Deliverables
The objective was to search for QTL containing regions that were identical by descent (IBD) among the pure breed resource populations (RP1, RP2, RP4, RP5). Toward this goal, we looked at chromosomal haplotypes in RP1 for the six regions chosen for intensive analysis by the BovMAS consortium. This led to analysis of linkage disequilibrium (LD) within BTA13 (chosen by the consortium for high resolution mapping studies) across all four purebred RPs.
The assumption was that high LD might identify regions of IBD. We also developed an efficient methodology: co-segregant pools analysis for determining sire haplotypes. Chromosomal haplotypes. Chromosome haplotypes for six QTLR regions chosen for more intensive analysis by the consortium, were constructed for 27 RP1 sires. In many instances the relationship among individuals sharing the same chromosomal haplotype, was not close; and some haplotypes were shared by 4, 5, and in one case 7 individuals.
This indicated the possibility of more extensive IBD in the studied regions. LD studies of BTA13. Materials and Methods: 411 individuals from RP 1, 4 and 5 were genotyped at 19 microsatellite markers distributed over BTA13; 273 individuals from RP2 were genotyped at 35 markers distributed over the same region (including 18 of above 19 markers). Haplotypes were constructed by the Simwalk2 and PowerMarker software. LD measures, calculated separately for each RP were: D, Monte Carlo approximation of Fisher exact test, and the standardized c2.
The c2 measure indicates the potential usefulness of marker-QTL LD as a means of selection. Results: Mean values for D' (known to give artefactually high values in multi-allelic situations) across the entire target region were 0.314, 0.249, 0.281, and 0.487 for RP1, 2, 4, 5, respectively; mean values of c2 were much less (0.124, 0.062, 0.096 and 0.092, respectively). LD decreased with increasing distance between loci according to the accepted expression LD = 1/(1+4bd), where d = distance in Morgans between loci, and b is a regression coefficient that relates to effective population size. For RP1, 2, 4, 5, the b coefficients were 25.6, 86.3, 28.0 and 45.1, respectively.
This is as expected, given the population structure and the intensive one-way selection of the two Holstein populations (RP1,4), as compared to the Brown Swiss (RP5) (selected for dual purpose till 40 years ago) and the dual-purpose selected Simmental (RP2). To search for a selection signature, a moving average of c2, taken 5 markers at a time along the chromosome, was calculated. All populations presented a maximum in the region from 35 to 50 cM, indicating a possible selection signature. Although the values obtained for c2 over 0-5 cM (0.193, 0.109, 0.218 and 0.133, for RP1, 2, 4, 5) were larger than the average values, they were still much too small to be a useful tool for selection. LD values between markers separated by 50 cM or more distributed according to chance expectations; there were no indications of long range LD.
Haplotype identification by co-segregant DNA pools. Haplotype identification of individual dairy sires is essential for interval mapping, and for tracing identified QTL in pedigrees for MAS. Parental genotypes are effective in determining sire haplotypes for informative markers (where sire and parent do not share the same genotype). Progeny genotypes can resolve all markers, but many daughters must be genotyped. In co-segregant pool analysis, progeny of the sire are divided into two groups according to the transmitted sire allele at an "index" marker in the haplotype.
DNA of the two marker allele groups is separately pooled. Markers belonging to one of the sire haplotype centered at the index markers will show high frequency in one of the pools, those belonging to the other haplotype will show high frequency in the alternative pool. In this way, a series of linked markers alleles can be rapidly assigned to the same haplotype. Materials and Methods.
These approaches were compared in a data set consisting of semen samples of 8 RP1 sires and their sires, and milk samples of their daughters. Results: Almost two-thirds of sire markers could be ordered in haplotypes on the basis of the genotypes of their sires. Daughter genotypes were able to resolve all markers, but this required about 50 daughter genotypes. Co-segregant pools were able to resolve sire haplotypes over regions of up to 30 cM to either side of the index marker.
Thus, co-segregant pools can serve as an efficient means for haplotype identification. Conclusions: The LD findings (low LD even across 5 cM wide distances, absence of long distance LD) are important for groups attempting to achieve high resolution LD mapping in dairy cattle or to use LD for MAS. Co-segregant pools can be useful to groups involved in interval and LD mapping, and in MAS of dairy cattle and other animal populations.
The objective was to examine the feasibility of tracing chromosomal regions containing a QTL (QTLR) from a grandsire for whom marker-QTL linkage phase has been established, to progeny and grandprogeny, i.e., to a son or grandson. The main question considered was the number of markers that need to be followed to achieve accurate tracing. For this, we had three data sets. The first consisted of DNA samples of three pedigrees of RP1 (Pedigrees A, B, and C) each consisting of a grandsire, sire and 1, 13, and 3 grandsons, respectively. The second consisted of DNA samples of eight pedigrees of RP4 consisting of 8 grandsires, sires, 33 sons, and 117 grandsons.
The grandsons in all cases, were young recently purchased male calves, candidates for progeny testing. For these pedigrees, the specific objective was to trace chromosomal regions that carry QTL alleles with positive or negative effects on milk protein percent, from the grandsire (S0) (having known marker-QTL linkage phase) to his sons (S1) and grandsons (S2).
In addition, genotyping data were available for a large number of three generation male-line pedigrees in RP1, each ending in a set of granddaughters (genetically equivalent to grandsons). These were used for more extensive analyses of tracing success as a function of the width of QTLR traced, and the number of markers used for tracing.
Pedigree data, RP1: Materials and Methods: For purposes of tracing we examined a total of 18 markers in a 40cM region of BTA13 running from 39.6 to 96.0cM, but not all markers were genotyped in all individuals. Haplotypes were derived manually from the genotypes of the individuals in the table, with some cross referencing within half-sib families, and also with the program PowerMarker.
Results: Pedigree A: 16 markers were monitored. S0 transmitted a recombinant haplotype to S1, but this was not transmitted to S2. Pedigree B: 9 markers were monitored. S0 transmitted the positive haplotype to S1; this haplotype was transmitted to 3 grandsons (S2), while 3 grandsons did not receive any part of it; five grandsons received recombinant S1 haplotypes containing 1 to 8 marker alleles of S0 haplotype. Pedigree C: 6 markers were monitored. S0 transmitted the negative haplotype to S1; this haplotype was transmitted to three grandsons (S2), one grandson did not receive the haplotype..
Pedigree data, RP4: Materials and Methods. Tracing chromosomal regions containing a QTLR from a grandsire for whom marker-QTL linkage phase has been established, to sire and grandsons, was carried out for four QTLR: one on BTA11, two regions on BTA13, one on BTA14 and one on BTA20.
Results: For BTA11 (3 markers spanning 92.2 to 115.4 cM) the grandsire haplotype was traced for 19 out of 21 sires; for BTA13, region 1 (5 markers spanning 51.7 to 71.1) the grandsire haplotype was traced for 24 out of 27 sires; for BTA13, region 2 (3 markers, spanning 84.4 to 98 cM) the grandsire haplotype was traced for 24 out 30 sires; for BTA14 (6 markers, spanning 60.7 cM to 100 cM) the grandsire haplotype was traced for 20 out of 39 sires, and for BTA20 (5 markers, spanning 8.2 cM to 33.4 cM) the grandsire haplotype was traced for 30 out of 42 sires. Thus, in almost all cases it was possible to trace the QTLR from grandsire to grandson, and identify recombinants in transmission from S0 to S1, or from S1 to S2.
Tracing from grandsire to granddaughter, Materials and Methods: Data were taken from available three generation genotypes for BTA 1, 7, 11 and 13. In each chromosome a number of haplotypes were defined according to QTLR width (0 to 5cM; 5 to 10cM; 10 to 20cM) and markers per haplotype (2 to 8, depending on QTLR width). For haplotypes of width 0 to 5cM, 5 to 10cM and 10 to 20cM, a total of 2008, 2895, and 3860 haplotypes were traced respectively. Results were virtually independent of the width of the haplotypes, namely: for haplotypes defined by 2, 3, 4, and 5 or more markers, proportion of traced haplotypes (including recombinants) were 0.93, 0.96, 0.97, and 0.98, respectively. Thus, these results strongly confirm the ability to trace selected phased QTLR from grandsire to grandson. Conclusions: It is eminently feasible to trace phased QTLR from grandsire to sons and grandsons. Thus, tracing of QTLR should not pose a problem for MAS. This information is useful to organizations that are considering the application of MAS in their breeding programs.
Most of QTL mapping experiments were performed in dairy cattle and thereof most studies focus on QTL in the highly selected purebred Holstein cattle. According to experiences with other domesticated animal and plant species the QTL mapped in one highly selected cosmopolitan breed or line presents only the smaller part of variability of complex traits. In spite of their inferiority for most of yield traits, the local cattle breeds can contribute positively to improve important yield and fitness traits in cosmopolitan breeds and inversely the cosmopolitan breeds can be used for targeted improvement (not replacement) of local breed.
The advanced backcross QTL strategy (AB-QTL) is one of the approaches to map and introgress distinct QTL between lines or breeds of domesticated species. Due to a long generation interval, there are limited possibilities to create a powerful QTL mapping or QTL introgression design in cattle. Over the last 30 years German-Austrian Fleckvieh breeders designed a large advanced backcross population, which is embedded within the Fleckvieh breed.
Our goal was to use this large advanced backcross population for learning about the feasibility of marker assisted introgression between cattle breeds and for retrospective use of collected mapping results for marker assisted accumulation of positive alleles and selection against negative alleles already introgressed by donor Red Holstein (RH) founder. Our mapping design has gone beyond traditional half-sib design to support the following requirements.
QTL mapping design should have high statistical power: large daughter design with more then 40000 sampled daughters.
The genotyping costs in large half-sib design should be acceptable: selective milk DNA pooling. " The origin of active QTL variants should be traceable: synchronic mapping in purebred Fleckvieh (FV, recipient) and advanced backcross Fleckvieh × Red Holstein (ABFV) families combined with haplotyping of all family-sires, all available ancestors including donor founder as well as most important founders of recipient population. By genome wide QTL mapping for Milk Yield (MY) and milk Protein Percentage (PP) in both FV and ABFV we detected 31 QTL distributed across 26 chromosomes. The genome wide genotyping and haplotyping of 69 animals including family-sires in both FV and ABFV, ancestors of family-sires and founder of both resource populations we made most QTL alleles traceable.
Most of QTL alleles are of FV origin: QTL segregate only in FV families or in FV and ABFV families but segregating ABFV families receive no or not detectable portion of appropriate chromosomal region from their RH founder.
Four QTL segregate most possibly in both resource populations: QTL segregate in both FV and ABFV families but all segregating ABFV families receive appropriate chromosomal region from their RH founder (BTA01, BTA05, BTA19 and BTA28).
One QTL was introgressed by RH founder into FV population (BTA10 proximal): QTL segregates only in three ABFV families, all three segregating ABFV families receive the most proximal chromosomal region of 35cM from their RH founder. The FV families as well as other ABFV families receiving no or other parts of BTA10 from the RH founder show also no QTL effect.
There can be other chromosomal regions with possible introgression but not detected by the used design: The most critical handicap of the used design is the low mapping resolution achieved by selective DNA pooling. Thus distinct QTL segregating on the same chromosome in two resource populations can appear as one QTL.
To test our ability to distinguish QTL alleles according to their breed origin we chose two QTL regions (BTA19 and BTA28) for fine mapping in a complex pedigree composed of 11 grand daughter design families of FV and ABFV origin. Using individual genotypes of 35 microsatellite markers across BTA19 and 20 microsatellite markers across BTA28 we want to resolve one-QTL versus two-QTL hypotheses. At the moment, it seems to us that this will be not a trivial challenge because of the necessity to account for complex relationships within a stratified population, to account for two or more QTL and for, possibly, not simply additive inheritance at one or two QTL.
Although two possible methods for estimating QTL allele frequency were proposed, Weller et al., (2002) presented an attractive method based on a modification of the granddaughter design (MGDD). In the MGDD, the maternal granddaughters of a QTL-heterozygous sire are genotyped, and the average trait value of three daughter genotype groups obtained:
- Group 1 that did not receive either of their MGS chromosomes for the region containing the QTL. These represent mean population value according to population allele frequencies at the QTL;
- Group 2 that received the positive QTL region from their MGS. These represent a group with higher frequency of the positive QTL allele than Group 1;
- Group 3 that received the negative QTL region from their MGS. These represent a group with higher frequency of the negative QTL allele than Group 1.
QTL allele frequencies are obtained by comparing mean trait values of the groups. A program was written in Fortran to trace haplotypes from the MGS to his granddaughters. Two data sets, from RP1 and RP4 were analysed. In addition, an extension of the MGDD design which included information on MGS that were homozygous at the QTL was developed and evaluated by simulation.
The haplotype tracing program. Methods: The program is a modification of Weller et al., 2002 [ 4], allowing for haplotypes of two or three markers. The haplotype transmitted to the granddaughter from the sire of the daughter is first determined, and from this the haplotype transmitted from the dam. In the next step, and differently from Weller et al. (2002), a probabilistic algorithm is used to define the probability that the granddaughter inherited either of the QTL alleles of the grandsire, or the haplotype transmitted from the granddam. If it is not possible to state with high probability either of those options, the granddaughter is considered non-informative and is not included in the overall analysis. Results: The program was able to determine the inherited MGS QTL allele for 45% of granddaughters in RP1, and 63% of the granddaughters RP4. QTL allele frequency estimation. Materials and Methods.
Two sires denoted C and S1, heterozygous at a QTL affecting PP on BTA20 were chosen for analysis. Genotypes were obtained for 431 grand daughters of sire S1 and 303 grand daughters of sire C; the program provided information on QTL transmission for 461 granddaughters across both MGS. Analyses were performed with proc GLM in SAS. QTL allele frequencies were estimated separately for each MGS, sires of daughters were fit as a fixed effect. The QTL allele frequency estimate of Q (positive allele) for RP4 based on one MGS was 0.43, contrary to expectation the second MGS appeared to be homozygous for the negative QTL allele. For RP1, genotypes at a QTL affecting PP on BTA13 were obtained for 156, 126, and 167 grandaughters for S1, S2 and S3, respectively, and the program provided information on 201 granddaughters (total) of the three sires. Sire 2 appeared to be homozygous for negative allele at the QTL; The QTL allele frequency estimate of Q from the other two sires was 0.79. An extension of the MGDD design. Weller et al. (2002) included only bulls heterozygous at the QTL in the analysis. The status of a bull (homozygous or heterozygous) had to be determined prior to inclusion in the design.
However, if a bull is homozygous for the QTL, the means of granddaughter groups receiving the two alternative marker haplotypes bracketing the QTL should both be either lower or higher than the mean of the granddaughter group having received none of the grandpaternal haplotypes, depending on whether the sire is homozygous for the positive or negative allele a t the QTL. Looking at several grandsires at the same time, if the means of offspring of one type of homozygous sire depart less from the means of offspring not having received a grandpaternal haplotype than the other type we may conclude that the population frequency of this QTL allele is higher.
A simulation study was performed extending Weller�s MGDD in this way, to allow for maternal grandsires (MGS) homozygous at the QTL on the assumption that zygosity state of the MGS would be determined from the granddaughter data. The proposed approach worked well when population allele frequencies of the QTL were moderate but was seriously biased when either of the QTL alleles was at high frequency. Future work will includes simulations based on prior knowledge on zygosity of the MGS, and may provide more reliable results.
Conclusions: Validation of the MGDD method of Weller et al. (2002) and the accompanying program for tracing MGS haplotypes to maternal granddaughters will be useful to breeding organizations in applying the MGDD for estimating allele frequency for QTL that are targets for MAS. Furthermore, MGDD is the only method now available to determine whether a sire is homozygous for positive or negative allele at a QTL, which is very important information for MAS.
When applying MABLUP methodologies in which QTL are included as random effects the variance explained by the polygenic component is estimated simultaneously with the variance explained by QTL and thus the problems addressed in WP12 are overcome. Other issues found necessary replaced the work months allocated to this issue.
We negotiated the scientific use of the MABLUP software developed at VIT Verden (Vereinigte Informationssysteme Tierhaltung w.V., Heideweg 1, 27283 Verden/Aller, Germany). All partners in collaboration with their national breeding organisations are currently producing data needed for MABLUP testruns. Results of the testruns will allow breeding organisations to evaluate implementation of MAS in their cattle populations, based on some of the QTL found in the BovMAS project.
An improved mapping methodology for selective DNA pooling data:
Approximate interval mapping (AIM) was developed: Single marker across sire test statistics (TS) are more or less strongly affected by the number and QTL status of the specific sires that are heterozygous for a given marker. Given the single marker TS an approximate multiple marker method was developed to predict TS for markers for which a sire was homozygous or at any other location on the chromosome, to extract maximum information on linkage. A simple selection index analogy was used to make multipoint predictions, exploiting the fact that the prediction of a TS at location l, given an observed TS on location i, is only a function of the genetic distance and hence the recombination rate. Power and map resolution of the proposed method were assessed by simulation. Power of AIM is higher and average bias of QTL location estimates smaller compared to single marker mapping. The advantage of AIM increases with decreasing marker informativity of the sires.
Power of different pooling strategies -index versus single trait mapping- were evaluated by stochastic simulation:. When mapping is focused on multiple traits the relative propitiousness of selective DNA pooling is reduced, even though genotyping costs are not crucial in this methodology. One approach of dealing with this limitation is to use a selection index for pool formation. Traits considered were: milk yield (MY), protein percent (PP), maternal fertility (FF) and somatic cell count (SCC) versus using a selection index, composed of all four traits, for pool assignment. Results indicate that the significant loss in power for single traits when mapping is on an index does not justify pool formation according to a selection index in selective DNA pooling, although power of QTL detection with favourable pleiotropic effects on the selection index was increased and cost savings were obtained. It seems to be advisable for most multi-trait studies to prepare separate pools for each trait instead of following the strategy of selection index pool formation. However, in situations with few, highly correlated traits and a shortage of resources, pooling on a specific mapping index might be an option.
To study different selection criteria for pool formation, real data of 21.616 Austrian RP2 &RP3 daughters were analysed for Protein Yield PY (h2 -0,30) and maternal fertility (non return rate after 90 days) (h2-0,02). High and low pools were created by means of 4 different selection criteria: yield deviation (YD), corrected yield deviation (CYD), which were corrected for half of the dam EBV, BLUP EBV and corrected EBV (CEBV), which were calculated by subtracting half of the dam EBV from the EBV of the daughter. Results showed that EBV are not appropriate selection criteria, because especially for low heritability traits there is a high selection pressure on the maternal side. Correction for the maternal influence is achieved using CYD and CEBV. Optimal selection criteria for pool formation are CYD since they represent unregressed values leading to unbiased estimations of QTL effects. Thereupon wherever possible CEBV were used as selection criteria in the genome scans of all RP. Following these analysis we evaluated the power of different selection criteria via stochastic simulation. Each sire was randomly to generate 2,000 female progeny per sire and selective DNA pooling was simulated with the 10 % best and worst daughters according to their EBV or CEBV for MY, PP, FF and SCC.
Furthermore MPFP (Marginal Proportion of False Positives) a criterion for setting significance levels was developed as a modification of the PFP (Mosig et al. 2001; Fernando et al. 2003). While the PFP for a particular test represents the PFP among all tests up to the (marginal) test in question; the MPFP represents the PFP of the last group of tests added to reach the marginal test. MPFP for the marginal test is inferred from the change in overall PFP produced by adding the last group of tests. MPFP will generally be greater than PFP, and provide more flexibility in deciding which tests are to be declared significant.
We deal with identifying marker-QTL phase in young elite sires on the basis of their initial progeny test, generally consisting of a small group of 100-200 daughters. Since elite young sires produce the next generation of candidate bulls, the gains from early information on marker-QTL phase in a young sire are two-fold:
- Shortening the generation interval between information and its use; and
- Increasing the proportion of progeny for which information is available.
In the direct progeny of a young sire, all QTL that are phased in the sire are informative, since the progeny must receive one or other of the sire haplotypes.
But only half of the grandprogeny receive one or other of the sire haplotypes. Hence, half of the QTL information in the grandsire is lost in transmission. Feasibility of identifying marker-QTL phase in young elite sires was investigated using a deterministic analysis and also by simulation based on daughter data from RP1 and RP4.
Deterministic analysis, Methods: We consider a situation where a QTL has been mapped to a known chromosomal region (QTLR) and a marker haplotype in close association with the QTL is known. A deterministic Bayesian analysis of power and proportion of false positives as a function of Type I error (a = 0.05, 0.10, 0.15 and 0.20), allele substitution effect in units of phenotypic standard deviation (d=0.2 and 0.3 ), and number of daughters in the progeny test (N=100,200,300,500,1000) was implemented.
The analysis assumed that the proportion of heterozygosity at the QTL was 0.40, and provided power, and proportion of false positives (PFP) among declaration of heterozygosity. Accepting a proportion of false positive is equivalent to diluting the anticipated effects of MAS by this proportion. Results: With N=100, and a=0.20, d=0.2, power was 0.57 with PFP = 0.34; with d=0.3, power was 0.75 with PFP = 0.29. With N=200, a=0.20, d=0.2, power was 0.72 with PFP = 0.29; with d=0.3, power was 0.90 with PFP= 0.25. For 300 or more daughters, power at a=0.20 was very high (0.82 or more), even for d=0.2, and PFP approaches its limit value of 0.23. Roughly similar results, showing power of about 0.40 for N= 150 and d= 0.3. were obtained by P3 by simulation based on a different set of assumptions. Simulation:
Simulations of ability of small samples to determine phase in young sires were carried out using data from two sires from RP1, 1392 daughters of S1, known heterozygous for a QTL affecting EBVPY on BTA4; 303 daughters of S2, known homozygous for the same QTL; 240 daughters of S3 from RP4, known heterozygous for a QTL affecting EBVPP on BTA13; and 368 daughters of sire S4 (RP4) , known heterozygous for a QTL affecting EBVPP on BTA14. The daughters were those in the high and low 10% of the total daughter population for the analyzed trait. Haplotypes spanning the QTLR were used to assign sire haplotypes to the daughters. Informativity of the haplotypes was 0.725, 0.751, 0.900, 0.690 for S1, S2, S3, and S4, respectively. To implement the simulations, the desired number of daughters were chosen at random from each of the two tails.
A t-test for significance (non-Bayesian) was then carried out. Power of the test was calculated as the proportion of tests reaching the given significance level. The simulation variables were (1) The number of daughters in each tail (N=40, 80, 160), and (2) the P value required for significance (a = 0.05, 0.10, 0.15 and 0.20). Each Nxa combination was run 100 times for each sire. With 20 daughters in each tail (equivalent to progeny test of 100 daughters), and a = 0.20, the proportion of significant results was 0.34 and 0.28 for S1 and S3, but only 0.16 (about equal to a) for S2. With 40 daughters in each tail, power was 0.29, 0.34 for S1 and S3, and again, 0.16, for S2. With 160 daughters in each tail (equivalent to progeny test of 400 daughters), power was 0.42, 0.47 and 0.23, respectively. These values are less than those obtained in the deterministic analysis.
However, application of the Bayesian approach used in the deterministic analysis to the young sire analysis, should bring the deterministic and simulated results closer. Increase in d, simulated by cross transfer of specific genotypes from high to low tails of the population distribution, strongly increased power and decreased PFP. This might be achieved in practice by multi-trait analysis.
Conclusion: It should be possible to determine marker-QTL phase in young progeny tested sires for known mapped QTL, with power of 0.50, and PFP of about 0.25. This information will be useful to organizations that are implementing MAS programs, encouraging them to implement marker-QTL phasing of their young elite sires.
The most important dual purpose cattle breed in Middle Europe is the Fleckvieh (Simmental) breed. So far no complete genome scan for quantitative trait loci (QTL) affecting any production trait has been performed in this Fleckvieh breed. Here we report on the first whole genome scan in the German-Austrian Fleckvieh (FV) for two milk production traits Milk Yield (MY) and Milk Protein Percentage (PP). To increase mapping power and reduce experiment costs we applied a selective DNA pooling strategy in a daughter design with ten large half-sib families. The average family size was 2341 daughters ranging from 1635 to 4043 per family. The selection criteria for DNA pooling were corrected breeding values calculated from the routinely performed, common genetic evaluation for South-Germany and Austria. The marker set consists of 220 microsatellite markers covering all 29 autosomes.
Totally 12569 pool genotypes were produced and combined into 2661 single marker linkage tests for both traits. Three families sampled in Austria by different logistic show high proportion of pool genotypes with inconsistent allele frequency patterns between two replicates and also high proportion of significant sire-by-marker tests results. Additional information (individual genotyping etc) indicated that a systematic error source in these three families cause inconsistent pool genotypes as well as highly significant mapping results at apparently consistent genotypes.
Therefore we excluded the three mentioned families from all further analysis. There were a total of 220 comparisonwise linkage tests at the individual marker level for both MY and PP. Combining false discovery rate and approximate interval mapping (AIM) thresholds we detected 29 QTL distributed across 25 chromosomes. Only four chromosomes (BTA10, BTA15, BTA17 and BTA21) are without indication for a QTL affecting PP or MY. Our pool results give consistent indications for segregation of two QTL on BTA04, BTA09, BTA13, and BTA20. Most of the detected QTL affect both PP and MY simultaneously but with higher significance for one of the two traits.
There are six PP-QTL which are partly independent from MY: BTA13 central, BTA14, BTA20 (GHR), BTA25, BTA28 and BTA29. Unlike to this, we found only a weak indication for a pure MY-QTL but there are eleven QTL affecting both traits but with higher significance for MY: BTA03, BTA04 (proximal), BTA07, BTA11, BTA12, BTA16, BTA18 (distal), BTA20 (proximal to GHR), BTA22, BTA23 and BTA26. The remaining twelve QTL showed effects for both traits too, but with higher significance for PP.
There are nine QTL (BTA02, BTA04 distal half, BTA11, BTA18 distal, BTA19, BTA22, BTA24, BTA25, and BTA28) which are possibly mapped for the first time in cattle and thus present novel information for genetic dissection of quantitative traits. Some recent crossing experiments in plant breeding indicated the local and global importance of genetic variability in presently less modern breeds or lines. The results of QTL analysis in different populations with underlying different adaptive evolution (alpine versus lowland) and different selection criteria (dual purpose versus dairy cattle) could be used to verify the role of selection in producing the observed divergence in phenotype, and may identify additional loci that underlie traits that are part of the same adaptive syndrome.
Therefore here produced QTL mapping results in Fleckvieh, as important dual purpose cattle breed from the alpine region, are valuable not only locally for implementation of marker assisted selection in this breed but also globally for molecular dissection of a major gene effect on a quantitative trait under natural and artificial selection.
The estimated positions of mapped QTL is rather inaccurate but provide valuable starting information for running studies aiming at the confirmation by interval mapping, fine mapping and use for MAS. Seven of the QTL detected here present starting points for ongoing fine mapping attempts with the ultimate goal to infer the genetic basis of traits under artificial selection in the Fleckvieh breed and to increase the sustainability of the European dual purpose cattle populations. Four of the QTL regions detected by this and other project have been selected as first choice for ongoing implementation of marker assisted selection in German-Austrian Fleckvieh cattle.
The aim of this comprehensive simulation study was a genetic and economic evaluation of applying marker assisted preselection (MAPS) in a realistic dairy cattle breeding program with a complex breeding goal composed of the 4 correlated traits - milk yield (MY), protein percent (PP), female fertility (FF) and somatic cell count (SCC). A complex genome was created and evolved by natural and artificial selection, aiming at a realistic distribution of the genetic variance among QTL.
Furthermore, a QTL mapping experiment was carried out applying selective DNA pooling. Detected QTL were fine mapped and effects on single traits and the index value were estimated. Subsequently, a realistic breeding program was simulated, in which selection was performed according to ?conventional breeding values, without the use of embryo transfer, which was compared with MAPS of test sire candidates among full-sibs generated by embryo transfer.
MAPS was applied as bottom up approach. For a constant test capacity the number of test sires was varied from 56, 80 to 168 with 150, 100 and 50 daughters per sire, respectively. To evaluate the impact of the success rate of embryo transfer on the genetic gain per generation, MAPS schemes with 2 and 4 male full-sibs were simulated. It was shown that MAPS could increase the competitiveness of breeding programs in dairy cattle breeding programs with complex breeding goals.
The simulated MAPS designs showed, due to the higher proportion of segregating QTL detected among elite sires, an improved efficiency with more daughters per test sire. Nevertheless, the selection steps following preselection of test sires, such as selection of proven sires and elite sires, yielded higher genetic gain in the population with smaller daughter groups and more test sires progeny tested. These opposite trends need to be optimised in an economic evaluation, which accounts for discounted return and costs. Genetic gain per generation in MAPS schemes for test sires entering progeny testing were remarkably increased compared to selection based on conventional breeding values in MY (1 to 27% across evaluated schemes) while the negative genetic trends in PP, as observed in the selection based on conventional breeding values, increased slightly. Genetic trends of FF and SCC in MAPS did not change substantially, compared to selection based on conventional breeding values.
The economic implications of MAPS among full-sibs and extra genetic benefits from MAPS compared to selection according to conventional breeding values were modelled with ZPLAN, a deterministic simulation program. Parameters used for the economic evaluation were annual monetary genetic gain (AMGG), discounted returns, discounted costs and discounted profit. Extra AMGG from MAPS, was moderate and mainly obtained in MY while the negative trend in PP, as observed in the conventional breeding schemes, increased slightly. AMGG increased between 5 and 9%, dependant on the evaluated MAPS scheme.
When test capacities were constant, the bottom up approach favoured, with regard to AMGG, large daughter groups (150 daughter records per test sire), while the selection stages applied to test sires after preselection were more efficient with a larger number of test sires. Since discounted costs increased when more test sires were evaluated each year, the highest discounted profit in breeding value assisted selection was obtained with 80 test sires.
The same number of test sires with four male full-sibs per embryo transfer available for preselection was the best MAPS scheme. Extra costs of MAPS were moderate compared to costs of the progeny testing breeding program, and partly reduced due to increased selection intensity among elite dam candidates.
An additional simulation was based on the real population structure of the Italian, Austrian and German Brown breed, using the 5 QTL with largest allele substitution effect for MY identified in the genome scan. The use of embryonic technologies such as Multiple Ovulation and Embryo Transfer is a determinant for a successful MAPS since it allows choosing superior young bulls out of a set of full brothers (advantages increase with increasing number of full sibs). As expected the overall advantage of MAPS based on 5 QTL influencing a single production trait evaluated with respect to the genetic gain achieved in a complex breeding goal is smaller as compared to its advantage on a single trait.
The results of these simulation studies can be used as indicators of the genetic gains to be expected by applying different scenarios of MAPS, for the Brown Swiss dairy cattle population in Italy and in Europe based on 5 QTL as found in the BovMAS project and the ultimate genetic and economic gains to be expected in a dairy cattle population with approximately 50.000 cows under milk recording by making use of all genetic variance, which was also traceable via segregation analysis to the test sires.
The recent experiments using advanced backcross QTL strategy (AB-QTL) in plant breeding showed that even wild progenitor species constitute a prominent source of still unfolded variability for complex traits in lines, which have been cultivated for over 8000 years.
These studies indicated the local and global importance of genetic variability in presently less modern breeds or lines. AB-QTL is one of the strategies to harness hidden potential and broaden the genetic diversity of the existing gene pool. Over the last quarter of a century German-Austrian Fleckvieh breeders designed a large advanced backcross population, which is embedded within the purebred Fleckvieh (FV).
Utilization of Red Holstein (RH) founders as donors in interbreed crosses was the strategy to introgress valuable traits/genes from the donor breed (RH) into the existing gene pool of the FV breed. The FV breeders used the advanced backcross method primarily to introgress quantitative trait loci relating to two complex traits and their components: Milk Yield and Udder Quality.
Most FV breeders wanted only the desired part of the genome of the RH population to be introgressed into that of the FV population, whilst the remaining genome of the FV population should be kept intact as much as possible.
The aim of this study was to use DNA marker based strategies to
- Simultaneously detect quantitative trait loci relating to milk yield (MY) and milk protein percentage (PP) in purebred FV and in the Advanced Backcross Fleckvieh (ABFV) population,
- To set up necessary information for marker assisted selection (MAS) regardless of allele origin and
- To prepare information for marker assisted accumulation of positive alleles and selection against undesired alleles introgressed by the most important donor founder.
For a whole genome scan in the ABFV population we sampled a large daughter design with eight half-sib families. The average family size was 2134 daughters, ranging from 1470 to 3329 per family. From 17073 stored milk samples we prepared selective DNA pools for nine traits: MY, PP, milk protein yield (PY), milk fat percentage, milk fat yield, milk somatic cell count, maternal fertility, calving difficulty, and stillbirth.
The selection criteria for pooling were corrected breeding values calculated from the routinely performed, common genetic evaluation for South-Germany and Austria. Each pool replicate consists of 101.5 the most extreme daughters. The whole genome scan was performed for two traits MY and PP using the same set of 220 microsatellite markers used for QTL mapping in purebred Bavarian-Austrian Fleckvieh too. Totally 9011 pool genotypes were produced and combined into 2034 single marker tests. Similarly as for purebred Fleckvieh two families sampled in Austria showed a higher proportion of inconsistent pool genotype results due to systematic sampling error.
Therefore we excluded these two families from all further analysis. According to results of adjusted false discovery rate and approximate interval mapping analysis we detected 22 QTL regions (QTLR) segregating in six advanced backcross families.
Most of the detected QTL affect both PP and MY simultaneously but with higher significance for one of the two traits. There are five PP-QTL which are partly independent of MY: BTA14 central, BTA24, BTA25, BTA28 and BTA29. Ten QTLR (BTA01, BTA02, BTA04, BTA05, BTA06, BTA07, BTA08, BTA09, BTA10, and BTA19) showed effects for both traits but higher significance for PP. The remaining seven QTLR showed effects for both traits but higher significance for MY: BTA03, BTA11, BTA12, BTA16, BTA22, BTA23 and BTA26.
According to the haplotype analysis in a complex pedigree including all family sires as well as all available ancestors and the founder of the ABFV population there are some indications for the introgression of the active QTL variants by the Red Holstein founder into the ABFV population. The QTL affecting MY and PP and segregating on the proximal region of BTA10 is most possibly introgressed by Red Holstein founder.
Additional to BTA10 there were some indications for introgressed active QTL variants on BTA01 BTA05, BTA19 and BTA28. For these four QTL regions we observed the significant QTL effects in the purebred Fleckvieh families too. The estimated QTL positions are rather inaccurate, thus it is not possible to make a clear conclusion about introgression but our results provide valuable starting information for running follow up studies aiming at the confirmation by interval mapping, use for fine mapping and marker assisted selection. Partial results have been presented to the final users, namely the Bavarian-Austrian Fleckvieh Breeder Associations and artificial insemination stations. These agreed on a program for implementation of the mapping results into marker assisted selection.
High resolution QTL mapping (HRM) in dairy cattle is important for MAS, and for cloning quantitative trait genes. HRM in dairy cattle is enabled by the large sire half-sib daughter families in these populations. Accessing this information for a specific QTL containing region (QTLR) requires genotyping daughter families for a dense battery of markers spanning the QTLR.
Genotyping all daughters for all markers, requires enormous numbers of genotypes. Two approaches for HRM that reduce required genotyping were investigated: Selective Recombinant Genotyping (SRG); and selective DNA pooling using a dense marker map and a new "Fractionated Pooling Design (FPD)".
In SRG, daughters are identified that received a recombinant QTLR haplotype from their sire, since only these daughters carry information for mapping within the QTLR.
This requires that the daughters are "informative" for the markers flanking the QTLR, i.e., that the specific marker allele transmitted from the sire is identified. When the genotype of the dam of the daughters is unknown, the daughters are informative if their genotype is readable and differs from that of their sire. For Israel Holsteins, mean informativity over 168 markers was 0.69, and proportion of readable genotypes about 0.85. Thus, the expected proportion of daughters informative at both flanking markers of a typical QTLR will be 0.36. In actual analysis of a number of QTLR, this proportion ranged from 0.09 to 0.60; with a mean of 0.28.
Thus, to identify recombinant daughters, it would be essential to genotype multiple markers at each flank of the QTLR. Because of low marker informativity, therefore, SRG may not be useful in animal populations; but will remain useful for F2 and BC populations where informativity is high. Consequently, SRG was not pursued further in the BovMAS program. Instead, we turned to the FPD approach. The FPD is based on DNA analysis of multiple sub-pools made up of independent subsets of individuals taken randomly from the trait distribution tails in each family.
In contrast to standard selective DNA pooling, the FPD estimates QTL position and effect, with confidence intervals based on re-sampling techniques. In addition, the FPD provides QTL detection based on permutation tests rather than asymptotic distributions of test statistics; joint analysis of multiple families and markers (even if markers are not shared among families); estimation of family specific QTL effects; and analysis of multiple linked QTL.
Simulations showed that the FPD procedure was able to locate QTL with high accuracy. Application of the FPD to BTA13 involved 8 sires heterozygous for a QTL affecting PP, and 6 sires heterozygous for a QTL affecting MY. Pools were genotyped at 19 markers spanning the region 23 to 99cM. FPD analysis suggested three QTL located at 30cM, 55cM and 80cM, respectively, with confidence intervals about 10 cM for each QTL. Application to BTA20 involved 10 sires heterozygous for QTL affecting PP or MY. Pools were genotyped at 21 markers spanning the region 0 to 83 cM. FPD analysis identified a QTL affecting PP at position 48cM, with confidence interval of 12 cM.
Validation of the FPD for genome-wide association studies of quantitative variation will allow highly reliable and cost-efficient large-scale QTL analysis, providing results unattainable by standard selective DNA pooling analytical procedures. SRG and FPD were proposed in the original BovMAS Technical Annex. While BovMAS was being implemented, however, a new methodology for HRM based on the variance component methodology for QTL mapping was proposed elsewhere and proved highly effective in some instances. The LMU partner applied this methodology for HRM of BTA13 and BTA20 in the Fleckvieh breed. HRM for BTA13 was based on five families comprising 1835 daughters with extreme phenotypes, genotyped for 32 markers; and augmented by 238 sons from four families of a granddaughter design typed for 42 markers.
Two QTL were uncovered; affecting milk and protein yield at 63.82cM, and protein and fat percentage at 85.45cM, respectively. HRM for BAT20 was based on 1,365 sons in a granddaughter design genotyped for eight or 17 markers. The study confirmed a QTL affecting milk fat and protein percentage linked to GHR, and identified a new QTL for protein yield at 26cM. Best results were obtained by linkage analysis alone or by combined linkage disequilibrium and linkage analysis.
The contribution of linkage disequilibrium to the mapping results was small compared to literature reports. This could be due to lower LD for the Fleckvieh, compared to highly selected milk breeds. LMU also produced genotyping data at high density in the large complex pedigree of Fleckvieh and advanced backcross Fleckvieh x Red Holstein populations, which should enable fine mapping of QTL on BTA09, 18, 19, 28 and 29.
A selective DNA pooling approach using milk samples was adopted to map QTL affecting milk yield (MK) and milk protein percentage (PP) in the Austrian, German, and Italian Brown Swiss dairy cattle populations. The scan was based on 139 markers evenly spaced on the genome. The mapping population consisted of half sib daughter families of 10 Brown sires with 1000 to 3600 individuals. Three families were sampled in Germany, three in Italy, one in Austria while the sampling for three sires occurred jointly in Austria and Italy.
No complete genome scan for productive traits was available in the Brown breed before this project, as no marker association with secondary traits (namely protein and fat yield, fat percentage and milk somatic cell count).
The markers resulted significant were 19 for PP, 29 for MK and 80 for both traits. Over the 840 sire x marker combinations 153 and 137 were significant for PP and MK respectively and 62 for both traits. Over the significant combinations for one or both the traits, MK and PP showed the same direction of the effect in 201 cases and opposite effect in 228 instances.
Among the QTLRs that showed a significant effect for PP and/or MY, five were chosen for more intensive study, located on BTA03, BTA11, BTA13, BTA14, BTA20. The proximal part of BTA14 is known to house DGAT1, a QTL with strong effect on FY and MY. The QTL segregating in some of the informative families confirmed the existence of the gene. In the Brown breed it seems that a second QTL region on BTA14 affecting both MY and PP can be identified between 70 and 110 cM.
Generally secondary traits tested show sire/marker significance with several QTL identified for MY and PP, while for milk somatic cell count there is fewer evidence of correlated effect at same location.
Main society interested in results are farmers and breeders of the Austria Germany and Italy. Dissemination of results was guaranteed thanks to seminars and round tables to farmers organization in Italy and in Austria. Articles for extended farmers audience have been produced and will be published in sector journals
Results on main and secondary traits arising from the BovMAS project allow the potential use of QTL knowledge in ongoing selection strategies.
New knowledge in the Brown dairy cattle population, open the possibility to enhance the breeding programs to new technologies and to deeper investigate in the close future the existence and the effects of genes related to economic traits in targeted genome regions, building up to existent know how.
The produced knowledge is ready to be integrated into the ongoing breeding schemes towards the application of technologies and skills developed in the project:
- Sampling of DNA from milk embedded in the ongoing national milk recording schemes;
- Genotyping at routine level to be carried out by breeders/farmers service labs;
- Integrating information from targeted QTL regions in genetic evaluations routinely from breeders organization. Benefits arising from a prompt application of QTL knowledge using within family MAS reside in paving the way to gene assisted selection as soon as information will became available.
The Italian Hostein-Friesian breed is the most important dairy cattle breed in Italy. So far no complete genome scan to identify QTL for milk production traits have been produced. We herein produced for the first time a complete map of QTL affecting milk yield (MY) and protein percentage (PP) in this breed using a selective milk DNA pooling approach in a daughter design.
It was estimated that the genome scan, that was conducted for eight half daughter families using 183 dinucleotide microsatellites distributed on all bovine autosomes (BTA) and the X chromosome, uncovered 35 QTL regions (QTLR) for MY and 32 QTLR for PP (about 1.1 and 1.2 QTLR for each autosome for PP and MY, respectively). For PP, 13 chromosomes (BTA01, BTA04, BTA05, BTA08, BTA09, BTA11, BTA16, BTA18, BTA22, BTA23, BTA24, BTA27 and BTA29) showed a single peak of significance, 5 chromosomes (BTA02, BTA06, BTA13, BTA17 and BTA21) showed two peak of significance and 3 chromosomes showed at least 3 distinct QTLR (BTA03, BTA14 and BTA20). For MY, 12 chromosomes showed only one clear peak of significance (BTA01, BTA02, BTA04, BTA05, BTA10, BTA12, BTA18, BTA19, BTA21, BTA25, BTA28 and BTA29), 6 showed the presence of two QTLR (BTA03, BTA08, BTA11, BTA13, BTA17 and BTA20) and 3 showed at least 3 QTLR (BTA06, BTA07 and BTA14). Moreover, it was estimated that about 95% and 92% of the PP and MY segregating QTL in the Italian Holstein population will have had some opportunity to be identified in the present genome scans.
For PP, taking the averaged substitution effect for the significant markers in each QTLR, allele substitution effect ranged from 0.008 to 0.025 EBV%, with mean 0.013793 EBV% and sum of 0.441379 EBV%. For MY, averaged allele substitution effect ranged from 241,4 to 664,0 kg (Daughter Yield Deviation), with mean 328,2 kg (DYD) and sum of 11487 kg (DYD). Five QTLR (BTA03, BTA11, BTA13, BTA14 and BTA20) were chosen to investigate the effect on other milk production or reproductive traits. In general, significant sire by marker tests were observed for most correlated traits (namely protein yield, fat yield and fat percentage) in the same regions.
No clear association of these five QTLR was evidenced with other traits like milk somatic cell count, calving difficulties and perinatal mortality. Protocols for marker analysis have been developed including new methods for single nucleotide polymorphism allele frequency estimation in milk DNA pools.
New mutations in candidate genes have been obtained. Partial publications of these results (mainly for BTA14 and BTA20) have been produced confirming the presence of QTL already identified by others on these chromosomes but also indicating the presence of additional QTL that may be considered for the use of these QTLR in marker assisted selection plans. Partial results have been illustrated to the final users, namely the Italian Holstein Breeder Association, semen centers and Provincial breeder associations.
The results provided in this study will represent the starting point to further investigate in detail QTLR in order to identify the causative mutations.
The objective was to develop a complete map of QTL affecting milk protein percent (PP) in the Israel Holstein cattle population, and investigate the effects of the detected loci on milk yield (MY) and protein yield (PY); and of selected loci on milk somatic cell counts (MSCC) and female fertility (FF). A daughter design, with selective DNA pooling based on milk samples was used. The study involved 18 sires, with a minimum of 1800 daughters per sire. PP, Materials and Methods:
All sires were included in the PP study. Based on estimated breeding values for milk PP (EBVPP), the highest and lowest 10-20% of daughters of each sire were sampled to make up the pools. 177 microsatellite markers distributed over all 29 bovine autosomes were used. Comparison-wise error rates (P-values) for sire by marker tests, and for marker tests across sires were according to Mosig et al. (2001). A Marginal PFP (MPFP) <= 0.05 was required for a declaration of significance at the marker level, and a MPFP <= 0.10 within significant markers, at the sire x marker level. The MPFP (see WP 12) is a modification of the "proportion of false positives (PFP)" criterion (Fernando et al., 2003); and differs from PFP in that significance is set according to the PFP within the last added group of tests. Results: Average heterozygosity among the QTL uncovered in this study was 0.44. Average standardized allele substitution effect of QTL was 0.24 (0.16-0.33). Differences in effects among QTL were not significant, and QTL with large effects were not found.
There were 101 significant markers. Chromosomes with a single QTL containing region (QTLR) were identified as a having group of significant markers flanked by non-significant markers; Multiple QTLR on a chromosomes were identified by groups of significant markers separated by non-significant markers. A total of 60 QTLR were defined; most chromosomes appeared to carry two or more QTL affecting PP; BTA15, 17, 24 and 28 did not carry any.
The total effect of all 60 putative QTL summed to 0.9 EBVPP and appear to account for the total genetic variation in EBVPP in the population. MY and PY, Materials and Methods: The study was based on ten sires. Pools were prepared and genotyped for 134 microsatellite markers distributed over 25 autosomes, many of which had been found previously to be significant for PP. Results: Of the 134 markers, 37 were significant for one trait only, 44 for two traits, and 50 for all three traits. Thus, many of the QTL affecting PP also affect MY and PY. Within the significant sire-marker-trait combinations, there were 342 combinations significant for one trait only, 110 significant for two traits, and 27 for all three traits. When the effects on PP and MY were both significant, 79% of sire-marker combinations with positive effects on PP, had negative effects on MY, but were almost equally divided between positive and negative effects on PY; 65% of sire-marker combinations with positive effects on MY had positive effects on PY.
The overall impression is that the major effect of the QTL is on the PP/MY axis, and effects on PY are a secondary consequence of minor variation in the relative effects of specific QTL on PP or MY. MSCC and FF, Materials and Methods: The study is based on six sires and concentrated on the six chromosomal regions selected for more intensive study by the BovMAS consortium: BTA3, 9, 11, 13, 14, and 20; 24 markers were tested. Results: Significant markers for PP, MSCC and FF were found on all six chromosomes. The 24 markers included a total of 313 sire-marker-trait combinations.
Very few sire-marker combinations were significant for more than one trait. Thus, the relationships between the three traits appear to be limited, compared to the relationships found among PP, PY and MY. Conclusions: The results of these three studies are of general scientific interest, in that they provide the first complete map of a quantitative trait (PP) in a segregating population, and show that essentially all of the additive genetic variation can be explained by mapped QTL.
The results provide a basis for high resolution mapping and cloning of the genes responsible for genetic variation in milk production traits and for MAS programs. The large number of identified QTL show that the potential gains from MAS will extend over a long period of time and result in major increases in productivity. This conclusion is of interest to cattle breeding organizations that are weighing whether to introduce a MAS program. Based on these results, there are ongoing discussions of implementation of a MAS program for the Israel Holstein population.
Searching for OpenAIRE data...
There was an error trying to search data from OpenAIRE
No results available