Complete mitochondrial cytochrome-b gene sequences (1,143 bp) were determined from the 14 extant species in the Diomedeidae (albatrosses and mollymawks) and in two outgroup species from the Procellariidae (petrels and shearwaters). Phylogenetic analysis using maximum parsimony and maximum likelihood methods identified a single best-supported hypothesis of evolutionary relationships within the Diomedeidae, namely that two lineages arose early in the evolution of the Diomedeidae. A further bifurcation in each of these lineages resulted in four monophyletic groups of albatrosses: (1) southern mollymawks, (2) sooty albatrosses, (3) North Pacific albatrosses, and (4) "great" albatrosses. Monophyly of the southern mollymawks (Diomedea bulleri, D. cauta, D. chlororhynchos, D. chrysostoma, and D. melanophris) and sooty albatrosses (Phoebetria fusca and P. palpebrata) indicates that Diomedea is paraphyletic. Resurrection of two genera, dropped historically in taxonomy of the Diomedeidae, results in a total of four genera. Calibrations based on the fossil record indicate that cytochrome-b evolutionary rates in albatrosses are slow compared with those of most mammals. Received 21 August 1995, accepted 10 May 1996.

XDepartment of Ornithology, American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024, USA; 2Percy FitzPatrick Institute of African Ornithology, University of Cape Town, Rondebosch 7700, South Africa; 3Centre National de la Recherche Scientifique, Centre D'Etudes Biologique de Chize, F79360 Villiers en Bois, France; 4Department of Conservation, Science and Research Division, Conservation Sciences Centre, Wellington, New Zealand; and 5Australian Antarctic Division, Channel Highway, Kingston, Tasmania, Australia THE ALBATROSSES AND MOLLYMAWKS (Family Diomedeidae) are the most familiar and best studied group of procellariiform (or tube-nosed) seabirds due largely to their highly philopatric nature and diurnal attendance at breeding lo- calities, where their surface nests are easily monitored (Warham 1990). The 13 traditionally accepted species of albatrosses are widely dis- tributed throughout the southern oceans, the North Pacific Ocean, and, in a single case, the tropical Pacific Ocean (Harris 1973, Harrison 1983, Marchant and Higgins 1990). Fossil evi- dence of albatross species present in the North Atlantic Ocean (Lydekker 1891a,b; Wetmore 1943), which are most similar to the extant Di- omedea albatrus of the North Pacific Ocean, in- dicates that the Diomedeidae once were truly cosmopolitan in oceanic distribution. E-mail: gnunn@amnh.org The discovery of a small population of an undescribed "great" albatross on Amsterdam Is- land in the Indian Ocean (D. amsterdamensis; Roux et al. 1983) brought the total number of species to 14 (Sibley and Monroe 1990). Much controversy exists regarding the exact taxonom- ic status of D. amsterdamensis. In particular, its relationship to subspecific taxa of D. exulans, which breed on low-latitude islands in the southern Pacific and Atlantic Oceans, is not well understood (Bourne 1989, Robertson and War- ham 1992). An affinity among these taxa seems likely because, to differing degrees, they share the retention of dark juvenal or immature plum- age as reproducing adults (plumage paedomor- phosis). In contrast, populations of the larger- sized D. exulans, which occur on subantarctic islands at higher latitudes, have a white plum- age as breeding adults. Additional genetic anal- ysis of individuals from all populations of ex- ulans will help resolve the much-debated tax- TABLE 1. The introduction and principal changes to described genera within the Diomedeidae. 785 Authority Genera and changes Linnaeus (1758) Reichenbach (1852) Coues (1866) Baird et al. (1884) Mathews (1912) Murphy (1917) Mathews (1934) Mathews and Hallstrom (1943) Mathews (1948) Boetticher (1949) in Jouanin and Mougin (1979) Alexander et al. (1965) Diomedea gert. nov. Thalassarche gen. nov.; Phoebastria gen. nov.; and Phoebetria gen. nov. SubBurned Thalassarche and Phoebastffa into Diomedea Thalassageron gen. nov. Resurrected Thalassarche; Diomedella gen. nov.; Nealbatrus gen. nov. Rhothonia subgen. nov. Resurrected Phoebastria SubBurned Rhothonia into Diomedea; transferred taxon from Phoebastria to Julietata gen. nov. SubBurned all albatrosses into Diomedea Galapagornis gen. nov.; Laysanornis gert. nov.; Penthirenia gert. nov. Standardized use of Diomedea and Phoebetria onomy of the amsterdamensis-exulans complex (G. Nunn unpubl. data). Based on elements of biogeographic distri- bution, a simple allometric relationship of wing and tail length, and characters of the divided plates making up the rhamphotheca of the bill, the 14 albatross species fall into four natural groups: (1) the southern mollymawks, (2) the North Pacific albatrosses, (3) the "great" alba- trosses, and (4) the sooty albatrosses (Warham 1990). On the basis of complete adult fuliginous plumage coloration, longer wedge-shaped tail, cuneate body form, and presence of a colored fleshy sulcus separating the ramicorns of the lower mandible (a morphological feature found in other procellariiforms), a traditional hypoth- esis of relationships within the Diomedeidae recognizes a simple demarcation of two genera: the "primitive" sooty albatross genus Phoebetria and a more comprehensive genus Diomedea that envelopes the North Pacific albatrosses, the "great" albatrosses, and the southern molly- mawks (Coues 1866, Peters 1931, Murphy 1936, Alexander et al. 1965, Jouanin and Mougin 1979, Warham 1990). Consideration of other morpho- logical features historically have led to the split- ting of current members of Diomedea into ad- ditional generic groups (see Table 1), although none has gained common acceptance. Indeed, Mathews (1948) went on to produce an entirely lumped albatross genus Diomedea comprised of all known species, including Phoebetria. In view of both the traditional hypothesis of albatross relationships based on a small number of morphological features in this conservative group, and the confusing taxonomy within the comprehensive genus Diomedea, we used mi- tochondrial DNA sequences to investigate higher-level phylogenetic relationships among the 14 extant species of Diomedeidae. The choice of the mitochondrial cytochrome-b (cyt-b) gene as an evolutionary marker for our study was based on several factors. First, the complete gene sequence, as well as flanking regions, are well characterized in birds (Desjardins and Morals 1990, Helm-Bychowski and Cracraft 1993, Kor- negay et al. 1993, Nunn and Cracraft 1996) and other vertebrate groups (Jermiin et al. 1994), enabling the design of oligonucleotide primers that amplify by polymerase chain reaction (PCR) in a broad phylogenetic range of birds. Second, the fast evolutionary rate of change in cyt-b has proven most suitable for studying recently di- vergent groups (Meyer 1994) and in birds has successfully resolved relationships from the species level (Richman and Price 1992, E. Smith et al. 1992, Blechschmidt et al. 1993) to generic and familial levels (Krajewski and Fetzner 1994, Lanyon and Hall 1994, Murray et al. 1994). Third, cyt-b is one of the larger protein-coding genes in tle avian mitochondrial genome (Desjardins and Morals 1990), and, so far, presents no prob- lem of alignment among birds. Finally, the ex- panding use of cyt-b gene sequences as a source of qualitative data for studies of avian system- atics ensures that in the near future a dense sampling of diverse taxa will be available, lead- ing to a common improvement of phylogeny- building within birds. Our genetic study explored several basic questions concerning the patterns and rates of evolution among extant albatross species: (1) Is the traditional classification congruent with a molecular phylogeny, i.e. is Phoebetria a sister- group to the remaining Diomedeidae, as ten- uously surmised by a handful of "primitive" characters (see Murphy 1936) that are shared with other petrels? (2) Does the molecular ev- idence offer any support for morphologically defined groups previously delimited within Di- omedea (Coues 1866)? (3) What absolute evolu- tionary rate calibration is suggested for cyt-b in the Diomedeidae, and how does this compare with other estimates of the molecular clock for this gene (Irwin et al. 1991, Martin et al. 1992)? METHODS Study organisms.--Samples of fresh blood or liver tissue were collected from the extant species of al- batrosses and two species of the Procellariidae. Most blood samples were collected from chicks or incu- bating adults at nesting colonies to ensure known breeding provenance, although samples were taken from adult birds at sea for three of the species. All tissue samples were stored in 100% ethanol in the field and transported without freezing. The binomen, or current trinomen where appropriate, and collec- tion locality of each taxon in this study are listed in the Appendix. Based on simple morphological com- parisons (Murphy 1936), our two outgroup species from the Procellariidae (Southern Giant Petrel [Ma- cronectes giganteus ] and Gray Petrel [Procellaria cinerea ]) represent members of the most likely sister group to the Diomedeidae. DNA isolation.--We extracted DNA suitable for en- zymatic amplification by boiling a minute piece of tissue (<5 g) or suspended blood (5 L) for 15 min in 500 L of a 5% w/v Chelex-bead suspension (Sing- er-Sam et al. 1989, Walsh et al. 1991). After brief vor- texing to break up the tissue, the beads were pulse- centrifuged for a few seconds, and 300 L of the su- pernatant were removed as a source of template DNA. Mitochondrial cytochrome-b gene isolation and sequenc- ing.--The cyt-b gene and short flanking regions were amplified and isolated as a single fragment using the PCR primers L14863 5'-TTTGCCCTATCTATCCT- CAT-3' situated at the end of ND5 (numbered follow- ing the chicken mitochondrial genome [Desjardins and Morais 1990]) and designed from a consensus of Procellariidae and Diomedeidae partial ND5 sequenc- es (unpubl. data), and H15915 in tRNA-threonine (Ed- wards and Wilson 1990). The human mitochondrial genome (Anderson et al. 1981) numbered primer H15915 is equivalent to chicken mitochondrial num- ber H16065 (Desjardins and Morais 1990). Of several experimental temperature-cycling parameters tried we found that a four-step cycle most efficiently and re- liably amplified this approximately 1.2-kilobase frag- ment in a Peltier-effect thermocycler (MJ Research): viz. i rain at 94讈, i rain at 40讈, 1 min at 60讈, and 3 rain at 72讈, for 35 cycles. Amplifications were per- formed in 50-L reaction volumes containing 67 mM Tris-HC1 (pH 8.8); 6.7 mM MgC12; 16.6 mM (NH,)2SO4; i0 mM -mercaptoethanol; ! mM each of dGTP, dATP, dTTP, and dCTP; ! M of each primer; i0-i,000 ng of complete genomic DNA; and 2.5 units of Taq poly- merase (Thermus aquaticus DNA polymerase, Perkin- Elmer-Cetus). The dsDNA products were visualized in a 2% NuSieve low-melting point agarose gel (FMC Bioproducts) containing 2 pg. ml ethidium-bromide (Maniatis et al. 1982). The dsDNA product was cut directly from the gel and resuspended in 300 L water by heating to 73讈 for 15 min. Further primer pairs were used to amplify double- stranded DNA subfragments from the isolated cyt-b gene (i.e. Li4863/H15104, Li4863/Hi5298, L14990/ H15298, L15236/H15505, Li5311/H15710, L15656/ H16065); for original primer descriptions see Helm- Bychowski and Cracraft (1993) and references there- in. A large degree of fragment overlap, as well as the sequencing of both DNA strands, ensured accurate data collection. An air thermocycler (Idaho Technol- ogies) was used to perform 10-L amplifications of dsDNA in glass microcapillary tubes using standard buffers described elsewhere (Wittwer et al. 1989, Wit- twer 1992). All subfragments were amplified with conditions: i sec at 94讈, 0 sec at 48讈 (i.e. a drop to 48讈 without pause before climbing to extension tem- perature), 10 sec at 72讈, and 35 cycles at slope 9 (machine-specific fastest temperature ramping rate available). Subfragments were visualized, isolated, and resuspended as described above. Concurrent negative and positive controls were performed for each ex- periment. Single-stranded DNA for direct sequencing was generated using i:100 dilutions of one primer in 50- L amplification reactions together with ! L of the resuspended subfragment of dsDNA (Gyllensten and Erlich 1988). Amplification reagents were the same as described above for initial gene isolation and were performed in a Peltier-effect thermocycler with con- ditions: i rain at 94讈, i rain at 52讈, 2 rain at 72讈, and 35 cycles. Products were concentrated and de- salted by spin-dialysis (Millipore 30,000 NMWL) be- fore sequencing by the Sanger termination-dideoxy method (Sanger et al. 1977) using Sequenase 2.0 (U.S. Biochemical). Sequencing products were sub- jected to denaturing gel electrophoresis followed by autoradiography. Corrected distance computation.-We computed codon third-position corrected distances for comparison with previously estimated values (Irwin et al. 1991, Thomas and Martin !993). Corrected distances were made us- ing DNADIST from the Phylip 3.5 set of programs (Felsenstein 1993), set to the ML substitution model (Felsenstein 1981) with a 10:i explicit transition bias (a conservative estimate for birds [Kocher et al. 1989]) and empirical nucleotide frequencies of the total an- alyzed gene sequences used in the distance compu- tation. Phylogenetic analysis.--Phylogenetic relationships were estimated using maximum parsimony and max- imum likelihood methods with bootstrapping to as- sess support for internal branches (Felsenstein 1985, Hillis and Bull 1993). The two methods have "basic equiprobable" (i.e. symmetrical) assumptions of the evolutionary process (i.e. cladistic parsimony [Farris 1983]) versus a suite of a priori assumptions implicit in a model of the evolutionary process at the DNA level (Felsenstein 1981). Parameterization of a model of evolution at the DNA level may be considered an appropriate approach in phylogeny building given mounting evidence for a high transition bias in the avian mtDNA mutational spectrum (Kocher et al. 1989, Edwards and Wilson 1990, E. Smith et aL 1992, Nunn and Cracraft 1996) and a general appreciation of the effects of base-compositional bias upon character-state change, i.e. character-state bias (Collins et al. 1995). Maximum parsimony analyses and bootstrapping (100 replicates) were accomplished with PAUP 3.1.1 (Swofford 1993). Ten replicate heuristic searches were performed with random addition of taxa to minimize input order bias. Branch-swapping was made by the tree bisection-reconnection (TBR) algorithm. Con- temporaneous changes were favored (i.e. parallelisms over reversals) by using delayed transformation (DELTRAN) optimization. These analyses included all characters and substitutions equally weighted. Maximum likelihood analyses and bootstrapping (100 replicates) were performed with the program fastDNAml 1.1.1 (Felsenstein 1981, Olsen et al. 1994). Heuristic searches were repeated with random ad- dition of taxa. Overall empirical base frequencies of the cyt-b gene sequences in this study, relative codon position evolutionary rates of 5:1:20 (lst:2nd:3rd), and a 10:1 transition (Ti) to transversion (Tv) substi- tution bias, were defined a priori as parameters of the nucleotide substitution model in the likelihood com- putations. Following phylogenetic analysis, we performed a parsimony-based "winning-sites" test (Templeton 1983) to compare possible alternate arrangements among the major groups established within the Di- omedeidae. Assuming a data set is potentially unin- formative, a winning-sites test can be used to compare character support among any number of opposing phylogenetic hypotheses (i.e. topologies) with an ex- pectation of stochastically equal support for each. When a comparison of character support for any two particular topologies departs from equality, a bino- mial test determines if one is significantly better-sup- ported than the other, and is therefore more likely to be the correct topological arrangement (Templeton 1983, Prager and Wilson 1988). This test is a conser- vative indicator of significance levels for tree support (Felsenstein 1988). Based on the constrained topolog- ical outcome of the winning-sites test, we explored the range of rooted hypotheses for the Diomedeidae by creating the five possible rooted arrangements in- cluding all 16 taxa. The most-parsimonious topologies were determined by searching for rearrangements within constrained groups using MacClade 3.1 (Mad- dison and Maddison 1993). The resultant topologies were then compared using parsimony as well as max- imum likelihood methods (Kishino and Hasegawa 1989). RESULTS Alignment of cyt-b sequences (1,143 bp) of the 14 albatross species and two members of the Procellariidae shows no nucleotide insertions or deletions among sequences (data available from Genbank, inclusive accession numbers U48940 to U48955). For a variety of reasons we consider the cyt-b gene sequences reported to be solely mitochondrial in origin. The entire cyt-b gene and flanking regions initially were isolated as a single, contiguous fragment. This procedure minimizes the potential amplifica- tion of smaller fragments that are more likely to be translocated into the nuclear genome (Quinn 1992, M. Smith et al. 1992, Kornegay et al. 1993). The gene sequences can be fully trans- lated using the chicken mitochondrial code (Desjardins and Morais 1990) without nonsense or intervening stop codons. Finally, the align- ment revealed neither a particular overabun- dance of first and second codon position changes, nor a shift in the typical avian mtDNA transition bias, phenomena known to occur in mtDNA sequences translocated to the nuclear genome (Arctander 1995). There are 333 variable nucleotide positions among the 16 gene sequences (29.1% of the total cyt-b gene sequence). Variable nucleotides pre- dominantly were at codon third positions (260 sites, 78.1%), followed by first positions (63 sites, 18.9%), and then second positions (10 sites, 3.0%). The distribution of variable sites reflects the majority of substitutions occurring at synony- mous sites (codon third positions and leucine codon first positions). Patterns of pairwise sequence divergence and sat- uration.--Uncorrected pairwise percentage dif- ference (Table 2) between albatross species ranged from 0.87% (D. amsterdamensis vs. D. ex- ulans ) to 11.20% ( D. irrorata vs. D. chlororhynchos ), i.e. the largest difference occurs within Diome- dea. Differences across the deepest node, i.e. from members of the Diomedeidae to the Procellar- iidae, ranged from 12.60% (both Phoebetria spe- cies vs. Macronectes giganteus) to 15.84% (D. epomophora and D. irrorata vs. Procellaria cinerea), potentially indicating some rate variability among the albatross lineages (� = 14.31 + SD of 0.81%, n = 28). The two procellariids (Ma- cronectes vs. Procellaria) differed by 11.37%, which was comparable to the largest difference found within the Diomedeidae. Within the Diomedeidae, corrected codon third-position distances ranged from 2.85% (D. exulans vs. D. amsterdamensis) to 47.05% (D. ex- ulans vs. D. chrysostoma), again emphasizing that the largest differences are within the traditional genus Diomedea (Table 2). From Diomedeidae to the Procellariidae, this computed value in- creased substantially to a mean value of 117.91% (n = 28, SD = 11.64) and identified the outgroup comparisons at codon third positions highly af- fected by multiple substitutions, i.e. within a zone of saturation. Pairwise empirical numbers of substitutions between the procellariiform cyt-b sequences partitioned into transitions (Ti) and transver- sions (Tv) revealed a consistent bias in favor of Ti changes (Table 3, Fig. 1). However, compu- tation of the ratio of Ti to Tv substitutions iden- tified a large range of variation in the relative contribution of each substitution class to these comparisons. Ratios of Ti:Tv ranged from a low of 2.1:1 between distantly related taxa (both Phoebetria species vs. Macronectes giganteus), to a difference composed entirely of Ti substitu- tions (31:0) between closely related taxa (D. chlo- rorhynchos vs. D. cauta). A comparison of all nu- cleotide positions showed that the growth of Ti differences between diverging taxa departs from linearity at approximately 11% uncorrected to- tal pairwise difference (Fig. 1A). The cluster of points greater than 11% comprised all pairwise comparisons of Procellariidae to Diomedeidae and indicated deeper comparisons have pro- gressed into a zone of saturation. Dependent upon codon position, however, there is a differing pattern of divergence in the cyt-b gene sequences. The growth of Ti substi- tutions at codon first positions increases linear- ly between all taxa in this study (Fig. lB) with- out evidence of a drop or plateau in divergence occurring as more remote comparisons are made. Codon second positions warrant no special at- tention because they do not contribute signif- icantly to overall differences. In contrast to co- don first positions, the codon third position Ti divergence increases linearly to a limit of ap- proximately 25% total difference (Fig. 1C) and then plateaus into the zone of saturation. The plateau effect is most pronounced at third po- TAIIE 3. Pairwise substitutional differences in cytochrome-b genes among Diomedeidae and Procellariidae. Transitions (Ti) above the diagonal, and transversions (Tv) below the diagonal. Species 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1 D. chlororhynchos -- 31 27 27 31 99 96 106 113 102 115 117 83 87 105 116 2 D. bulleri 3 -- 24 28 16 97 94 98 104 98 109 113 79 79 112 115 3 D. chrysostorna 3 6 -- 18 24 98 94 100 103 94 109 11 79 81 109 120 4 D. rnelanophris 5 8 4 -- 27 92 90 98 103 94 108 109 77 81 109 111 5 D. cauta 0 3 3 5 -- 95 92 99 106 98 111 111 82 81 111 116 6 D. exulans 17 20 20 22 17 -- 8 33 66 61 69 64 85 95 111 119 7 D. arnsterdarnensis 19 22 22 24 19 2 -- 36 64 59 68 66 84 94 110 120 8 D. epomophora 14 17 17 19 14 3 5 -- 68 69 69 74 98 100 121 132 9 D. imrnutabilis 12 15 15 17 12 11 13 8 -- 19 49 40 100 103 117 129 10 D. nigripes 11 14 14 16 11 10 12 7 1 -- 46 39 95 97 118 121 11 D. irrorata 13 16 16 18 13 12 14 9 5 4 -- 47 108 112 124 135 12 D. albatrus 10 13 13 15 10 9 11 6 2 1 3 -- 106 108 118 132 13 P. palpebrata 6 9 9 11 6 15 17 12 10 9 11 8 -- 22 97 107 14 P. fusca 6 9 9 11 6 15 17 12 10 9 1! 8 2 -- 97 111 15 M. giganteus 47 48 48 48 47 50 50 49 47 46 48 45 47 47 -- 104 16 Pr. cinerea 47 50 48 50 47 50 50 49 47 46 46 45 47 47 26 -- sitions because of the preponderance of syn- onymous changes that can occur at these sites. In view of the overall pattern of divergence among these sequences, we believe that third- position comparisons to the outgroup Procel- lariidae will almost certainly exhibit some effect of site saturation. Among the Diomedeidae, however, comparisons should be largely unaf- fected by saturation at any given site. Base compositional bias.--The pattern of com- positional bias (Prager and Wilson 1988) at each codon position in procellariiform cyt-b (Table 4) is almost identical to that found in mammals (Irwin et al. 1991) and other birds (Kornegay et al. 1993, Nunn and Cracraft 1996). First posi- tions are little-biased (C = 0.088), being G-poor ( = 21.1 ___ 1.15%) and slightly C-rich (� = 30.5 + 0.64%). Second positions are more biased than first (C = 0.219), again G-poor ( = 12.9 + 0.00%) but T-rich ( = 39.6 + 0.17%). The highest com- positional bias is found at third positions (C = 0.428), which have very low G ( = 3.8 ___ 0.77%) and low T ( = 14.1 + 1.41%) content, and are rich in A ( = 37.4 ___ 1.13%) and C (? = 44.7 + 1.65%). The overall GC content of each sequence ( = 0.468 ___ 0.009, range 0.450 [Procellaria ci- nerea] to 0.485 [Diomedea irrorata]) falls within the range of other known bird cyt-b genes and averages slightly higher than available passer- ine sequences and lower than the Muscovy Duck (Cairina moschata) and most phasianids (sum- marized in Jermiin et al. [1994]). Cytochrome-b protein sequence variation.--There was a low number of amino-acid replacements among the translated cyt-b protein sequences (following chicken codon usage [Desjardins and Morais 1990]). In total, 46 (12.1%) of the 380 amino acid sites were variable among the 16 translated sequences. Variable residues largely were confined to the transmembrane regions of the molecule. A majority of amino acid replace- ments involved exchanges between hydropho- bic residues--a replacement pattern similar to that found in mammals and other organisms (Irwin et al. 1991, Degli Esposti et al. 1993). Phylogenetic analysis.--An identical branch- ing topology was best supported by both max- imum parsimony (single most-parsimonious tree L = 574 steps, CI [excluding uninformative characters] = 0.591) and maximum likelihood (LnL = -4103.65) analyses (phylogeny shown in Fig. 2). Further parsimony analyses including a greater number of outgroup taxa from the Procellariidae, Hydrobatidae, and Pelecanoidi- dae (a total of 85 taxa) resulted in no change in root location or branching pattern within the Diomedeidae (data not shown). The "phylo- genetic tree" indicated an initial bifurcation in the Diomedeidae. The main lineages each sub- divided once more, resulting in four phyloge- netic groups: (1) southern mollymawks, (2) sooty albatrosses, (3) "great" albatrosses, and (4) North Pacific albatrosses. The full description for this tree is: ((((Diomedea chlororhynchos, ((D. bulleri, D. cauta), (D. chrysostoma, D. melanophris, (Phoe- betria palpebrata, P. fusca)), ((D. epomophora, (D. amsterdamensis, D. exulans)), ((D. immutabilis, D. nigripes), (D. irrorata, D. albatrus) ) ) ), (Macronectes giganteus, Procellaria cinerea ) ). Maximum parsimony (MP) and maximum 150 125 lOO 75 50 � 25- o! A Transitions Transversions 2% 4% 6% 8% 10% 12% 14% I % 18% 40 B l& lOO ] 75 i � o ! g v' 12% 0% 10% 20% 30% 40% FIG. 1. Empirical numbers of transition (Ti) and transversion (Tv) substitutions (on y axis) plotted against total uncorrected pairwise percentage difference (on x axis) for (A) all codon positions (i.e. 1,143 sites), (B) codon first positions (i.e. 381 sites), and (C) codon third positions (i.e. 381 sites). likelihood (ML) bootstrap analyses identified broadly concordant levels of support for branches within the phylogeny. Bootstrap sup- port for the first lineage, comprised of southern mollymawks and sooty albatrosses, was rela- tively low (79% MP, 52% ML) compared with values for other deep branches. Complete boot- strap support (i.e. from all replicates) was found for monophyly of the second lineage, contain- ing "great" and North Pacific albatrosses (100% MP, 100% ML). In addition, the four monophy- letic groups within the Diomedeidae all were highly supported: southern mollymawks (100% MP, 100% ML; five taxa), sooty albatrosses (100% MP, 100% ML; two taxa), great albatrosses (100% MP, 100% ML; three taxa), and North Pacific albatrosses (99% MP, 99% ML; four taxa). Further branching events within the three groups with more than two members (i.e. ex- cluding sooty albatrosses) were robustly sup- ported. Within southern mollymawks, the para- phyletic branching of D. chlororhynchos was October 1996] Albatross Molecular Phylogeny 791 6ti9 Diomedea immutabilis Diomedea nigripes palpebrata � Procellariidae Great albatrosses North Pacific albatrosses (Phoebastria ) Sooty Albatrosses Southern Mollymawks (Thalassarche) FIG. 2. Phylogenetic relationships among the Diomedeidae based on maximum parsimony (MP) and maximum likelihood (ML) analyses of cytochrome-b gene sequences (rooted to the outgroup Procellariidae). Identical branching patterns were determined by both analyses (most parsimonious tree L = 574, CI [excluding uninformative characters] = 0.591; maximum likelihood tree LnL = -4103.65). Percentage bootstrap support found in phylogenetic analyses are shown to the left of internal branches (MP/ML). marginally supported (65% MP, 61% ML) as oc- curring before the origin of the remaining four species in this group. Among the remaining four taxa, a sister-taxa relationship of D. bulleri and D. cauta was highly supported (96% MP, 93% ML), and D. chrysostoma and D. melanophris also formed sister taxa supported by relatively high bootstrap values (73% MP, 87% ML). In- spection of replicates revealed minor support for the branching of D. chlororhynchos between the two species-pairs described above, i.e. res- olution at the base of the three lineages (chlo- rorhynchos, cauta/bulleri, chrysostoma/melano- phris) could be considered problematic based on the current data set. Similarly, among the North Pacific albatrosses most replicates supported monophyly of D. irrorata and D. albatrus (60% MP, 69% ML), although the remaining repli- cates supported a paraphyletic branching order for these taxa (D. irrorata basal) before the well- supported monophyletic apical origin of D. im- mutabilis and D. nigripes (96% MP, 94% ML). To further assess the higher-level relation- ships determined from our phylogenetic tree we tested character support for an unrooted net- work of one taxon sampled from each of the four high bootstrap-supported groups shown in Figure 2. Assuming monophyly of the four groups and including all 14 albatross sequences, 120 possible combinations of taxa exist for this test. We extracted from the data set and tested separately each possible combination of four taxa. Character support was assessed for the three topological arrangements possible among each four-taxon combination (i.e. we computed 360 separate values; results summarized in Fig. 3A). For all 120 combinations of taxa, a monophy- letic origin of sooty albatrosses and southern mollymawks (i.e. Topology I; Fig. 3A) was best supported in comparison to the two alternative arrangements (i.e. Topologies II and III; Fig. 3A). Four combinations of sampled taxa tied equally highest support with a ratio of 42 char- acters (38 or 39 transitions and 3 or 4 transver- sions) supporting the Topology I arrangement to 7 characters (only transitions) supporting a best alternative arrangement (three combina- tions for Topology II and one combination for Topology III). The character support for these four identical higher-level topologies when compared with the nearest scoring alternative topology (i.e. 42:7) was highly significant (P < A Range of total character support Topology I Topology II southern great southern great sooty North Pacific North Pacific sooty albatross albatross albatross albatross 32-42 4-11 Character partition transitions 29-39 4-11 transversions 2-4 0 Topology m southern sooty mollymawk albatross great North Pacific albatross albatross 4-10 4-10 0 B Tree 1 Tree 2 Tree 3 Best tree Length=578 steps (+4) Length=579 steps (+5) Length=574 steps LnL=-4106.65 Ln -L=-4111.48 LnL=-4103.65 (difference=-2.99, sd=8.33) (difference=-7.83, sd=6.98) Procellariidae j Procellariidae j Procellariidae great albatross ' great albatross great albatross North Pacific albatross sooty albatross southern mollymawk Tree 4 Length=592 steps (+18) LnL=-4129.61 (difference=-25.96, sd=9.42) Tree 5 Length=592 steps (+18) LnL=-4129.61 (difference=-25.96, sd=9.42) Procellariidae j Procellariidae great albatross j North Pacific albatross sooty albatross southern mollyroawk southern mollymawk "' sooty albatross North Pacific albatross great albatross Fig. 3. (A) Four-taxon test of character support for the internal branch of topological arrangements among sampled taxa of the higher-level groups and (B) the five possible rooted arrangements of Fig. 3A. Topology I. Most-parsimonious trees include all 16 taxa and were determined based on the higher-level group constraints shown (i.e. rearrangements were allowed only within these groups). Exact tree descriptions are: Tree 1, best tree as described in Results; Tree 2, ( ( ( ( ( Diomedea epomophora, ( D. amsterdamcrisis, D. exulans ) ), (fiD. immut abilis, D. nigripes), D. albatrus), D. irrorata)), (Phoebetria palpebrata, P. fusca)), (D. chlororhynchos, ((D. bulleri, D. cauta), (D. chrysostoma, D. melanophris)))), (Macronectes giganteus, Procellaria cinerea)); Tree 3, ((((D. chlororhynchos, ((D. bulleri, D. cauta), (D. chrysostoma, D. melanophris, ((D. epomophora, (D. amsterdamcrisis, D. exulans)), (((D. immutabilis, D. nigripes), D. albatrus), D. irrorata))), (P. palpebrata, P. fusca)), (Macronectes giganteus, Procellaria cinerea)); Tree 4, (((((P. palpebrata, P. fusca), (D. chlororhynchos, ((D. bulleri, D. cauta), (D. chrysostoma, D. melanophris)))), ((D. immutabilis, D. nigripes), (D. albatrus, D. irrorata))), (D. epomophora, (D. amsterdamcrisis, D. exulans))), (Macronectes giganteus, Procellaria cinerea)); and Tree 5, (((((P. palpebrata, P. fusca), (D. chlororhynchos, ((D. bulleri, D. cauta), (D. chrysostoma, D. melanophris)))), (D. epomophora, (D. amsterdamcrisis, D. exulans))), ((D. immutabilis, D. nigripes), (D. albatrus, D. irrorata))), (Macronectes giganteus, Procellaria cinerea)). 0.001) based on a binomial test (Templeton 1983). The four-taxon higher-level topology was con- gruent in branching pattern with the phylo- genetic tree derived from analyses of the com- plete data set. We estimated the root location to Topology I by creating complete 16-taxa trees constrained to this topology. A branch to the Procellariidae outgroup was attached to the five possible branch positions of Topology I and parsimony and likelihood support computed for the dif- ferent higher-level group arrangements (Fig. 3B). As expected, our phylogenetic tree (Tree 1; L = 574 steps) is the most parsimonious and most likely rooted hypothesis of relationships among these birds (Fig. 3B). Based on a parsi- mony criterion, the four alternative trees (i.e. Tree 2 to Tree 5) are rejected as hypotheses of relationships because they are less parsimoni- ous (L = 578-592 steps). Of the four rejected trees, numbers 4 and 5 are 18 steps longer (L = 592 steps) and also are significantly rejected based on log-likelihood comparisons (i.e. the log-likelihood difference is outside a 95% con- fidence interval compared with the maximum likelihood tree [Kishino and Hasegawa 1989]). Although rejected by a parsimony criterion, the two remaining trees (2 and 3), at four steps lon- ger (L = 578), are not significantly rejected based on log-likelihood differences. We note, how- ever, that the inclusion of many invariant sites-- in the case of mtDNA primarily the first and second codon positions--creates a considerable inflation of the estimate of likelihood variance. This inflation in turn affects the potential re- jection of alternative hypotheses by unrealist- ically widening the confidence interval based on the computed standard error (e.g. see the discussion of codon first and second position characters grouped together as "class-2 sites" by Hasegawa and Kishino [1989]). Interestingly, however, Tree 2, which postulates southern mollymawks as the sister-group to remaining albatrosses, is more likely (LnL = -4105.15) than Tree 3, which postulates the sooty alba- trosses as the sister-group (LnL = -4111.48). Further tree-searching determined the exis- tence of an additional 10 trees (not including Tree 1, the most parsimonious tree) that were more parsimonious than Tree 2 and Tree 3 (i.e. in the range L = 575,577 steps; LnL = -4104.81, -4116.65). These 10 trees were congruent with the phylogenetic tree in terms of their higher- level branching pattern (i.e. a topology of Tree . 1 but with trivial apical rearrangements occur- ring within the higher-level groups). DISCUSSION Genera of albatrosses.--The status and number of genera in the Diomedeidae (Table 1) have varied widely since the formal description of the first known taxon Diomedea exulans (Lin- naeus 1758). Nearly a century elapsed before Reichenbach (1852) introduced three additional genera (i.e. Phoebetria, Phoebastria, and Thalas- sarche), assigning a member of Diomedea into each of the four. In a taxonomic synopsis of the petrels that followed Reichenbach's work, Coues (1866) presented unique morphological char- acters that established monophyly of known species of Diomedeidae among the procellari- iforms. Further, he (1866:187-188) developed a hierarchically defined classification based on the presence or absence of well-described morpho- logical characters among albatrosses. Although Coues presented evidence for "morphological- groups" within Diomedea, he did not name them formally at that time. In fact, Coues arbitrarily rejected two of Reichenbach's genera (Phoe- bastria and Thalassarche) and adopted a more conservative arrangement, admitting only Dio- medea (southern mollymawks, North Pacific al- batrosses, and "great" albatrosses) and Phoe- betria (for the single sooty albatross known at that time; see Table 1). The discovery of new albatross taxa, partic- ularly from the southern oceans and the Aus- tralia/New Zealand region, led to an increased interest in the taxonomy of the group, and sev- eral genera either were reintroduced or created (see Table 1). Generic-level revisions finally cul- minated in Mathews' (1934) treatment of the family Diomedeidae, which admitted all eight genera that had been described previously (Ta- ble 1). Subsequently, however, Mathews and Hallstrom (1943) subsumed the monotypic ge- nus Rhothonia back into Diomedea, reuniting the morphologically similar "great" albatrosses. In addition, Mathews and Hallstrom elected a new monotypic genus Julietata for the unique equa- torial species Phoebastria irrorata based on mor- phological differences from other North Pacific albatrosses (albatrus, immutabilis, nigripes), which again resulted in a total of eight genera for the 18 known taxa. Shortly thereafter, in a wave of lumping by avian taxonomists (including G. M. Mathews), the "generically oversplit" taxono- mies of Diomedeidae (Mathews 1934, Mathews and Hallstrom 1943) were abandoned in favor of a single, all-encompassing genus Diomedea (Mathews 1948). In response to the chaotic and in many cases confusing changes that Mathews' introduced into the ornithological literature (e.g. see Mur- phy 1945, Serventy 1950), a standardization of procellariiform taxonomy arose (Alexander et al. 1965). The taxonomic revision of Alexander et al. (1965) essentially was an "agreed state- ment" among a committee of leading seabird taxonomists to reject the plethora of superfluous naming in the "Mathews" classifications in fa- vor of prior taxonomic treatments of the order Procellariiformes (e.g. Alexander 1928, Peters 1931, Murphy 1936). Although Mathews' work certainly suffered from nomenclatural zealous- ness, numerous higher-level procellariiform groups were returned to less-sophisticated tax- onomic treatments despite being founded on well-defined morphological features. In the process of circumventing Mathews' "inconsis- tent new classifications," Alexander et al. (1965) returned the genera of albatrosses to the earlier classification that essentially had been devised by Coues (1866) and in which Reichenbach's genera Thalassarche and Phoebastria had been subsumed into Diomedea. Since the revision by Alexander et al. (1965), little research has oc- curred on albatross evolutionary relationships, and the generic designations advocated have persisted in the literature (e.g. Jouanin and Mougin 1979, Sibley and Monroe 1990). Sug- gestions for defining "subgeneric" groups within the comprehensive genus Diomedea have resurfaced, however (e.g. Wolters 1975, War- ham 1990). A phylogenetic classification of the Diomedei- dae.--Our results from phylogenetic analysis of cyt-b gene sequences revealed very clear evi- dence of four higher-level species-groups in the Diomedeidae (see Fig. 2). These phylogenetic species-groups are congruent with some tradi- tional morphologically defined groups of al- batrosses (i.e. current or discarded genera, or unnamed morphological groups). However, the traditional taxonomic framework for albatross relationships, i.e. sooty albatrosses Phoebetria versus all other Diomedea (e.g. Coues 1866, Mur- phy 1936, Alexander et al. 1965) is not sup- ported based on most-parsimonious rooting of the arrangement among the well supported higher-level groups (see Fig. 2). Our cyt-b phy- logenetic hypothesis supports the notion of an early basal dichotomy among members of the Diomedeidae. This initial dichotomy led to a lineage limited in distribution to the southern oceans comprised of two groups, southern mol- lymawks and sooty albatrosses, versus a more geographically widespread lineage also with two groups, the North Pacific albatrosses and the "great" albatrosses from the southern oceans. The molecular phylogeny provides for the first time evidence of monophyly of sooty albatross- es and southern mollymawks and indicates that the traditional genus Diomedea is paraphyletic. The current taxonomy of albatrosses, traceable to Coues (1866), provides an illustrative avian example of an "either-or" type of classification. Historically, the genus Diomedea received all al- batross taxa that were not Phoebetria, regardless of the clear morphological affinities that united groups within Diomedea. Indeed, Coues pre- sented the morphological evidence for these groups but refrained from providing formal names. Unfortunately, the later and more so- phisticated taxonomies of albatrosses that rec- ognized these affinities and provided named groups (e.g. Mathews 1934, Mathews and Hall- strom 1943) were rejected during the taxonomic "cleansing" of the Procellariiformes (Alexander et al. 1965). It is interesting to note that the molecular phylogeny is concordant with monophyly of Coues' morphologically defined groups within the paraphyletic traditional genus Diomedea. Based on discrete characters of the bill and sim- ple diagnostic allometric measurements, Coues proposed two natural groups within the Di- omedea (i.e. with the prior exclusion of Phoebe- tria). Coues' "Genus I. Diomedea: Group A" (i.e. D. exulans, D. albatrus, D. irrorata [the first de- scription of a partial cranium of this species and correctly assigned to the group], and D. nigripes) is concordant with a primary lineage in our phylogeny. This group, some of which were later transferred to the resurrected genus Phoe- bastria (albatrus, immutabilis, nigripes, and irrorata; Mathews 1934), contains the largest albatrosses and can be characterized by features including a relatively short tail, laterally broadened bill (particularly a wide boss-like broadening of the culminicorn posterior of the nostrils), as well as large, wide nostrils. The later division of taxa to Phoebastria also is congruent with the cyt-b phylogeny. Coues' second major group "Genus Diomedea: Group B" (i.e.D. melanophris, D. cauta, TABLE 5. A phylogenetic classification of the Dio- medeidae. Family Diomedeidae Genus Thalassarche Species Genus Phoebetria Species Genus Diomedea Species Genus Phoebastria Species T. chlororhynchos T. bulleri T. cauta T. chrysostoma T. melanophris P. palpebrata I. fusca D. exulans D. amsterdamensis D. epomophora P. immutabil P. nigripes P. &rorata P. albatrus D. chlororhynchos, and D. chrysostoma) is the sec- ond clade within the other primary cyt-b lin- eage. This group is now more commonly known as the southern mollymawks (including D. bul- leri, discovered after this date) and can be char- acterized in comparison with Group A by a lat- erally compressed and much weaker bill with narrow culminicorn, and a relatively longer and slightly rounded tail. Interestingly, Phoebetria shares some of these morphological features with the southern mollymawks (Thalassarche), providing useful corroborative evidence for their monophyly. For example, even Coues (1866) remarked on the unclear distinction in bill morphology between Thalassarche and Phoe- betria, pointing out similar general features such as extreme overall lateral compression and the acute narrowing of the culminicorn posterior to the nostrils. Indeed, Coues considered the bill of Phoebetria "hardly separable" from that of some members of Diomedea (i.e. Thalassarche) but eventually relied upon additional "features radically distinct from...those presented by Diomedea proper" to distinguish at a generic level the single known species of sooty alba- tross (Phoebetria fuliginosa [palpebrata]) estab- lished at that time. It seems clear that the sup- posedly primitive morphological features that Coues used to define Reichenbach's genus Phoe- betria (i.e. complete fuliginous plumage, pres- ence of a sulcus in the lower mandible, and acuminate elongation of the rectrices) are in fact plesiomorphic in origin, as they can be found in various combinations in several other petrel lineages. The evolutionary relationships among alba- trosses inferred from their traditional taxonomy presents a subjective hypothesis of groupings. We therefore recommend a formal revision of the taxonomy of Diomedeidae to achieve a clas- sification congruent with the new phylogenetic hypothesis of relationships. Our revision con- stitutes four genera of coordinate phylogenetic rank, each equivalent to one of the higher-level phylogenetic groups, and eliminates paraphyly of the traditional genus Diomedea. We resurrect two previously described genera: (1) Thalas- sarche Reichenbach 1852 (type species melano- phris Temminck 1828), by original designation, to include the southern mollymawks; and (2) Phoebastria Reichenbach 1852 (type species brachyura Temminck 1829 [synonym albatrus Pallas 1769]), by original designation, to include the North Pacific albatrosses. These genera have historical precedence over later synonyms (see Table 1). Thus, a total of seven traditional spe- cies-level taxa are transferred from the genus Diomedea to the resurrected genera. This clas- sification leaves the "great" albatrosses as the sole members of the genus Diomedea Linnaeus 1758 and retains the genus Phoebetria Reichen- bach 1852 for the sooty albatrosses. The adop- tion of phylogenetically defined higher-level species-groups refines the nomenclature of the Diomedeidae and establishes four comparable biological units within the family. The recom- mended traditional species-level taxa within the genera are given in Table 5. Albatross nest building.--The evolution of be- havioral and life-history (BLH) characters in birds has been shown to map closely to phy- logenetic relationships (Prum 1990, Winkler and Sheldon 1993, Paterson et al. 1995). Members of Diomedeidae possess a diversity of stereo- typed courtship and specialized nest-building behaviors that are well described (e.g. Marchant and Higgins 1990). The inspection of a complex nest-building character among the albatrosses in our study provides a valuable example of the benefits gained from a comparative phyloge- netic approach to analysis of BLH characters (Brooks and McLennan 1991). Within the Di- omedeidae, only sooty albatrosses and southern mollymawks build a tall pedestal-shaped nest made primarily from earth but with occasional rock and plant material included. Birds return each breeding season to the previous year's ped- estal-nest, which is both repaired and increased in size, and from which both members of the pair perform elaborate courtship and territorial displays. This contrasts with the low nest mounds of gathered vegetation of the "great" albatrosses and the scantily lined nest scrapes of North Pacific albatrosses, which in both cases are rebuilt each breeding season (Warham 1990). Given that no other petrel builds a nest ped- estal, the traditional taxonomic arrangement of albatrosses would lead us to expect either that an ancestor of all albatrosses built a pedestal nest and subsequently was lost from a lineage within the genus Diomedea, or that both Phoe- betria and a lineage within Diomedea indepen- dently gained the pedestal nest-building be- havior from an ancestor that did not build a pedestal nest. However, it seems more likely that the complex and stereotyped nest-building behavior arose only once. Our cyt-b phylogeny resolves a single, i.e. monophyletic, origin of this behavioral character in an ancestor of Phoe- betria and Thalassarche that has persisted throughout this major albatross lineage. Similar to some BLH characters analyzed by Paterson et al. (1995), it appears that building a pedestal nest has remained stable over a considerable evolutionary period. It is likely that the evo- lution of a laterally compressed bill morphol- ogy also arose in the ancestor of Thalassarche and Phoebetria. The bill morphology is of par- ticular importance because nest building is ac- complished by a lateral plastering action of the bill while the bird rotates on top of the pedestal nest (G. Nunn pets. obs.). Further comparative analyses of other such characters among alba- trosses may provide valuable insight into the patterns of evolution of complex BLH charac- ters in seabirds (e.g. see Paterson et al. 1995). Calibration of evolutionary rate.--Considerable evidence suggests that molecular evolutionary rates vary among taxonomic lineages (Britten 1986, Li et al. 1987). In particular, the assump- tion of a molecular clock is unlikely to extend throughout a phylogeny whose members ex- hibit diverse reproductive or metabolic rates, the latter being inversely correlated with body size (Martin et al. 1992, Martin and Palumbi 1993). Within the Diomedeidae, however, both basal metabolic rate (Adams and Brown 1984) and age at first breeding (Jouventin and Wei- merskirch 1988) vary by much less than an or- der of magnitude, and evolutionary-rate cali- brations among the now well-established phy- logenetic lineages are possible. Fossil remains of seabirds, mostly from the Northern Hemisphere (Warheit 1992), indicate the existence of a diverse albatross fauna during the late Tertiary period. A stratigraphically dated fossil from California suggests the presence of an albatross that shared affinities with D. alba- trus (= Phoebastria) in the North Pacific Ocean as early as the mid-Miocene, approximately 15 million years before present (MYBP; Miller 1962). Relatively few procellariiforms have been discovered from Tertiary deposits bordering the southern oceans (Olson 1985a,b), in contrast to the abundance of Northern Hemisphere fossils. However, the discovery of a cranial fragment of a southern mollymawk of surprisingly mod- ern appearance in Australia indicates the pres- ence of a member of the genus Thalassarche at the Miocene-Pliocene boundary about 10 MYBP (Wilkinson 1969). Previous assessment of corrected mitochon- drial cyt-b codon third-position divergence in homeotherms suggested an evolutionary rate of approximately 10% per million years in a num- ber of mammalian lineages (Irwin et al. 1991); this value has been corroborated by further studies of ground squirrels (Thomas and Martin 1993). On the basis of the fossil evidence out- lined above, we were able to calibrate cyt-b third- position rate estimates for the phylogenetically defined higher-level groups Phoebastria and Thalassarche, given well-supported relation- ships to their respective sister groups, i.e. Di- omedea and Phoebetria. The corrected third-po- sition pairwise divergence between members of (1) Phoebastria and Diomedea (� = 23.62 + 1.21%, n = 12) and (2) Thalassarche and Phoebetria (� = 28.58 + 0.88%, n = 10) leads to rate cali- brations of 1.58% and 2.86% per million years, respectively. These independently derived es- timates are not perfectly concordant with one another but are lower than estimates derived in mammals (Irwin et al. 1991). The inclusion of a wider array of mammalian lineages suggests that rates may be substantially slower in those of larger body size (i.e. lower basal metabolic rate) such as the baleen whales (Martin and Palumbi 1993). Low metabolic and reproductive rates of albatrosses may be causal factors ex- plaining the low cyt-b rate calibrations, but these observations are not idiosyncratic. Rates of mtDNA evolution established from the split be- tween the anseriform genera Anser and Branta (Shields and Wilson 1987) suggest that, in com- parison with their body size, geese have a slower mtDNA substitution rate than other homeo- therms (Martin and Palumbi 1993). Similarly, comparative restriction fragment analyses of mtDNA evolutionary rates among a selection of passefine and nonpasserine groups also sup- port a rate slow-down in comparison with non- avian vertebrate groups (Kessler and Avise 1985). The surprisingly low cyt-b calibrations obtained here support the hypothesis that avian mitochondrial genomic evolutionary rates are considerably slower in birds than in mammals. ACKNOWLEDGMENTS Financial support for this research came from a Frank M. Chapman Postdoctoral Fellowship to G. B. Nunn. Principal tissue collections were made under the aus- pices of the South African Department of Environ- mental Affairs and Tourism, New Zealand Depart- ment of Conservation, Australian Department of En- vironment (Antarctic Division), and the Galapagos National Park; we thank each of these agencies for granting permission to conduct collecting and for providing logistical support. We thank P. Arctander (University of Copenhagen, Denmark) for providing tissue-collecting equipment, and the staff of the Ga- lapagos Research Station for providing expert logis- tical support in the Galapagos Islands. In addition, we thank H. Hasegawa, N. Brothers, the Burke Mu- seum, and the Louisiana State University Museum of Natural Science for the generous donations and loans of tissue samples for this study. G. Olsen provided valuable assistance on implementation of the com- puter program fastDNAml. We gratefully acknowl- edge G. Barrowdough, J. Bates, W. R. P. Bourne, R. Brooke, J. Cracraft, C. Griffiths, J. Groth, S. Hackett, N. Klein, S. Stanley, and W. L. N. Tickell for their comments and suggestions on this manuscript and for discussion of saturation rates in mtDNA. A. Baker, C. Krajewski, and an anonymous reviewer also pro- vided thoughtful comments on the manuscript. The research reported in this paper is a contribution from the Lewis B. and Dorothy Cullman Research Facility at the American Museum of Natural History and has received generous support from the Lewis B. and Do- rothy Cullman Program for Molecular Systematics Studies, a joint initiative of The New York Botanical Garden and The American Museum of Natural His- tory. LITERATURE CITED ADAMS, N.J., AND C. R. BROWN. 1984. Metabolic rates of sub-Antarctic Procellariiformes: A compara- tive study. Comparative Biochemistry and Phys- iology 77A:169-173. ALEXANDER, W.B. 1928. Birds of the ocean. Putnam, New York. ALEXANDER, W. B., R. A. FALLA, C. JOUANIN, R. C. MURPHY, F. SALOMONSEN, K. H. Voous, G. E. WATSON, W. R. P. BOURNE, C. A. FLEMING, N. H. KURODA, M. K. ROWAN, D. L. SERVENTY, W. L. N. TICKELL, J. WAHM, AND J. M. WIN-RSOTTOM. 1965. The families and genera of petrels and their names. Ibis 107:401-405. ANDERSON, S., A. T. BANKIER, B. O. BARRELL, M. H. L. DE BRUIIN, A. R. COULSON, J. DROUIN, I. C. EPERON, D. P. NIERLICH, B. A. ROE, F. SANGER, P. H. SCHREIER, A. J. H. SMrrH, R. STADEN, AND I. G. YOUNG. 1981. Sequence and organisation of the human mitochondrial genome. Nature 290:457- 465. ARCTANDER, P. 1995. Comparison of a mitochondrial gene and a corresponding nuclear pseudogene. Proceedings of the Royal Society of London Se- ries B 262:13-19. BAIRD, S. F., T. M. BREWER, AND R. RIDGWAY. 1884. The water birds of North America, vol. 2. Little, Brown and Company, Boston. BLECHSCHMIDT, K., H.-U. Pm'ER, J. DE KORTE, M. WINK, I. SERoOLD, AND A. J. HELBIG. 1993. Untersu- chungen zur loekularen Systematik der Raub- toowen (Stercorariidae). Zoologische Jahrbuech- er Abteilung fuer Systematik Oekologie und Geographie der Tiere 120:379-387. BOURNE, W. R. P. 1989. The evolution, classification and nomenclature of the great albatrosses. Ger- faut 79:105-116. BaITTEN, R.J. 1986. Rates of DNA sequence evolu- tion differ between taxonomic groups. Science 231:1393-1398. BROOKS, D. R., AND D. A. MCLENNAN. 1991. Phylog- eny, ecology, and behavior: A research program in comparative biology. University of Chicago Press, Chicago. COLLINS, T. M., P. H. WIMBERGER, AND G. J.P. NAYLOR. 1995. Compositional bias, character-state bias, and character-state reconstruction using parsimony. Systematic Biology 43:482-496. COUES, E. 1866. Critical review of the Family Pro- cellariidae, part V, embracing the Diomedeinae and the Halodrominae. Proceedings of the Acad- emy of Natural Sciences of Philadelphia 18:172- 197. DEGLI ESPOSTI, M., S. DE ValEs, M. CaiMI, A. GHELLI, T. PATARNELLO, AND A. MEYER. 1993. Mitochon- drial cytochrome b: Evolution and structure of the protein. Biochimica et Biophysica Acta 1143: 243-271. DESIARDINS, P., AND R. MORAIS. 1990. Sequence and gene organisation of the chicken mitochondrial genome: A novel gene order in higher verte- brates. Journal of Molecular Biology 212:599-634. EDWARDS, S. V., AND A. C. WILSON. 1990. Phyloge- netically informative length polymorphism and sequence variability in mitochondrial DNA of Australian songbirds (Pomatostomus). Genetics 126: 695-711. FARRIS, J.S. 1983. The logical basis of phylogenetic analysis. Pages 7-36 in Advances in cladistics (N. I. Platnick and V. A. Funk, Eds.). Columbia Uni- versity Press, New York. FELSENSTEIN, J. 1981. Evolutionary trees from DNA sequences: A maximum likelihood approach. Journal of Molecular Evolution 17:368-376. FELSENSTEIN, J. 1985. Confidence limits on phylog- enies: An approach using the bootstrap. Evolu- tion 39:783-791. FELSENSTEIN, J. 1988. Phylogenies from molecular sequences: Inference and reliability. Annual Re- view of Genetics 22:521-565. FELSENSTEIN, J. 1993. PHYLIP (Phylogeny Inference Package) 3.55c. Distributed by the author. Uni- versity of Washington, Seattle. GYLLENSTEN, U. B., AND H. A. ERLICH. 1988. Gener- ation of single-stranded DNA by the polymerase chain reaction and its application to direct se- quencing of the HLA-DQA locus. Proceedings of the National Academy of Sciences USA 85:7652- 7656. HARRIS, M.P. 1973. The biology of the Waved Al- batross Diomedea irrorata of Hood Island, Gala- pagos. Ibis 115:483-510. HARmSON, P. 1983. Seabirds: An identification guide. Houghton Mifflin, Boston. HASEGAWA, M., AND H. KISHINO. 1989. Confidence limits on the maximum-likelihood estimate of the hominoid tree from mitochondrial-DNA se- quences. Evolution 43:672-677. HELM-BYCHOWSKI, K., AND J. CRACRaFT. 1993. Re- covering phylogenetic signal from DNA se- quences: Relationships within the cotvine assem- blage (Class Aves) as inferred from complete se- quences of the mitochondrial DNA cytochrome- b gene. Molecular Biology and Evolution 10:1196- 1214. HILLIS, D. M., AND J. J. BULL. 1993. An empirical test of bootstrapping as a method for assessing con- fidence in phylogenetic analysis. Systematic Bi- ology 42:182-192. IRWIN, D. M., T. D. KOCHER, AND A. C. WILSON. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32:128-144. JERMIIN, L. S., D. GRAUR, R. M. LOWE, AND R. H. CROZIER. 1994. Analysis of directional mutation pressure and nucleotide content in mitochondrial cyto- chrome b genes. Journal of Molecular Evolution 39:160-173. JOUANN, C., AND J.-L. MOUCIN. 1979. Order Procel- lariiformes. Pages 48-121 in Check-list of birds of the world, vol. 1, 2nd ed. (E. Mayr and G. W. Cottrell, Eds.). Harvard University Press, Cam- bridge, Massachusetts. JOUVENTIN, P., AND H. WEIMERSKIRCH. 1988. Demo- graphic strategies of southern albatrosses. Pages 857-868 in Acta XIX Congressus Internationalis Ornithologici (H. Ouellet, Ed.). Ottawa, Ontario, 1986. National Museum of Natural Science, Ot- tawa. KESSLER, L. G., AND J. C. AvISE. 1985. A comparative description of mitochondrial DNA differentia- tion in selected avian and other vertebrate gen- era. Molecular Biology and Evolution 2:109-125. KISHINO, H., AND M. HASEGAWA. 1989. Evaluation of the maximum likelihood estimate of the evolu- tionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29:170-179. KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. ED- WARDS, S. PAABO, F. X. VILLABLANCA, AND A. C. WILSON. 1989. DynamicsofmitochondrialDNA evolution in mammals: Amplification and se- quencing with conserved primers. Proceedings of the National Academy of Sciences USA 86: 6196-6200. KORNEGAY, J. R., T. D. KOCHER, L. A. WILLIAMS, AND A. C. WILSON. 1993. Pathways of lysozyme evo- lution inferred from the sequences of cyto- chrome b in birds. Journal of Molecular Evolution 37:367-379. KRAIEWSKI, C., AND J. W. FETZNER. 1994. Phylogeny of cranes (Gruiformes: Gruidae) based on cyto- chrome-b DNA sequences. Auk 111:351-365. LANYON, S. M., AND J. G. HALL. 1994. Reexamination of barbet monophyly using mitochondrial-DNA sequence data. Auk 111:389-397. LI, W.-H., M. TANIMURA, AND P.M. SH,RP. 1987. An evaluation of the molecular clock hypothesis us- ing mammalian DNA sequences. Journal of Mo- lecular Evolution 25:330-342. LINNAEUS, C. 1758. Systema naturae per regna tria naturae, secundum classes, ordines, genera, spe- cies cum characteribus, differentils, synonymis, locis. Laurentii Salvii, Stockholm. LYDEKKER, R. 1891a. Catalogue of the fossil birds in the British Museum. British Museum of Natural History, London. LYDEKKER, R. 1891b. On British fossil birds. Ibis Se- ries 6:394-395. MADDISON, W. P., AND D. R. MADDISON. 1993. MacClade 3.1 Interactive analysis of phylogeny and character evolution. Sinauer, Sunderland, Massachusetts. MANIATIS, T., E. F. FRITSCH, AND J. SBROOK. 1982. Molecular cloning: A laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. MARCHANT, S., AND P. J. HIGGINS. 1990. Handbook of Australian, New Zealand and Antarctic birds, vol. 1. Ratites to Ducks. Oxford University Press, Oxford. MARTIN, A. P., G. J. P. NAYLOR, AND S. R. PALUMBI. 1992. Rates of mitochondrial DNA evolution in sharks are slow compared with mammals. Nature 357:153-155. MARTIN, A. P., AND S. R. PALUMBI. 1993. Body size, metabolic rate, generation time, and the molec- ular clock. Proceedings of the National Academy of Sciences USA 90:4087-4091. MATHEWS, G. M. 1912. Birds of Australia, vol. 2. H. F. & G. Witherby, London. MATHEWS, G.M. 1934. Remarks on albatrosses and mollymawks. Ibis Series 13(4):807-816. MATHEWS, G.M. 1948. Systematic notes on the pet- rels. Bulletin of the British Ornithologists' Club 68:155-170. MATHEWS, G. M., AND E. J. L. HALLSTROM. 1943. Notes on the order Procellariiformes. The Federal Cap- ital Press of Australia Limited, Canberra. MEYER, A. 1994. Shortcomings of the cytochrome b gene as a molecular marker. Trends in Ecology and Evolution 9:278-280. MILLER, L. 1962. A new albatross from the Miocene of California. Condor 64:471-472. MURPHY, R.C. 1917. A new albatross from the west coast of South America. Bulletin of the American Museum of Natural History 37:861-864. MURPHY, R.C. 1936. Oceanic birds of South America, vol. 2. American Museum of Natural History, New York. MURPHY, R. C. 1945. Review. Notes on the Procel- lariiformes. Auk 62:153. MURRAY, B. W., W. B. McGILLIVRAY, J. C. BARLOw, R. N. BEECH, AND C. TROSECK. 1994. The use of cytochrome b sequence variation in estimation of phylogeny in the Vireonidae. Condor 96:1037- 1054. NUNIX/, G. B., AND J. CRACRAFT. 1996. Phylogenetic relationships among the major lineages of the Birds-of-Paradise (Paradisaeidae) using mito- chondrial DNA gene sequences. Molecular Phy- logenetics and Evolution 5:445-459. OLSEN, G. J., H. MATSUDA, R. HAGSTROM, AND R. OVER- SEEK. 1994. FastDNAml: A tool for construction of phylogenetic trees of DNA sequences using maximum likelihood. Computing for the Ap- plied Biosciences 10:41-48. OLSON, S. L. 1985a. An early Pliocene marine avi- fauna from Duinefontein, Cape Province, South Africa. Annals of the South African Museum 95: 147-164. OLSON, S.L. 1985b. Early Pliocene Procellariiformes (Aves) from Langebaanweg, southwestern Cape Province, South Africa. Annals of the South Af- rican Museum 95:123-145. PATERSON, A.M., G. P. WALLIS, AND R. D. GRAY. 1995. Penguins, petrels, and parsimony: Does cladistic analysis of behavior reflect seabird phylogeny? Evolution 49:974-989. PETERS, J. L. 1931. Check-list of birds of the world, vol. 1. Harvard University Press, Cambridge, Massachusetts. PRAGER, E. M., AND A. C. WILSON. 1988. Ancient origin of lactalbumin from lysozyme: Analysis of DNA and amino acid sequences. Journal of Mo- lecular Evolution 27:326-335. PRUM, R. O. 1990. Phylogenetic analysis of the evo- lution of display behavior in the Neotropical manakins (Aves: Pipridae). Ethology 84:202-231. QUINN, T. W. 1992. The genetic legacy of Mother Goose: Phylogeographic patterns of Lesser Snow Goose Chen caerulescens caerulescens maternal lin- eages. Molecular Ecology 1:105-117. REICHENBACH, L. 1852. Avium Systema Naturale. Leipzig, Germany. RICHMAN, A.D., AND T. PRICE. 1992. Evolution of ecological differences in the Old World leaf war- blers. Nature 355:817-821. ROSERTSON, C. J. R., AND J. WAIU-IAM. 1992. Nomen- clature of the New Zealand Wandering Alba- trosses Diomedea exulans. Bulletin of the British Ornithologists' Club 112:74-81. Roux, J.-P., P. JOUVENTIN, J.-L. MOUGIN, J.-C. STAHL, AND H. WEIMERSKIRCH. 1983. Un nouvel alba- tross Diomedea amsterdamensis n. sp. dcouvert sur File Amsterdam (37�50'S, 77�35'E). Oiseau 53:1- i1. SANGER, F., S. NICKLEN, AND A. R. COULSON. 1977. DNA sequencing with chain termination inhib- itors. Proceedings of the National Academy of Sciences USA 74:5463-5467. SERVENTY, D.L. 1950. Taxonomic trends in Austra- lian ornithology with special reference to the work of Gregory Mathews. Emu 49:259-267. SHIELDS, G. F., AND A. C. WILSON. 1987. Calibration of mitochondrial DNA evolution in geese. Jour- nal of Molecular Evolution 24:212-217. SIBLEY, C. G., AND B. L. MONROE, JR. 1990. Distri- bution and taxonomy of birds of the world. Yale University Press, New Haven, Connecticut. SINGER-SAM, J., R. L. TANGUAY, AND A.D. RIGGS. 1989. Use of Chelex to improve the PCR signal from a small number of cells. Amplifications: A forum for PCR users 3:11. SMITH, E. F. G., P. ARCTANDER, J. FJELDS.., AND O. G. AMIR. 1992. A new species of shrike (Laniidae: Laniarius) from Somalia, verified by DNA se- quence data from the only known individual. Ibis 133:227 -235. SMITH, M. F., W. K. THOMAS, AND J. L. PATRON. 1992. Mitochondrial DNA-like sequence in the nuclear genome of an akodontine rodent. Molecular Bi- ology and Evolution 9:204-215. SWOFFORD, D. L. 1993. Phylogenetic analysis using parsimony, 3.1.1. Distributed by the author. Smithsonian Institution, Washington, D.C. TEMPLETON, A. R. 1983. Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and the apes. Evolution 37:221-244. THOMAS, W. K., AND S. L. MARTIN. 1993. A recent origin of marmots. Molecular Phylogenetics and Evolution 2:330-336. WALSH, P.S., D. A. METZGER, AND R. HIGUCH. 1991. Chelex 100 as a medium for simple extraction of. DNA for PCR-based typing from forensic mate- rial. Biotechniques 10:506-513. WARHAM, J. 1990. The petrels: Their ecology and breeding systems. Academic Press, London. WAP, HEIT, K. I. 1992. A review of the fossil seabirds from the Tertiary of the North Pacific: Plate tec- tonics, paleoceanography, and faunal change. Pa- leobiology 18:401-424. WETlYlORE, A. 1943. Fossil birds from the Tertiary deposits of Florida. Proceedings of the New En- gland Zoological Club 22:59-68. WILKINSON, H.E. 1969. Description of an upper Mio- cene albatross from Beaumaris, Victoria, Austra- lia, and a review of fossil Diomedeidae. Memoirs of the National Museum Victoria 29:41-51. WINKLER, D. W., AND F. H. SHELDON. 1993. Evolution of nest construction in swallows (Hirundinidae): A molecular phylogenetic perspective. Proceed- ings of the National Academy of Sciences USA 90:5705-5707. WITTWR, C.T. 1992. Buffers and reaction compo- nents for rapid cycling. The Rapid Cyclist 1:6-8. WITTWO, C. T., G. C. FILLMORE, AND D. R. HaLLYARD. 1989. Automated polymerase chain reaction in capillary tubes with hot air. Nucleic Acids Re- search 17:4353-4357. WOLTERS, H.E. 1975. Die vogelarten der Erde: Eine systematische Liste mit Verbreitungsangaben sowie deutschen und englischen Namen. Parey, Berlin. APPENDIX. Individual taxa sequenced (with binomen or current trinomen for traditional polytypic species) and their collection localities and dates. Taxonomic scheme of genera follows our suggested phylogenetic classification of higher-level groups in Diomedeidae. Taxon Collection locality and date Diomedeidae Thalassarche (southern mollymawks) Thalassarche chlororhynchos chlororhynchos (Yellow-nosed Mollymawk) Thalassarche bulleri bulleri (Buller's Molly- mawk) Thalassarche chrysostoma (Gray-headed Mol- lymawk) Thalassarche melanophris melanophris (Black- browed Mollymawk) Thalassarche cauta cauta (Shy Mollymawk) Phoebetria (sooty albatrosses) Phoebetria palpebrata (Light-mantled Sooty Albatross) Phoebetria fusca (Dark-mantled Sooty Alba- tross) Diomedea "great" albatrosses) Diomedea exulans dabbenena (Wandering Al- batross) Diomedea amsterdamensis (Amsterdam Alba- tross) Diomedea epomophora sanfordi (Royal Alba- tross) Phoebastria (North Pacific albatrosses) Phoebastria immutabilis (Laysan Albatross) Phoebastria nigripes (Black-footed Albatross) Phoebastria irrorata (Waved Albatross) Phoebastria albatrus (Short-tailed Albatross) Procellariidae Macronectes giganteus (Southern Giant Pe- trel) Procellaria cinerea (Gray Petrel) Gough Island, Atlantic Ocean; October 1990. Snares Islands, New Zealand; March 1993. Marion Island, Indian Ocean; April 1993. Near Cape Town, South Africa; April 1992. Pedra Branca, Tasmania; August 1992. Marion Island, Indian Ocean; April 1993. Marion Island, Indian Ocean; April 1993. Gough Island, Atlantic Ocean; October 1990. Ile Amsterdam, Indian Ocean; July 1993. Forty-fours, Chatham Islands, New Zealand; March 1992. North Pacific Ocean (40�06'N, 161�30'E); August 1991. North Pacific Ocean (Washington State, USA); 1988. Isla Espafiola, Galapagos Islands; October 1993. Torishima, Japan; April 1993. Marion Island, Indian Ocean; October 1990. Gough Island, Atlantic Ocean; October 1990.