Molecular Systematics of the Scaled Quail Complex (Genus Callipepla)
Total Page:16
File Type:pdf, Size:1020Kb
The Auk 115(2):394-403, 1998 MOLECULAR SYSTEMATICS OF THE SCALED QUAIL COMPLEX (GENUS CALLIPEPLA) ROBERT M. ZINK 1 AND RACHELLE C. BLACKWELL JamesFord Bell Museum of NaturalHistory, 100 EcologyBuilding, University of Minnesota, St. Paul, Minnesota 55108, USA ABSTRACT.--Weobtained 1,040 bp of sequencefrom the mitochondrialDNA (mtDNA) genescytochrome b (cyt b; 736 bp) and NADH-subunit2 (ND2; 304 bp) to addressphylo- geneticrelationships among the four speciesin the ScaledQuail complex.California Quail (Callipeplacalifornica) and Gambel'sQuail (C. gambelii)were sistertaxa, whereasthe rela- tionshipsof theElegant Quail (C. douglasii)and Scaled Quail (C.squamata) were unclear; they might be sisterspecies, or ElegantQuail might be the sisterto California plus Gambel's quails.A third, less-likelyalternative predicts a contemporaneousorigin of ElegantQuail, ScaledQuail, and the ancestorof Californiaand Gambel'squail. The latterphylogenetic hy- pothesis,however, matches Hubbard's (1973) biogeographic model. Irrespective of which biogeographichypothesis is correct,calibration of mtDNA geneticdistances suggests that the speciationevents are mucholder than the late Pleistocenedates given by Hubbard.Cal- ibrationof the rate of mtDNA (cyt b, ND2) evolutionbased on datingof fossilremains of the extinctspecies Cyrtonyx cooki suggested a rate of 2% per millionyears. Northern Bobwhite (Colinusvirginianus), Mountain Quail (Oreortyxpictus), and Montezuma Quail (Cyrtonyx montezumae)were successivelymore distantly related to the ScaledQuail complex.Phylo- genetictrees derived from allozymes(Gutierrez et al. 1983) and mtDNA sequenceswere topologicallyidentical, suggesting that bothtypes of genetrees recover the speciestree. Re- ceived28 March 1997, accepted9 October1997. THE SCALEDQUAIL COMPLEX(Hubbard 1973) Slowinski 1995, Klicka and Zink 1997). Some- includes four species:California Quail (Calli- what unusual,at least by modernstandards, peplacalifornica), Gambel's Quail (C. gambelii), was the initial suppositionof a basaltrichoto- ScaledQuail (C. squamata),and ElegantQuail my. (C. douglasii).These species occur throughout We used sequencedata from the mitochon- arid regionsof the southwesternUnited States drial DNA (mtDNA) cytochromeb (cyt b) and and Mexico north of the Isthmus of Tehuante- NADH-subunit 2 (ND2) genes to test Hub- pec (Fig. 1). Hubbard (1973) proposedan evo- bard's(1973) hypothesis. Our data confirma lutionaryscenario involving at leasttwo glacial close relationshipbetween California Quail cyclesto accountfor the evolutionand biogeo- and Gambel'sQuail, and a possiblesister-taxon graphichistory of thesequail. During the Illi- relationshipbetween Scaled Quail andElegant noian Glacial a widespread ancestor was Quail. We also assessedcongruence between thoughtto havebeen isolated into three taxa phylogenetichypotheses derived from mtDNA termed " pre-californica/ gambelii," " pre-doug- sequencesand allozymevariation (Gutierrez et lasii," and "pre-squamata."Their rangesex- al. 1983). pandedduring the Sangamoninterglacial and during the WisconsinGlacial californicaand METHODS gambeliisplit into distinct species,whereas douglasiiand squamatadid not becomefurther We sequencedthree CaliforniaQuail (from Baja subdivided.Thus, Hubbardpresented an ex- California),three Gambel's Quail (New Mexico),four plicit and testablehypothesis for the evolution ScaledQuail (New Mexico),and three ElegantQuail and distributionof thesefour species.Typical (Sonora,Mexico), and one eachof the followingout- of suchexplanations was the assumptionthat group species:Northern Bobwhite(Colinus virgini- the lastbout of glaciationplayed a rolein spe- anus),Montezuma Quail (Cyrtonyxmontezumae), and ciation (Berminghamet al. 1992, Zink and MountainQuail (Oreortyxpictus). Specimen voucher numbersand detailson collectinglocalities are given with sequenceinformation in the Genbankaccession E-mail: rzink@biosci.umn.edu (nos.AFO28750 to 28782;see Table 2). Sequencedata 394 April 1998] MolecularSystematics ofQuail 395 'rnia Quail (C. californica) J Gambel'sQuail (C.gambelii) Scaled Quail (C. squamata) Elegant Quai ( C. douglasii) F•c. 1. Distributionof speciesin the ScaledQuail complex(from Hubbard 1973). for the aboveingroup and outgroupspecies consti- 1996) were performed and run on 6% acrylamide tute Data Set A. For comparisonwith allozymedata gels.Cyt b wassequenced in bothdirections with the of Gutierrezet al. (1983),we usedonly sequencedata amplification primers and one internal primer, for a 736 bp segmentof cyt b from Kornegayet al. H15299 (Hackett 1996). ND2 was sequencedonly (1993;Genbank accession code) for Chukar (Alectoris with H5578, because L5215 did not work as a se- chukar;L08378), JapaneseQuail (Coturnixcoturnix; quencingprimer. Sequences were alignedwith the L08377), and Silver Pheasant(Lophura nycthemera; published chicken sequence (Desjardins and Morais L08380,as a substitutefor theRing-necked Pheasant 1990);no gaps were detected. [Phasianuscolchicus]), and we obtainedsequence data We computed basic sequencestatistics, Kimura's of LesserPrairie-Chicken (Tympanuchus pallidicinc- (1980) two-parameterdistances (K2P), and neigh- tus)from J. G. Grothand G. E Barrowclough(unpubl. bor-joining(NJ) trees with MEGA (Kumar et al. data). Sequencedata for thesespecies plus thosein 1993). We used PAUP (Swofford 1993) to conduct Data Set A constitute Data Set B. We also determined maximumparsimony searches (branch and bound) the sequencefor a smallsegment of cyt b andthe ad- of thedata, with weightsof 1:1and 1:2 for transitions joining t-RNA•eu from samplesof study skinsfor two and tranversions,to bootstrap(Felsenstein 1985) the additional species,Banded Quail (Philortyxfasciatus; unweightedcharacters 1,000 times, and to compute Bell Museumornithology catalogue number 10947) g-valuesas measuresof phylogeneticsignal (Hillis and Singing Quail (Dactylortyxthoracicus; Bell Mu- 1991, Hillis and Huelsenbeck1992, Kallersjoet al. seum 14958), to assesstheir relationshipsto the 1992). PAUP* (Swofford pers. comm.) was used to ScaledQuail complex. computelog-determinant distances (log-der), a con- We used standard methods to extract DNA from servative estimate of mtDNA genetic distances tissueand study skins (Ellegren1992, Hillis et al. (Swofford et al. 1996). Maximum-likelihoodtrees 1996,Zink et al. 1997)and the polymerasechain re- were estimated with PHYLIP (Felsenstein1993). action (PCR; Saiki et al. 1988, Palumbi 1996) to am- Competingtree topologieswere tested with the plify mtDNA. PrimersL14841 (Kocheret al. 1989) Kishino-Hasegawa(1989) maximum likelihood test. and H4a (Harshman1996) amplified an approxi- MacClade (Maddison and Maddison 1992) was used mately 1,100bp segmentof cyt b. A 400 bp portion to evaluatethe lengthsof alternativetopologies. We of ND2 wasamplified with L5215and H5578 (Hack- used Mantel's (1967) test as implemented in NTSYS ett 1996).Manual sequencingreactions (Hillis et al. (Rohlf 1992) to comparematrices of Rogers'(1972) 396 ZINK AND BLACKWELL [Auk, Vol. 115 TABLE 1. Distribution of variation at nucleotide spectively)were: Adenine (27.1%, 34.0%), Thy- sitesfor cytochromeb (Cyt b) and NADH-subunit 2 (ND2); TS transition, TV - transversion. mine (25.2%, 22.8%), Cytosine (33.7%, 32.2%), and Guanine (14.0%, 11.0%). Of the 1,040 First Second Third alignedbases, 241 were variable,and 135 were Gene region TS TV TS TV TS TV potentiallyphylogenetically informative; vari- able sites were mostly third positiontransi- Cyt b 30 7 8 0 92 33 ND2 6 3 7 2 42 11 tions (Table 1). A total of 33 amino acids was variable (11 were parsimonyinformative). Sequencedivergence (K2P distances;Table 2) ranged from 0.0019 (within California Quail genetic distances derived from allozyme compari- and Gambel's Quail) to 0.1635 (Montezuma sons (Gutierrez et al. 1983) and 1og-detdistances; Quail vs. Mountain Quail). Among the four significancevalues were basedon 9,999 randomma- trix permutations.The absolutenumbers of transi- speciesof the ScaledQuail complex,sequence tions and transversionsat first and third positionsof divergenceaveraged 0.054 -+ SD of 0.016, and codonswere plotted against1og-det distance to eval- ranged from 0.022 (California vs. Gambel's uate saturation.To assessmolecular-rate heteroge- quail)to 0.069(Scaled vs. Gambel'squail). The neity,we comparedthe log likelihoodsof maximum- Northern Bobwhite was closer to the Scaled likelihoodtrees computed with and withoutthe as- Quail complex(average K2P distance = 0.085 sumption of a molecular clock (PHYLIP routines + 0.002) than was the Mountain Quail (0.115 _+ DNAMLK and DNAML, respectively)by doubling the difference between the two likelihood estimates 0.004) or the MontezumaQuail (0.151 + 0.004). Saturation was not evident at first and third and determiningits chi-squareprobability (df = n - 2, where n = number of species;Felsenstein 1993). positions(Fig. 2), nor at the nearly invariant secondposition (not shown).Log likelihoods RESULTS derived from analyseswith and without the as- sumption of a molecular clock did not differ DataSet A.--We obtained1,040 bp of mtDNA significantly(P > 0.10), indicating an absence sequencefor theScaled Quail complexand out- of molecular-rateheterogeneity among taxa. groups,including 304 bp from ND2 and 736bp A significantg•-value of -0.63 suggestedsig- from cyt b. Fromthe 17 individualssequenced, nal in the data(Kallersjo et al. 1992).The short- we identified 11 haplotypes.Percentage base est tree among 1,000 random trees, 460 steps, compositionwas biasedin a manner typical of was significantlylonger