| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Animal Science, Iowa State University, Ames 50011-3150
 |
Abstract |
|---|
 Top Abstract IntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
Key Words: Breeding Programs • Markers-Assisted Selection • Quantitative Trait Loci
|
Introduction |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
|
Principles of Marker-Assisted Selection |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
Methods to detect these types of loci were described in Andersson (2001). The LE markers can be readily detected on a genome-wide basis by using breed crosses or analysis of large half-sib families within the breed. Such genome scans require only sparse marker maps (15 to 50 cM spacing, depending on marker informativeness and genotyping costs; Darvasi et al., 1993) to detect most QTL of moderate to large effects. Many examples of successful applications of this methodology for detection of QTL regions are available in the literature (see Andersson, 2001). The LD markers are by necessity close to the functional mutation for sufficient population-wide LD between the marker and QTL to exist (within 1 to 5 cM, depending the extent of LD, which depends on population structure and history). The LD markers can be identified using candidate genes (Rothschild and Soller, 1997) or fine-mapping approaches (Andersson, 2001). Direct markers (i.e., polymorphisms that code for the functional mutations) are the most difficult to detect because causality is difficult to prove and, as a result, a limited number of examples are available, except for single-gene traits (Andersson, 2001).
The three types of marker loci differ not only in methods of detection, but also in their application in selection programs. Whereas direct markers and, to a lesser degree, LD markers, allow for selection on genotype across the population because of the consistent association between genotype and phenotype, use of LE markers must allow for different linkage phases between markers and QTL from family to family. Thus, the ease and ability to use markers in selection is opposite to their ease of detection and increases from direct markers to LD markers and LE markers. In what follows, selection on these three types of markers will be referred to as gene-assisted selection (GAS), LD markers-assisted selection (LD-MAS), and LE marker-assisted selection (LE-MAS).
Traits
Molecular markers have been used to identify loci or chromosomal regions that affect single-gene traits and quantitative traits. Single-gene traits include genetic defects, genetic disorders, and appearance. For the purposes of QTL detection and application, quantitative traits can be categorized into a) routinely recorded traits; b) difficult to record traits (feed intake, product quality); and c) unrecorded traits (disease resistance). Each of these can be further subdivided into traits that are i) recorded on both sexes; ii) sex-limited traits; and iii) traits that are recorded late in life. The ability to detect QTL depends on the availability of phenotypic data and decreases in the order a, b, c and within each of those in the order i, ii, and iii. For related reasons, genome scans, which require more phenotypic data than candidate gene analyses, are often used to detect QTL for traits in Category a, whereas candidate gene approaches are more often used to identify QTL for traits that are not routinely recorded (b and c). Potential extra genetic gains from MAS or GAS are in inverse proportion to the ability to make genetic progress using conventional methods and are greatest for traits in Category c and lowest for traits in Category a, in particular for traits that are routinely recorded on both sexes prior to selection (Meuwissen and Goddard, 1996).
Strategies for Use of Molecular Data in Selection
For the purpose of genetic improvement, markers can be used to enhance within-breed selection based on GAS, LD-MAS, or LE-MAS, or to enhance programs to capitalize on between-breed variation by selection within a cross. A specific form of the latter is marker-assisted introgression (MAI), which will be discussed later. Because of the extensive LD in crosses, markers can be used as LD markers without requiring close linkage (Lande and Thompson, 1990). Most applications of markers in livestock, however, are based on within-breed selection.
In principle, all applications of molecular genetic information for genetic improvement involve selection on a molecular score, although the composition of this score differs from application to application (Dekkers and Hospital, 2002). For example, the molecular score could be based on the presence or absence of certain alleles or genotypes, as for MAI, or on estimates of marker or QTL effects, which can be summed over loci when multiple QTL regions are selected for (Lande and Thompson, 1990). In general then, three strategies for can be distinguished for the use of the molecular score (MS) in selection, in combination with phenotype, or EBV derived from phenotypic information. These apply to GAS, LD-MAS, and LE-MAS, and to MAS using within- and between-breed variation: I) tandem selection, with selection of candidates on MS followed by selection on phenotype or EBV; II) index selection on a combination of MS and phenotype or EBV: I = b1MS + b2EBV (Lande and Thompson, 1990); III) preselection on MS (or an index of MS and EBV) at a young age, followed by selection on an updated EBV at a later age (Lande and Thompson, 1990).
Selection on total EBV, as the sum of an estimate of the breeding or genetic value for the QTL and an estimate of polygenic EBV, as would be obtained from including molecular data as fixed or random effects in a BLUP animal model genetic evaluation model (e.g., Van Arendonk et al., 1999), is equivalent to index selection (Strategy II) with index weights equal to one (I = MS + EBV). For other cases, and if the objective is to maximize response over multiple generations, index weights will differ from one (e.g., Dekkers and van Arendonk, 1998). This is further complicated by the fact that many genes or QTL affect multiple traits and that selection most often is for multiple traits. Use of MAS for multiple trait selection was addressed by Lande and Thompson (1990) and Weller (2001).
Extra Response from Marker-Assisted Selection.
The basic objective of selection programs is to improve the population for a comprehensive multiple-trait breeding goal. This goal can, in principle, be formulated as a combination of genetic traits. To affect progress in the overall goal, a finite amount of available selection pressure, which is limited by characteristics of the breeding program and population (e.g., reproductive rate), must be divided among traits. Increased emphasis on one trait diverts emphasis away from other components, but the joint effect on all traits determines the success of the breeding program. The challenge of the design of a breeding program is to balance selection emphasis among traits to maximize response in the overall objective. With the availability of genetic markers and tests, this is further complicated by the need to balance emphasis on molecular vs. quantitative genetic information. This also holds for selection against genetic defects, the emphasis on which must be balanced against selection on quantitative traits. Extra genetic gains from MAS, therefore, depend on the effect of direct selection on individual loci on genetic progress at other loci (polygenes) and for other traits that affect overall genetic merit. This is the case even for selection against genetic defects and in the absence of pleitropic effects of such loci.
How much response in polygenes and other traits is affected by selection on markers depends on the selection strategy used. Although tandem selection results in the most rapid fixation of the gene(s) that are targeted by the molecular score, it results in the greatest loss in response for polygenes and for traits that are not included in the molecular score and may therefore result in less response in the trait and the overall breeding goal. In theory, index selection results in the greatest overall response to selection for a given selection stage, in particular if weights on the molecular score are optimized (e.g., Dekkers and van Arendonk, 1998). Figure 1
illustrates differences in response from tandem vs. index selection on direct markers. These results apply to selection on a single trait and selection on an overall breeding goal. For the latter, effects of the gene are expressed in terms of genetic standard deviations for the breeding goal.
|
). Lost response decreased to zero as the effect of the gene increased because tandem selection is equivalent to index selection if the QTL has a very large effect. These results indicate that it is important to incorporate molecular information in an index and to avoid tandem selection. This also holds for selection against genetic defects and single-gene traits, which requires assigning an economic value to such traits to enable their incorporation into an overall breeding goal. The choice between tandem and index selection (and other alternatives) also depends on other factors, such as market and cost considerations. For example, rapid fixation of the targeted gene (e.g., by tandem selection) will reduce costs of genotyping over generations and may be desirable from a marketing perspective. This can, however, also be achieved by increasing the weight on the molecular score in an index, as has been demonstrated by Settar et al. (2002).
Tandem and index selection apply to the use of molecular information in a given stage of selection. If selection of candidates is over multiple stages, the impact on response for polygenes and other traits can be minimized if molecular information is used at an early age when limited or no phenotypic information is available to distinguish selection candidates (preselection, Strategy III; Lande and Thompson, 1990). An example is preselection among full-sib dairy bulls for entry into progeny testing programs (Kashi et al., 1990; Mackinnon and Georges, 1998). Application of MAS at this level, however, requires family sizes large enough that selection room is available to apply MAS on a within-family basis. This strategy also minimizes the effect on other selection stages and therefore minimizes the risk of losing response to routine selection on phenotypic information and has been the preferred approach for initial applications of LE-MAS in dairy cattle.
Strategies for Marker-Assisted Introgression.
Marker-assisted introgression programs are based on tandem selection in a multigenerational backcrossing program, in which a MS based on the presence of donor breed alleles at or around the target gene is used in the first selection step (foreground selection), followed by background selection on a MS based on presence or absence of recipient alleles at markers spread over the genome, on phenotype, or an index of the two (e.g., Visscher et al., 1996). Although tandem selection has been implicit to gene introgression programs, the selection on an index of molecular score and phenotypic information in these programs should be considered, especially for quantitative traits, unless the gene has a very large effect. Although this could result in selection of some parents that do not carry the target allele, overall response is expected to be greater. In particular, if multiple genes or QTL regions must be introgressed simultaneously, the requirement that selected parents carry the target allele for all QTL is infeasible in livestock and not necessary for successful introgression (Chaiwong et al., 2002).
|
Industry Application of Marker-Assisted Selection |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
, with tests categorized by the type of trait and the type of marker. A substantial number of genetic tests are available. Some applications of selection for individual genes occurred prior to the era of molecular genetics, including selection on observable genetic defects and appearance, the halothane test as a physical test for the RYR gene, and use of the B-blood group as a physiological LD marker for selection for disease resistance in poultry, which started in the 1960s (Hansen et al., 1967; Hansen and Law, 1970; J. Arthur, Hy-Line Int., Dallas Center, IA, personal communication). Several tests are used for within-house selection only (e.g., PICmarq markers used by the Pig Improvement Co., G. Plastow, Sygen Int., Berkeley, CA, personal communication), whereas others are available through commercial genotyping services. To date, the majority of publicly available tests are for direct or LD markers.
|
). An alternative is to follow a breed-cross QTL analysis with an LE QTL analysis within commercial lines in identified regions, which has shown to be successful in pigs (Evans et al., 2003) and has been used in beef cattle (J. Hetzel, Genetic Solutions, Brisbane, Australia, personal communication). This results in identification of LE markers that can be used for selection. An important exception to the use of experimental crosses for first-phase QTL detection is dairy cattle, for which genome scans are based primarily on the large paternal half-sib families that are available in the industry, using the daughter or grand-daughter designs (Weller et al., 1990; Bovenhuis and Schrooten, 2002). This has resulted in the availability and use of LE markers for several QTL regions (Boichard et al., 2002; Spelman, 2002; Bennewitz et al., 2003; Khatkar et al., 2004; D. Funk, American Breeders Service, DeForrest, WI, personal communication; E. Mullaart, Holland Genetics, Arnheim, The Netherlands, personal communication). Several issues related to the transfer of initial results from genome scans to applications in breeding programs were discussed in Spelman and Bovenhuis (1998).
Examples of Applications of MAI or MAS
Marker-Assisted Introgression Programs.
Marker-assisted introgression has been the main approach for utilization of genetic markers in plant breeding and successes and limitations of these applications have been documented by Hospital et al. (2002). Because of longer generation intervals, lower reproductive rates, and greater rearing costs, introgression is only feasible in livestock for genes of large effect. However, some examples of MAI in livestock are available. Hanset et al. (1995) reported on the successful introgression of the halothane normal allele into a Piétrain line that had a high frequency of the halothane-positive allele. They used LD foreground selection on markers linked to the RYR locus. Yancovich et al. (1996) used marker-assisted background selection to speed up the recovery of the broiler genome when introgressing the naked-neck gene from a rural low-BW breed into a commercial broiler line. Gootwine et al. (1998) reported on MAI of the Booroola gene (FecB) into dairy sheep breeds using LD markers for foreground selection. In developing countries, programs for the introgression of disease resistance or tolerance genes are being considered for cattle (J. Gibson, ILRI, Nairobi, Kenya, personal communication).
Marker-Assisted Selection.
The main application and potential for use of markers to enhance genetic improvement in livestock is through within-breed selection. This requires markers that trace within-breed variability. Although several genetic tests are available to effect such selection, as documented in Table 1, the extent to which they are used in commercial breeding programs is unclear, as is the manner in which they are used (i.e., Strategies I, II, or III), and whether their use leads to greater responses to selection. Direct and LD markers have been primarily used, as evidenced by their abundance among available tests (Table 1). Direct and LD markers allow selection on markers across the population, which facilitates their use. Some of the earlier applications of MAS in livestock were prior to the era of molecular genetics (e.g., selection for disease resistance in poultry using an Elisa tests for the B-blood group as an LD marker; J. Arthur personal communication) and selection against the halothane gene in pigs using the halothane test. Subsequently, several genetic tests have been used to select against carriers of recessive genetic defects in livestock species, as reflected in Table 1. One of the first examples of use of an LD marker for a quantitative trait was the test for the estrogen receptor gene (ESR; Rothschild et al., 1996; Short et al., 1997), which has been used in several commercial lines to enhance selection for litter size (G. Plastow, personal communication). Plastow et al. (2003) and G. Plastow (personal communication) also reported the use of more than 15 proprietary direct and LD markers (PICmarq) for traits associated with reproduction, feed intake, growth, body composition, and meat quality in pigs. M. Lohuis (Monsanto, Co., St. Louis, MO, personal communication) reported the recent in-house use of a combination of LE and LD markers in commercial lines of pigs, and fine-mapping efforts to replace LE markers with LD markers for important QTL regions. In dairy cattle, in addition to direct markers for genetic defects and milk protein variants (Table 1), an LD marker near the prolactin gene (Cowan et al., 1990) that is segregating in one prominent Holstein sire family, has been used for preselection of young bulls since (Cowan et al., 1997). M. Cowan (Genetic Visions, Middleton, WI, personal communication) reported the use of several additional direct, LD, and LE markers for selection of bull dams and preselection of young bulls. In-house selection programs using LE markers have been conducted in several dairy cattle breeding programs, including for pre-selection of young bulls in the US based on QTL studies reported by Georges et al. (1995) and Zhang et al. (1998) (D. Funk, personal communication) and in New Zealand (Spelman, 2002) and The Netherlands (E. Mullaart, personal communication) based on QTL results reported by Spelman et al. (1996), Arranz et al. (1998) and Coppieters et al. (1998). Establishment of national genetic evaluation programs using LE markers to provide information for in-house use by dairy cattle breeding organizations has been reported for France (Boichard et al., 2002) and Germany (Bennewitz et al., 2003). In beef cattle, several direct and LD markers are commercially available (Table 1) and used by individual breeders (J. Hetzel, S. Schmutz, University of Saskatchewan, Saskatoon, Canada, personal communication). Programs for LE MAS are being initiated in some cases (J. Hetzel, personal communication). In sheep, J. McEwan (AgResearch, Invermay, New Zealand, personal communication) reported on an LD-MAS program for a 5-cM region around the Carwell gene (Nicoll et al., 1998). An animal model with the Carwell genotype included as a fixed effect is used. In addition, several direct or LD markers associated with reproduction and disease, including scrapie, are being used (S. Dominik CSIRO, Armidale, Australia, G. J. Nieuwhof Meat and Livestock Commission, Milton Keynes, U.K., personal communication).
Evaluation of the Success of Commercial MAS
The effect of MAS or MAI on genetic improvement is difficult to quantify, even under experimental conditions, because differences in response are not expected to be large, especially when considering that traits selected for using MAS are part of a multiple-trait breeding goal, and appropriate controls are often not available. Responses to MAS or MAI can, however, be evaluated at different levels, as summarized in Table 2: 1) changes in gene frequencies for the selected marker locus; 2) changes in gene frequencies in the targeted locus (if different from the selected loci); 3) effect of the targeted locus or region on the trait in the target population; and 4) improvement of the population or selected individuals in overall genetic level for trait(s) of interest.
|
For LE markers, the effect of MAS on the marker locus cannot be evaluated by its population frequencies because the desired marker allele differs by family. Some examples of the ability to select on LE marker(s) on a within-family basis, and the logistical limitations of implementing such selection, have been documented (e.g., Spelman, 2002) and will be discussed later.
Changes in Gene Frequencies for the Target Locus.
For direct markers, changes in marker frequencies are equivalent to frequency changes in the target locus. For LD markers, effects on frequencies of alleles at the targeted locus will depend on the extent of LD between the marker(s) and the causative locus, which can differ between populations and can change over generations because of recombination. These associations, and therefore the effect of LD-MAS on allele frequencies at the target locus, can only be evaluated indirectly based on marker-trait associations. The impact of selection on LE markers on the target locus can also only be evaluated indirectly through marker-trait analysis. Because of the need to evaluate effects on a within-family basis, monitoring marker-trait associations requires much more data for LE than for LD markers.
Phenotypic Effect of the Target Locus or Region.
Success of MAS also depends on the consistency of QTL effects across populations and environments. The effect of the target locus or region on the trait can differ in the selected or target population from its initial effect or its estimate before selection. Evaluation of introgression programs in plants has found that effects tend to be consistent for major genes that control simple traits but not for QTL for complex traits (e.g., yield; Hospital et al., 2002). Inconsistent effects have also been observed for some well-studied genes in livestock. For example, for the ESR gene for litter size in pigs (Rothschild et al., 1996), significant associations were demonstrated in multiple commercial lines in one of the largest studies of a candidate gene in livestock conducted to date (Short et al., 1997). However, some subsequent studies have found no effect of this LD marker on the trait and interactions with line and environment have been identified also (Rothschild and Plastow, 2002). Potential reasons for inconsistent results across studies and populations include statistical anomalies such as false positive or negative results (small sample sizes) and overestimation of significant QTL effects, as well as true effects, such as inconsistent marker-QTL linkage phases across populations for LD markers, genotype x environment interactions, and epistatic effects (Beavis, 1994).
Effects of the target locus or region in the target population can be readily evaluated for major genetic defects. For example, Rothschild and Plastow (1999) reported a reduction in mortality to zero and an improvement in meat quality from removal of the halothane stress gene. Effects, however, require careful analysis of marker–trait associations for more complex traits. This can be done at the population level using a random nonpedigreed sample for direct and LD markers, but it must be done on a within-family basis for LE markers. The latter requires substantially more data and a defined pedigree structure. Such analyses are not only needed to evaluate and monitor the success of MAS, but are also required to develop and modify QTL effect estimates and selection criteria. Thus, implementation of MAS requires continuous monitoring and reevaluation of gene or QTL effects in the target population and environment. This requires continuous emphasis on phenotypic recording in both nucleus and field populations.
Genetic Merit of the Population.
As described previously, MAS diverts selection emphasis away from polygenes and traits without marked QTL, and the ultimate success of MAS is determined by its impact on total genetic merit. It has also been shown that the impact of MAS on other loci and traits differs between the three selection strategies, and is greatest for tandem selection, followed by index selection, and preselection. It is unclear to what extent each of these strategies is used in commercial applications of MAS.
Because appropriate controls are often not available, success of MAS based on improvements in overall genetic merit of the population or selected individuals is very difficult to quantify in commercial breeding programs, let alone at an experimental level. Because of this and other reasons, few reports are available and these are not well documented. For example, Rothschild and Plastow (1999) reported an increase in response by up to 30% in litter size by incorporating the ESR genotype in selection indices for dam lines in PIC nucleus herds. This, however, represented the increase in genetic superiority for litter size of selected animals over a relatively short period of time, with limited accuracy.
Use of markers in preselection, as for entry of young dairy bulls into progeny test programs, does provide opportunities to assess the success of MAS. For example, by correlating EBV following progeny test with the preselection criterion, or by comparing progeny test EBV of preselected bulls to those of their full brothers, which may have been progeny tested by other organizations. To date, such studies have not yet been conducted in a comprehensive manner but several indirect assessments have been made. For example, Cowan et al. (1997) found an increase in mean EBV and in the number of progeny tested dairy bulls returned to service by preselection of young bulls based on the 
-casein locus and on the prolactin marker (Cowan et al., 1990). This was, however, based on limited numbers. Recently, M. Cowan (personal communication) reported a similar impact on graduation rates from subsequent use of these and other LE and LD markers for preselection of young bulls and bull dams. D. Funk (personal communication), however, reported limited initial evidence of an effect on the number of bulls returned to service from one of the first applications of LE-MAS for preselection based on QTL regions identified by Georges et al. (1995) and Zhang et al. (1998) of 70 now progeny-tested dairy bulls. Apart from small numbers, this apparent lack of success may reflect the limitations of the data and markers used in this early application of MAS for within-family selection, rather than the potential of MAS (D. Funk, personal communication). In a more fully documented example, Spelman (2002) described limited success from 2 yr of preselection of young bulls using 25 LE markers for six QTL regions in New Zealand. Lack of success was due to the limited number of bulls that could be preselected because of lack of selection room within families, which was due to the limited success of reproductive technologies used to increase full-sib family size from bull dams. Similar limitations in creating selection room for MAS have been identified in other programs (D. Funk and M. Cowan, personal communication) and indicates that the success of MAS depends not only on the accuracy of QTL estimates, but also on the ability to integrate technologies that are required to effectively implement a MAS program (see Integration of Marker-Assisted Selection in Breeding Programs).
|
LE vs. LD vs. Direct Markers |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
summarizes the relative requirements and opportunities for detection and application of these three types of markers. These comparisons, which will be further discussed below, also provide insight into the reasons for the extent of success and limitations that have been experienced in different commercial applications of MAS.
|
; Andersson 2001). Whereas LE markers can be random anonymous markers, direct markers require identification of the causative mutation, and LD markers require close linkage with the causative mutation, either identified as targeted candidate gene polymorphisms or by high-density marker maps. In addition, LE markers allow for genome-wide analysis of QTL based on a limited number of markers at 15- to 50-cM spacings. Genome-wide analysis is also possible for LD markers, but this will require a very dense marker map, depending on the extent of LD in the population (e.g., Meuwissen et al., 2001). The latter seems to be wide in livestock populations (Farnir et al., 2000; McRae et al., 2002), such that informative markers every 1 or 2 cM may be sufficient to detect most QTL. Associations between direct or LD markers and traits can be identified based on a limited number of phenotyped and genotyped individuals, without a specific population or family structure. Detection of QTL using LE markers, however, requires the presence of LD that extend over 20 or more centimorgan. Such LD can be created by crossing lines or found within families in outbred populations. The latter requires large numbers of phenotyped individuals with a specific family structure (e.g., Weller et al., 1990). The same is true for estimation and confirmation of LE vs. LD marker effects in other (outbred) populations (e.g., Spelman and Bovenhuis, 1998), resulting in greater phenotyping and genotyping requirements at this stage for LE markers, in particular if the initial QTL detection was based on a cross between lines.
Marker-Assisted Genetic Evaluation.
Requirements for integration of marker data in routine genetic evaluation procedures are also much greater for LE than for LD or direct markers, both with regard to requirements for the number and which individuals that must be phenotyped and genotyped, and with regard to methods of analysis (Table 3). Use of LE markers in an outbred population requires the phenotyping and genotyping of selection candidates and/or their relatives because effects must be estimated on a within-family basis. The extent of family data needed depends on recombination rates between markers and QTL. Less data will be needed and can be from more distant relatives if recombination rates are low. Direct and LD markers require the genotyping of only selection candidates because estimates of genotype effects can be obtained from prior information or from a sample of individuals that have both genotype and phenotype information.
Data from LE markers can be incorporated into BLUP animal model genetic evaluations using the approach of Fernando and Grossman (1989), by fitting random effects for each QTL and allowing for different QTL effects within families. This method, or extensions thereof, has been applied to several commercial situations in dairy cattle (Boichard et al., 2002; Bennewitz et al., 2003; E. Mullaart, personal communication). Application of these procedures requires substantial modification of existing animal model genetic evaluation procedures, estimation of variance components, and extensive computing resources. Data from LD or direct markers on the other hand, can be incorporated in existing genetic evaluation procedures as fixed genotype or haplotype effects (Van Arendonk et al., 1999). If not all animals are genotyped, which will be the case in practice, marker data must be supplemented with genotype probabilities, which can be derived using pedigree and marker data (e.g., Israel and Weller, 2002). Nevertheless, computational requirements for incorporating LD or direct markers in genetic evaluation are much less than for LE markers. Genetic evaluation requirements are slightly greater for LD than for direct markers because LD markers require identification and analysis of marker haplotypes and confirmation of marker-QTL linkage phases.
Whereas the previous refers to requirements for a given QTL, LE-MAS allows for genome-wide analysis and evaluation of QTL with a limited number of markers. This is also possible for LD-MAS with high-density genotyping. Meuwissen et al. (2001) demonstrated that EBV of high accuracy could be obtained based on a Bayesian mixed-model analysis of marker haplotypes with high-density genotyping and phenotyping of a limited number of individuals. Costs of genotyping limits the application of high-density genotyping at present, but these are expected to decrease in the future.
Implementation Logistics.
Because of the greater requirements for phenotyping, genotyping, genetic evaluation, and within-family selection, the logistical demands for implementation of MAS are also considerably greater for LE than for LD or direct markers. Logistical problems associated with implementation of LE-MAS were described previously and has led several commercial programs focusing on the use of LD or direct instead of LE markers Spelman, 2002; Plastow, 2003; M. Lohuis, personal communication).
Genetic Gain.
Opportunities for increases in genetic gain through MAS on a given QTL differ depending on whether the QTL is marked by LE, LD, or direct markers. Villanueva et al. (2002) showed that even when all individuals in the population are phenotyped and genotyped, extra genetic gains from MAS are lower for LE markers than for direct markers. The difference is caused by the accuracy of estimates of the molecular score, which is lower for LE markers because of the limited information that is available to estimate effects on a within-family basis, whereas for direct markers, effects are estimated from data across families. Differences were reduced but far from eliminated when marker spacing was reduced to 1 or even 0.05 cM. Adding prior data to QTL effect estimates resulted in nearly equivalent gains for LE and direct markers, indicating that accuracy of molecular scores was the causative factor (Villanueva et al., 2002). Prior data could come from previous generations if marker–QTL distances are short. Greater differences between the two types of markers are expected if phenotypic and/or genotypic data is not available on all individuals, which will limit the accuracy of molecular scores based on LE markers for individuals in families with limited data, in particular if marker-QTL distances are considerable.
The LD markers also enable use of phenotypic and genotypic data across families to estimate marker scores but accuracies may be slightly lower than for direct markers as a result of incomplete marker–QTL LD and a greater number of effects that must be estimated. Hayes et al. (2001) found that haplotypes of 4 and 11 markers in a 10-cM region that captures the QTL were associated with 64 and 98% of the QTL variance for levels of LD that may be expected in livestock populations. Accuracy of estimates of molecular scores based on data from 1,000 individuals were 0.66 and 0.79 for haplotypes of 4 and 11 markers. Increasing the number of markers from 4 to 11 increased accuracy, but to a greater degree if more progeny were evaluated. Increasing the number of markers increases the extent of LD between the haplotype and the QTL, which increases accuracy, but also increases the number of effects to be estimated, which decreases accuracy (Hayes et al., 2001). The latter is less important if the number of individuals evaluated is greater.
Commercialization.
Final considerations regarding the use of LE vs. LD vs. direct markers involve opportunities for marketing and protection. A detailed discussion of intellectual property issues related to molecular genetics in livestock is found in Rothschild and Newman (2002). It is clear that opportunities for intellectual property protection through patents are greatest for direct markers, substantial for LD markers (especially if based on candidate genes), and limited for LE markers. Direct markers and, to a lesser degree, LD markers for candidate genes, also enable product differentiation in the market based on presence or absence of specific genotypes. These opportunities are again nearly absent for LE markers because knowledge on identity of the QTL is limited.
|
Integration of Marker-Assisted Selection in Breeding Programs |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
and 3. Implementation of MAS requires development and integration of procedures and logistics for DNA collection and storage, genotyping and storage, and for data analysis (Figure 2). This must be supported by a systematic approach to quality control and must support day-to-day decision making (e.g., on which animals to genotype or regenotype in case of errors, which animals to phenotype, etc.).
|
|
). The latter will be mostly driven by genetic gain but are ultimately determined by economics.
Economic Aspects of MAS
Commercial application of MAS requires careful consideration of economic aspects and business risks. Economic analysis of MAS requires a comprehensive approach that aims to evaluate the economic feasibility and optimal implementation of MAS. An excellent example of such an analysis is in Hayes and Goddard (2003), who conducted a comprehensive economic analysis of the implementation of LE-MAS in the nucleus breeding program of an integrated pig production enterprise. Detection of QTL and MAS on identified QTL regions for a multitrait breeding goal and associated genotyping costs and extra returns from the production phase of the integrated enterprise were considered in the economic assessment. They concluded that implementation of LE-MAS was feasible for the assumed cost and price parameters. They also found that, in particular if QTL detection was based on small sample sizes, stringent thresholds should be set during the QTL detection phase such that genotyping costs during the implementation phase are reduced and selection of false positives is minimized. An economic analysis of introgression of the Booroola gene into dairy sheep breeds is given in Gootwine et al. (2001) and of MAS preselection in dairy cattle in Brascamp et al. (1993), Mackinnon and Georges (1998), and Spelman and Garrick (1998).
Whereas Hayes and Goddard (2003) evaluated economic returns from MAS from increased profit at the production level, which is proportional to extra genetic gain, most commercial breeding programs derive profit from increased market share of breeding stock or germplasm. In general, implementation of MAS will have a greater impact on market share than on genetic gain. An example is given in Figure 4, which evaluates the effect of GAS preselection of young dairy bulls in a competitive market. A deterministic model of a mixture of two normal distributions to represent sons that received alternate QTL alleles from their heterozygous sire was used. Extra response from preselection of sons from heterozygous sires depends on the variation that is still present among selection candidates for polygenic EBV for the overall selection criterion, which is based on pedigree information only. If stringent selection on EBV has been applied to bull dams and bull sires, this variation will be limited and the accuracy of polygenic EBV to further differentiate selection candidates will be small (Dekkers, 1992). In Figure 4, correlations estimated with true polygenic breeding values among young bulls of 0.0 and 0.1 are evaluated and preselection is on an index of the MS and polygenic EBV.
|
). Mean EBV and market share increased with effect of the QTL and decreased with increasing accuracy of the polygenic EBV at the time of preselection. Effects on market share were, however, always greater than effects on mean EBV. This does not imply that the economic feasibility of MAS is greater in a competitive market than when returns are derived from commercial production, as was evaluated by Hayes and Goddard (2003). Economic feasibility not only depends on the proportional increase in the objective, but also on the absolute returns associated with a percentage increase in genetic gain vs. market share; in fact, Brascamp et al. (1993) showed that economic returns from increased market share were less than from increased production for a preselection situation similar to that considered here. Nevertheless, it is important that economic analysis is conducted in relation to business and market realities and goals. Computational approaches using genetic algorithms (e.g., Kinghorn et al., 2002) can be used to develop selection and mating strategies based on multiple sources of information, including markers that meet multiple business goals and constraints. Weller (1994) provides further discussion of alternative criteria to economically evaluate alternative breeding programs.
Other Opportunities
Optimal implementation of MAS involves careful consideration of alternative selection strategies, business goals, and integration of molecular with other technologies (e.g., reproductive technologies following Georges and Massey, 1991). Opportunities also exist to implement LD-MAS in synthetic lines, capitalizing on the extensive disequilibrium that exists in crosses and their power to detect QTL (Zhang and Smith, 1992). In addition, strategies must be developed to estimate gene effects at the commercial level for nucleus breeding programs, in particular if they involve crossbreeding. This also opens opportunities to use markers to capitalize on nonadditive effects and assignment of specific matings.
Genetic markers can also be used to control inbreeding, parental verification, and product tracing. Pedigree verification is an important aspect of the use of molecular markers in several breeding programs (e.g., Spelman, 2002; M. Cowan, personal communication) and can lead to substantial opportunities for increasing accuracy of EBV and genetic gain (e.g., Van Arendonk et al., 1998; Israel and Weller, 2000), but these are beyond the focus of this article. The use of markers for product tracing has been implemented in some industries (Plastow, 2003).
|
Implications |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
|
Footnotes |
|---|
2 This article was presented at the 2003 Joint ADSA-ASAS-AMPA meeting as part of the Breeding and Genetics symposium "Molecular Genetics."
3 Correspondence: 225C Kildee Hall (phone: 515-294-7509; fax: 515-294-9150; e-mail: jdekkers{at}iastate.edu).
Received for publication October 16, 2003. Accepted for publication February 11, 2004.
|
Literature Cited |
|---|
|
TopAbstractIntroductionPrinciples of Marker-Assisted...Industry Application of Marker...LE vs. LD vs....Integration of Marker-Assisted...ImplicationsLiterature Cited |
|---|
g). Selection is for a quantitative trait with selected proportions 10 and 25%, and accuracies of polygenic EBV of 0.8 and 0.5 for sires and dams, respectively.
