Royal Society Publishing


Domesticated cattle were one of the cornerstones of European Neolithisation and are thought to have been introduced to Europe from areas of aurochs domestication in the Near East. This is consistent with mitochondrial DNA (mtDNA) data, where a clear separation exists between modern European cattle and ancient specimens of British aurochsen. However, we show that Y chromosome haplotypes of north European cattle breeds are more similar to haplotypes from ancient specimens of European aurochsen, than to contemporary cattle breeds from southern Europe and the Near East. There is a sharp north–south gradient across Europe among modern cattle breeds in the frequencies of two distinct Y chromosome haplotypes; the northern haplotype is found in 20 out of 21 European aurochsen or early domestic cattle dated 9500–1000 BC. This indicates that local hybridization with male aurochsen has left a paternal imprint on the genetic composition of modern central and north European breeds. Surreptitious mating between aurochs bulls and domestic cows may have been hard to avoid, or may have occurred intentionally to improve the breeding stock. Rather than originating from a few geographical areas only, as indicated by mtDNA, our data suggest that the origin of domestic cattle may be far more complex than previously thought.


1. Introduction

The domestication of wild animals and plants allowed the shift from nomadic and hunter–gatherer behaviour to a settled agrarian way of life (Childe 1957; Diamond 2002). The question of whether animal domestication was limited geographically, or if it was characterized by multiple and independent events and/or local backcrossing with wild ancestors, is of general relevance for understanding the anthropological processes associated with the cultural transition from hunting to farming. The aurochs, or the wild ox (Bos primigenius), extinct since 1627, was once widespread throughout Europe, northern Africa, and southern Asia, where Palaeolithic rock and cave paintings indicate that it was important to humans as prey and perhaps also in rituals (Clutton-Brook 1999). Cattle were domesticated from aurochsen about 10 000 years ago and, as for most other domestic animal species, domestication was probably limited to a few regions including the Near East (Bos taurus) and Asia (Bos indicus) (Loftus et al. 1994; Bar-Yosef & Belfer-Cohen 1996; MacHugh et al. 1997; Clutton-Brook 1999). This is consistent with the observation that in cattle, as well as in the majority of domestic animal species, only a few major mitochondrial DNA (mtDNA) lineages exist (Bruford et al. 2003); the only exception to this situation is that found in the horse (Vilá et al. 2001, Jansen et al. 2003).

The appearance of domestic cattle in Europe coincides with the Neolithisation and the proposed human migration associated with it (Renfrew 1987); cattle became of great economic importance in Central Europe around 5500 BC with the Linearbandkeramik (LBK) culture (Benecke 1994). A predominant view, supported by mtDNA data, is that European cattle descend from aurochsen domesticated in the Near East and brought to Europe by the first farmers (Bököny 1974; Bailey et al. 1996; Troy et al. 2001). More generally, this model reflects domestication as a rare and difficult process that, once it has occurred, generally spreads through trade or migrating human populations rather than by the use of additional wild ancestors in new areas.

Sex differences in, for example, the feasibility of taming wild ancestors, the way early domesticates were exploited and the likelihood for participation in backcrosses are factors that may have left domestic animals with separate genetic legacies for sex-specific markers (MacHugh & Bradley 2001). Analyses based on Y chromosome data may therefore reveal previously unrecognized patterns of animal domestication (Lindgren et al. 2004). To this end we set out to analyse the origin of domestic cattle in Europe using Y chromosome markers.

2. Material and methods

(a) Analysis of modern samples

Male samples from modern cattle breeds were as specified in table 1. Eight introns, in total 3.5 kb, from the Y chromosomal genes DBY, UBE1Y, UTY and ZFY were amplified and sequenced using conserved exonic primers (Hellborg & Ellegren 2003); DNA from female cattle served as control in order to ensure Y-specific amplification. One male each of Bos indicus, bison Bison bison, and gaur Bos frontalis were also sequenced. Polymorphic sites were identified from sequence alignments using AutoAssembler and Sequence Navigator (Applied Biosystems). Subsequent genotyping of segregating sites in modern samples was based on re-sequencing or, for an indel polymorphism, on fragment size analysis. Median joining network was constructed using Network v.

View this table:
Table 1

Distribution of Y chromosome haplotypes among cattle breeds. Breeds are listed by country of current origin. Note that some breeds, e.g. Finnish and Swedish Ayrshire, Finnish Holstein-Friesian, and Swedish Jersey, are likely to represent breeds that have been imported in modern times.

(b) Ancient samples

Nearly all known finds of Holocene aurochsen and early domesticated cattle in central Europe come from archaeological excavations and are often found as isolated broken elements. A large series of find complexes from several excavations with reasonable to good bone preservation was reviewed and units originating from culturally well-dated and archaeologically closed contexts were selected for further analysis (table 2). Most samples originated from Saxony, a restricted region between the cities of Dresden and Leipzig in Germany (stored at Landesamt für Ärschäologie mit Landesmuseum für Vorgeschichte in Dresden). This area belongs to the heartland of the Linearbandkeramik (LBK) culture, the first Early Neolithic farmers (Benecke 1994), and can be regarded as having been a topographically and culturally coherent region during this period. Moreover, six samples of Pleistocene and early Holocene aurochsen were available from southern Scandinavia, five from northern Italy (Natural History Museum in Stockholm; Ekström 1993), and one from Austria (Wien Naturhistorisches Museum, Archäologisch—Zoologische sammlung), respectively. The material was measured according to the standard used in archaeozoology (von den Driesch 1976). The distinction between aurochsen and domestic cattle is based on the size and robustness of the bones only, although during the Neolithic there was a large overlap in the size-distribution of the two. In this study, only those bones that fall clearly outside the accepted size-range of prehistoric domesticates in the region (Dohle 1994, Elburg 1999) are classified as B. primigenius. Note that Scandinavian and Italian samples are from Pleistocene or early Holocene sediments and are thus pre-domestic.

View this table:
Table 2

Description of ancient specimen and results from single nucleotide polymorphism (SNP) typing.

(c) Genetic analysis of ancient samples

Because amplification of nuclear DNA in ancient remains is a challenging task, we sought to increase amplification success by selecting samples only from bones of excellent gross morphological preservation, using a DNA preparation method based on selective enrichment and concentration of the target sequence, and by the design of very short amplicons (42–55 bp). Moreover, we used pyrosequencing as a means for single nucleotide polymorphism (SNP) detection with which method it is possible directly to monitor traces of contamination. A detailed description of the DNA extraction protocol from prehistoric samples is provided as electronic supplementary material.

Authentication of the results is a key aspect in analyses of ancient DNA. To this end we followed widely accepted guidelines and recommendations for stringent ancient DNA research (Cooper & Poinar 2000). Thus, all samples were independently replicated in the Uppsala laboratory as well as in a laboratory in Madrid (Centro Mixto UCM-ISCIII de Evolución and Comportamiento Humanos). Common measures to prevent contamination were used, such as a separated area for working with pre-PCR and ancient DNA, the wearing of protective clothing, UV-sterilization of all reagents, and cleaning of all working surfaces and tools with HCl and sodium hypochlorite. Primers were designed in a way so that they would not anneal to human DNA, and this was verified by amplification.

3. Results and discussion

(a) Y chromosome haplotypes in modern cattle breeds

We screened 3.5 kb of non-coding Y chromosome sequence in 20 male cattle from 12 European breeds, selected to represent a suite of breeds indigenous to different parts of Europe. Two co-segregating sites were found: an A/C SNP in UTY intron 19 and a 2 bp insertion–deletion polymorphism in ZFY intron 5 (table 3). These two markers, together forming two haplotypes (Y1 and Y2), were subsequently genotyped in a total of 180 domestic cattle from 45 European and three Anatolian breeds (table 1), revealing overall haplotype frequencies of 0.48 and 0.52, respectively (no intermediate was observed). However, it was obvious that haplotypes were non-randomly distributed among breeds. For 38 out of the 48 breeds only one haplotype was identified, indicating a narrow paternal genetic basis of most breeds. Importantly, there was a clear geographic structure in the distribution of the two haplotypes with Y2 at very high frequency (Iberia, France, and Switzerland) or fixed (Italy) in south European breeds, as well as in Anatolian breeds (figure 1). In contrast, Y1 dominated in north European breeds while central European breeds showed intermediate haplotype frequencies.

View this table:
Table 3

Description of Y chromosome haplotypes.

Figure 1

Map showing the distribution of the Y1 (open) and Y2 (filled) Y chromosome haplotypes among modern cattle breeds in Europe, defined by country of origin. Size of the sectors is defined by the number of animals identified within each region with either haplotype. Due to small sample size, data from France, Spain, and Portugal are combined into a single chart.

In theory, several explanations of the north–south gradient across Europe in the occurrence of the Y1 and Y2 haplotypes are possible. Adaptive genetic differences along latitudinal gradients (or between certain geographical areas; Beja-Pereira et al. 2004) as a response to selection can lead to clines in allele frequencies, in this case potentially manifested in Y chromosome haplotype frequencies. However, most genes on the mammalian Y chromosome are involved in male reproduction and it is not obvious how artificial selection during domestication would imply different selection regimes related to such traits across a geographical gradient in Europe. Another possibility is that Y1 and Y2 distributions are the result of two different migrations of stock from the Near East, along the northern Danubian route and an alternative Mediterranean southern route. The alternating prevalence of Y1 and Y2 in these routes could result from sampling and bottleneck effects, but would not explain why Y1 is absent in modern breeds from Anatolia.

A remaining interpretation is that the first domestic cattle brought to Europe from the Near East during Neolithisation carried Y2. When these cattle subsequently spread northwards through Europe, local hybridization with aurochsen bulls may have introgressed Y1 in the breeding population. Geographical structure in the distribution of Y chromosomal haplotypes among ancient aurochsen, as has been found for mtDNA haplotypes (Troy et al. 2001), would add support to this interpretation. Specifically, we should expect to see Y1 at high frequency among European aurochsen.

(b) Ancient DNA analysis

To test this idea we attempted amplification of Y chromosome polymorphisms in DNA prepared from 39 ancient bone samples of European Bos (from Germany, Sweden, Italy and Austria) dated from Late Pleistocene, i.e. pre-domestication, to Bronze Age (>9500–1000 BC). Twenty-one of the 39 samples successfully amplified for the UTY SNP distinguishing Y1 and Y2 (table 2). Out of these, 11 were aurochsen (>9500–4400 BC) based on pre-domestic dates or morphology, five were morphological intermediates between aurochsen and domesticates (4900–4400 BC) and six were early domesticates or of unknown status (5500–1000 BC). Twenty showed the UTY ‘C allele’ of Y1 while only one showed the ‘A allele’ of Y2 (table 2). This indicates that Y1 was at high frequency among European aurochsen prior to the arrival of domestic cattle and that it was also frequent among aurochsen contemporary with the first domestic cattle.

A phylogenetic analysis including Bos indicus (full-bred Sahiwal), bison and gaur reveals Y1 to be the derived form (see figure 3 in the electronic supplementary material). However, the observation of Y1 in Pleistocene and Neolithic European aurochsen excludes the possibility that Y1 has an origin in post-domestic mutation from Y2 haplotype, occurring and becoming fixed in northward migrating domestic populations. Moreover, the finding of Y1 in early domesticates from Germany (table 2) shows that the occurrence of Y1 in central European domestic cattle is not a recent consequence of breed structure or artificial selection. Overall, these data thus support the idea of European aurochs' introgression of haplotype Y1 into domestic cattle through local hybridization.

Four substitutions (in DBY and ZFY), plus a microsatellite length polymorphism (DBY), were found to distinguish the indicus (haplotype Y3) from all European taurus (table 3). This lends support to the independent origin, from separate domestication events, of taurine and indicine cattle, a pattern for which data from maternal (Loftus et al. 1994; MacHugh et al. 1997) and paternal (Hanotte et al. 1997) markers are thus concordant.

Neighbour-joining trees based on interpopulation divergence among European aurochsen, north and south European domesticates, and Bos indicus summarize the contrasting patterns of domestication revealed by Y chromosome and mtDNA data (figure 2). These trees show two aspects of the data. First, both mtDNA and Y haplotype data exhibit three divergent lineages. Second, the distribution of these lineages differ for maternal and paternal ancestry. In each case, the most divergent lineage is represented by Bos indicus. However, while domestic cattle from northern and southern Europe share the same mtDNA lineage, domestic cattle from northern Europe show closer affinity with aurochsen Y chromosome haplotypes sampled locally than with domestic southern European or Anatolian populations. Thus, while hundreds of assayed European domesticates show no maternal contribution from European aurochs (e.g. Troy et al. 2001), north European cattle Y chromosomes seem predominantly to be a local legacy of wild ox. It would be valuable to analyse aurochsen from Near East to formally test that Y1 was absent or rare in this region. Unfortunately, aurochs remains from Near East tend to be poorly preserved and amplification of mtDNA has proven difficult (D. G. Bradley, unpublished data).

Figure 2

Neighbour-joining trees based on interpopulation pairwise Fst divergence among European aurochsen, north and south European domestic cattle, and Bos indicus. The left-hand tree used Y SNP data and the right-hand mtDNA control region sequence. The Y SNP haplotypes used were: the aurochs sample described herein including those from animals of intermediate morphology; a north Europe domestic cattle sample of haplotypes from Sweden, Finland, Britain, Holland, and Germany; a south European domestic cattle sample from Switzerland, France, Portugal, Spain, Italy and Anatolia; and three pure African zebu populations were used as a Bos indicus outgroup. The mtDNA data were taken from and Cymbron et al. (1999) and Troy et al. (2001).

(c) Cattle domestication

Early domestic cattle were probably kept with little control compared to modern standards (Clutton-Brook 1999). There is little archaeological evidence of fencing and cattle barns, and strategies for the keeping of cattle are likely to have involved free-roaming herds without confined paddocks. To provide the animals with access to adequate sources of fodder, herds are likely to have been driven between suitable pastures created by clearing and burning woodland. During times without close cattle control, the possibility of unintentional mating between aurochsen and domestic cattle may have been difficult to avoid. If such hybridization took place, a genetic imprint on domestic cattle is most likely to have been left by crosses of aurochs bulls with domestic cows; calves from opposite crosses would not have been integrated with the domestic stock. Moreover, crosses may have been deliberately arranged to improve or increase the breeding stock.

Domestication has been regarded as a difficult process limited to a few regions from which domesticates subsequently spread (Clutton-Brook 1999; Diamond 2002). For most domestic animals, this model is supported by data from mtDNA where most haplotypes cluster in a few lineages only (Bruford et al. 2003). Our observations from European cattle suggest a more complex scenario in which local backcrosses and hybridization with wild ancestors, in areas other than that of the origin of domestication, have diversified the domestic gene pool. There is evidence that many domestic mammals, including cattle, show extensive levels of nucleotide diversity in autosomal DNA sequences (Vilá et al. 2005), seemingly at odds with a narrow genetic basis as indicated by mtDNA. Hybridization and backcrossing with wild ancestors may thus have been a common phenomenon during domestication, with the prevailing direction of such events being crosses of wild males with domestic females, meaning that it is not detected by analyses of mtDNA. Indeed, this is known in contemporary bovids such as the mithan (domesticated gaur), females of which are deliberately mated with wild gaur bulls.


Ancient samples were kindly provided by Judith Oexle, Uwe Reuter, Erich Pucher, Jonas Ekström, and Thomas Moers. Modern cattle samples were obtained from the European Cattle Genetic Diversity Consortium (Isaäc J. Nijman, Johannes A. Lenstra, John Williams, Paolo Ajmone, Georg Erhardt, Gaudenz Dolf, Juha Kantanen, Clemen Rodellar), Elizabeth Glass, Olivier Hanotte, and Kaj Sandberg. Jan Ekman, Jennifer Leonard, Per Persson and Carles Vilá are acknowledged for helpful discussion. This work was financed by The Swedish Research Council, The Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning, Arkeologisamfunde, German Academic Exchange Service, and a European Union Marie Curie fellowship to C.S. H.E. is a Royal Swedish Academy of Sciences research fellow supported by a grant from Knut and Alice Wallenbergs foundation.



View Abstract