The Auk 115(1):188-195, 1998 CYTOCHROME-B EVIDENCE FOR VALIDITY AND PHYLOGENETIC RELATIONSHIPS OF PSEUDOBULWERIA AND BULWERIA (PROCELLARIIDAE) VINCENT BRETAGNOLLE,•'5 CAROLE A3VFII•,2 AND ERIC PASQUET3'4 •CEBC-CNRS, 79360 Beauvoirsur Niort, France; 2Villiers en Bois, 79360 Beauvoirsur Niort, France; 3Laboratoirede ZoologieMammi•res et Oiseaux,Museum National d'Histoire Naturelie, 55 rue Buffon,75005 Paris, France; and 4Laboratoirede Syst•matiquemol•culaire, CNRS-GDR 1005, Museum National d'Histoire Naturelie, 43 rue Cuvier, 75005 Paris, France ABSTRACT.--Althoughthe genus Pseudobulweria was described in 1936for the Fiji Petrel (Ps.macgillivrayi), itsvalidity, phylogenetic relationships, and the number of constituenttaxa it containsremain controversial. We tried to clarifythese issues with 496bp sequencesfrom the mitochondrialcytochrome-b gene of 12 taxa representingthree putative subspecies of Pseudobulweria,seven species in six othergenera of the Procellariidae(fulmars, petrels, and shearwaters),and onespecies each from the Hydrobatidae(storm-petrels) and Pelecanoidi- dae (diving-petrels).We alsoinclude published sequences for two otherpetrels (Procellaria cinereaand Macronectesgiganteus ) and use Diomedeaexulans and Pelecanuserythrorhynchos as outgroups.Based on thepronounced sequence divergence (5 to 5.5%)and separate phylo- genetichistory from othergenera that havebeen thought to be closelyrelated to or have beensynonymized with Pseudobulweria,we conclude that the genusis valid, and that the MascarenePetrel (Pseudobulweria aterrima) and the Tahiti Petrel (Ps. rostrata) are distinct spe- cies.In treesconstructed with maximumparsimony and maximumlikelihood, Pseudobul- weriais the sistertaxon to Puffinusand Calonectris, and these genera in turn aremost closely relatedto Bulweria(and Procellariain the maximum-parsimonytree). Pseudobulweria is not closelyrelated to Pterodromain either tree. Because Ps. r. trouessartifrom New Caledonia, and Ps.r. rostratafrom Polynesia differ by only0.6%, these taxa do notdeserve species status and shouldbe regardedas valid subspecies.Received 7 October 1996, accepted 23 July1997. THE GENUSPSEUDOBULWERIA, first proposed (Atti• et al. 1997),and Ps.rupinarum already is by Mathewsin 1936for the Fiji Petrel(Ps. mac- extinct (Olson 1975). Therefore, systematic gillivrayi),remains one of the least-knowngen- studieson this grouphave relied nearly exclu- era of petrels(Warham 1996, Atti• et al. 1997). sivelyon Ps. r. rostrataand on morphological at- Althoughit hasbeen synonymized with Pter- tributesassessed from museumspecimens (e.g. odroma(e.g. Jouanin and Mougin 1979, Warham Jouanin 1970, 1987; De Naurois and Erard 1979; 1990,Del Hoyo et al. 1992),some systematists Imber 1985). have reinstatedit as a separategenus (Imber Using availabledata as well as new infor- 1985,Warham 1996),either closeto Pterodroma mation provided by DNA sequencing,we ex- (Sibley and Monroe 1990) or to Bulweriaand amined:(1) the validity of the taxon aterrima Procellaria(Imber 1985).In his reappraisalof comparedwith rostrata,because these two taxa the petrels,Imber (1985)also included in Pseu- have been suggestedto be conspecific(e.g. dobulweriathe MascarenePetrel (Ps. aterrima), Jouanin1970); (2) thevalidity of thegenus Pseu- the Tahiti Petrel(Ps. rostrata), and a fossiltaxon dobulweria;and (3) thephylogenetic positions of (Ps.rupinarum). Beck's Petrel (Ps. rostrata becki) Pseudobulweriaand Bulweriawithin the family is knownfrom onlytwo specimensand may be Procellariidae.For this latter aspect,we paid extinct. Similarly, Ps. macgillivrayiis known particularattention to Bulweria,which has been from only two specimens(Watling and Lewa- claimedto be closeto (Kuroda1954) or syn- navanua1985), Ps.aterrima is known from sev- onymouswith Pterodroma(Olson 1975), and to- en specimensand is on the vergeof extinction getherwith Pseudobulweriais thought to be an- cient(Bourne 1975; Imber 1985;Warham 1990, E-mail: [email protected] 1996).We usedthe mtDNA sequence(496 bp 188 January1998] RelationshipsofPetrels 189 segment) at the cytochrome-blocus because TCCTGTTTCGTGGAGGAAGGT-3').These primers this segmenthas proved valuable for phyloge- amplify a DNA segmentof 496 bases.PCR was per- netic analysisat the species,generic, and fa- formed in a 50-p•Lvolume using ca. 0.3 p•gof tem- milial levelsin birds (e.g.Lanyon 1994, Baker plateDNA and 50 picomolesof eachof the primers. et al. 1995,Helbig et al. 1995,Arctander et al. The PCRmix (final concentrations)contained 20 mM Tris-HCL, pH 8.55, 16 mM (NH4)2SO•, 2.5 mM 1996, Friesenet al. 1996, Krajewski and King MgC1v 150 p•g/mL BSA, 330 p•M dNTP, and 0.3 1996). (1.5 units) of GoldstarTaq DNA polymerase(Euro- genetec).The PCRprofile for amplificationswas 35 MATERIAL AND METHODS cyclesof 60 s at 93øC,40 s at 50øC,and 40 s at 72øC. PCR productswere opened under a speciallyde- Speciesand sampling techniques.--We obtained sam- signedhood and checkedby electrophoresisin 1% ples from sevenprocellariid genera as well as one agarose-BETand TBE buffer (Sambrooket al. 1989) speciesof storm-petreland onespecies of diving-pe- with the molecularweight marker VI (Boehringer). trel: Tahiti Petrel (Ps. r. trouessarti;New Caledonia, PCR productswere clonedusing the PCRscriptTM blood sampleand Ps. r. rostrata;Gambier Is., blood), SK(+) cloningkit (Stratagene)according to recom- Mascarene Petrel (R•union I., tissue), Barau'sPetrel mendedspecifications. A classicalwhite/blue selec- (Pterodromabaraui; R•union I., tissue),Black-winged tion (Sambrooket al. 1989) was used for screening Petrel (Pt. nigripennis;New Caledonia,blood), Bul- recombinantclones. Four white coloniesper cloning wer's Petrel (Bulweriabulwerii; Canary Is., blood), were grown overnightin L-broth at 37øC,the pha- Cory's Shearwater (Calonectrisdiomedea; Malta, gemidic DNA extracted (Sambrooket al. 1989), and blood), Wedge-tailedShearwater (Puffinus pacificus; the insertwas checkedby digestionof the recombi- New Caledonia,blood), Northern Fulmar (Fulmarus nant phagemidicDNA with BssHII.Sequencing on glacialis;Brittany, tissue), Snow Petrel(Pagodroma ni- microtiter plates was performed with the T7 se- vea;Ad•lie Land, Antarctica,blood), Leach'sStorm- quencingkit (Pharmacia),using the method of ter- Petrel (Oceanodromaleucorhoa; French Guyana, tis- minator dideoxynucleotides(Sanger et al. 1977). sue), and Common Diving-Petrel (Pelecanoidesuri- Two coloniesper cloningwere sequencedwith ex- natrix;New Zealand,tissue). We alsoused published ternalvector primers KS and T3 to checkfor artifacts sequencesfrom Procellariacinerea and Macronectesgi- introduced by cloning of PCR products. Differences ganteus(Nunn et al. 1996), making availabledata betweenclones occurred in only two samples(both from9 of the 13genera in thisfamily (sensu Warham one-transitiondifference in the 496 bp), and a third 1996). Diomedeaexulans (Nunn et al. 1996) and Pele- colony was sequencedto confirm the correctse- canuserythrorhynchos (Avise et al. 1994)were used as quence. outgrouptaxa. Bloodsamples (ca. 1 mL) were taken Data ana[ysis.--Sequenceswere read and entered from the tarsusfor the larger speciesand from the twice usingthe computerpackage MUST (Philippe wing for the smallerspecies and preservedin a buf- 1993) and were easily aligned becauseno insertions feredsolution of 4M guanidine-thiocyanate(see Lau- or deletions were found. Relative transitional satu- lier et al. 1995). Tissuesamples (liver) were taken ration was examinedby plotting transitionalagainst from specimensfrozen at -18øC. One specimenwas transversionalpairwise raw differences.Maximum- sampledfor eachspecies. likelihood (ML) analyses(Felsenstein 1981) were DNA extractionand PCR amplification.--Asmall performed with program FastDNAml (Olsen et al. amount of tissue or blood (ca. 0.1 g) was powdered 1994).The optionsF (empiricalfrequencies) and G in liquid nitrogenand suspendedwithin CTABbuff- (global rearrangements)were used, and 10 random- er containing proteinaseK (Doyle and Doyle 1987, input order of specieswere done for eachmodel to Winnepenninckxet al. 1993). Digestion was at 60øC minimize input-order bias. We used three models for 1 h, and protein isolationwas carried out with differingin the numberof categoriesof substitution chloroform-isoamylalcohol. In addition,0.5 unitsof rates(CSR). In 1-CSR,the three codonpositions had RNAasewere added to the secondaqueous phase equal substitution rates; in the 2-CSR model, first and incubated at 37øC for 30 min to remove RNA. To- and secondcodon positions had equalrates whereas tal genomicDNA wasprecipitated with isopropanol. the third differed; and in the 3-CSR model, each co- DNA concentrationand qualitywere evaluated with don positionhad a differentrate of substitution.Two a spectrophotometer.Extracted DNA was amplified sets of relative codon-positionevolutionary rates with thefollowing two primers,whose numbers cor- (lst:2nd for the 2-CSRmodel, and lst:2nd:3rdfor the respondto the locationof the 3' end of the primer 3-CSR model) were used: 1:5 and 1:10, and 2:1:5 and respectivelyin the full sequenceof human mtDNA 2:1:10, respectively.Three transition-transversion (Andersonet al. 1981, in bold) and chickenmtDNA ratios(TS:TV) were used and thusdefined a prioriin (Desjardinsand Morais 1990):L14841/L14990 (5'-CA- the likelihoodcomputations: 2:1, 5:1, and 10:1.The TCCAACATCTCTGCTTGATGAAA-3')defined by best likelihood values were obtained for 2-CSR and Kocher et al. (1989), and H15338/H15487 (5'-GA- 3-CSR models with TS:TV=5:I, and with evolution- 190 BRETAGNOLLE,ATTIf:,