Introduction

Mitochondrial DNA (mtDNA) variation has been used extensively for examining large-scale questions of human evolution, population structure and demographic history (Cann, Stoneking & Wilson 1987, Vigilant et al., 1991). Recently, however, the ability of mtDNA to resolve the close genetic relationships within relatively homogeneous populations such as the Europeans has been questioned (Bertranpetit et al., 1994; Pult et al., 1994). Using a new phylogenetic method we believe, on the contrary, that it is possible to reveal considerable structure within Europe and, by comparing geographical affinities and divergences between groups of mitochondrial lineages, examine different hypotheses about European colonization.

Mitochondria are maternally inherited and non-recombining, and the effectively haploid genome accumulates mutations faster than nuclear DNA. The most variable region of the mitochondrial genome is the 1122 bp non-coding control region between bp 16024 and 00576 (numbering after Anderson et al., 1981) within which the variation is concentrated in two regions (I and II) (Stoneking et al., 1991). As a population genetics method it differs from allele-frequency based surveys of nuclear encoded variants in several respects: i) the variation is very extensive and is not scrambled by recombination; ii) the effective population size is roughly one quarter that for nuclear variants which enhances the effect of drift; iii) being maternally inherited, only female lineages are relevant; iv) deducing the phylogenetic relationships within and between haplotype clusters is relatively straightforward and allows divergence and expansion times to be estimated.