Abstract

We used complete mitochondrial cytochrome b sequences, largely obtained from sequences deposited with GenBank, supplemented by published sequences, from most genera and most species among the Procellariiformes to infer their phylogeny and molecular taxonomy. We analysed both issues of higher-level relationships within the order, and questions of the correct classification of taxa at the levels of genus, subgenus, species and subspecies. Nucleotide sequence data of cytochrome b are insufficient to resolve all higher-relationship issues, which must await the analysis of additional mtDNA and nuclear sequence data, but they do suggest some striking new findings. Sequence and distance data allow us to make judgments about the boundaries between taxa at various levels that are less arbitrary in these matters than those based on morphological or phylogenetic data alone. Working within the multidimensional Biological Species Concept, we reject the recently proposed splits among albatross species, and lump D. amsterdamensis as a subspecies of Diomedea exulans. A strong relationship of the storm-petrels to the albatrosses is apparent. We subdivide the storm-petrels into two subfamilies, Hydrobatinae and Oceanitinae. Presently, Oceanodroma is paraphyletic, and is regrouped into four genera: Hydrobates (of which Oceanodroma becomes a junior synonym), Cymochorea, Halocyptena, and Thalobata. In the fulmar clade, Macronectes halli should be merged with M. giganteus. The shearwaters formerly assigned to Puffinus apparently cluster into two major clades at the generic level: Puffinus and Ardenna. Puffinus creatopus should become a subspecies of P. carneipes. Lugensa is a distinct genus, with its closest affinities to Pachyptila; and the evidence suggests reducing the prions to two species: P. turtur and P. vittata. However, the prion–Lugensa group remains incertae sedis. Bulweria groups with Pseudobulweria and Procellaria. Our data reveal that Pterodroma has internal structure at the subgeneric level; they establish the subgenera Pterodroma, Hallstroma and probably Cookilaria, but final analysis will require data from nearly all gadfly-petrel species. Amino acid distances are used to estimate times of divergence for the various branchings.