| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Animal Breeding and Genetics Group, Wageningen University, PO Box 338, 6700 AH Wageningen, The Netherlands
2 Holland Genetics, CR Delta, PO Box 5073, 6802 EB Arnhem, The Netherlands
Corresponding author: Chris Schrooten; e-mail: schrooten.c{at}hg.nl.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Closed nucleus breeding schemes in dairy cattle that use information on QTL were evaluated by deterministic simulation. In the base scheme, the selection index consisted of pedigree information and own performance (dams), or pedigree information and performance of 100 daughters (sires). In alternative breeding schemes, information on a QTL was accounted for by simulating an additional index trait. The fraction of the variance explained by the QTL determined the correlation between the additional index trait and the breeding goal trait. Response in progeny test schemes relative to a base breeding scheme without QTL information ranged from +4.5% (QTL explaining 5% of the additive genetic variance) to +21.2% (QTL explaining 50% of the additive genetic variance). A QTL explaining 5% of the additive genetic variance allowed a 35% reduction in the number of progeny tested bulls, while maintaining genetic response at the level of the base scheme. Genetic progress was up to 31.3% higher for schemes with increased embryo production and selection of embryos based on QTL information. The challenge for breeding organizations is to find the optimum breeding program with regard to additional genetic progress and additional (or reduced) cost.
Key Words: genetic evaluation • marker-assisted selection • quantitative trait locus • dairy cattle
Abbreviation key: MAS = marker-assisted selection.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
To map QTL, genetic markers have been used. Markers for which linkage to specific QTL has been established can be used for marker-assisted selection (MAS). They give information on the inheritance of the QTL alleles and thus give information about the genetic potential of an animal. This information can be used to increase genetic progress, either by increasing the reliability of EBV, by reduction of the generation interval, or by increasing the selection intensity.
Current dairy cattle breeding schemes rely on progeny testing to assess the genetic value of bulls. Progeny testing increases the generation interval, but this shortcoming is compensated for by an increase in accuracy. Progeny testing would not be necessary if markers were available that explained a substantial fraction of the genetic variance. This ideal will not be realized in the near future, however, because breeding goals in dairy cattle consist of many traits. The MAS of sires will therefore most likely be applied as an initial selection step in a multistage selection setting. Marker information will increase the accuracy of selection, especially in the first stage. Application of MAS in the initial stage can be used to increase rate of gain through preselection or to decrease costs, by progeny testing of fewer young bulls while maintaining the rate of gain.
Soller and Beckmann (1982) suggested screening of sires to be progeny tested, based on genetic polymorphisms. Since then, various studies have evaluated a range of breeding schemes with MAS. Some of these studies looked at 2-stage selection schemes where candidate young bulls were selected based on pedigree and marker information (e.g., Kashi et al., 1990; Mackinnon and Georges, 1998; Spelman and Garrick, 1998). Another group of studies looked at selection in a single stage, combining all available information, including marker information (e.g., Meuwissen and van Arendonk, 1992; Meuwissen and Goddard, 1996), and allowing selection of parents across all age classes. Estimates for additional gain ranged from –6% (Spelman and Garrick, 1997) to +105% (Spelman et al., 1999). Profitability of MAS breeding schemes depends on several factors, such as the fraction of the genetic variation explained by the QTL, the frequencies of QTL alleles, the design of the breeding scheme, and the time horizon of evaluation.
Genotyping techniques can be applied to newborn animals, but they can also be applied to embryos (Bredbacka, 2001). Selection at the embryo stage offers additional opportunities, because the number of implanted embryos can be reduced, compared with a scheme based on selection of calves. To date, these opportunities have not been explored in detail.
Marker-assisted selection can be part of multistage selection: markers are typed early in life and used in the first stage of selection, for example, in combination with pedigree information. In the second stage, the remaining selection candidates are selected on a combination of first-stage information, and phenotypic information that has come available in the interim. The aim of this study was to evaluate dairy cattle breeding schemes regarding additional genetic response with MAS, and regarding possible changes to the structure of the breeding program when MAS is implemented in the breeding program.
Because dairy cattle breeding schemes are set up as multistage breeding schemes, and MAS is likely to be applied in such a multistage setting, this study examined 1) general aspects of multistage selection; and then addressed 2) the potential increase in genetic progress from MAS of newborn animals; 3) the possibilities to reduce the number of progeny tested young bulls in a MAS scheme while maintaining a constant rate of gain; and 4) the additional genetic progress from MAS in breeding schemes with increased embryo production and genotyping of embryos.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
In a third set of alternatives, the selected fraction in the first stage (p1) was varied from 0.01 to 1.0, while keeping the number of selection candidates and the selected fraction in stage 2 (p2) at a fixed level. Accuracies were equal to accuracies in the second set of alternatives. The third set of alternatives represented situations where the number of candidates available before selection in stage 1 increased with the intensity of first-stage selection. Selected fraction in stage 2 was 0.01, 0.02, 0.05, 0.10, or 0.20.
Responses were computed using an exact method developed by Ducrocq and Colleau (1986), based on principles described by Tallis (1961) and Dutt (1973).
To translate some of the general characteristics of multistage selection to practical breeding schemes, dairy cattle breeding schemes utilizing MAS were simulated.
Breeding Schemes Applying MAS
The response in a number of schemes applying MAS preselection was compared with the response in a breeding scheme with preselection on pedigree information only. The comparison was carried out by deterministic simulation of multistage selection schemes, using the program SelAction (Rutten et al., 2002). This program predicts the rate of genetic gain based on advanced selection index theory. SelAction accounts for reduction in variance due to selection (Bulmer, 1971), and corrects selection intensities for finite population size and for the correlation between index values of family members (Meuwissen, 1991). Full pedigree information as with BLUP selection is accounted for (Villanueva et al., 1993), and the program is therefore an accurate approximation of stochastic simulation with BLUP. Deviations from normality after the first stage of selection are dealt with, and selection response is predicted for the equilibrium situation (Rutten et al., 2002), using the exact method of Tallis (1961) and Ducrocq and Colleau (1986). Features of SelAction and the theoretical background are described by Rutten and Bijma (2001a,b) and by Rutten et al. (2002).
Base breeding program.
The breeding program was a closed nucleus scheme, with discrete generations. The breeding goal consisted of one trait, with heritability of 0.35 and phenotypic variance of 100. Selection of males and females in stage 1 was for an index based on pedigree information only. In stage 2, selection in females was for an index based on pedigree information and phenotypic information of the candidate and its half sibs and full sibs. In males, selection in stage 2 was for an index based on pedigree information, progeny information (100 daughters in the population), and sib information. This situation is comparable to selection for a milk production trait in dairy cattle. Each generation, 20 sires and 200 dams were selected to produce 4000 embryos. Half of these embryos resulted in live calves, with a sex ratio of 0.50. In stage 1, 200 males were selected out of 1000 to enter the progeny test. In stage 2, 20 males were selected out of 200 to produce the next generation. Selection in females reduced numbers from 1000 to 700 in stage 1 and from 700 to 200 in stage 2. Involuntary culling and death were ignored. The base breeding scheme is summarized in Figure 1
.
|
and
, respectively, where q is the fraction of the genetic variance explained by the QTL. Phenotypic correlations of the breeding goal trait with the QTL and polygenic component were 
and
, respectively, where 
is the heritability of the overall trait. The heritability of the polygenic q component was 
Schemes were simulated with a fixed number of candidates, or with an increased number of candidates before preselection (relative to the base population).
Fixed number of candidates.
The number of candidates before selection in stage 1 and the number of selected animals were fixed. Selection in stage 1 was on an index with information on parents (BLUP breeding values) and information on the QTL. In stage 2, information on sibs and own performance was added to the index for females, and information on sibs and progeny was added to the index for males.
Preselection in males.
The level of preselection in bulls was varied from p1,m = 0.20 (preselection equal to preselection in the base scheme) down to p1,m = 0.02 (all selection in stage 1), with steps of 0.01. In this way, the level of preselection where genetic progress was equal to genetic progress in the base scheme, could be determined. The selected fractions for the females in stage 1 and stage 2 were equal to selected fractions in the base scheme. For the scheme with all selection in stage 1 (p1,m = 0.02), generation intervals were reduced, because there was no need to wait for progeny test results. In comparing results for this scheme with other schemes, genetic progress was corrected for the difference in generation intervals. The generation interval was assumed to be 3 in the scheme with p1,m = 0.02, and 5 in all other schemes. Schemes are referred to as MQ05, MQ10, MQ20, and MQ50, where M indicates varying selection in males.
Preselection in females.
The advantage of additional information is expected to be different for dams and sires. To quantify this difference, schemes were simulated where the level of preselection in females was varied from p1,f = 0.70 (equal to preselection in the base scheme) down to p1,f = 0.20 (all selection in the first stage), with steps of 0.05. The selected fractions for the males in stage 1 and stage 2 were equal to the selected fractions in the base scheme. These schemes are referred to as FQ05, FQ10, FQ20, and FQ50, where F indicates varying selection in females.
Increased numbers of candidates.
For the previously mentioned comparisons, the number of embryos (i.e., number of candidates before selection) was kept constant. However, application of genotyping techniques on embryos offers additional opportunities for preselection without increasing the number of transferred embryos. This practice was modeled by simulating breeding schemes with increased reproductive rates, i.e., 8000 instead of 4000 embryos were produced by 20 sires and 200 dams. The number of progeny tested bulls was kept constant. Therefore, selection in the first stage was more intense, i.e., p1,m = 0.10 for males, and p1,f = 0.35 for females. These schemes are referred to as EQ05, EQ10, EQ20, and EQ50, where E indicates that extra embryos were produced.
Overview.
The general characteristics of the schemes are summarized in Table 1. Results of these schemes were compared with a corresponding base scheme with the same characteristics, but without a QTL. The value of the QTL information was assessed in 2 ways: 1) the increase in genetic gain; 2) the reduction in the number of sires required to be progeny tested to achieve the same rate of gain as in the base scheme.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
shows the response in multistage selection schemes with a high ratio between accuracies in stage 1 and stage 2 (i.e., a relatively high accuracy in stage 1). The response is expressed as a fraction of the response obtained when all selection was in stage 2 (R0), which is the maximum response. Note that the horizontal axis contains values for 1 – p1, i.e., larger values indicate stronger selection in stage 1.
|
, when the overall selected fraction was 0.01, response dropped below 95% when the selected fraction in stage 1 equaled 0.05 (1 – p1 = 0.95). This means that 95% of the candidates did not need to be retained until the second selection step, with only 5% loss in response. When the overall selected fraction was 0.20, response dropped below 95% when the selected fraction in stage 1 equaled 0.42 (1 – p1 = 0.58, Figure 2).
The reduction in response depends on the ratio between accuracies of the selection index in stage 1 and stage 2. This can be seen by comparing results in Figure 2 with results in Figure 3. The ratio of accuracies of the selection index in stage 1 and stage 2 was 0.69 in Figure 2 and 0.22 in Figure 3, respectively. In Figure 3, a reduction of 5% of the response for overall selected fractions of 0.01 and 0.20 was obtained for selected fractions in stage 1 of 0.44 (1 – p1 = 0.56) and 0.78 (1 – p1 = 0.22), respectively. The difference between Figures 2 and 3 depends on the ratio of accuracies, and not on the absolute level of accuracies (Tallis, 1961).
|
and 3, the selected fractions in stage 1 and stage 2 were varied, keeping the overall selected fraction constant. In Figure 4, selection in stage 2 was kept at the original level, and the number of selection candidates at the beginning of the selection process was increased. This resulted in increased overall selection intensity. When the selected fraction was 0.01 (selection in one stage), the increase in response was 8.5% when an additional preselection step was introduced with a selected fraction of 0.50 (Figure 4). For example, initially, 10 out of 1000 animals were selected in one stage. With preselection, 1000 out of 2000 animals were selected in stage 1, and 10 out of 1000 in stage 2. The increase in response was 24.1% when preselection with a selected fraction of 0.50 was introduced in a scheme with a selected fraction of 0.20 (Figure 4). Thus, when the selected fraction was already small, increase in response was relatively small when a preselection step was introduced. In the latter case (i.e., scheme with selected fraction 0.20), the increase in overall selection intensity was relatively higher than in the first case (scheme with selected fraction 0.01).
|
to 4 show that
and 3). and 3). and 3). ).
MAS Breeding Schemes with Varying Preselection in Males
Figure 5 shows genetic response per generation relative to response in the base scheme for varying levels of preselection in bulls and various sizes of the QTL effect. Information on the QTL was available in both males and females, but preselection was applied only in the males. When the QTL explained 5% of the genetic variance (MQ05), genetic progress was 4.5% higher for selected fractions equal to that of the base scheme (p1,m = 0.20). For MQ10, MQ20, and MQ50, genetic response per generation was 7.7, 12.5, and 21.2% higher, respectively (Figure 5).
|
, increasing the selection intensity in stage 1 while decreasing selection intensity in stage 2 led to decreased genetic progress. However, the decrease was small when halving the number of progeny tested animals (i.e., p1,m decreased from 0.20 to 0.10), especially for large QTL. Genetic progress compared with the base scheme was –3.9, +0.3, +6.7, and +19.0%, for MQ05, MQ10, MQ20, and MQ50, respectively (Table 2, Figure 5), when the selected fraction in stage 1 was 0.10. Generation interval in males and females was assumed to be 5 for all values of p1,m, except for p1,m = 0.02, where the generation interval was assumed to be 3. Due to the different generation interval when p1,m = 0.02 (all selection in males was applied in stage 1, i.e., no progeny testing of young bulls was needed), these results deviated from the curve describing all other results (Figure 5). Responses for the schemes without progeny testing ranged from –1.7% for MQ05 to +68.3% for MQ50, relative to the base scheme.
|
and 3, respectively, for an overall selected fraction of 0.02. For the schemes in Figure 2 and Figure 3, the ratios between accuracies in stage 1 and stage 2 were 0.69 and 0.22, respectively. The ratio between accuracies in stage 1 and stage 2 was 0.72 for MQ50 and 0.31 for MQ05 (Table 2). Table 2 shows genetic progress at 2 levels of preselection in males (p1,m = 0.20 and p1,m = 0.10). Also, the level of preselection resulting in responses equal to response in the base scheme is listed, and the contribution of females to genetic progress is shown.
For MQ05, genetic response was equal to response in the base scheme when the selected fraction in stage 1 was 0.13 (Table 2). This means that progeny testing 130 bulls in a MAS scheme with a QTL explaining 5% of the genetic variance would yield genetic response equal to that of a nonMAS scheme where 200 bulls were progeny tested. For MQ10 and MQ20, response was equal to response in the base scheme when selected fractions in stage 1 were approximately 0.10 and 0.06, respectively, corresponding to 100 and 60 progeny tested bulls. Genetic response for MQ50 was always higher than genetic response in the base scheme.
In the base scheme, 28.0% of genetic progress was contributed by selection in females. When information on a QTL was used, the contribution of selection in females increased slightly. For MQ05, MQ10, MQ20, and MQ50, the contribution of selection in females was 28.2, 28.5, 29.1, and 31.6% respectively, when selected fractions were equal to selected fractions in the base scheme. The addition of QTL information to the selection index led to increased accuracy of selection in stage 1 and stage 2. The increase in accuracy in stage 1 was equal for males and females, whereas the increase in accuracy in stage 2 was relatively higher for females than for males. This resulted in a relatively higher contribution of females to the genetic progress for schemes with a QTL explaining a higher percentage of the additive genetic variance.
MAS Breeding Schemes with Varying Preselection in Females
Figure 6 shows genetic response relative to the base scheme when preselection in females was varied while keeping the selected fraction in stage 1 of males at 0.20. Information on the QTL was available in both males and females, but preselection was applied only in the females. This represents breeding schemes where preselection leads to fewer females that need to be performance tested, to reduce the size of the nucleus and reduce the costs of recording. Curves are relatively flat, which indicates that genetic response was only slightly decreased with increased preselection of females. The ratio between accuracies in stage 1 and stage 2 was relatively high, ranging from 0.50 for alternative FQ05 to 0.91 for alternative FQ50. Consequently, a relatively flat curve was expected, based on the general characteristics of multistage selection. With a selected fraction in stage 1 of 0.35 instead of 0.70 (i.e., halving the size of the nucleus), the genetic progress compared with the base scheme was –0.1, +4.2, +10.3, and +21.0%, for FQ05, FQ10, FQ20, and FQ50, respectively. When the selected fraction in stage 1 of males was 0.10 instead of 0.20, curves looked similar, but genetic progress was at a lower level. When p1,f in females was 0.35, the genetic progress compared with the base scheme was –8.5, –3.3, +4.5, and +18.8%, for FQ05, FQ10, FQ20, and FQ50, respectively.
|
).
|
). To obtain responses equal to a base scheme with p1,m = 0.10, the number of progeny-tested young bulls could be reduced from 200 in the base scheme to 120, 80, or 40 in schemes EQ05, EQ10, and EQ20, respectively. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Information sources for the index in stage 1 were included in the index for stage 2 as well, resulting in a correlation between index 1 and 2. This correlation is equal to the ratio of the accuracies of index 1 and index 2 (Saxton, 1983). Xie and Xu (1998) compared the efficiency of multistage MAS with efficiency of conventional phenotypic selection. They applied a similar procedure, but to avoid numerical multiple integration, they imposed a constraint on the solutions for multistage MAS by forcing the covariance between the index in stage 1 and stage 2 to be 0. This has an effect similar to that of a restricted index, resulting in less than optimal gain. They did not account for the effect of finite population size on selection intensities and for reduction in variance due to selection. The program SelAction (used here) accounts for these effects, and constructs an optimum index using full pedigree information with an animal model (Villanueva et al., 1993).
Schemes for MAS tend to have high responses initially, due to strong emphasis on marked genetic variance (e.g., Spelman and Garrick, 1997). Depending on initial allele frequencies, the effect of the QTL, and selection intensity on the QTL, the variation due to the QTL decreases after the initial generations of MAS but decreases at a slower rate in later generations. This reduction in QTL variance leads to decreasing marginal response over the base scheme without QTL information across time. In our study, we calculated equilibrium rates of response, assuming an infinitesimal model (or a relatively large number of small QTL). Therefore, it was not possible to evaluate response on a per year basis, and to look at differences between short-term response and long-term response, as discussed by Gibson (1994). The response shortly after the implementation of MAS was not predicted, but this response is expected to be higher than the equilibrium response presented here. According to this line of reasoning, response in later generations may be overestimated, unless new QTL are detected continuously.
In this study, the QTL was modeled as a trait correlated with the breeding goal trait. We assumed that the QTL behaves as a polygenic trait that remains normally distributed over time. This assumption is probably only valid if the QTL is, in fact, not one QTL, but several unlinked QTL, where each QTL has multiple alleles. The continuous detection of new QTL over time would make the assumption more realistic, because it would allow for a certain fraction of the genetic variance to be explained by QTL at any point in time.
The simulation model ignores inbreeding (Rutten et al., 2002). Ideally, schemes with equal rates of inbreeding should be compared. The rate of inbreeding is proportional to the square of the long-term genetic contributions of selected animals. This latter term is likely to differ between the schemes evaluated in this study, resulting in different rates of inbreeding. It is a considerable challenge, however, to determine the expected rate of inbreeding for multistage selection schemes that are simulated deterministically.
Genetic Response in MAS Schemes
Several authors have studied MAS schemes for dairy cattle. Responses from MAS obtained in these studies varied widely, depending on size of the QTL, type of breeding scheme, and the way of calculating response, e.g., equilibrium rates of response or response from one round of selection, response in young bulls or response in population females. In the following section, results from a number of MAS-studies that looked at preselection will be discussed.
The use of markers to select young bulls was first suggested by Soller and Beckmann (1982). Extra gain in their scheme was estimated (Smith and Simpson, 1986) at 46% for direct selection on 20 QTL explaining 50% of the genetic variance. Kashi et al. (1990) reported an increase of up to 26% in genetic merit of bulls entering the progeny test, resulting from one round of marker-assisted selection of candidate bulls before entering the progeny test. In this case, selection was based on haplotypes of polyallelic markers associated with 20 segregating QTL explaining all the genetic variance. They did not account for the loss in genetic variation as a result of MAS. Mackinnon and Georges (1998) reevaluated the approach by Kashi et al. (1990). They found an 8% increase in response in young bulls from one round of selection in a bottom-up scheme when the QTL explained 10% of the genetic variance. For a QTL explaining 50% of the genetic variance, increase in response was 23%. In the current study, equilibrium response increased by 7.7% and 21.2% for QTL explaining 10% and 50% of the genetic variance, respectively, which is in the same range as the response obtained by Mackinnon and Georges (1998). Spelman and Garrick (1998) found a 9% increase in rate of gain from one round of MAS in a combination of a top-down and a bottom-up scheme (Mackinnon and Georges, 1998). The QTL explained 16.7% of the genetic variance, and bull dams produced 40 progeny. Only one bull per full-sib group was selected. The QTL information was used on both the dam and the sire selection paths, in a population that had been selected for 1 to 2 generations. In the current study, 5 male progeny were produced per bull dam and selection was across families, not distinguishing between full sibs and unrelated animals. When the QTL explained 10 or 20% of the genetic variance, equilibrium response increased by 7.7 and 12.5%, respectively, which is comparable to the response observed by Spelman and Garrick (1998).
Implications for the Breeding Program
Results in Figures 5 and 6 indicate, that increased accuracy of selection in stage 1 facilitates a reduction of the number of animals entering into progeny or performance testing. For example, a QTL explaining 5% of the genetic variance would allow a reduction from 200 to 130 progeny-tested young bulls without affecting the genetic response. Doubling the number of embryos produced in combination with preselection for a QTL explaining 5% of the genetic variance would require the progeny testing of only 80 young bulls to obtain the genetic response of the base breeding scheme. Figure 5 also shows that schemes without progeny testing would be more beneficial than the base breeding scheme for QTL explaining from approximately 6% of the genetic variance and over, especially due to the huge decrease in the generation interval from 5 to 3 units.
In the studied alternatives, the intensity of preselection was increased in one of the sexes to look at differences between the effects of increased preselection in either sex. However, increased preselection may be applied in both sexes. Additional calculations (results not shown) on a few situations with increased preselection in both sexes were carried out, to look at implications for the potential reduction in progeny-tested bulls. For example, when the selected fraction for females in stage 1 was 0.40 instead of 0.70, the number of progeny-tested bulls could be reduced only to 180 instead of 130, to have the same genetic progress as in the base scheme.
These examples demonstrate that QTL information can be used in various ways, and the costs and benefits differ accordingly. Depending on which scheme is adopted, a scheme utilizing QTL information will require additional investment for production of extra embryos, collection of DNA, genotyping of embryos or newborn animals and their relatives, etc. Cost savings can be made with regard to cost for collection of information on progeny test daughters (labor, incentives paid to farmers), variable cost for buying, feeding, and housing of young and waiting bulls, fixed cost for housing (when the number of progeny-tested bulls is reduced substantially, less housing facilities are needed), etc. The overall selection intensity and ratio of accuracies in the various stages of selection determine the "cost" of increasing selection intensity in the first stage, as was concluded from the study on the general characteristics of multistage selection. The question of whether it is beneficial to implement MAS or not depends on the net result of additional cost, savings, and benefits from increased genetic progress of the chosen breeding program.
The QTL information can be measured accurately on newborn animals or on embryos (Bredbacka, 2001), which offers good opportunities for preselection and reduction of the number of progeny-tested or performance-tested animals by MAS. Schemes with increased selection intensity for males in stage 1 seem most promising, because reducing the number of progeny-tested young bulls can reduce the cost of the breeding program considerably. Comparing Figure 6 with Figure 5, a relatively higher reduction in the number of performance-tested females would be possible, but this is likely to have less impact on cost reduction.
Breeding goals in dairy cattle consist of a variety of traits. Although QTL have been mapped for quite a few of these traits, in some cases explaining approximately 50% of the genetic variation of a particular trait (e.g., Grisart et al., 2002), the fraction of the variation of the overall breeding goal explained by QTL is likely to be moderate. For breeding schemes in the current study, the fraction of the genetic variance in the breeding goal explained by the QTL was assumed to be 5% or higher. Even for the QTL explaining 5% of the genetic variance, there was a considerable impact of marker-assisted selection on the genetic response.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication June 19, 2004. Accepted for publication December 6, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Duangjinda, D. Buayai, V. Pattarajinda, Y. Phasuk, S. Katawatin, T. Vongpralub, and A. Chaiyotvittayakul Detection of bovine leukocyte antigen DRB3 alleles as candidate markers for clinical mastitis resistance in Holstein x Zebu J Anim Sci, February 1, 2009; 87(2): 469 - 476. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Neuner, R. Emmerling, G. Thaller, and K.-U. Gotz Strategies for Estimating Genetic Parameters in Marker-Assisted Best Linear Unbiased Predictor Models in Dairy Cattle J Dairy Sci, November 1, 2008; 91(11): 4344 - 4354. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Ansari-Mahyari, A. C. Sorensen, M. S. Lund, H. Thomsen, and P. Berg Across-Family Marker-Assisted Selection Using Selective Genotyping Strategies in Dairy Cattle Breeding Schemes J Dairy Sci, April 1, 2008; 91(4): 1628 - 1639. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. M. Dekkers Marker-assisted selection for commercial crossbred performance J Anim Sci, September 1, 2007; 85(9): 2104 - 2114. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. F. Kearney, P. R. Amer, and B. Villanueva Cumulative Discounted Expressions of Sire Genotypes for the Complex Vertebral Malformation and {beta}-Casein Loci in Commercial Dairy Herds J Dairy Sci, December 1, 2005; 88(12): 4426 - 4433. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
