Phylogeographic inference can be a powerful tool in reconstructing species' evolutionary histories; however, although inferred phylogeographic patterns should depend in part on the underlying types and rates of mutations, the effects of different types of mutations have seldom been quantified. In this study we identified two chloroplast minisatellites in the common reed Phragmites australis, and showed that these are more variable than chloroplast microsatellites. We then recreated parsimony networks of the global phylogeography of P. australis based on data that either included or excluded repetitive sequences (minisatellites and microsatellites), thereby illustrating the influence that these repetitive sequences can have on large-scale phylogeographic inference. The resulting networks differed in the numbers of mutational steps, degrees of uncertainty, and total numbers of haplotypes. In addition, the suggested ancestor-descendant relationships among lineages changed substantially depending on whether repetitive sequences were included. We therefore caution against the inclusion of repetitive sequences in large-scale networks because of their high potential for homoplasy. Nevertheless, we advocate the inclusion of repetitive sequences in other analyses: specifically, we show that the ratio of mutations in repetitive vs. non-repetitive regions can provide insight into the relative ages of lineages.
© 2010 Blackwell Publishing Ltd.