Characterizing homozygosity across United States, New Zealand and Australian Jersey cow and bull populations
© Howard et al. 2015
Received: 27 October 2014
Accepted: 18 February 2015
Published: 15 March 2015
Dairy cattle breeding objectives are in general similar across countries, but environment and management conditions may vary, giving rise to slightly different selection pressures applied to a given trait. This potentially leads to different selection pressures to loci across the genome that, if large enough, may give rise to differential regions with high levels of homozygosity. The objective of this study was to characterize differences and similarities in the location and frequency of homozygosity related measures of Jersey dairy cows and bulls from the United States (US), Australia (AU) and New Zealand (NZ).
The populations consisted of a subset of genotyped Jersey cows born in US (n = 1047) and AU (n = 886) and Jersey bulls progeny tested from the US (n = 736), AU (n = 306) and NZ (n = 768). Differences and similarities across populations were characterized using a principal component analysis (PCA) and a run of homozygosity (ROH) statistic (ROH45), which counts the frequency of a single nucleotide polymorphism (SNP) being in a ROH of at least 45 SNP. Regions that exhibited high frequencies of ROH45 and those that had significantly different ROH45 frequencies between populations were investigated for their association with milk yield traits. Within sex, the PCA revealed slight differentiation between the populations, with the greatest occurring between the US and NZ bulls. Regions with high levels of ROH45 for all populations were detected on BTA3 and BTA7 while several other regions differed in ROH45 frequency across populations, the largest number occurring for the US and NZ bull contrast. In addition, multiple regions with different ROH45 frequencies across populations were found to be associated with milk yield traits.
Multiple regions exhibited differential ROH45 across AU, NZ and US cow and bull populations, an interpretation is that locations of the genome are undergoing differential directional selection. Two regions on BTA3 and BTA7 had high ROH45 frequencies across all populations and will be investigated further to determine the gene(s) undergoing directional selection.
The widespread use of dense single-nucleotide polymorphism (SNP) assays for genomic prediction has led to the creation of large reference populations across multiple countries and breeds [1,2]. Previous studies have utilized these assays to identify and characterize regions of the genome that have undergone positive selection, referred to as selection signatures [3-9]. Selection signatures are characterized by distributions of nucleotides around favorable mutations that differ statistically from that expected purely by chance due to directional selection increasing the frequency of the favorable allele over time . Nucleotides linked to the favorable mutation also tend to increase in frequency a phenomenon referred to as “hitchhiking” . A recent study by Kemper et al.  provided evidence that locating signatures of selection is difficult for complex traits due to hundreds of loci associated with the trait undergoing weak selection. Even though a selection signature is difficult to detect for complex traits, selection does change the allele frequency of loci associated with the trait. Turchin et al.  showed that specific alleles of SNP associated with human height were at a higher frequency in northern than southern Europe, which mirrors observations of differences in height in European populations.
A potential alternative to detect signatures of selection for complex traits may be to characterize regions of the genome that have a higher likelihood of occurring within a continuous run of homozygosity (ROH) [13,14]. A ROH is generated when an individual receives a haplotype that is identical by descent from each parent . Furthermore, parents can pass on identical chromosomal segments to a child even when the relationship between them is a very distant one, which creates a continuum of homozygous length, depending on the degree of shared ancestry and its age . In dairy cattle the use of artificial insemination has allowed elite bulls to produce thousands of progeny, resulting in a high frequency of familial relationships within the pedigree, potentially giving rise to a high and non-uniform distribution of ROH frequency across the genome within a given population. Utilizing a ROH based metric, referred to as locus autozygosity, on United States (US) Holstein sires, Kim et al.  showed that differences in the location and distribution of ROH regions varied across groups that underwent different degrees of selection pressure. Furthermore, multiple regions that they declared as different were found to be associated with milk yield traits.
In general dairy cattle breeding programs have similar breeding objectives, regardless of country, that are driven by traits of economic importance such as milk, fat and protein yield along with fertility, longevity and conformation. The environments and management conditions in which individual animals perform differ greatly across countries. This is confirmed by genetic correlations varying from 0.75 to 0.80 between US and New Zealand (NZ) and US and Australia (AU) for milk, fat and protein yield . Furthermore, the relative importance of a given trait varies across countries potentially giving rise to different selection pressures across the genome. Selection in North America is mainly practiced in environments where confinement and total mixed ration are typical management settings, in comparison to NZ and AU where performance is predominantly in pasture based systems. Different management systems may lead to variation in the importance of a given genomic region, thereby differentially increasing the frequency of favorable alleles. For example, Kolver et al.  confirmed that North American Holstein-Friesian cows have a greater capacity to convert feed to milk when fed a total mixed ration type diet in comparison to cows from NZ. Studies involving Holsteins have confirmed that the NZ Holstein is genetically different than Holstein derived from other European and North American countries [18-20]. Recently Pryce et al.  combined genotype panels on Holstein animals from multiple research herds (North America, Europe, AU, NZ) and conducted a PCA analysis on the genomic relationship matrix (GRM) and found slight differences across research herds with the greatest difference arising in the NZ population compared to the other research herds.
A limited number of studies have investigated genetic differences across countries within the Jersey breed . Characterizing what causes these subtle changes at the genomic level within the Jersey dairy cattle breed is worthwhile because of the higher levels of inbreeding and smaller effective population size in Jerseys when compared to the Holstein breed [22,23]. Furthermore, lower correlations of production and fertility traits evaluated in northern (US) and southern (NZ and AU) hemisphere countries have been estimated for the Jersey breed in comparison to the Holstein breed , which could make detection of regions of the genome that are under differential selection across countries more insightful. Also, in comparison to the Holstein breed, there has been somewhat less international gene exchange, therefore characterizing differences across population could allow for more efficient collaborations to enhance genomic improvement. The objective of this study was to characterize differences and similarities in the location and frequency of homozygosity related measures of Jersey dairy cows and bulls from the United States US, AU and NZ.
Population stratification and average homozygosity across the genome
Number of animals by birth year within each population 1 for cows and bulls
Average (±SD) ROH homozygosity by population 1
AU (n = 886)
US (n = 1047)
AU (n = 306)
US (n = 736)
NZ (n = 768)
Characterizing the frequency of autozygosity across and within populations
Regions of the genome that have different ROH45 frequencies across bull 1 and cow 2 populations
ROH45 Difference (Maximum Location 4 )
Largest Interval 4
US vs AU Cows
US vs AU Bulls
US vs NZ Bulls
AU vs NZ Bulls
Characterizing the change in autozygosity within each population
The change of locus autozygosity (∆ROH45) across time was modeled using logistic regression of autozygosity on year of birth, where there were at least 40 genotyped animals. Initially the analysis was conducted on both bulls and cows, but no regions were found to be significant for the cows, possibly due to the narrow range in birth year of cows (Table 1) in comparison to the bulls. Therefore only the bull results are presented.
Multiple regions have undergone changes in autozygosity across time for the US and AU bull population, although no regions were significant for NZ (Additional file 1: Figure S1). The US bull population had 3 regions that have undergone autozygosity change across time and they are located on BTA1 (49.0-54.1; 68.5-75.7 Mb) and BTA11 (38.0-45.8 Mb). The two regions located on BTA1 were also shown to have different ROH45 frequencies across populations. The AU bull population had 2 regions that have undergone autozygosity change across time and they are both located on BTA9 (44.9-51.6; 61.3-68.7 Mb). Figure 4 clearly displays that differences across the genomes in the frequency of ROH regions exist, although the exact mechanism by which these occur, such as selection or drift cannot be disentangled.
Effect of regions of high autozygosity or large differences across populations on yield traits
Regions that had differential ROH45 frequencies across population, high ROH45 frequencies in common across all population, and regions that have undergone significant autozygosity change across time were further investigated (N = 4849 SNP) to determine if SNP within these regions are associated with traits of economic importance. Yield deviations (YD) for cows that were derived from standardized lactation milk, fat and protein yield were weighted according to Garrick et al.  and a single marker regression model on the subset of SNP was used to describe the association between a trait and SNP. Markers with p-values smaller than 0.001 were declared significant. The false discovery rate (FDR) was calculated for each trait according to Benjamini and Hochberg .
Multiple regions contained SNP that were associated with milk, fat and protein yield and the FDR for milk, fat and protein yield was 0.30, 0.17, and 0.60, respectively. The region with the largest number of SNP was on BTA7 (38.6 – 58.0 Mb) and included 17 SNP associated with fat yield. Furthermore, a region on BTA17 (16.4 – 18.9 Mb) had 5 SNP associated with fat yield and a region on BTA3 had 6 SNP associated with milk yield. A complete list of the regions in addition to candidate genes are presented in Additional file 2: Table S2. A gene network analysis revealed a network involved in immune system function for milk yield (FDR = 12.7 percent) involving 11 genes that are outlined in Additional file 3: Figure S2 with 6 genes below the 0.001 threshold on BTA2 (LCK), BTA7 (IL3; IL4; MKNK2; CSF2) and BTA18 (CEBPG) and the remaining 5 below the 0.01 threshold.
The current study characterized the frequency and distribution of ROH across cow and bull populations derived from US, AU and NZ. Previous reports across multiple dairy breeds have similarly found that the NZ population is genetically different from other dairy cattle populations [18-21]. The correlations published by Interbull for milk, fat and protein yield  for the Jersey breed between US and NZ is weaker, i.e. further away from 1, in comparison to AU and NZ. As the genetic correlation deviates from 1 it indicates that the expression of the trait is different across environments . A traditional method to examine the degree of differentiation is to compute Wright’s Fst statistic across two populations. The use of this measure is advantageous when large differences in allele frequencies occur, such as across cattle breeds. Within a breed, small differences in allele frequencies are expected across populations and particularly when there is some degree of genetic exchange, as is the case of the Jersey population. Due to this the usefulness of Fst to determine regions that are different within a breed is reduced, therefore alternative methods were used.
One such alternative method to characterize the genomic differences across populations is to compute the average or a specific region’s ROH frequency. The ROH metric has previously been used to examine population history  and as an alternative inbreeding metric [32,33]. Recently, Kim et al.  characterized the variation in ROH frequency in US Holstein dairy cattle utilizing an unselected Holstein population compared to two heavily selected Holstein populations. The mean number of ROH per individual was significantly lower in the unselected population than the two selected populations . This study confirms that there are also differences in ROH levels across populations within the same breed, which may be due to different selection intensities across countries or different thresholds on the levels of allowable consanguineous matings.
Furthermore, Kim et al.  found that several of the regions that had differing levels of ROH across populations were associated with economically important traits including milk, fat and protein yield. The same approach was utilized in this study to detect signatures of selection in common and different across populations. Two regions on BTA3 and BTA7 were found to have high ROH45 frequencies across all populations. Previous studies have also found selection signatures on the same region of BTA3 [7,9], which contains the SLC35A3 gene at 43.4 Mb. A mutation in this gene is known to cause a lethal recessive mutation in Holstein dairy cattle known as complex vertebral malformations (CVM) . A lethal mutation would not give rise to the high level of autozygosity surrounding the CVM mutation, although selection at a nearby linked locus could potentially cause the region to have high levels of autozygosity. The selection signature on BTA7 confirms the findings of Kemper et al. in several cattle breeds  and Qanbari et al. in Fleckvieh cattle  and harbors multiple olfactory genes. Olfactory receptors detect and identify a wide range of odors, providing a cue for the animal to interact with its environment. Furthermore, gene duplications within the beef cattle genome tend to encode genes that interface with the external environment such as olfactory receptors , suggesting that they may be under strong selection for newly evolving functions.
Multiple regions of the genome displayed different autozygosity and interestingly, regions that were different across the US and NZ bull populations are similar to the results described by Kim et al.  where the comparison was between selected versus unselected Holstein populations. The regions include BTA1 (48.9-55.4 Mb), BTA2 (119.2-129.4 Mb), BTA9 (57.0-74.3 Mb), BTA14 (72.4-79.8 Mb), BTA16 (51.9-60.0 Mb) and BTA21 (25.2-33.6 Mb). This suggests that selection for yield has resulted in similar regions of high ROH45 frequency across different breeds.
Furthermore, SNP within regions that have undergone a significant autozygosity change have previously been reported to be associated with milk yield traits. The SNP with the largest significance within BTA9 (44.9-51.6) and BTA1 (49.0-54.1; 68.5-75.7) were within 1 Mb of SNP that have been previously shown to be associated with milk, fat and protein yield and fat and protein percentage, respectively . It is unsurprising that no regions have undergone a significant autozygosity change in the NZ population, given their rather low and relatively uniform level of autozygosity across the genome in comparison to the greater variability in length and location of ROH in AU and US.
Functional analysis of genes within 500 kb in both directions of the significant SNP revealed regions involved in behavior (NBEA), milk fat synthesis (FABP3), fatty acid metabolism (ACSL6), and metabolism (KCTD15). A previous study that investigated selection signatures across multiple beef and dairy breeds found a sweep region on BTA12 containing NBEA , which could be associated with traits associated with behavior . Fatty acid binding proteins such as FABP3 are one of the key intracellular FA transporters and is highly expressed in the mammary gland . In general the favorable allele that was associated with the yield trait based on estimated SNP effects from a single marker regression model using the current dataset had a higher frequency in the US population in comparison to AU or NZ, which has lower levels of milk production, although other reasons may have caused allele frequencies to drift other than solely selection such as random genetic drift.
The gene network involving immune function is unsurprising due to a strong selection emphasis towards traits involving milk production which has led to a more pronounced negative relationship with metabolic, reproduction and health fitness traits . In a study by Parker-Gaddis et al.  using US Holstein data, the genetic correlation for fitness traits such as ketosis, lameness, mastitis, metritis, and retained placenta were all negative with the US net merit index . Furthermore, the particular environment that an animal is managed in may differentially compromise the host immunity and increase the incidence of variety of diseases [44,45] thereby augmenting the selection pressure on a given region. For the genes involved in immune function the frequency of the favorable allele based on estimated SNP effects from a single marker regression model using the current dataset was not consistently higher in a particular population.
Combining SNP assay data across multiple countries was initially aimed at increasing the reliability of genomic breeding value estimates [1,2]. Nonetheless, other potential uses can be garnered from the multi-country collaboration. One example, may be to use this a priori knowledge of the location of these genomic differences in mating schemes in order to decrease the level of homozygosity in the progeny at the genomic level. The availability of a multi-country reference population allows for the detection of a diverse set of haplotypes, which could potentially be exploited using methods such as optimum-contribution selection methodologies [46,47] that weights selection response versus future inbreeding. Furthermore, a multi-country reference population increases the likelihood of detecting selection candidates with favorable but different combinations of chromosomal segments . Relationship matrices that characterize the similarity of haplotype segments  may allow for a more effective progeny inbreeding penalty. A sizable body of literature exists on using genomic information to constrain parental relationships and control the rate of inbreeding or level of homozygosity [50-54]. In general these methods constrain relationships averaged across the genome, although Pryce et al.  confirmed that certain regions have a larger impact on inbreeding depression than other regions. Therefore, optimum-contribution selection algorithms that incorporate this a priori knowledge of regions that have a large impact on inbreeding depression and different levels of ROH across countries, may be more effective in controlling homozygosity at the genomic level and minimizing inbreeding depression. In order for genomic information to be utilized in mating designs whole herd genotyping is required, which currently is not a common practice. As the technology improves and the cost of low-density genotyping platforms decreases, mating designs that utilize genomic information could assist producers in managing their herd at the genomic level.
Some limitations of the current study regarding ROH distribution and frequency could stem from the MAF threshold chosen, that may have resulted in removing SNP that have a low MAF in one population and higher MAF in another population, or SNP near fixation in all populations. As a consequence, potential regions that are similar and different across populations may not be detected if they are near fixation in one or more populations. This editing procedure was used because the detection of “hard sweep” selection signatures involving breed defining traits such as coat color or polledness was not the primary emphasis in this study. Also, it has been shown that the medium density SNP panel is not sensitive enough for the precise determination of short ROH segments . A high density SNP panel was not available across all populations, but we anticipate that denser panels (or sequence data) should help to disentangle the selective history for short segments. Lastly, criteria used for inclusion of individuals in the genotyped populations may not be similar across populations, which may have resulted in genotyped animals in some of the populations not necessarily being representative of the animals within the given country. Multiple editing procedures were here used to minimize this phenomenon in order to make comparisons meaningful.
Regions that displayed differential ROH45 frequencies across bull and cow resource populations from US, AU and NZ were characterized and the largest difference was between the US and NZ population which was in line with the PCA analysis. Regions of the genome that had high levels of autozygosity across all populations were found on BTA3 and BTA7. Furthermore a proportion of the regions that were different across populations were associated with milk yield traits. These subtle populations differences could potentially be exploited at the animal level in order to design mating schemes, that are tailored toward maximizing the level of heterozygosity along with superior additive genetics in the progeny, which will be the focus of future research.
Animal and genotypes
No animal care approval was required for the present manuscript because all records came from field data. The US resource population utilized in the study included genotypes obtained from the American Jersey Cattle Association while the AU and NZ resource population was provided by the Australian Dairy Herd Improvement Scheme (ADHIS; Melbourne, Australia). The majority of the US cows (n = 7458) were genotyped with a low density chip, either GGP (GeneSeek, Lincoln, NE), BovineLD (Illumina, San Diego, CA) or Bovine3K (Illumina, San Diego, CA)) and imputed to medium density (n = 61,013 SNP). The remaining cows (n = 777) were genotyped using the Illumina BovineSNP50 BeadChip (Illumina, San Diego, CA). The AU cows (n = 4075) were part of the Australian genomic reference population and were genotyped by the Dairy Futures CRC (Melbourne, Australia) with the Illumina BovineSNP50 BeadChip (Illumina, San Diego, CA). Bull genotypes from the Illumina BovineSNP50 BeadChip (Illumina, San Diego, CA) were also obtained from the American Jersey Cattle Association (n = 2394) and the Australian Dairy Herd Improvement Scheme (AU = 1069 bulls; NZ = 1748 bulls). The NZ population comprised of bulls genotyped by Livestock Improvement Corporation (Hamilton, New Zealand).
Genotype quality control, imputation and phasing were done within each population. For the US population genotype quality control included removing animals that had less than 90% of the SNP called, SNP with a minor allele frequency (MAF) below 0.01 or a p-value of a chi-square test for Hardy-Weinberg equilibrium less than 0.001. Full details of the quality control methods for the AU and NZ populations are described in detail in  and are similar to the rules applied to the US populations. The SNP unmapped to the Bovine Genome Build 4.0 (http://bovinegenome.org/cgi-bin/gbrowse/bovine_UMD31/) and SNP on sex chromosomes were excluded from the analysis. Missing SNP within the USA population were imputed using Beagle  and SNP with an imputation accuracy of less than 97.5% were removed. We recognize that a MAF threshold may result in removing SNP that have a low MAF in one population and higher MAF in the other population or SNP near fixation in all populations, nonetheless imputation accuracy was greatly impacted by MAF. The remaining SNP that passed quality control for the cow and bull groups were then combined resulting in 31,431 and 27,927 SNP in common between the groups, respectively.
In order to make comparisons across populations as equitable as possible a subset of the complete set of genotypes that met certain criteria were used to characterize difference across populations. To minimize the possible time trend effects and selective genotyping in a particular population cows and bulls included were selected that were born within a similar time frame. For the cow analysis, animals born within a three-year (2007–2009) period were used to make comparisons across and within populations. The AU cow resource population was created by selecting animals that had a large amount of individual phenotypic data and is tailored to represent the diversity across the AU Jersey population. To eliminate herds from the US that genotyped only a few of their elite cows, only herds that had greater than 20 genotyped individuals within a given year were used. The US animals (n = 1047) selected for the comparison came from herds that had genotyped on average 31 animals per year while all AU cows (n = 886) were used within the given years. For the bull analysis, animals born within a six-year period (2001–2006) were used. No criterion was used for the bulls on the number of genotyped animals within a year and herd, as the bulls were representative of the progeny tested bulls in each population. The total number of bulls was 306, 736, 768 for AU, US and NZ bulls respectively. The use of the same year classes in the analysis across bulls and cows was not possible due to fewer number of genotyped animals within a given year for the bulls in comparison to the cows. The number of animals by year class is outlined in Table 1.
Principal component analysis
where N is the number of SNP, pm is the allele frequency of SNPm and xm is the genotype at SNPm. A PCA was conducted on the GRM matrix using the R function eigen . The resulting matrix is a matrix of eigenvectors, referred to as principle components (PC), ordered by descending eigenvalues, where PC1 had the largest eigenvalue. The first two PC were plotted and annotated by country to determine the degree of genetic differentiation across the populations and the variance explained by the PC1 was calculated as the variance attributed to PC1 divided by the total variance.
Characterizing the homozygosity across and within populations
Cows and bulls born between 2007 to 2009 and 2001 to 2006, respectively, were used to characterize the homozygosity across and within populations. Homozygosity characteristics for each population were measured as the overall genomic homozygosity (proportion of SNP that were homozygous across the entire genome) as well as the proportion of genome contained within a ROH. Using a sliding window approach with a fixed SNP length, a ROH was declared when a set number of contiguous homozygous SNP with no heterozygotes was observed. The sliding window approach started with the first SNP on a chromosome and combined all SNP within a given SNP number into a window, then the window was shifted by one SNP to form a new window and this process was repeated until the end of a chromosome. The SNP lengths considered were 45 (average ± SD = 3.44 ± 0.92 Mb), 70 (average ± SD = 5.45 ± 1.25 Mb) and 95 (average ± SD =7.47 ± 1.54 Mb). These SNP lengths were chosen to provide a range of ROH lengths. A minimum heterozygous threshold was not utilized here as it has been shown that setting a threshold for the number of heterozygous SNP within a ROH region potentially leads to inaccurate ROH calling at the boundaries of a ROH region . The proportion of the genome contained with an ROH was estimated by the sum of ROH lengths (Mb) of an individual divided by the total Mb length across all 29 autosomes .
Differences across the genome in the location and length of stretches of homozygosity were investigated utilizing the method outlined by Kim et al. . Briefly, the ROH45status of a SNP was defined based on whether it belonged to a ROH of at least 45 SNP. The ROH45 of a SNP was tagged as 1 if the SNP was in a ROH and 0 otherwise. A length of 45 was chosen for the ROH45 metric based on the average Mb length of 3.44 and a previous study has used a similar SNP length value . The ROH45 metric is advantageous compared to the conventional ROH since it is able to capture the number of times a SNP is in a ROH without declaring the beginning and end of a ROH. The ROH45 value for each SNP was compared across the two populations using a chi-square test with 1 degree of freedom. A statistical threshold was determined using a permutation test (n= 1,000 samples) . Briefly, within each analysis the populations were combined and animals were randomly allocated into groups that were the same size as the original data (n=2 for cows, n=3 for bulls). The ROH45 value for each SNP by group was calculated and significance was reported as the number of times the observed difference was greater than the permutation sample difference across all SNP. The presence of differential autozygous regions was declared as contiguous significant differences (P < 0.001) of at least 45 SNP for regions greater than 4 Mb in length. The presence of regions with high levels autozygosity in common across all populations was declared as contiguous SNP within the top 2.5% of at least 45 SNP and greater than 4 Mb in length.
Change of autozygosity across time
Where ROH45 is the autozygous status of a locus (0, 1), YB is the year of birth of an individual, α is the intercept of the model and β is the change of annual locus autozygosity. Statistical thresholds were determined using a permutation test  (n = 1,000 samples) similar to the one previously discussed. The presence of autozygosity change across time was again declared for contiguous significant differences (P < 0.001) of at least 45 SNP in regions of at least 4 Mb.
Effect of regions of high autozygosity or large differences across populations on yield traits
where h 2 refers to the heritability, r 2 refers to the repeatability and l refers to the parity. The values used for h2 and r2 were 0.25 and 0.43 and were averages across all three traits. The p-values that were smaller than 0.001 were declared as significant and the false discover rates (FDR) were calculated according to Benjamini and Hochberg .
Cow positional candidate genes using Bos Taurus assembly (UMD3.1; Ensemble 68) with regions declared significantly different and similar across the populations were obtained for functional characterization and the identification of gene ontology terms using DAVID [60,61] and gene network work analysis using GeneMANIA . Regions surrounding SNP associated with milk yield traits were extending 125 kb in both directions for characterization. Furthermore, previously identified QTL from CattleQTLdb  and a tabulated list of QTL for milk production and mastitis  were used to locate previously known QTL affecting traits of economic importance.
We would like to acknowledge Dairy Futures Cooperative Research Centre (Melbourne, Australia) for funding and the American Jersey Cattle Association (Reynoldsburg, OH), Dairy Records Management Systems (Raleigh, NC) and Council on Dairy Breeding (Reynoldsburg, OH) for providing the US genotypes and phenotypes. The Australian Dairy Herd Improvement Scheme provided the Australian and New Zealand phenotypes and genotypes. We gratefully acknowledge the Dairy Futures Cooperative Research Centre (Melbourne, Australia) for providing funding to genotype the Australian females and Dr. Richard Spelman from Livestock Improvement Corporation (Hamilton, New Zealand) for providing the genotypes of the New Zealand bulls used in this study. Lastly, visiting scholar expenses at the Department of Environment and Primary Industries and La Trobe University were funded by the USDA-Food and Agricultural Sciences National Needs Graduate and Postgraduate Fellowship (NNF) Grants Program international travel section.
- Lund MS, de Ross SPW, de Vries AG, Druet T, Ducrocq V, Fritz S, et al. A common reference population from four European Holstein populations increases reliability of genomic predictions. Genet Sel Evol. 2011;43:43.PubMed CentralPubMedView ArticleGoogle Scholar
- Muir B, Doormaal BV, Kistemaker G. International genomic cooperation – north american perspective. Interbull Bull; Paris. 2010;41:71–6.Google Scholar
- Hayes BJ, Lien S, Nilsen H, Olsen HG, Berg P, Maceachern S, et al. The origin of selection signatures on bovine chromosome 6. Anim Genet. 2008;39:105–11.PubMedView ArticleGoogle Scholar
- Flori L, Fritz S, Jaffrezic F, Boussaha M, Gut I, Heath S, et al. The genome response to artificial selection: a case study in dairy cattle. PLoS One. 2009;4(8):e6595.PubMed CentralPubMedView ArticleGoogle Scholar
- MacEachern S, Hayes BJ, McEwan J, Goddard M. An examination of positive selection and changing effective population size in Angus and Holstein cattle populations (Bos taurus) using a high density SNP genotyping platform and the contribution of ancient polymorphism to genomic diversity in Domestic cattle. BMC Genomics. 2009;10:181.PubMed CentralPubMedView ArticleGoogle Scholar
- Qanbari S, Pimentel ECG, Tetens J, Thaller G, Lichtner P, Sharifi AR, et al. A genome-wide scan for signatures of recent selection in Holstein cattle. Anim Genet. 2010;41:377–89.PubMedGoogle Scholar
- Stella A, Ajmone-Marsan P, Lazzari B, Boettcher P. Identification of selection signatures in cattle breeds selected for dairy production. Genetics. 2010;185:1451–61.PubMed CentralPubMedView ArticleGoogle Scholar
- Kim E, Cole JB, Huson H, Wiggans GR, Van Tassell CP, Crooker BA, et al. Effect of artificial selection on runs of homozygosity in U.S. Holstein cattle. PLoS ONE. 2013;8(11):e80813.PubMed CentralPubMedView ArticleGoogle Scholar
- Kemper KE, Saxton SJ, Bolormaa S, Hayes BJ, Goddard ME. Selection for complex traits leaves little or no classic signatures of selection. BMC Genomics. 2014;15:241.View ArticleGoogle Scholar
- Kim Y, Stephan. Detecting a local signature of genetic hitchhiking along a recombining chromosome. Genetics. 2002;160:765–77.PubMed CentralPubMedGoogle Scholar
- Maynard-Smith JM, Haigh J. The hitch-hiking effect of a favorable gene. Genet Res. 1974;23(1):23–5.View ArticleGoogle Scholar
- Turchin MC, Chiang CWK, Palmer CD, Sankararaman S, Reich D, Hirschhorn JN. Evidence of widespread selection on standing variation in Europe at height- associated SNPs. Nat Genet. 2012;44(9):1015–9.PubMed CentralPubMedView ArticleGoogle Scholar
- Broman KW, Weber JL. Long homozygous chromosomal segments in reference families from the centre d’Étude du polymorphisme humain. Am J Hum Genet. 1999;65:1493–500.PubMed CentralPubMedView ArticleGoogle Scholar
- MacLeod IM, Meuwissen TH, Hayes BJ, Goddard ME. A novel predictor of multilocus haplotype homozygosity: comparison with existing predictors. Genet Res (Camb). 2009;91:413–26.View ArticleGoogle Scholar
- Kirin M, McQuillan R, Franklin CS, Campbell H, McKeigue PM, Wilson JF. Genomic runs of homozygosity record population history and consanguinity. PLoS One. 2010;5:e13996.PubMed CentralPubMedView ArticleGoogle Scholar
- Interbull: Interbull routine genetic evaluation for dairy production traits, April 2014. http://www.interbull.org/web/static/mace_evaluations_archive/eval/prod-apr14.html. Accessed July. 19, 2014.
- Kolver ES, Roche JR, DeVeth MJ, Thorne PL, Napper AR. Total mixed rations versus pasture diets: evidence for a genotype x diet interaction in dairy cow performance. Proc NZ Soc Anim Prod. 2002;62:246–51.Google Scholar
- Pryce JE, Johnston J, Hayes BJ, Sahana G, Weigel KA, McParland S, et al. Imputation of genotypes from low density (50,000 markers) to high density (700,000 markers) of cows from research herds in Europe, North America, and Australasia using 2 reference populations. J Dairy Sci. 2014;97:1799–811.PubMedView ArticleGoogle Scholar
- Tyrisevä AM, Meyer K, Fikse W, Ducrocq V, Jakobsen J, Lidauer M, et al. Principal component approach in variance component estimation for international sire evaluation. Genet Sel Evol. 2011;43:21.PubMed CentralPubMedView ArticleGoogle Scholar
- Pryce JE, Arias J, Bowman PJ, Davis SR, Macdonald KA, Waghorn GC, et al. Accuracy of genomic predictions of residual feed intake and 250-day body weight in growing heifers using 625,000 single nucleotide polymorphism markers. J Dairy Sci. 2012;95:2108–19.PubMedView ArticleGoogle Scholar
- de Roos AP, Hayes BJ, Spelman RJ, Goddard ME. Linkage disequilibrium and persistence of phase in Holstein-Friesian, Jersey and Angus cattle. Genetics. 2008;179(3):1503–12.PubMed CentralPubMedView ArticleGoogle Scholar
- Stachowicz K, Sargolzaei M, Miglior F, Schenkel FS. Rates of inbreeding and genetic diversity in Canadian Holstein and Jersey cattle. J Dairy Sci. 2011;94:5160–75.PubMedView ArticleGoogle Scholar
- Haile-Mariam M, Bowman PJ, Goddard ME. A practical approach for minimising inbreeding and maximising genetic gain in dairy cattle. Genet Sel Evol. 2007;39:369–89.PubMed CentralPubMedView ArticleGoogle Scholar
- Van Raden PM, Olson KM, Null DJ, Sargolzaei M, Winters M, van Kaam J B.C.H.M: Interbull: Reliability Increases from Combining 50,000- and 777,000-Marker Genotypes from Four Countries, May 2012. https://journal.interbull.org/index.php/ib/article/view/1266. Accessed July 19, 2014.
- Yang J, Benyamin B, McEvoy BP, Gordon S, Henders AK, Nyhold DR, et al. Common SNPs explain a large proportion of the heritability for human height. Nat Genet. 2010;42:565–9.PubMed CentralPubMedView ArticleGoogle Scholar
- Weir BS, Cockerham CC: Estimating F-statistics for the analysis of population structure. Evolution 1984, 38 L1358-1370.Google Scholar
- Doerge RW, Churchill GA. Permutation tests for multiple loci affecting a quantitative character. Genetics. 1996;142:285–94.PubMed CentralPubMedGoogle Scholar
- Garrick DJ, Taylor JF, Fernando RL. Deregressing estimated breeding values and weighting information for genomic regression analyses. Genet Sel Evol. 2009;41:55.PubMed CentralPubMedView ArticleGoogle Scholar
- Benjamini Y, Hochberg Y. Controlling the false discovery rate—a practical and powerful approach to multiple testing. J R Stat Soc B. 1995;57:289–300.Google Scholar
- Falconer DS, Mackay TFS. Introduction to quantitative genetics. 4th ed. New York, NY: Longman Scientific and Technical; 1996.Google Scholar
- Purfield DC, Berry DP, McParland S, Bradley DG. Runs of homozygosity and population history in cattle. BMC Genet. 2012;13:70.PubMed CentralPubMedView ArticleGoogle Scholar
- Ferenčaković M, Sölkner J, Curik I. Estimating autozygosity from high-throughput information: effect of SNP density and genotyping errors. Genet Sel Evol. 2013;45:42.PubMed CentralPubMedView ArticleGoogle Scholar
- Bjelland DW, Weigel KA, Vukasinovic N, Nkrumah JD. Evaluation of inbreeding depression in Holstein cattle using whole-genome SNP markers and alternative measures of genomic inbreeding. J Dairy Sci. 2013;96:4697–706.PubMedView ArticleGoogle Scholar
- Thomsen B, Horn P, Panitz F, Bendixen E, Petersen AH, Holm L-E, et al. A missense mutation in the bovine SLC35A3 gene, encoding a UDP-N-acetylglucosamine transporter, causes complex vertebral malformation. Genome Res. 2006;16(1):97–105.PubMed CentralPubMedView ArticleGoogle Scholar
- Qanbari S, Pausch H, Jansen S, Somel M, Strom TM, Fries R, et al. Classic selective sweeps revealed by massive sequencing in cattle. PLoS Genet. 2014;10(3):e1004148.PubMed CentralPubMedView ArticleGoogle Scholar
- Elsik CG, Tellam RL, Worley KC. The genome sequence of taurine cattle: a window to ruminant biology and evolution. Science. 2009;324:522–8.PubMed CentralPubMedView ArticleGoogle Scholar
- Cole JB, Wiggans GR, Ma L, Sonstegard TS, Lawlor TJ, Crooker BA, et al. Genome-wide association analysis of thirty one production, health, reproduction and body conformation traits in contemporary U.S. Holstein cows. BMC Genomics. 2011;12:408.PubMed CentralPubMedView ArticleGoogle Scholar
- Ramey HR, Decker JE, McKay SD, Rolf MM, Schnabel RD, Taylor JF. Detection of selective sweeps in cattle using genome-wide SNP data. BMC Genomics. 2013;14:382.PubMed CentralPubMedView ArticleGoogle Scholar
- Castermans D, Wilquet V, Parthoens E, Huysmans C, Steyaert J, Swinnen L, et al. The neurobeachin gene is disrupted by a translocation in a patient with idiopathic autism. J Med Genet. 2003;40:352–6.PubMed CentralPubMedView ArticleGoogle Scholar
- Bionaz M, Loor JJ. Gene networks driving bovine milk fat synthesis during the lactation cycle. BMC Genomics. 2008;9:366.PubMed CentralPubMedView ArticleGoogle Scholar
- Rauw WM, Kanis E, Noordhuizen-Stassen EN, Grommers FJ. Undesirable side effects of selection for high production efficiency in farm animals: A review. Livest Prod Sci. 1998;56:15–33.View ArticleGoogle Scholar
- Parker Gaddis KL, Cole JB, Clay JS, Maltecca C. Genomic selection for producer-recorded health event data in US dairy cattle. J Dairy Sci. 2014;97:3190–9.PubMedView ArticleGoogle Scholar
- Cole JB, VanRaden PM, Multi-State Project S-1040. Net merit as a measure of lifetime profit: 2010 revision. In: AIPL Research Report 2010, NM$4 (12–09). Beltsville, MD: USDA Animal Improve- ment Programs Laboratory (AIPL); 2010.Google Scholar
- Sordillo LM. Factors affecting mammary gland immunity and mastitis susceptibility. Livest Prod Sci. 2005;98:89–99.View ArticleGoogle Scholar
- Hogan J, Smith LK. Coliform mastitis. Vet Res. 2003;34:507–19.PubMedView ArticleGoogle Scholar
- Wray NR, Goddard ME. Increasing long-term response to selection. Genet Sel Evol. 1994;26:431–51.PubMed CentralView ArticleGoogle Scholar
- Meuwissen THE. Maximising the response of selection with a pre- defined rate of inbreeding. J Anim Sci. 1997;75:934–40.PubMedGoogle Scholar
- Henryon M, Berg P, Sørensen AC. Invited review: animal-breeding schemes using genomic information need breeding plans designed to maximise long-term genetic gains. Livest Sci. 2014;166:38–47.View ArticleGoogle Scholar
- Hickey JM, Kinghorn BP, Tier B, Clark SA, van der Werf JHJ, Gorjanc G. Genomic evaluations using similarity between haplotypes. J Anim Breed Genet. 2013;130:259–60.PubMedView ArticleGoogle Scholar
- de Cara MAR, Fernádez J, Toro MA, Villanueva B. Using genome-wide information to minimize the loss of diversity in conservation programmes. J Anim Breed Genet. 2011;128:456–64.PubMedView ArticleGoogle Scholar
- Engelsma KA, Veerkamp RF, Calus MPL, Windig JJ. Consequences for diversity when prioritizing animals for conservation with pedigree or genomic information. J Anim Breed Genet. 2011;128:473–81.PubMedView ArticleGoogle Scholar
- Pryce JE, Hayes BJ, Goddard ME. Novel strategies to minimize progeny inbreeding while maximizing genetic gain using genomic information. J Dairy Sci. 2012;95:377–88.PubMedView ArticleGoogle Scholar
- Sonesson AK, Woolliams JA, Meuwissen TH. Genomic selection requires genomic control of inbreeding. Genet Sel Evol. 2012;44:27.PubMed CentralPubMedView ArticleGoogle Scholar
- Clark SA, Kinghorn BP, Hickey JM, van der Werf JHJ. The effect of genomic information on optimal contribution selection in livestock breeding programs. Genet Sel Evol. 2013;45:44.PubMed CentralPubMedView ArticleGoogle Scholar
- Pryce JE, Haile-Mariam M, Goddard ME, Hayes BJ. Identification of genomic regions associated with inbreeding depression in Holstein and Jersey dairy cattle. Genet Sel Evol. 2014;46:71.PubMed CentralPubMedView ArticleGoogle Scholar
- Erbe M, Hayes BJ, Matukumalli LK, Goswami S, Bowman PJ, Reich M, et al. Improving accuracy of genomic predictions within and between dairy cattle breeds with imputed high-density single nucleotide polymorphism panels. J Dairy Sci. 2012;95:4114–29.PubMedView ArticleGoogle Scholar
- Browning SR, Browning BL. Rapid and accurate haplotype phasing and missing data inference for whole genome association studies using localized haplotype clustering. Am J Hum Genet. 2007;81:1084–97.PubMed CentralPubMedView ArticleGoogle Scholar
- R: A language and environment for statistical computing. [http://www.R-project.org/]
- Gilmour AR, Gogel BJ, Cullis BR, Thompson R: ASReml User Guide Release 3.0. 2009 Hemel Hempstead, UK: VSN International Ltd.Google Scholar
- Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID Bioinformatics Resources. Nature Protoc. 2009;4(1):44–57.View ArticleGoogle Scholar
- Huang DW, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37(1):1–13.PubMed CentralView ArticleGoogle Scholar
- Warde-Farley D, Donaldson SL, Comes O, Zuberi K, Badrawi R, Chao P, et al. The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res. 2010;38(Suppl):W214–20.PubMed CentralPubMedView ArticleGoogle Scholar
- Hu ZL, Park CA, Wu XL, Reecy JM. Animal QTLdb: an improved database tool for livestock animal QTL/association data dissemination in the post-genome era. Nucleic Acids Res. 2013;41:D871–9.PubMed CentralPubMedView ArticleGoogle Scholar
- Ogorevc J, Kunej T, Razpet A, Dovc P. Database of cattle candidate genes and genetic markers for milk production and mastitis. Anim Genet. 2009;40:832–51.PubMed CentralPubMedView ArticleGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.