuously develop the specific breeding programmes. Because the whole-genome sequence of the chicken became available in 2004, intense and mostly private research activities have also included genomics as an option to improve selection and increase the genetic gain in both layers and broilers (González-Recio et al., 2009). However, not all of the results have been made publicly available yet, indicating the impact of commercialization particularly in chicken breeding (Fulton 2012). Breeding goals included many traits difficult to measure and improve (disease and parasite resistance, longevity, late production and production persistence traits) as well as sex-limited traits in the laying lines and traits with marked sex dimorphism in the broilers, so chicken breeding should principally be able to benefit substantially from GS via improved options for more accurate selection at young age and within families. After a period of limited accessibility of whole-genome data, high density SNP chips have recently become commercially available, enabling routine work with genomic information in the chicken industry. For the future, it can be expected that genomic data will be used by the breeding companies to ensure long-term competitiveness. With cost-efficient strategies for phenotyping and genotyping, optimization of genomic tools with regard to the specific requirements of poultry breeding will be propelled to allow transfer of knowledge to optimized breeding and success on the market. Challenges to be met include refinement of breeding programmes with possible adjustments of breeding goals to the demands of customers and producers (new traits relating to animal welfare, health and functionality, food conversion and efficiency, sustainability) and finding ways to restore some variability in the highly specialized commercial lines (Fulton 2012).

Growing experience in poultry genomics, primarily based on the intense research in chicken, could facilitate adoption of the new tools in turkey breeding. Full genome sequence information can already be assessed in the public domain (Table 1), but not yet the extensive SNP data that have been collected (Aslam et al. 2012). Given the increased speed of genomic advances and existing networks in the poultry industry, soon advent of GS in turkey breeding can be surmised (Ralph 2012).

Horses

The role and structure of the equine industry differ considerably between countries, but there is little doubt that it has considerable economic strength. In contrast to the horse, meat production sector which is economically important in some countries, but lacks structures for genomically enhanced breeding programmes, race and sport horse industries have clear organization structures and well-developed breeding programmes around the world. Routine genetic evaluations have been established for riding horses in several riding horse populations with a focus on performance, whereas further important breeding goal traits are not yet considered sufficient because of testing difficulties (Koenen and Aldridge 2002). The long generation interval in the equine implies large potential of GS to enhance the genetic gain and at the same time optimize the spectrum of selection traits. The available SNP chip has already been used for research (Schröder et al. 2012; Petersen et al. 2013), and routine use of whole-genome data for breeding purposes may be expected in the near future. Collection of enough high-quality phenotype data is likely to be the major limiting factor, but possible solutions have already been presented (Van Grevenhof et al. 2012). However, collaborations between breeding organizations will be essential for assembly of the reference population, and the networks in the equine industry may not yet be as close as in dairy cattle in the pre-genomics era, where international information exchange via Interbull was already established.

Other mammalian species

The genomes of further economically important mammalian species have been sequenced, providing the basis for systematic work with whole-genome data. Rabbits may be mentioned as an example for animals that are of economic relevance in the food production sector of only some countries. Breeding programmes are usually based on mass selection for production traits like litter size, growth rate and meat characteristics (Gyovai et al. 2012), implying similar options for enhancing the genetic gain by GS through earlier and particularly more accurate selection as, for example in pigs or chicken. However, the organization structure and overall weak financial power of commercial rabbit breeding make developments towards routine use of genomics unlikely in the near future. Considering the important role of rabbits as experimental animals, science-driven advances can be expected, so SNP chips may become available, which could allow GS at least in some larger and market-oriented populations. Given the low value of individual breeding animals, affordable tools for whole-genome screening would facilitate stronger focus on the important low-heritability traits like disease and parasite resistance (De Rochambeau et al. 2006).

Dogs are used by humans for a broad variety of tasks, including traditional and new fields of work (e.g. herding, therapy), and represent an important model species for research. Accordingly, the canine genome was fully sequenced already in 2003, followed by the development of SNP arrays that have been used in different context for research (Parker and Ostrander 2005; Ke et al. 2011; Oliete 2012). In 2011, the opportunities to efficiently select against disease conditions using GEBV were illustrated, implying possible move of dog breeding from science to routine use of whole-genome data (Guo et al. 2011). Dog breeding programmes usually consider several moderately to highly heritable conformation traits, but frequently also include strategies to reduce disease prevalences and to improve traits impacting the intended use of the dogs (e.g. behaviour, specific working abilities). For this category of traits, which have relatively low heritabilities and require considerable efforts for phenotype recording, GS promises considerable enhancement of the genetic gain through more accurate selection at young age. When compared to the livestock species, the cost factor for routine implementation of genomics may be less problematic, because many dog breeders already spend large amounts of money for increasing numbers of single-gene test, which may become dispensable with whole-genome screening. Canine GEBV may therefore become available for different populations and types of traits in the near future. With a view to the unique diversity of breeds, created through prolonged artificial selection for often primarily morphological characteristics, experiences from dog breeding how to use whole-genome data for population management and coping with negative effects of diversity loss may be highly valuable for other species and the whole animal breeding sector.

Aquaculture

Aquaculture is a growing business with considerable needs for long-term strategies to ensure efficient and sustainable production, but organized breeding has been established for only few species. Breeding programmes are mainly based on mass selection and testing of siblings of candidates, so early and more accurate prediction of the genetic merit of individuals promises considerable increases in the genetic gain. However, the prediction formula for GS would most likely be derived from direct linking of phenotype and genotype records, and under these conditions, considerable investments are needed for both genotyping and collection of large enough quantities of phenotype data. Furthermore, GEBV reliabilities may drop quickly without regular re-estimation of effects, which implies substantial running costs (Sonesson and Meuwissen, 2009). Although cost calculations need to consider the relatively low value of the breeding stock, aquaculture may in the long-term still benefit from using whole-genome data, which allows departure from the family-oriented breeding and selection structure. Inbreeding is an important issue in aquaculture, and options for thorough management of genetic diversity may be particularly beneficial to avoid detrimental effects of the currently applied breeding practices. Genome sequencing of economically important fish species has been recently tackled by international research consortia, and with the development of SNP chips of appropriate format and pricing, GS may become feasible in commercial aquaculture.

Conclusions

  1. Top of page
  2. Contents
  3. Introduction
  4. Development and General Aspects of Genomic Selection
  5. Applications of Genomic Selection and Specific Challenges for Breeding
  6. Conclusions
  7. Conflict of interest
  8. Author contributions
  9. References

Whole-genome data of many important animal species can today be accessed easily and relatively cheaply, implying options for routine use in breeding. Benefits of GS are largest in constellations with long general interval, expensive and onerous performance testing, high value of individual elite breeding animals and existing organization structures that allow accessing reliable phenotype information or genetic merit predictions for parts of the population. Dairy cattle fits this pattern, explaining the fast adoption and broad acceptance of GS that now impose challenges of efficient and successful long-term use of genomic data. In other species, GS is less developed implying current focus on implementation issues, with intensified collaboration of stakeholders as possible key factor of success. With the refined selection methods that allow more tailored breeding programmes, the importance of trait definitions and reliable flow of phenotype data has increased, so the era of genomics may be likewise seen as the era of phenomics.

References

  1. Top of page
  2. Contents
  3. Introduction
  4. Development and General Aspects of Genomic Selection
  5. Applications of Genomic Selection and Specific Challenges for Breeding
  6. Conclusions
  7. Conflict of interest
  8. Author contributions
  9. References
  • Aslam ML, Bastiaansen JWM, Elferink MG, Megens H-J, Crooijmans RPMA, Blomberg LA, Fleischer RC, van Tassell CP, Sonstegard TS, Schroeder SG, Groenen MAM, Long JA, 2012: Whole genome SNP discovery and analysis of genetic diversity in Turkey (Meleagris gallopavo). BMC Genomics 13, 391.
  • Boysen T-J, Heuer C, Tetens J, Reinhardt F, Thaller G, 2013: Novel use of derived genotype probabilities to discover significant dominance effects for milk production traits in dairy cattle. Genetics 193, 431442.
  • Brookes AJ, 1999: The essence of SNPs. Gene 234, 177186.
  • Brown DJ, Huisman AE, Swan AA, Graser H-U, Woolaston RR, Ball AJ, Atkins KD, Banks RG, 2007: Genetic evaluation for the Australian sheep industry. Proc Assoc Advmt Anim Breed Genet 17, 187194.
  • Cervantes I, Meuwissen THE, 2011: Maximization of total genetic variance in breed conservation programmes. J Anim Breed Genet 128, 465472.
  • Chesnais JP, 2010: How is the AI industry using genomic tools in practice? In: Proc Interbull Int Workshop, Paris, France; Interbull Bull 41; Interbull Centre, Uppsala, Sweden; pp. 5962.
  • De Cara MAR, Fernández J, Toro MA, Villanueva B, 2011: Using genome-wide information to minimize the loss of diversity in conservation programmes. J Anim Breed Genet 128, 456464.
  • De Haas Y, Calus MPL, Veerkamp RF, Wall E, Coffey MP, Daetwyler HD, Hayes BJ, Pryce JE, 2012: Improved accuracy of genomic prediction for dry matter intake of dairy cattle from combined European and Australian data sets. J Dairy Sci 95, 61036112.
  • Dekkers JC, 2004: Commercial application of marker- and gene-assisted selection in livestock: Strategies and lessons. J Anim Sci 82(E-Suppl), E313E328.
  • De Rochambeau H, Licois D, Gidenne T, Verdelhan S, Coudert P, Elsen JM, 2006: Genetic variability of the resistance for three types of enteropathy in the growing rabbit. Livest Sci 101, 110115.
  • De Roos APW, Schrooten C, Veerkamp RF, Van Arendonk JAM, 2011: Effects of genomic selection on genetic improvement, inbreeding, and merit of young versus proven bulls. J Dairy Sci 94, 15591567.
  • Ducrocq V, Santus E, 2011: Moving away from progeny test schemes: Consequences on conventional (inter)national evaluations. In: Proc Interbull Int Workshop, 27-28 February 2011, Guelph, Canada; Interbull Bull 43. Available: http://www.interbull.org/images/stories/Ducrocq_copy.pdf (accessed 1 February 2013).
  • Dürr J, Philipsson J, 2012: International cooperation: the pathway for cattle genomics. Anim Front 2, 1621.
  • Eggen A, 2012: The development and application of genomic selection as a new breeding paradigm. Anim Front 2, 1015.
  • Engelsma KA, Veerkamp RF, Calus MPL, Windig JJ, 2011: Consequences for diversity when prioritizing animals for conservation with pedigree or genomic information. J Anim Breed Genet 128, 473481.
  • Falconer DS, 1989: Introduction to Quantitative Genetics, 3rd edn. Longman Scientific and Technical, New York.
  • Fan B, Du ZQ, Gorbach DM, Rothschild MF, 2010: Development and application of high-density SNP arrays in genomic studies of domestic animals. Asian-Aust J Anim Sci 23, 833847.
  • Fogarty NM, Banks RG, van der Werf JHJ, Ball AJ, Gibson JP, 2007: The Information Nucleus – a new concept to enhance sheep industry genetic improvement. Proc Assoc Advmt Anim Breed Genet 17, 2932.
  • Fulton JE, 2012: Genomic selection for poultry breeding. Anim Front 2, 3036.
  • Goddard M, 2009: Genomic selection: prediction of accuracy and maximization of long term response. Genetica 136, 245257.
  • González-Recio O, Gianola D, Rosa GJM, Weigel KA, Kranis A, 2009: Genome-assisted prediction of a quantitative trait measured in parents and progeny: application to food conversion rate in chickens. Genet Sel Evol 41, 3.
  • Guo G, Zhou Z, Wang Y, Zhao K, Zhu L, Lust G, Hunter L, Friedenberg S, Li J, Zhang Y, Harris S, Jones P, Sandler J, Krotscheck U, Todhunter R, Zhang Z, 2011: Canine hip dysplasia is predictable by genotyping. Osteoarthr Cartil 19, 420429.
  • Gyovai P, Nagy I, Gerencsér Z, Matics Z, Radnai I, Donkó T, Bokor Á, Farkas J, Szendrő Z, 2012: Genetic parameters for litter weight, average daily gain and thigh muscle volume measured by in vivo Computer Tomography technique in Pannon White rabbits. Livest Sci 144, 119123.
  • Hayes BJ, Bowman PJ, Chamberlain AJ, Goddard ME, 2009: Genomic selection in dairy cattle: progress and challenges. J Dairy Sci 92, 433443.
  • Hypor, 2012: Hypor moves into a new era of genomic selection. Available: http://www.hypor.com/News/Hypor%20Moves%20Into%20A%20New%20Era%20 Of%20Genomic%20Selection.aspx (assessed 1 February 2013).
  • Ibañez-Escriche N, Gonzalez-Recio O, 2011: Review. Promises, pitfalls and challenges of genomic selection in breeding programs. Span J Agric Res 9, 404413.
  • Ke X, Kennedy LJ, Short AD, Seppälä EH, Barnes A, Clements DN, Wood SH, Carter SD, Happ GM, Lohi H, Ollier WER, 2011: Assessment of the functionality of genome-wide canine SNP arrays and implications for canine disease association studies. Anim Genet 42, 181190.
  • Kemper KE, Bowman PJ, Pryce JE, Hayes BJ, Goddard ME, 2012: Long-term selection strategies for complex traits using high-density genetic markers. J Dairy Sci 95, 46464656.
  • Koenen EPC, Aldridge LI, 2002: Testing and genetic evaluation of sport horses in an international perspective. 7th WCGALP, 19-23 August 2002, Montpellier, France.
  • Kwok P-Y, Gu Z, 1999: Single nucleotide polymorphism libraries: why and how are we building them? Mol Med Today 5, 538543.
  • Liu Z, 2011: Use of MACE results as input for genomic models. In: Proc Interbull Int Workshop, 27-28 February 2011, Guelph, Canada; Interbull Bull 43. Available: http://www.interbull.org/images/stories/Liu.pdf (accessed 1 February 2013).
  • Liu Z, Seefried FR, Reinhardt F, Rensing S, Thaller G, Reents R, 2010: Dairy cattle genetic evaluation enhanced with genomic information. 9th WCGALP, 1-6 August 2010, Leipzig, Germany.
  • Meuwissen THE, Hayes BJ, Goddard ME, 2001: Prediction of total genetic value using genome-wide dense marker maps. Genetics 157, 18191829.
  • Meuwissen THE, Hayes BJ, Goddard ME, 2013: Accelerating improvement of livestock with genomic selection. Annu Rev Anim Biosci 1, 221237.
  • Miglior F, Chesnais J, van Doormaal B, 2012: Genetic improvement: a major component of increased dairy farm profitability. 38th Ann Meet ICAR, 28 May to 1 June 2012, Cork, Ireland.
  • Montaldo HH, Casas E, Ferraz JBS, Vega-Murillo VE, Romá-Ponce SI, 2012: Opportunities and challenges from the use of genomic selection for beef cattle breeding in Latin America. Anim Front 2, 2329.
  • Muir WM, 2007: Comparison of genomic and traditional BLUP-estimated breeding value accuracy and selection response under alternative trait and genomic parameters. J Anim Breed Genet 124, 342355.
  • Oliete JQ, 2012: Application of Genome-Wide Single Nucleotide Polymorphism Arrays to Understanding Dog Disease and Evolution. PhD thesis. University of Barcelona, Barcelona.
  • Parker HG, Ostrander EA, 2005: Canine genomics and genetics: running with the pack. PLoS Genet 1, e58.
  • Patry C, Ducrocq V, 2011: Evidence of biases in genetic evaluations due to genomic preselection in dairy cattle. J Dairy Sci 94, 10111020.
  • Petersen JL, Mickelson JR, Rendahl AK, Valberg SJ, Andersson LS, Axelsson J, Bailey E, Bannasch D, Binns MM, Borges AS, Brama P, da Câmara Machado A, Capomaccio S, Cappelli K, Cothran EG, Distl O, Fox-Clipsham L, Graves KT, Guérin G, Haase B, Hasegawa T, Hemmann K, Hill EW, Leeb T, Lindgren G, Lohi H, Lopes MS, McGivney BA, Mikko S, Orr N, Penedo MCT, Piercy RJ, Raekallio M, Rieder S, Røed KH, Swinburne J, Tozaki T, Vaudin M, Wade CM, McCue ME, 2013: Genome-wide analysis reveals selection for important traits in domestic horse breeds. PLoS Genet 9, e1003211. doi:10.1371/journal.pgen.1003211.
  • Pryce JE, Daetwyler HD, 2012: Designing dairy cattle breeding schemes under genomic selection: a review of international research. Anim Prod Sci 52, 107114.
  • Pryce JE, Hayes B, 2012: A review of how dairy farmers can use and profit from genomic technologies. Anim Prod Sci 52, 180184.
  • Pryce JE, Gredler B, Bolormaa S, Bowman PJ, Egger-Danner C, Fuerst C, Emmerling R, Sölkner J, Goddard ME, Hayes BJ, 2011: Genomic selection using a multiple-breed, across-country reference population. J Dairy Sci 94, 26252630.
  • Pryce JE, Hayes B, Goddard ME, 2012: Novel strategies to minimize progeny inbreeding while maximizing genetic gain using genomic information. J Dairy Sci 95, 377388.
  • Ralph J, 2012: Future uses of genomics in the poultry industry. Available: http://www.aviagenturkeys.com/media/204651/future_uses_of_genomics_in_the_poultry_industry.pdf (assessed 8 February 2013).
  • Schaeffer L, 2006: Strategy for applying genome-wide selection in dairy cattle. J Anim Breed Genet 123, 218223.
  • Schefers JM, Weigel KA, 2012: Genomic selection in dairy cattle: integration of DNA testing into breeding programs. Anim Front 2, 49.
  • Schröder W, Klostermann A, Stock KF, Distl O, 2012: A genome-wide association study for quantitative trait loci of show-jumping in Hanoverian warmblood horses. Anim Genet 43, 392400.
  • Sellner EM, Kim JW, McClure MC, Taylor KH, Schnabel RD, Taylor JF, 2007: Applications of genomic information in livestock. J Anim Sci 85, 31483158.
  • Simianer H, Chen J, Erbe M, 2011: Animal breeding in the genomics era: Challenges and opportunities for the maintenance of genetic diversity. In: Proc 62nd Ann Meet EAAP, Stavanger, Norway; EAAP, Rome, Italy; p. 76 (abstract).
  • Smith C, 1967: Improvement of metric traits through specific genetic loci. Anim Prod 9, 349358.
  • Sonesson AK, Meuwissen THE, 2009: Testing strategies for genomic selection in aquaculture breeding programs. Genet Sel Evol 41, 37.
  • Toosi A, Fernando RL, Dekkers JCM, 2010: Genomic selection in admixed and crossbred populations. J Anim Sci 88, 3246.
  • Van der Werf JHJ, 2009: Potential benefit of genomic selection in sheep. Proc Assoc Advmt Anim Breed Genet 18, 3841.
  • Van Grevenhof EM, Van Arendonk JAM, Bijma P, 2012: Response to genomic selection: The Bulmer effect and the potential of genomic selection when the number of phenotypic records is limiting. Genet Sel Evol 44, 26.
  • Weigel KA, Van Tassell CP, O'Connell JR, Van Raden PM, Wiggans GR, 2010: Prediction of unobserved single nucleotide polymorphism genotypes of Jersey cattle using reference panels and population based imputation algorithms. J Dairy Sci 93, 22292238.
  • Weigel KA, Hoffmann PC, Herring W, Lawlor TJ, 2012: Potential gains in lifetime net merit from genomic testing of cows, heifers, and calves on commercial dairy farms. J Dairy Sci 95, 22152225.