Aim Deciphering the complex colonization history of island archipelagos is greatly facilitated by comprehensive phylogenies. In this study we investigate the phylogeny and biogeography of the insular reed-warblers (genus Acrocephalus) of the tropical Pacific Ocean, from Australia to eastern Polynesia. Location Oceania. Methods We used sequences of mitochondrial DNA (cytochrome b, ND2 and ATP8 genes) to infer the colonization patterns of reed-warblers endemic to Pacific islands and Australia. We sampled all known taxa of Acrocephalus in the Pacific except A. luscinius nijoi, for which no sample was available. Most taxa were represented by toe-pad samples from museum specimens collected in the 19th and 20th centuries. With a few exceptions, several specimens per taxon were sequenced independently in two institutions (Smithsonian Institution and Natural History Museum of Geneva). Results Our data indicate that Pacific reed-warblers do not form a monophyletic group, because A. luscinius luscinius from Guam falls outside the main Pacific radiation. The remaining Pacific taxa are divided into two clades: one clade includes all the reed-warblers from Micronesia (except Guam) and Australia, and two Polynesian taxa from the Line Islands and the southern Marquesas; the other clade includes all remaining Polynesian taxa. The taxa endemic to three archipelagos (Mariana, Marquesas and Society islands) are polyphyletic, suggesting several independent colonizations. Main conclusions Our results provide evidence for a complex pattern of colonization of the Pacific by reed-warblers. Calibration analyses suggest that reed-warbler lineages are much younger than the ages of the islands they occupy. Several remote archipelagos were colonized independently more than once. Consequently, we infer that the colonization of reed-warblers in the Pacific did not follow a regular, stepping-stone-like pattern. The phylogeny also suggests a previously undetected case of reverse colonization (from island to continent) for the Australian lineage and indicates that A. luscinius, as currently defined, is not monophyletic. We discuss the supertramp strategy of reed-warblers in the Pacific and show that, although Pacific reed-warblers meet some of the supertramp criteria in their aptitude for colonizing remote archipelagos, their life history characteristics do not fit the model.
