Table 1 shows variable positions of aligned sequence haplotypes recovered from ancient samples. The dominant feature among ancient haplotypes is the numerical and topological predominance of the European modal sequence T3. Thirty samples show identity over at least the most variable 240bp of the control region, four samples exhibit 1bp difference and four samples two or more mismatches. T3, and its one or two base pair derivatives, has predominated in every modern European regional group studied thus far (Cymbron et al. 1999; Troy et al. 2001). Thus, this most basic element in the pattern of European bovine mtDNA diversity is one with a time depth stretching back at least to the Neolithic and cannot be ascribed to more recent aspects of breed or population history.

Whereas individual haplotypes may be shared between multiple populations and show limited geographical restriction, the affinities between regional samples may be illustrated by treating these as populations. Here, we calculated genetic distances (pairwise FST) among 10 B. taurus population groupings from European regions, Africa, the Near and Far East, plus the two ancient European sample groups. These were graphed in two dimensions using multidimensional scaling (figure 2). The Neolithic sequence sample, which comprises primarily northern European finds (30 samples with complete sequence over all 240bp), is placed adjacent with modern British, mainland European and Scandinavian samples, suggesting regional continuity of mtDNA phylogeography within the continent. The positioning of the Bronze Age sample is insecure due to small sample size (5).

Figure 1 graphs detailed patterns of diversity within the Neolithic and Bronze Age populations using reduced median networks (Bandelt et al. 1995). Samples from mainland Europe and the Near East are added for reference (Troy et al. 2001). In the Neolithic, as with modern European samples, the great majority of sequences root back to the phylogeny through T3, in a classic starburst pattern (figure 1d). As described previously, this pattern is a subset of that observed in the Near East, the Far East and Africa, where the T3 cluster is one component of a more complex array of at least five, regionally distributed B. taurus mtDNA clusters (Troy et al. 2001; Mannen et al. 2004). Thus, the basic pattern of northern European diversity, with a predominance of the T3 cluster and a minority of T samples, is one with a time depth stretching to the initial establishment of domestic populations on the continent.

It has been argued that the five clusters within modern B. taurus mtDNA each represent a single domesticated sequence type, with derived haplotypes displaying mutations accumulated since domestication. The networks around these are starlike, suggesting that they are the products of population expansion, which is supported by significantly negative Tajima's (1996) D and Fu's (1996) F_s values (table 2); samples grouped by population display similar values (not shown). A reasonable hypothesis is that the population expansions recorded were induced by domestication. This would predict that the time to the common ancestor within each cluster should be of the order of 10000 years. This may be calculated using the statistic ρ, the mean number of mutations from the central founder sequence to lineages within each cluster. This gives an unbiased estimate, and a central 95% credible interval may be calculated assuming a true starlike phylogeny and a Poisson distribution for the mutational process (Richards et al. 2000). Calibration of the control region mutation rate is difficult, but has previously been given as 38% per million years (Myr) for this region using a comparison with bison sequence (Troy et al. 2001). This accords well with a recent calibration of bison control region evolution (685bp fragment, 32%Myr⁻¹, 95% CI of 23–41%), which used a more informed Bayesian phylogenetic analysis of an extensive radiocarbon-dated dataset (Shapiro et al. 2004).