The Auk 113(4):784-801, 1996 EVOLUTIONARY RELATIONSHIPS AMONG EXTANT ALBATROSSES (PROCELLARIIFORMES: DIOMEDEIDAE) ESTABLISHED FROM COMPLETE CYTOCHROME-B GENE SEQUENCES GARY B. NUNN, •'6JOHN COOPER,2 PIERREJOUVENTIN, 3 CHRIS J. R. ROBERTSON,4 AND GRAHAM G. ROBERTSONs XDepartmentof Ornithology,American Museum of NaturalHistory, Central Park West at 79th Street, New York, New York 10024, USA; 2PercyFitzPatrick Institute of AfricanOrnithology, University of CapeTown, Rondebosch7700, South Africa; 3CentreNational de la RechercheScientifique, Centre D'Etudes Biologique de Chize, F79360 Villiers en Bois,France; 4Departmentof Conservation,Science and Research Division, ConservationSciences Centre, Wellington, New Zealand;and 5AustralianAntarctic Division, Channel Highway, Kingston, Tasmania, Australia ABSTR•Cr.--Completemitochondrial cytochrome-b gene sequences(1,143 bp) were deter- mined from the 14 extant speciesin the Diomedeidae(albatrosses and mollymawks)and in two outgroupspecies from the Procellariidae(petrels and shearwaters). Phylogenetic analysis using maximumparsimony and maximumlikelihood methods identified a singlebest-sup- ported hypothesisof evolutionaryrelationships within the Diomedeidae,namely that two lineagesarose early in the evolution of the Diomedeidae.A further bifurcationin each of theselineages resulted in four monophyleticgroups of albatrosses:(1) southernmollymawks, (2) sootyalbatrosses, (3) North Pacificalbatrosses, and (4) "great" albatrosses.Monophyly of the southernmollymawks (Diomedea bulleri, D. cauta,D. chlororhynchos,D. chrysostoma, and D. melanophris)and sootyalbatrosses (Phoebetria fusca and P. palpebrata)indicates that Diomedea is paraphyletic.Resurrection of two genera,dropped historically in taxonomyof the Diome- deidae, resultsin a total of four genera. Calibrationsbased on the fossil record indicate that cytochrome-bevolutionary rates in albatrossesare slow comparedwith thoseof mostmam- mals. Received21 August1995, accepted 10 May 1996. THE ALBATROSSESAND MOLLYMAWKS(Family The discovery of a small population of an Diomedeidae) are the most familiar and best undescribed"great" albatrosson Amsterdam Is- studiedgroup of procellariiform(or tube-nosed) land in the Indian Ocean (D. amsterdamensis; seabirdsdue largely to their highly philopatric Roux et al. 1983) brought the total number of nature and diurnal attendance at breeding lo- speciesto 14 (Sibley and Monroe 1990). Much calities, where their surface nests are easily controversyexists regarding the exacttaxonom- monitored (Warham 1990). The 13 traditionally ic status of D. amsterdamensis.In particular, its acceptedspecies of albatrossesare widely dis- relationship to subspecifictaxa of D. exulans, tributed throughout the southern oceans,the which breed on low-latitude islands in the North Pacific Ocean, and, in a single case,the southern Pacific and Atlantic Oceans, is not well tropical Pacific Ocean (Harris 1973, Harrison understood (Bourne 1989, Robertson and War- 1983, Marchant and Higgins 1990). Fossil evi- ham 1992). An affinity among thesetaxa seems dence of albatrossspecies present in the North likely because,to differing degrees, they share Atlantic Ocean (Lydekker 1891a,b; Wetmore the retention of dark juvenal or immatureplum- 1943), which are most similar to the extant Di- age as reproducingadults (plumage paedomor- omedea albatrus of the North Pacific Ocean, in- phosis). In contrast, populations of the larger- dicates that the Diomedeidae once were truly sized D. exulans, which occur on subantarctic cosmopolitanin oceanicdistribution. islandsat higher latitudes, have a white plum- age asbreeding adults.Additional geneticanal- ysis of individuals from all populations of ex- E-mail:[email protected] ulanswill help resolve the much-debated tax- 784 October 1996] AlbatrossMolecular Phylogeny 785 TABLE1. The introductionand principalchanges to describedgenera within the Diomedeidae. Authority Generaand changes Linnaeus (1758) Diomedeagert. nov. Reichenbach (1852) Thalassarchegen. nov.; Phoebastriagen. nov.; and Phoebetriagen. nov. Coues (1866) SubBurned Thalassarche and Phoebastffa into Diomedea Baird et al. (1884) Thalassagerongen. nov. Mathews (1912) ResurrectedThalassarche; Diomedella gen. nov.; Nealbatrusgen. nov. Murphy (1917) Rhothoniasubgen. nov. Mathews (1934) Resurrected Phoebastria Mathews and Hallstrom SubBurnedRhothonia into Diomedea;transferred taxon from Phoebastriato (1943) Julietatagen. nov. Mathews (1948) SubBurned all albatrosses into Diomedea Boetticher (1949) in Galapagornisgen. nov.; Laysanornisgert. nov.; Penthireniagert. nov. Jouaninand Mougin (1979) Alexander et al. (1965) Standardized use of Diomedea and Phoebetria onomy of the amsterdamensis-exulanscomplex (G. higher-level phylogeneticrelationships among Nunn unpubl. data). the 14 extantspecies of Diomedeidae.The choice Based on elements of biogeographic distri- of the mitochondrial cytochrome-b(cyt-b) gene bution, a simple allometricrelationship of wing as an evolutionary marker for our study was and tail length, and charactersof the divided basedon severalfactors. First, the completegene plates making up the rhamphothecaof the bill, sequence,as well as flanking regions, are well the 14 albatrossspecies fall into four natural characterizedin birds (Desjardins and Morals groups: (1) the southern mollymawks, (2) the 1990, Helm-Bychowskiand Cracraft 1993, Kor- North Pacificalbatrosses, (3) the "great" alba- negay et al. 1993, Nunn and Cracraft 1996) and trosses,and (4) the sooty albatrosses(Warham other vertebrate groups (Jermiin et al. 1994), 1990).On the basisof completeadult fuliginous enabling the design of oligonucleotideprimers plumage coloration, longer wedge-shapedtail, that amplifyby polymerasechain reaction (PCR) cuneate body form, and presenceof a colored in a broad phylogenetic range of birds. Second, fleshy sulcus separating the ramicorns of the the fast evolutionary rate of changein cyt-bhas lower mandible (a morphologicalfeature found proven most suitable for studying recently di- in other procellariiforms),a traditional hypoth- vergent groups (Meyer 1994) and in birds has esis of relationships within the Diomedeidae successfullyresolved relationships from the recognizesa simple demarcationof two genera: specieslevel (Richman and Price 1992, E. Smith the "primitive" sooty albatrossgenus Phoebetria et al. 1992, Blechschmidtet al. 1993) to generic and a more comprehensivegenus Diomedea that and familial levels (Krajewskiand Fetzner 1994, envelopes the North Pacific albatrosses,the Lanyonand Hall 1994,Murray et al. 1994).Third, "great" albatrosses,and the southern molly- cyt-bis one of the larger protein-codinggenes mawks (Coues1866, Peters 1931, Murphy 1936, in tl•e avian mitochondrial genome(Desjardins Alexanderet al. 1965,Jouanin and Mougin 1979, and Morals 1990), and, so far, presentsno prob- Warham 1990).Consideration of other morpho- lem of alignment amongbirds. Finally, the ex- logicalfeatures historically have led to the split- panding use of cyt-bgene sequencesas a source ting of current members of Diomedeainto ad- of qualitative data for studiesof avian system- ditional generic groups (seeTable 1), although atics ensures that in the near future a dense none has gained common acceptance.Indeed, sampling of diversetaxa will be available,lead- Mathews (1948)went on to producean entirely ing to a common improvement of phylogeny- lumped albatrossgenus Diomedeacomprised of building within birds. all known species,including Phoebetria. Our genetic study explored several basic In view of both the traditional hypothesisof questionsconcerning the patterns and rates of albatrossrelationships based on a small number evolution among extant albatrossspecies: (1) Is of morphological features in this conservative the traditional classificationcongruent with a group, and the confusingtaxonomy within the molecularphylogeny, i.e. is Phoebetriaa sister- comprehensivegenus Diomedea,we used mi- group to the remaining Diomedeidae, as ten- tochondrial DNA sequences to investigate uously surmisedby a handful of "primitive" 786 Nu ET^•. [Auk, Vol. 113 characters(see Murphy 1936) that are shared i0 mM •5-mercaptoethanol;i mM eachof dGTP,dATP, with other petrels?(2) Does the molecularev- dTTP, and dCTP; ! M of each primer; i0-i,000 ng of idence offer any support for morphologically completegenomic DNA; and 2.5 units of Taq poly- definedgroups previously delimited within Di- merase(Thermus aquaticus DNA polymerase,Perkin- omedea(Coues 1866)? (3) What absolute evolu- Elmer-Cetus). The dsDNA products were visualized in a 2% NuSieve low-melting point agarosegel (FMC tionary rate calibrationis suggestedfor cyt-bin Bioproducts)containing 2 pg. ml • ethidium-bromide the Diomedeidae,and how doesthis compare (Maniatis et al. 1982). The dsDNA product was cut with other estimates of the molecular clock for directlyfrom the gel and resuspendedin 300/xLwater this gene (Irwin et al. 1991,Martin et al. 1992)? by heating to 73øCfor 15 min. Further primer pairswere usedto amplify double- METHODS strandedDNA subfragmentsfrom the isolatedcyt-b gene (i.e. Li4863/H15104, Li4863/Hi5298, L14990/ Studyorganisms.--Samples of fresh blood or liver H15298, L15236/H15505, Li5311/H15710, L15656/ tissue were collected from the extant speciesof al- H16065); for original primer descriptionssee Helm- batrossesand two speciesof the Procellariidae. Most Bychowskiand Cracraft(1993) and referencesthere- blood samples were collected from chicks or incu- in. A large degree of fragment overlap, as well