Introduction

Cattle history, inferred from mitochondrial DNA (mtDNA) markers (Bradley et al., 1996; Loftus et al., 1994, 1999; Troy et al., 2001; Edwards et al., 2007a; Achilli et al., 2008), suggests that cattle (Bos taurus and Bos indicus) domestication involved at least two genetically distinct auroch (Bos primigenius) species in the Near East and great Indus Valley. Most European cattle resulted from the expansion of a small cattle population from the Near East after domestication, with one (T3) of the four major haplogroups (T, T1, T2 and T3) identified in the Near East predominating in the mainland continent. However, the history of European cattle may be more complex than what was previously thought. Recent studies have identified new mitochondrial haplogroups in European cattle (Achilli et al., 2008), and probable introgression events of local wild aurochs into domesticated B. taurus cattle (Beja-Pereira et al., 2006; Achilli et al., 2008). African cattle mtDNA lineages are mainly assigned to the T1 haplogroup, which is very rare in the Middle East and Anatolia. This could suggest an independent domestication event, although the hypothesis of an African demographic expansion of a few T1-like cattle domesticated in the Near East is still arguable (Beja-Pereira et al., 2006; Bradley and Magee, 2006). Regarding zebu cattle, mtDNA sequences allowed the identification of two major haplogroups: I1 and I2 (Baig et al., 2005; Lai et al., 2006; Lei et al., 2006; Achilli et al., 2008; Chen et al., 2010). These two mtDNA lineages are well represented in India, suggesting that zebu domestication involved at least two different wild female populations (Baig et al., 2005) or, more likely, a single domestication event in the Indus Valley with a subsequent introgression process of wild (I2) females into proto-domesticated herds (Chen et al., 2010).

The DNA studies regarding cattle Y-chromosomes are comparatively rare (Verkaar et al., 2004) and have mainly focused on the assessment of male-mediated migration patterns and admixture between B. taurus and B. indicus (Hanotte et al., 2000; Anderung et al., 2007; Edwards et al., 2007b) or the assessment of differences in diversity (Ginja et al., 2009; Kantanen et al., 2009). Recent analyses have identified five polymorphic sites on the cattle Y-chromosome (Götherström et al., 2005), allowing the identification of three clusters (Y1, Y2 and Y3) in contemporary cattle, with Y1 being more frequent in B. taurus from north-western Europe, Y2 being dominant in B. taurus found in southern Europe and Anatolian cattle, and Y3 being exclusive to B. indicus. The Y2 haplotypes would represent the taurine cattle domesticated in the Fertile Crescent, whereas the Y1 haplogroup would represent European aurochs. This would indicate a greater genetic influence of the primitive European aurochs on the formation of modern cattle breeds in Europe (Beja-Pereira et al., 2006; Achilli et al., 2008). However, ancient DNA does not generally support significant crossbreeding between local European wild cattle and domesticated cattle (Edwards et al., 2007a; Bollongino et al., 2008). In addition, medieval Scandinavian samples largely belonged to haplogroup Y2 and, overall, this scenario would not be consistent with a conserved signal from the time of domestication (Svensson and Götherström, 2008). Therefore, the strong geographical differentiation between the Y haplogroups found in modern cattle breeds in Europe requires further explanation.

Cattle Y-chromosome studies are also affected by a lack of powerful sources of information. There are limited numbers of informative segregating sites and polymorphic Y-specific microsatellites (Götherström et al., 2005; Ginja et al., 2009). Liu et al. (2003) reported the existence of several interspersed multilocus microsatellites (IMMs). These IMMs are loci that yield several amplified bands differing in size from the same male individual, using a single PCR reaction performed using a single primer pair. These loci are assumed to be interspersed and a result of the repetitive nature of the Y-chromosome. The male-specific region of the Y-chromosome is a mosaic of heterochromatic sequences and three classes of euchromatic sequences: X-transposed, X-degenerate and ampliconic (Skaletsky et al., 2003). The absence of recombination with the X-chromosome causes rapid degeneration by mutation, deletion and transposon invasion that accumulate through time due to the larger number of cell divisions required to produce male gametes, and due to inefficient repair (for example, Muller's ratchet) and inefficient selection (Charlesworth et al., 2005; Graves, 2006; Bachtrog and Andolfatto, 2006; Van Laere et al., 2008). The repetitive nature of the Y-chromosome has been hypothesised to be a mechanism for maintaining Y-chromosome genes in a non-recombining environment and may arise by a number of mechanisms, including sexual antagonism, genomic conflict and hemizygous exposure (Vallender and Lahn, 2004; Murphy et al., 2006). When a section is duplicated, it may result in a new locus that could generate new alleles at the locus site over time (Butler et al., 2005). In the hemizygous state, the latter will be difficult to identify and, therefore, only its presence or absence can be identified. The new allele may be present in some individuals but absent in others, allowing each locus to be treated as a polymorphic site. Therefore, the polymorphism can be assessed as the presence/absence of a given band across individuals. However, the value of IMMs in population genetics studies has not been explored yet (Luo et al., 2007). Here, we tested a set of IMMs to ascertain their usefulness for cattle phylogeny. The consistency of the information provided by the IMMs, along with the cattle history inferred from previous genetic analyses, will be discussed.

Materials and methods

Selection of IMMs

A total of five microsatellite loci, previously reported as IMMs by Liu et al. (2003), were tested in both taurine and zebu samples for male specificity, paternal compatibility and repeatability scoring to ascertain their performance in cattle Y-chromosome variability assessment (Pérez-Pardal et al., 2009a). The IMM behaviour was consistent across subspecies. When amplification occurred using female DNA or identical band amplification results were not obtained with different PCR conditions, the IMM was eliminated and was not considered for further analyses (Figure 1 and Supplementary Figure S2). The PCR amplification conditions are described in Table 1. The description of the pedigrees used to test the markers is given in Supplementary Figure S2. Laboratory analyses were carried out using Automatic Sequencers ABI-310 and ABI-3130 (Applied Biosystems, Foster City, CA, USA). Size characterisation of PCR products was carried out using GenMapper software (Applied Biosystems).

Figure 1
figure 1

Electropherograms of cattle Y-specific interspersed multilocus microsatellites (IMMs) UMN2405 and UMN2303. We show haplotypes J30, J34 and J8 (see Supplementary Table 2) that belong to haplogroups Y1, Y2 and Y3, respectively. Grey bars represent the loci assessed as polymorphic across all samples used. Polymorphism is assessed as the presence/absence of a given band across individuals. Examples of genotyping that were carried out are shown by zooming in on parts of the electropherograms. Band sizes are shown at the top of the zoomed-in areas. The presence or absence of a given band in a sample is assessed as 1 (at the top of the peaks) or 0.

Full size image
Table 1 Description of IMMs tested
Full size table

Statistical analyses

Observed haplotypes were analysed as follows:

  1. 1)

    Application of the Bayesian procedures implemented in the programme MrBayes 3.1. (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003). Two MCMC runs, starting from different random trees, were completed; each with 20 000 000 generations and four chains. All sample points before reaching convergence were discarded as burn-in samples. The remaining samples were used to generate a majority rule consensus tree, in which the percentage of samples recovering any particular clade represented the clade's posterior probability (Huelsenbeck and Ronquist, 2001).

  2. 2)

    Construction of biologically meaningful phylogenies of all genotyped samples using the program Network 4.5 (available at http://www.fluxus-engineering.com/). To avoid reticulation, a reduced median algorithm (Bandelt et al., 1995) was used to generate a .rmf file, and the median joining network method (Bandelt et al., 1999) was applied to this file. Following the authors recommendations (Forster et al., 2000), the ‘frequency >1’ option was applied to discard singly occurring Y types. Same weights were assigned to each haplotype.

  3. 3)

    Computation of correspondence analysis using Proc Corresp of the SAS/STAT package (SAS Institute, Cary, NC, USA).

Unbiased estimates of haplotype diversity and their variances (V(h)) were calculated for the whole sample and for each identified haplogroup as mentioned in the study by Nei (1987) (formulas 8.5 and 8.13 therein). The s.e. value of h was calculated by taking the square root of V(h).

The program Arlequin 3.11 (Excoffier et al., 2005) was used to calculate the analysis of molecular variance.

Divergence times were estimated using the ρ statistic (the average number of mutations from derived haplotypes to a haplotype designated as ancestral for the haplogroup; Forster et al., 1996) using the program NETWORK 4.5. The average distance to the node of interest (ρ) was transformed to absolute time estimates by multiplication (ρ × years per mutation), assuming the phylogenies constructed using median-joining algorithms, as implemented in NETWORK. As a phylogeny-based statistic, ρ offers the advantage of being unbiased by demographic processes. The sampling error of ρ was approximated as √ρ/n, where n denotes the sample size. No published mutation rate for IMMs is available. As IMMs are defined as dominant markers, they should most probably be considered as slowly mutating markers. Therefore, we computed divergence times using the mutation rate (0.00018 per generation) reported by Forster et al. (2000) for slowly mutating Y-chromosome markers. The generation interval was fitted to 4.84 years. This value is the mean value of 15 different estimates reported by Gutiérrez et al. (2003) in eight breeds of cattle; most of them managed under very traditional conditions and not subject to sire selection programs.

The determination of ancestral haplotypes was performed by computing an internal number of square differences (d2), computed as the average overall IMMs typed of where d2A is the parameter for the IMM A, ja and jb are the lengths in repeat units of alleles a and b at IMM A, and n is the total number of bands at which IMM A was scored (here, 25 for IMM UMN2403 and 23 for UMN2303). Assuming a strict stepwise mutation model, the lower the value of d2, the deeper the presence of the haplotype in the pedigree.

Results

We tested five IMMs reported by Liu et al. (2003) for the following: (i) amplification only with male DNA; (ii) polymorphism; (iii) paternal compatibility; and (iv) correct and repeatable scoring. Two of the tested IMMs (UMN2405 and UMN2303) fulfilled these requirements and were used further in this study. The IMM UMN2405 had a total of 30 bands, 25 of which were polymorphic. Of the 38 sites, 23 were polymorphic for UMN2303 (Table 1).

A total of 453 male individuals belonging to 52 cattle populations of Europe (351 samples), Africa (44 samples) and Asia (58 samples; Supplementary Table S1) were genotyped for the two selected IMMs. Altogether, they yielded a total of 146 haplotypes (haplotype diversity=0.321±0.005). Genotypes, haplotypes and their frequencies are provided in Supplementary Table S2.

Haplotypes were analysed by constructing evolutionary trees using Bayesian, network and correspondence analyses. All procedures allowed for the identification of three main haplogroups (Figure 2); two of these groups included only taurine samples, whereas the third was composed of only zebu samples. The consistency of these haplogroups with those previously identified by Götherström et al. (2005) using Y-specific single nucleotide polymorphisms was tested (see Supplementary Table S3). They were fully coincident with our haplogrouping. Therefore, the IMM-derived haplogroups will be referred to as Y1, Y2 (taurine) and Y3 (zebu). The Bayesian analysis provided a statistically high confidence value (0.76 between the taurine clusters and 1.0 among the taurine and zebu clusters). The observed haplogroups also showed high geographic consistency. The Y1 B. taurus haplogroup included those breeds located on the European Atlantic coastline and British Islands, and the Y2 B. taurus haplogroup included those breeds spread across continental Europe, the three European (Iberian, Italian and Balkan) peninsulas and Africa. The taurine Y-chromosomes are the only cattle Y-chromosome present in Europe and are the most frequent cattle chromosomes in Africa and Northern Asia (from the Eastern Caspian Sea to Mongolia and Japan). The Y3 B. indicus haplogroup was the only haplogroup present in India and was well represented in Western Africa.

Figure 2
figure 2

Graphical representation of genetic variability. (a) The phylogenetic tree constructed from the 146 identified haplotypes using the Bayesian procedures implemented in the program MrBayes 3.1. (b) A network tree constructed using the program Network 4.5. (c) The two dimensions were calculated on the polymorphism of the 146 Y-chromosome haplotypes identified through correspondence analysis. The three identified haplogroups were named Y1 (in blue), Y2 (in green) and Y3 (in orange). (d) Geographic map and frequency of haplogroups (Y1: blue; Y2: green; and Y3: orange) by country (see Supplementary Table S1). The numbers of samples and haplotypes included in each haplogroup were as follows: Y1: 114/54; Y2: 281/77; and Y3: 58/15.

Full size image

Haplogroup Y1 had the highest haplotype diversity (0.482±0.011), whereas both the zebu (Y3) and the main taurine (Y2) haplogroups had substantially lower haplotype diversities (0.276±0.006 and 0.268±0.014, respectively). The Y2 haplotypes identified in African cattle are not present on other continents. Interestingly, the only Japanese Kobe and Mongolian samples available to us tended to cluster separately within haplogroup Y2 (statistical confidence: 0.98; Supplementary Figure S1) with haplotypes identified in European breeds. Most Y1 haplotypes identified in Asian and African cattle are represented in Friesian cattle, suggesting that these haplotypes did not originate on these continents.

The average number of locus differences within and between haplogroups was calculated using the program Arlequin 3.11 (Excoffier et al., 2005). The higher mean pairwise differences within haplogroups for taurine were Y1 (5.08±2.48) and Y2 (5.00±2.44), whereas the difference was 2.58±1.40 in zebu (Y3) haplotypes. The corrected mean pairwise differences between Y1 and Y2 with Y3 were 16.67 and 17.39, respectively. The difference between Y1 and Y2 was 12.06.

Analysis of molecular variance (Excoffier et al., 2005) was carried out at the haplogroup, continent and geographic area levels (Table 2). Analysis of molecular variance showed that intercontinental genetic differentiation was relatively weak (explaining approximately one-third of the variability), whereas variability was greater than 50% among the defined geographical areas. Breeds accounted for less than 30% of the variation.

Table 2 Geographical structuring assessed using AMOVA
Full size table

Divergence times were estimated using ρ procedures (Forster et al., 1996) assuming three different mutation rates (see Materials and methods) and a generation interval of 4.84 years (Table 3). The estimated divergence times between haplogroups, Y3 and Y1, varied from 131 761.0±5378.0 to 658 780.5±26 889.0 years before present (yBP). However, divergences between haplogroups Y1 and Y2 (varying from 77 981.0±3802.8 to 389 890.5±19 013.4 yBP) and between Y2 and Y3 (varying from 75 292.0±3802.8 to 376 446.0±19 013.4 yBP) were substantially lower.

Table 3 Divergence times between ancestral haplotypes
Full size table

Discussion

To the best of our knowledge, we have, for the first time, used Y-specific IMMs to ascertain the history of an animal species from the male side. The benefits of using IMMs in cattle are clear; when compared with the available Y-specific microsatellites and single nucleotide polymorphisms (Götherström et al., 2005; Ginja et al., 2009; Kantanen et al., 2009; Pérez-Pardal et al., 2009a, 2009b), IMMs have higher numbers of polymorphisms and higher resolution. Very recently, using classical Y-specific microsatellites we have been able to distinguished two different Y2 haplotypic subfamilies in taurine cattle; one of them restricted to the African continent (Pérez-Pardal et al., 2009b). However, this set of markers was not powerful enough to correctly identify and differentiate haplogroups Y1 and Y2. This result underscores the need to identify new Y-specific markers to assess genetic relationships between different bovine Y-specific haplogroups (Pérez-Pardal et al., 2009b).

Moreover, the interest of this study extends beyond the bovine species, as IMMs have been identified in the Y-chromosomes of many other mammals, such as the family Felidae (Luo et al., 2007) and the genus Equus (Wallner et al., 2004). However, the use of IMMs has some constraints. First, their use is not straightforward: their selection, amplification, genotyping and scoring must be performed carefully. The second, and more important, problem is that their mutation model is currently unknown, which limits their use in evolutionary studies.

In this study, IMMs yield information about unanswered questions of cattle history from male legacies in light of previously available information: (i) what is the genetic influence of local aurochs on the formation of domesticated cattle?; (ii) do African cattle result from a local domestication event?; and (iii) are zebu cattle genetically uniform at the sire level?

Within Y2, an intriguing subcluster was identified. It includes cattle Y-chromosomes sampled in Mongolia, Japan, Northern Italy and Northern Atlantic Europe. Recently, Achilli et al. (2008) identified the new mtDNA haplogroup Q in Northern Italian and Korean cattle samples, possibly resulting from the introgression of local wild aurochs. These are largely the same geographical areas in which our subcluster was found. Therefore, a likely explanation for the Y2-subcluster identified in this study is the introgression of wild sires into domesticated cattle. This introgression process would have involved poorly differentiated wild sire populations and would have happened in very distant geographical areas. This is consistent with mtDNA data that demonstrate the existence of a wild cattle population sharing many genetic features from Europe to North Asia (Achilli et al., 2008). Overall, the recruitment processes of wild cattle are likely to have occurred in some parts of Europe and Northern Asia, giving genetic signatures at both mtDNA and Y-chromosome levels.

On the basis of the strong geographic signal obtained from modern samples, Götherström et al. (2005) proposed the possible introgression of local aurochs into European domesticated cattle. However, more recent evidence obtained from ancient and medieval samples does not support this hypothesis (Bollongino et al., 2008; Svensson and Götherström, 2008), and the alternative model of a sudden introduction of domestic cattle into Central Europe without significant crossbreeding with local wild cattle remains unchallenged (Edwards et al., 2007a; Bollongino et al., 2008). Therefore, practical fixation of Y1 haplotypes in Northern Atlantic European cattle breeds would more likely be the result of drift processes that occurred just before and at the moment of creation of modern cattle breeds approximately 200 yBP (Svensson and Götherström, 2008). However, this argument has an intuitive weakness because just after the formation of modern cattle breeds (approximately 100 yBP), the intense and sustained spreading of Y1-fixed cattle breeds (Friesian and British cattle) was facilitated by improved possibilities for long-distance trade and artificial insemination (approximately 50 yBP). The latter occurred all over the world, particularly across Europe. This dramatic and uncontrolled spreading process would have unavoidably erased differences among European taurine breeds at the Y-haplogroup level. It is noteworthy, that the Y1 samples identified in Africa and Northern Asia (Figure 2) are more likely to be the result of a recent introgression of European cattle rather than the expansion of a genetically heterogeneous sire population domesticated in the Fertile Crescent. Moreover, our results do not allow for the rejection of the hypothesis that introgression of local male aurochs partially influenced the formation of European cattle. This hypothesis is supported by two facts. The first is that the haplotypic diversity in the Y1 cluster is substantially higher than that observed for haplogroups Y2 and Y3, which are expected to be the result of domestication. This would suggest that the occurrence of a few local introgression events of Y1-related sires rather than a domestication process resulting in a much more intense bottleneck. The second fact is that ancient cattle management differences between northern and southern European cattle holders are supported by observed differences between northern and southern European cattle populations at the mtDNA level (Beja-Pereira et al., 2006), and evidence of co-evolution of cattle milk protein genes and human lactase in Northern Europe (Beja-Pereira et al., 2003). Our results support the idea that cattle management in Atlantic Europe might include the use of wild males with domesticated cows.

The question of the origins of both African taurine and zebu cattle has been the subject of intense debate and remains controversial. Our study confirms the existence of a Y2 haplotypic subfamily in African cattle (Pérez-Pardal et al., 2009b). Consistent with some archaeological findings (Wendorf and Schild, 1994, Bradley and Magee (2006) claim that African taurine cattle are not a simple subset of those domesticated in the Near East, due to the following three facts: (i) the mtDNA composition of cattle from these two areas is qualitatively different, with African samples belonging to the T1 haplogroup and showing a large proportion of unique haplotypes (Loftus et al., 1994; Bradley et al., 1996; Troy et al., 2001); (ii) there is a significant component of microsatellite variation within African cattle that seems to be indigenous in origin (Hanotte et al., 2002); and (iii) microsatellite variation also suggests that gene flow from the Near East into Northeast Africa was predominantly male-driven (MacHugh et al., 1997). Our results support this hypothesis because: (i) most haplotypes identified in African taurines are assigned to haplogroup Y2 and are not present in other continents; and (ii) the African Y2 samples show high haplotypic diversity (0.708±0.069). At the mtDNA level, African cattle do not present the lack of variation associated with a severe foundation bottleneck, which would be necessary to fix a haplotype (T1) family after the domestication event occurred in the Fertile Crescent (Bradley and Magee, 2006). This evidence can also be applied to our findings regarding the Y-chromosome.

The origin of African cattle is also linked to humped zebu cattle. Archaeological evidence suggests introgression of the humped zebu-like cattle into Africa approximately 3000 yBP (Payne, 1970). However, a major wave of B. indicus introgression in Africa may have started around the end of the seventh century AD in the Arab settlements along the East coast of Africa (Epstein, 1971), with subsequent spreading through zebu male-mediated interbreeding with African taurine females (Hanotte et al., 2000, 2002). Four of the six Y3 haplotypes identified in West African cattle tend to cluster together and are not present in Asian zebu sires (Supplementary Figure S1). This finding indicates a non-recent introgression of zebu cattle into Africa. Indeed, West African cattle share haplotypes with Asian zebu samples. However, West African cattle are often regarded as modern descendants of the founding African B. taurus cattle, which have been less affected by recent B. indicus introgression into the continent (Hanotte et al., 2002; Edwards et al., 2007b). Therefore, West African sanga cattle (the technical designation for the cross between taurine and zebu cattle) could be representative of a first ancient introgression of zebu cattle into Africa.

The scenario seen in our Asian zebu samples is quite different from that of cattle on the other continents. Most Y3 haplotypes observed in Asian zebu sires are present in the Indian populations, which leads to low haplotypic diversity (0.256±0.017). At the maternal genetic level, a major component of zebu cattle diversity is retained by two distinct mitochondrial haplogroups (I1 and I2), suggesting two different domestication events or a single domestication process involving two different wild female populations (Baig et al., 2005; Lai et al., 2006; Chen et al., 2010). In this study, the sampling size for Asian zebu sires was limited. However, Indian samples were obtained in areas in which both I1 and I2 mtDNA clades are well represented (Chen et al., 2010). Despite this, haplotypic diversity is extremely low, suggesting high genetic homogeneity of Asian male zebus. This would support the hypothesis of a single domestication event for B. indicus. Recently, Chen et al. (2010) suggested that I1-like zebu cattle were domesticated in a single event in the Indus Valley (approximately 6000 yBP), with subsequent recruitment of mitochondrial I2 wild zebu females approximately 3500 yBP. Our results suggest that this introgression event of zebu wild females into proto-domesticated herds would have occurred with previously domesticated zebu sires.

To the best of our knowledge, our IMM data, for the first time, allowed the estimation of divergence times between cattle subspecies (B. taurus and B. indicus) on the male genetic path. Recent analyses of the complete B. taurus and B. indicus mitochondrial genome sequences estimated their divergence times to be 1.7–2.0 million years (Hiendleder et al., 2008). However, using a fragment of the HVI region of mtDNA, most divergence times between the two subspecies were estimated to be 200 000–1 000 000 yBP (Loftus et al., 1994 and others). This mtDNA fragment gives information that is comparable in amount to that given by our IMMs. Current estimates are highly consistent with the generally accepted range of divergence times (Loftus et al., 1994). However, divergence times estimated between the Y1–Y2 and Y2–Y3 haplogroups are consistently lower than the estimated time between Y1–Y3. Overall, this would indicate a recent fragmentation of the wild populations that predated domestication, and a lack of differentiation between geographically distant wild sire populations.

In summary, we have demonstrated the usefulness of Y-specific IMMs to complement and add information to mtDNA-based studies regarding the history of the species. The divergent sire lineages, characterised here by Y-haplogroups, align with geographic areas and support previously suggested origins of cattle and directions of spread (Bradley et al., 1996; Troy et al., 2001; Baig et al., 2005; Lai et al., 2006; Edwards et al., 2007a; Achilli et al., 2008). In light of the above evidence, the most parsimonious interpretation of our data is that cattle have multiple paternal origins, possibly arising through multiple independent domestications. However, it is also possible that many of the paternal lineages identified in this study originated through introgression and not through separate domestication events.