The Auk 118(3):611-623, 2001

A CYTOCHROME-b PERSPECTIVE ON PASSERINA BUNTING RELATIONSHIPS

JOHN KLICKA,L3 ADAM J. FRY,TM ROBERT M. ZINK, • AND CHRISTOPHERW. THOMPSON2,5 •J.E BellMuseum of NaturalHistory, University of Minnesota, 1987 UpperBuford Circle, St. Paul,Minnesota 55108, USA; and 2BurkeMuseum and Department of ZoologyDB-10, University of Washington,Seattle, Washington98195, USA

ABSTRACT.--Wesequenced the completemitochondrial cytochrome-b gene (1,143 nucle- otides)for representativesof eachspecies in thecardinalid genera Passerina (6 species),Guir- aca(1 species),and Cyanocompsa(3 species), and useda variety of phylogeneticmethods to addressrelationships within and amonggenera. We determinedthat Passerina,as presently recognized,is paraphyletic.Lazuli Bunting(P. amoena) is sisterto the muchlarger Blue Gros- beak (Guiracacaerulea). Indigo Bunting(P. cyanea) and Lazuli Buntingare not sistertaxa as generallythought. In all weightedparsimony trees and for the gamma-correctedHKY tree, Indigo Buntingis the sisterof two sistergroups, a "blue" (Lazuli Buntingand Blue Gros- beak)and a "painted" (Rosita'sBunting [P. rositae], Orange-breasted Bunting [P. leclancherii], VariedBunting [P. versicolor], and Painted Bunting [P. ciris]) clade. The latter two speciesform a highly supportedsister pair of relativelymore recent origin. Uncorrected (p) distancesfor ingroup (Passerinaand Guiraca)taxa range from 3.0% (P.versicolor-P. ciris) to 7.6%(P. cyanea- P.leclancherii) and average6.5% overall. Assuming a molecularclock, a bunting"radiation" between4.1 and 7.3 Mya yielded four lineages.This timing is consistentwith fossilevidence and coincideswith a late-Miocenecooling during which a varietyof westerngrassland hab- itatsevolved. A reductionin sizeat that time may haveallowed buntings to exploitthat new foodresource (grass seeds). We speculate that the BlueGrosbeak subsequently gained large size and widespreaddistribution as a result of ecologicalcharacter displacement. Received 25 October1999, accepted 16 September2000.

THE BUNTINGSof the avian genusPasserina ently recognized (Sibley and Monroe 1990, (family Cardinalidae) include six species AOU 1998),form a natural group.Nevertheless, whose compositerange includesMexico, the some authors (Phillips et al. 1964, Blake 1969, United States,and parts of southernCanada. Mayr and Short1970) place the monotypicBlue They are small, heavy-billed finches,most of Grosbeak(Guiraca caerulea) within this genus, which (see Thompson and Leu 1995) exhibit and others(Paynter 1970) further expandPas- pronouncedsexual dichromatism with brightly serinato include additional genera.Thus, de- coloredmales and mostlybrownish or olive fe- spite the phenotypic similarities of the six rec- males. Traditional museum studies relying ognized species, monophyly of Passerina upon comparisonsof studyskins (e.g. Ridgway remains unresolved. 1901),as well asa morerecent study employing Relationshipswithin Passerinaalso remain numericalphenetic analyses of both skinsand unclear. Phillips et al. (1964) consider the In- skeletons(Hellack and Schnell 1977), have con- digo Bunting(E cyanea)and the Lazuli Bunting cludedthat the membersof Passerina,as pres- (P. amoena)conspecific. Hybridization on the GreatPlains between P. cyanea and P.amoena has 3 Present address: Barrick Museum, Box 454012, been well documented(Sibley and Short 1959, University of Nevada Las Vegas, 4505 Maryland Emlen et al. 1975, Kroodsma et al. 1975). Sim- Parkway, Las Vegas, Nevada 89154-4012, USA. ilarly, the Painted Bunting (P.ciris) and Varied E-mail: [email protected] Bunting (P. versicolor)are known to have hy- 4Present address: Department of Ecology and bridized (Storer 1961). Whereas such hybrid- EvolutionaryBiology, Box G-W, Brown University, Providence, Rhode Island 02912, USA. izationmight be takenas evidence of sister-spe- sPresent address: Washington Department of Fish ciesrelationships, a known pairing betweenP. and Wildlife, 16018 Mill Creek Boulevard, Mill cyaneaand P. ciris (Taylor 1974) muddles the Creek, Washington98012, USA. picture.The ability to hybridizemight be a mis-

611 612 KLICKAET AL. [Auk, Vol. 118 leading indicator of relationship (Zink and (outgroup). Those three generawere thus targeted McKitrick 1995), therefore the construction of for completecyt-b genesequencing. Because adding an independentphylogeny is warranted. more taxa to the sistergroup increasesthe chanceof Our understandingof the speciesin Passerina obtaining the correcttree (Smith 1994), single rep- resentatives of all three members of Cyanocompsa is uneven.Passerina cyanea is among our best were subsequentlyused. To obtaina measureof in- known songbirdswith song,display behaviors, traspecificgenetic variation and to provide a check mating strategies,molt, and migration having for potentialsequencing errors, each of sixspecies of been studied extensively(see Payne's[1992] Passerinaand the monotypic G. caerulea(ingroup comprehensivereview for references).At the taxa) were representedby two differenttissue spec- other extreme are the endemic forms of south- imens (Table 1). ern Mexico, Leclancher's(Orange-breasted) Laboratoryprotocols.--Total DNA was extracted Bunting (P.leclancherii) and Rosita's(Rose-bel- from fragments(-100 mg) of muscleor liver tissues lied) Bunting(P. rositae), about which very little followingeither a standardphenol/chloroform pro- tocol (Hillis et al. 1996) or via incubationin Chelex/ is known. A few comparativestudies have in- Proteinase K, a modification of Ellegren's (1992) volved most or all members of Passerina, ex- method. Overlapping sequencefragments were am- amining molts and plumages(summarized in plified via the polymerasechain reaction(PCR) us- Thompsonand Leu 1994), songs (Thompson ing variouscombinations of the followingpublished 1968), and morphology (Hellack and Schnell primers:L14851 (Kornegayet al. 1993),L14841 (Ko- 1977). However,interpretation of thosestudies cher et al. 1989), H15299 (Hackett 1996), H4A is compromisedby the lack of an explicitphy- (Harshman 1996), LCBA, LCBB, LCBC, HCBC logenetic hypothesis (Brooks and McLennan (Klicka et al. 1999);and two designed(J. Klicka) for 1991). general use on New World nine-primariedoscines: In this study,we usedDNA sequencesfrom HCBRB (GGAGAATGACTCCGAYGTTTCA), and the completemitochondrial (mtDNA) protein HCBJT (GGCTGGGGTGAAATTTTCTGGGTCT). A map of primer locationsis availableby request.PCR coding cytochrome-b(cyt-b) gene to assessthe reactionswere performedin 50 bL total volumesus- monophyly of Passerinaas presently recog- ing a 1.5 mM concentrationof MgC12,1 •M concen- nized and to constructa phylogenetichypoth- trations of eachprimer, 0.2 mM final concentration esis for the constituentsof this group. Cyt-b of dNTPs, 0.5 •L of Tfi polymerase(Thermusfiavus, and its flanking regionsare well characterized Epicentre), 2.2 •L of 20x Tfi buffer (400 mM in birds and its rate of evolutionis particularly [NH412SO4and 1 M Tris-HC1 [pH 9.0], Epicentre)and well suited for avian studiesat the specieslevel 10-1,000 ng (2 •L) of DNA template.A typical dou- (Moore and DeFilippis 1997). The resulting ble-strandedamplification began with a 2 min de- phylogenetictree providesa historicalframe- naturationat 94øCfollowed by 38 cyclesof denatur- work for a better understandingof the evolu- ing (94øC,45 s), annealing(54øC, 1 min, 30 s), and primer extension(72øC, 1 min, 30 s) with a final ex- tion of variouslife-history traits and for inves- tensionof 72øCfor 10 min. The presenceof the de- tigating historicalbiogeography. siredproduct was verified by electrophoresisof 5 bL in a 1% agarosegel (SeakemLE, FMC) followedby METHODS Ethidium Bromide staining. Excess primers and dNTPs were then removed by enzymatic treatment Taxon sampling.--Outgroupchoice is a critical of the PCR product with Exonuclease1 and Shrimp component of molecular phylogenetic analysis Alkaline Phosphatase(U.S. BiochemicalCorp.). To (Smith 1994).Outgroups that are too distantly relat- ensuresequencing accuracy, large fragment overlaps ed to the group of interest lead to spuriousrooting were used and wherever possible both light and of the ingroup topology (Wheeler 1990). Relation- heavy strandsof the treatedproduct were sequenced ships among Cardinalid genera are poorly under- using either the original or nestedprimers. Sequenc- stood, thereforean initial genericlevel phylogenetic ing reactions(Hillis et al. 1996)were donemanually surveyof the groupwas performed.A cyt-b (430bp) (productno. 70130,U.S.B.), run out on 6% acrylam- segmentfrom at leastone representativeof eachpu- ide (Gene-Page,Amresco) gels, and visualizedby au- tative Cardinalid genus (exceptingCyanoloxia) was toradiography (3sS). Sequences (GenBank nos. sequenced,as were representativesof severalThrau- AF301446-AF301462) were aligned visually using pid genera,which were used as outgroups.The de- published chicken (Gallus;Desjardins and Morals tailed resultsof that initial study are to be published 1990) and Snow Goose (Chen [Anser] caerulescens; elsewhere;however, in all phylogeneticanalyses the Quinn and Wilson 1993) sequencesfor comparison. Passerinabuntings and G. caeruleawere determined Data analysis.--Parsimony(uniform and differen- to constitutea clade (ingroup), sisterto Cyanocompsa tial weighting schemes)and maximum-likelihood April 2001] PasserinaBunting Paraphyly 613

z methodswere used to estimatephylogenetic trees For comparison,and to allow quick visual assess- ment of relative geneticdistances, we alsoconstruct- ed neighbor-joiningtrees. All analyseswere done us- ing PAUP* (version 4.062; Swofford 1999). To assess saturation, absolute numbers of transitions and transversionsat first, second,and third codonposi- tionswere plotted againstKimura 2-parameter(K2- P, Kimura 1980)distances for all pairwisecompari- sons.Deviations from linearity(see below) indicated saturationeffects, suggesting that the weightingof the more slowly evolvingand presumablyless ho- moplastic transversionswas warranted. We used an array of maximumparsimony (MP) weightingschemes to explorethe dynamicsof cyt-b evolution(Voelker and Edwards1998). All analyses used branch-and-bound or full heuristic searches with 10 randomaddition sequence replicates. Initial parsimonyanalyses including 1:1 (equalweights), 2: 1, and 5:1 weighting of transversionsover transitions resulted in two nearly identical topologieswhose unweightedtree lengthsdiffered by a single step. Compositeand codonposition-specific sequence sta- tisticswere generatedin PAUP*using the weighted topology (but equal weights). Relative rates of changeamong codon positions, percentage nucleo- tide composition,and transition:transversionratios were calculated in MacClade (Maddison and Mad- dison 1992) by independentlyreconstructing, for varioussequence partitions (see below), the average number of eachtransformation type onto the same tree. Subsequently,an analysiswas performedin which weightswere assignedon the basisof the ra- tios calculatedfor eachcodon partition; thus, first, second,and third positionsites were weighted8.6, 1.5, and 4.9, respectively.The "native" transitionto transversionratio (Sturmbauerand Meyer 1992)was also approximatedby plotting transitionsversus transversionsfor third position sites (plot not shown,approximate ratio 9.5:1).Following Yoder et al. (1996),we alsoused a 5:1 weightingof third po- sition transversionsonly. Six-parameterweighting schemesaccount for an additionalsource of error,base composition biases. The logic is identical to that behind transversion weighting;frequent classes of characterchange are consideredmore likely than rare classesto have ex- periencedhomoplasy (Cunningham 1997). Each of six possible transformations (Ae•G, Ce•T, Ae•C, Ae•T, Ce•G, Te•G) wasassigned a weighton theba- sis of the observedfrequency in the data (Williams and Fitch1989). These weights were derivedby first mapping changes(average of all possiblereconstruc- tionsused) onto the equal-weight(1:1, [nearly iden- ticalresults were obtained using the weightedtree]) parsimony tree using MacClade (Maddison and Maddison 1992). The frequenciesof each of the six classesof changewere then calculated and negative- 614 KLICKAET AL. [Auk, Vol. 118

TABLE2. Cytochrome-bsubstitution type propor- Node supportfor MP analyseswas determinedvia tions shown above the diagonal; corresponding bootstrapping(Felsenstein 1985) with 500 replications six-parameterweights shown below. Valueswere using full heuristicsearches and random additionof obtainedby reconstructingthe dataonto the short- taxa. One hundredsuch replicates were donefor ML est equally weighted tree. Correctionsfor triangle inequalityshown in parentheses. analyses.Decay indices (Bremer 1994) were generated for the mosthighly supportedparsimony-derived to- A C G T pology using TREEROT (Sorenson1996). MacClade (Maddisonand Maddison1992) was usedto compare A -- 11.6% 23.5% 4.2% tree lengthsfor competingtree topologies.Alternative C 4.1 -- 1.0% 59.3% topologieswere testedusing the Kishino-Hasegawa G 2.8 8.8 (6.9) -- 0.4% T 6.0 (5.1) 1.0 10.6 (7.9) -- (1989)maximum-likelihood ratio test. Biogeographicinterpretations are enhancedby in- voking a molecularclock. The validity of assuming clock-like rates of evolution for those taxa was tested natural-log (-In) transformedto obtain the weights via a likelihood-ratio test (Huelsenbeckand Rannala enteredinto a stepmatrix (Table2). 1997) by comparinglog-likelihood values from ML Mutation rates are known to vary throughoutthe trees constructed with and without a molecular clock cyt-b gene, presumably due to functional con- constraint using the HKY85/gamma(F)-corrected straints at the amino acid level (e.g. Howell 1989, model of sequenceevolution. To allow direct com- Krajewski and King 1996, Griffiths 1997). We ad- parisonwith the Fleischeret al. (1998)songbird-spe- dressedthat problem by using MP and a differen- cific clock calibration,we generateda matrix of K2- tial weighting schemein which transitionslocated P F-corrected pairwise distances following their in putatively more variable regions were down- protocol.The otvalue used (0.300)was estimatedvia weighted. Three rate classesof residues were de- the method of Sullivan et al. (1995) in PAUP* from fined using Howell's (1989) structural model for the shortestunweighted MP tree. mousecyt-b in which individual residueswere clas- sified as evolutionarilyconserved, intermediate, or var- RESULTS iable.Although similar, thesepartitions do not cor- respond exactly to those used in other published Sequencevariation.--All 1,143 nucleotides studies(e.g. Griffiths 1997). We performed an anal- were aligned with no insertions,deletions, or ysis in which the sequenceconstituting each of nonsensecodons in evidence.Although the am- these three partitions was weighted according to the inherent transition:transversion ratio (calculat- plificationof nuclearpseudogenes of mitochon- ed in MacClade as described above) in that data drial origin can confoundphylogenetic analy- partition. The result was transversionsweighted sis (Zhang and Hewitt 1996), a number of 3.0:1, 6.2:1, and 4.8:1 over transitions for conserved, factorsmake it unlikely that our sequencesare intermediate, and variable sequencepartitions re- nuclear copies.In addition to straightforward spectively. Standard tree statistics, including alignment and lack of insertionsor deletions, length, consistencyindex (CI), retention index (RI), heme-ligating histidines and other conserved and rescaledconsistency index (RC) were generat- residues(Howell 1989) were identified, and no ed for all parsimony reconstructions. coamplified DNA product (i.e. ambiguous All maximum-likelihood(ML) analysesused em- sites)was apparent on the sequencinggels. For pirical basefrequencies, with full heuristicsearches. An initial analysiswas basedon Felsenstein's(1981, birds, nuclearcopies are most often associated [F81]) model of DNA evolution which corrects for with the use of blood as a DNA source(Johnson unequalbase frequencies. The datawere subsequent- and Sorenson 1998); we used muscle or liver ly analyzedwith the HKY85 (Hasegawaet al. 1985) tissueexclusively. Furthermore, the nucleotide model which additionally accommodatestransition bias observedin gene partitionsof this data set biases.That model in particular has proven to be ef- is consistentwith that reported for other birds fective in describingthe dynamicsof many genes (Moore and DeFilippis 1997). Lastly, our re- (Goldman 1993, Yang 1996). This analysis used a suits are reasonablefrom both a biologicaland transversiontransition ratio of 6.9 and a gamma dis- geographicalperspective, as we show below. tribution with a shapeparameter (or) of 0.106.These Of the 1,143aligned sites,240 were variable, parameterswere estimatedusing the stable(1:1, 2:1) MP treesfollowing the recommendationof Swofford and of those,195 were potentiallyphylogenet- et al. (1996). Becausethe models used are nested, we ically informative(Table 3). Among thosesites, were able to assessstatistically their relative values 164 (84%) were at third positions,28 (14.5%) at by comparinglog-likelihood values using likelihood first positions,and only 3 (1.5%) at secondpo- ratio tests (LRT, Huelsenbeckand Rannala 1997). sition sites. Relative rate calculations indicate April 2001] PasserinaBunting Paraphyly 615

TABLE3. Overall and codonposition-specific dynamics of the cylochrome-bgene for all taxa. Mean base compositionis averagedover all sequencesusing PAUP*. Transition-transversion ratio (Ts/Tv) valuesare the averagenumber of changesreconstructed on the weightedparsimony topology.

Parsi- mony Posi- Number Variable inform- Relative tion of sites sites ative rate %A %C %G %T Ts/Tv

All 1143 240 195 16.0 27.3 34.7 13.3 24.7 5.08 0.106 1st 381 33 28 6.6 25.6 30.0 23.4 21.0 8.60 0.007 2nd 381 5 3 1.0 20.7 25.4 12.6 41.3 1.50 0• 3rd 381 202 164 40.4 35.5 48.8 3.8 11.9 4.91 1.07

that third-positionsites are evolving40 x faster For these data, log-likelihoodsderived from than are second-positionsites. Nucleotide com- analyseswith and without the assumptionof a positional bias was most acute at third codon molecular clock (Table5) did not differ signif- positions (0.173), followed by second (0.058) icantly (0.5 > P > 0.1) suggestingthat these and first (0.006)positions. That was due mainly taxa display an approximatelyuniform rate of to the expectedstrong bias againstG (3.8%) at substitutionsuch that inferenceof divergence that site. Overall, base composition(Table 3) times is appropriate. and base-compositionbiases are similar to Phylogeneticanalyses. From the many anal- those recovered in other avian cyt-b studies ysesperformed, two fully resolvedtopologies (e.g. Kornegayet al. 1993, Nunn et al. 1996). were consistentlyrecovered (Fig. 2A, C). They Most empirical gamma-shapeparameter es- differ only in the placementof P. leclancherii. timates (o0 vary from 0.1 to 0.5 (Yang1996) and Equally weighted parsimony,distance, and the the compositesequence estimate for thosedata F81 maximum-likelihood analysis yield a to- falls within this range. However, o• estimates pology in which the Mexican endemicsP. le- for first and third positionsdiffer by two orders clancheriiand P. rositaeare sisters (Fig. 2A), of magnitude (0.007 vs. 1.07), suggestingthat whereas in all weighted parsimony and other severeamong-site rate heterogeneityexists at likelihood analyses,P. leclancherii is sisterto the first positions,whereas the higher value asso- well supported P. ciris-P.versicolor clade (Fig. ciated with third positions suggestsa more 2C). With the data weighted equally, those to- even distribution of substitutions.Despite the pologiesdiffer by a singlestep (Fig. 2A: length fact that the preponderanceof phylogenetically [1] = 419, CI; 0.649, RI; 0.742, RC = 0.481; informativesites are at third positions,we note 3c: 1 = 420, CI = 0.648, RI; 0.740, RC = 0.479; that the overall o•estimate (0.106) imposed dif- Kishino-Hasegawatest [1989],P = 0.3950).Be- fers from the estimate for only third positions causethe Figure 2C topologywas recoveredin by an order of magnitude. all weighted-parsimonyanalyses and in the Uncorrectedpercentage sequence divergence best fit maximum-likelihoodanalysis, we con- (p, Table 4) between Passerinaspecies ranges sider this our best estimate of relationships from 3.0% (P. versicolor-P.ciris) to 7.6% (P. cy- within that group. Lanyon (1993) advisesthat anea-P.leclancherii) and averages6.5 (_1.0%, n systematistsidentify both a "best estimate"and = 84). Minimum within-speciespairwise val- [italics his] a "reliable estimate" of phyloge- ues range from 0.1% (P.ciris) to 0.4% (P.cyanea netic relationships.In that spirit, we consider and G. caerulea).Overall, intraspecific diver- the more conservative,bootstrapped equally gencewas 0.24 (+0.1%, n = 7), less than the weighted MP tree (Fig. 2B) as our mostreliable empiricalvalue of 0.7% that was calculatedfor estimate. a rangeof aviantaxa (Moore1995). Despite the In both fully resolvedtrees, P. cyanea is sister relatively low divergence values, saturation to all other ingroup taxa, althoughthat node is plots for the combineddataset (Passerina, G. ca- poorly supported in several analyses. Two erulea,and Cyanocompsa)provide evidencethat main cladesare supportedin both trees,a pre- multiple substitutionshave begun to accumu- dominatelytemperate zone ("blue"; G. caeru- late in third position transitions (Fig. 1, other lea + P. amoena)clade and a more sedentary [linear] plots not shown). group ("painted" = P.rositae + P.leclancherii + 616 KLICKAET AL. [Auk, Vol. 118

ß Passerina x Passedna e x Cyanocompsa ß x Gallus

0 0.05 0.1 0.15 0.2 0.25 0.3 Corrected Sequence Divergence

F•G.1. Saturationplot for the cytochrome-bgene, third-position transitionsonly. Within-speciescom- parisonsare included. Similar plots of other codon positions and transformation types (not shown) were approximatelylinear.

P. versicolor+ P. ciris) centered in the aridlands of the southwestern United States and western Mexico. Levels of support as measuredby de- cay indices(Fig. 2A) and bootstrapvalues vary among nodes (Table5) and betweenanalytical methods.Perhaps the most extremecase of the latter is the node which unites the painted clade (C on Fig. 2C). Under the HKYF model, that node has 78% bootstrap support yet its counterpart on the most parsimoniousequal- weights topology has a decay index of 1 and <50% bootstrap support. Similarly, support for a basalP. cyanea (node B) is moderatelyhigh under the equaland 2:1parsimony analysis but is lacking under the HKYF maximum likeli- hood model.

...... Both treesdemonstrate that the monotypicG. caerulea,despite having a massapproximately twice that of mostPasserina buntings, is embed- ded within Passerina,rendering the genuspar- aphyletic.In fact, in all analysesG. caeruleaand P. amoenawere sister taxa with relatively high support from both bootstrapand decayvalues (Table 5). ConstrainingP. cyaneaand P. amoena to be sisters(the traditional arrangement)with G. caeruleasister to all of Passerina,requires 17 additional (equalweight) stepsand resultsin a significantly worse topology (Kishino-Hase- gawa [1989]test, P = 0.0498).Likewise, forcing G. caeruleato be sisterto the similarly sizedspe- April 2001] ?asserinaBunting Paraphyly 617

TABLE5. Levelsof bootstrapsupport for selectnodes (see Fig. 2c) for combineddata usingmaximum par- simonyand maximum likelihood analyses.All MP (x 500) and ML (x 100) replicatesincluded full heuristic searches.For ML estimates,empirical nucleotidefrequencies were used and ot(=0.106) and T,/Tv (=6.92) were estimated from the data under the HKY model of evolution.

Parsimony'• Likelihood b Node =W x2 x5 x9.5 Y 6P CS How F81 HKY HKYF HKYF + c

A 99 99 98 95 96 95 99 98 100 100 100 -- B 71 71 58 <50 58 64 59 54 74 60 51 -- C -- 52 71 75 72 61 74 67 -- 63 78 -- D -- 52 75 82 76 <50 70 75 -- 55 62 -- E 96 99 97 91 97 96 100 94 95 98 99 -- F 99 99 99 100 99 97 100 98 100 100 99 -- G 73 81 80 84 84 83 82 71 69 80 82 -- -Ln 3,984.3* 3,756.6* 3,568.4 3,576.4 length 419 CI 0.649 0.671 0.715 0.747 0.703 0.688 0.719 0.717 RI 0.742 0.756 0.786 0.809 0.781 0.767 0.791 0.791 RC 0.481 0.508 0.562 0.605 0.549 0.528 0.569 0.567

• For MP, x9.5 = native rate weighting,Y = (Yoder) x5 weightingof third positiontransverions only, 6P = six parameterweighting, CS codonspecific weighting, How = weightedaccording to Howell's(1989) cyt-b variability categories (see text). bLRTs indicate that HKY is a significantimprovement over F81 (null hypothesis:transition rate equalstransversion rate is rejected[-2 logA - 455.4, df 1, P << .001]);Hkyl? is significantlybetter than uncorrectedHKY (null hypothesis:equal rates amongsites is rejected[-2 log A - 376.4, df = 1, P << .001]). LRT for molecularclock is not significantlydifferent (-2 log A = 16, df = 15, 0.5 < P < 0.01).Unconstrained ML = 2985.44.

A B C 23[P' cyanea 2 1001P.cyanea2 '- P. cyanea 1 1001--P. P.cyanea 21 '-P. cyanea 1 • G.caerulea 1 G. caerulea 1 G.caerulea 1 G.caerulea 2 ßcaerulea 2 P. amoena 1 ßarnoeRa 1 99•1••G.P. amoenacaerulea 22 ßarnoena 2 P. rositae 1 r•P. rositae1 '• P.rositae 2 P.rositae 1 11 ' P.rositae 2 P.rositae 2 / 30FP' leclanchrii 1 P. versicolor 2 P.versicolor 2 1• •P. leclancherii 2 P.versicolor 1 I i•P.versicolor2 P. ciris 2 P.ciris 2 I 101Lp. versicolor 1 99•1••P. versicolorciris 1 1 100rP.P.cidsleclancherii1 1 1001P. leclancherii 21 ß P. leclancherii 2

C. parellina C. parellina C. parellina

C. brissonii

L•3C. brissonfiC. cyanoides 73[C. brJssoniicyanoides • C. cyanoides FIG. 2. Resultsof analyses.(A) Topologyderived from equally weighted MP analysis;decay indices are shownabove nodes. Branch lengths shown are proportionalto geneticdistances derived from an F-corrected (or= 0.106) Tamura-Nei (1993) neighbor-joiningalgorithm. (B) bootstrapped(x500) equal weightsMP tree. (C) Bootstrapped(x 100) ML tree with best likelihoodscore (HKYF); ot= 0.106, Ts/Tv = 6.93.Lettered nodes refer to Table 5. 618 KLICKAET AL. [Auk,Vol. 118 cies of Cyanocompsayields an even worse tree isonsprovided additional evidencein support (33 steps,P = 0.0023).Thus, the mitochondrial of that relationship.Among membersof this genetree is unambiguousregarding the place- genus,only adult males of P.amoena and G. ca- ment of G. caerulea.The other strongly sup- eruleahave conspicuous, complete wingbars in ported terminal taxonpair is P.versicolor-P. cir- definitiveplumage, and only in the plumageof is, receiving no less than 97% bootstrap this pair does a cinnamon (closest to Mikado support in any analysis. Brown [no. 121C], of Smithe 1975) color occur (similar to that of Cyanocompsafemales). Black DISCUSSION streakingdorsally is alsomost conspicuousin this pair althoughit occursto a lesserdegree Passerinaparaphyly.--A surprising finding of in P.cyanea. Guiraca caerulea (nearest Smalt Blue this study was the high bootstrap support [no. 70], Smithe 1975) and P. amoena(nearest acrossall analyses(average of 96.5%) for a G. TurquoiseBlue [no. 65], Smithe1975) are rather caerulea-P.amoena sister relationship, demon- different shades of blue. Curiously, both of stratingthat G. caeruleais derived from within those shadesoccur in P. cyanea,the blue of G. the Passerinaassemblage. This result indicates caeruleaoccurring on the head and throat and that a mergingof the monotypicgenus Guiraca the blue of P. amoenaoccurring on the lower into Passerina is warranted. The conventional body (rump). "superspecies"groupings of P. cyanea-P.amo- The "painted" clade (node C, Fig. 2C) is enawith P.versicolor (Mengel 1970) or P.cyanea- united similarly by plumage characteristics. P.amoena with P.versicolor and P.ciris (Mayr and All havesome blue plumage,but it is the pres- Short 1970) are rejected by our data. Those enceof patchesof bright carotenoidpigments workerswho, on the basis of morphological (yellow, pink, scarlet) and conspicuouseye and behavioral evidence, had considered G. ca- rings that setsthem apart. It is selectionon pre- eruleato be a memberof the genusPasserina cisely these sorts of sexually and sociallyse- (despiteits muchlarger size;e.g. Phillips et al. lected traits that is presumed to lead to accel- 1964,Blake 1969,Mayr and Short 1970)are vin- erated divergence(West-Eberhard 1983, Zink dicated;although, we know of no one who hy- 1996), speciation,and ultimately taxonomicdi- pothesized a G. caerulea-P.amoena sister rela- versity. The short internode distanceswithin tionship.Our result conflictswith conclusions the "painted" clade (e.g.node D, Fig. 2A, C) is drawn from a family-level allozyme study consistent with that interpretation, that is, (Tamplin et al. 1993).Those authors concluded closelyspaced speciation events. that G. caeruleamight be moreclosely related to Ecologicalimplications.--The Passerina bun- Pheucticusthan to either Passerinaor Cyano- tings (exclusiveof G. caerulea)possess some of compsa;however, they noted that the strong the smaller body sizes in the Cardinalidae,as plumage and vocal similaritiesbetween G. ca- shownby the morphometricphenetic analyses eruleaand thePasserina buntings, seem unlikely (PCA plots) of that group (Hellack and Schnell to be merely shared primitive characters.In- 1977, Tamplin et al. 1993). With the addition of deed, their (Tamplinet al. 1993;Fig. 2) consen- G. caerulea,Passerina becomes composed of two sus of 172 most parsimonioustrees does not discrete size classes("bunting" and "gros- support a Guiraca-Pheucticusrelationship. Al- beak," seeFig. 3). Notably,P. amoena and G. ca- thoughmtDNA and allozymesurveys are often eruleaare closelyrelated sisterspecies, yet the congruentat the intragenericlevel, the appar- massof the former is approximatelyone-half ent lack of resolution in the Tamplin et al. that of the latter.A comparablesize dichotomy (1993) study may reflect a taxon sampling is apparentwithin the sistergenus Cyanocompsa problem or the limited resolvingpower of al- where C. parellina(15.9 _+1.23 g, n -- 19) and lozymes for higher (intergeneric) taxonomic C. cyanoides(35.8 _+2.77 g, n = 13) also exhibit levels. a bunting versus grosbeaksize relationship We suggestthat if G. caeruleawere the size of whereasa third Cyanocompsamember, brissonii, the otherbuntings (adult males, approximately is reportedly of an intermediatesize. Collec- 15-20 g, seebelow), it would have long been tively, these observationssuggest that body classifiedas a member of Passerina,perhaps size (mass) is not a particularly conservative even as sisterto P.amoena. Study-skin compar- trait within this assemblageof birds. April 2001] PasserinaBunting Paraphyly 619

direct result of competition with congeners P. cyanea (ecologicalcharacter displacement) or if it oc- Indigo Bunting curred in response to alternative selection pressures.Character displacementas a result of interspecificcompetition is difficult to dem- Blue Grosbeak onstrate(Schluter et al. 1985)but may be tested by correlating shifts in morphology with de- gree of sympatry(or allopatry)with competi- G.P. amoenacaerulea tors. Guiraca caerulea, with seven described sub- Lazuli Bunting species(Storer and Zimmerman 1959),varies in size acrossits range and shouldbe studied in P. rositae that context. A hypothesis of character dis- Ros•ta'sBunting placementpredicts that the largest G. caerulea should occur where the speciesis sympatric with high Passerinadensities and the smallest Vaned Bunting in regionsunoccupied (or sparselypopulated) by any Passerinaspecies. Little is known about intragenericcompeti- • P.P.versicolor ciris tive interactions for this group.Examination of Painted Bunting breeding-bird survey maps (Price et al. 1995) suggeststhat actual overlapsin breeding dis- P. leclancherii tribution may be lessextensive than that typi- Orange-breasted Bunbng cally depicted in maps of breeding distribu- tions (suchas thoseshown, Fig. 3), and where FIG. 3. Approximate breeding distributionsand overlapdoes occur, little is known aboutpar- "sizes" of all Passerinamembers shown in a phylo- titioning of habitat and food resources.In areas geneticcontext. Note that the distributionof Passer- of overlapon the westernGreat Plains, the sim- ina rositaeis restrictedto the PacificSlope of the Isth- ilarly sized but nonsisterspecies P. cyaneaand mus of Tehuantepec.Silhouette sizes reflect relative P.amoena defend interspecific territories (Emlen averagemass, obtained from museum labels of adult male specimens:P. cyanea(14.8 _+0.8 g, n = 31); P. et al. 1975).Interspecific territoriality doesnot amoena(16.8 _+2.23 g, n = 4); P. rositae(20.07 _+0.16 occur between the more broadly overlapping g, n = 61); P.leclancherii (13.70 _+0.19 g, n = 105);P. pair P. cyaneaand P. ciris (Forsythe1974), sug- versicolor(13.0 _+1.78 g, n = 104);P. ciris (16.47 _+2.11 gesting perhaps a diminution of competitive g, n = 9); and G. caerulea(32.6 + 2.87 g, n = 15). interactions over time and phylogenetic dis- tance.Also, P.cyanea does not defendterritories The phylogenyshown (Fig. 2C) indicatesthat againstthe visually similar but muchlarger G. "bunting" size is the ancestralstate for Passer- caerulea(Payne 1992). Indeed, the emphasison ina with G. caeruleasecondarily acquiring a breeding distributionsmay be overstated;com- "grosbeak" size. Similarities in the size and petitive interactionsbetween species of Passer- shape of Passerinabuntings, taken together ina might be bestunderstood by a winter study with theirrelative lack of rangeoverlap (Fig. 3), in southwestern Mexico where six of the seven suggestthe possibility that many, or all, are taxa occur in sympatry. "ecological equivalents" and by extension, Approximatetimes of divergence.--Atemporal competitorsfor resources.A compositemap of interpretation of the phylogeny requires a all bunting breeding distributions(not shown) roughly uniform substitutionrate amongtaxa. indicates that virtually all suitable bunting The likelihood-ratio test for molecular clock in- habitat throughoutNorth Americais occupied dicatesbase substitutionsare accumulatingin by one bunting speciesor another.The rela- a sufficientlyclock-like manner to allow such tively huge and overlappingrange of G. caeru- interpretations.We acknowledgethat applica- leawith respectto thoseof thesmaller buntings tion of a molecular clock remains controversial (Fig. 3) is likely a consequenceof its now dis- (e.g.Hillis et al. 1996,Klicka and Zink 1997)but similar size and food resource needs. We do not we are also aware of the heuristic value of as- know whetherthat morphologicalshift wasthe signingtentative divergence dates to phyloge- 620 KLICKAET AL. [Auk, Vol. 118 netic branching events. We use two indepen- (i.e. the last 250,000 years) that reconfigure dentlyderived clock calibrations. The commonly habitats in such a way that populationsfrag- used rate of 2% divergenceper million yearsis ment and subsequentlyspeciate while in iso- the rate obtainedby independentfossil calibra- lation. This "Late PleistoceneOrigins" model tions from an array of avian orders (Klicka and was rejected using molecular data (Klicka and Zink 1997). That this same rate is also derived Zink 1997, 1998), a conclusionsupported by for mammals (Wilson et al. 1985) and arthro- this study. Here we show that P. amoena-P.cy- pods (Brower1994) lends some credibility to its anea are not even sister taxa and that it is un- generality. Becausevariation in evolutionary likely that any membersof this genusoriginat- rates has been shown to be related to variation ed as a consequence of Late Pleistocene in body size and its correlates(e.g. Martin and glaciations. Palumbi 1993), the rate of 1.6% per million Collectively,the lack of consistentbootstrap yearscalculated by Fleischeret al. (1998)is rel- support at nodesB, D, and E (Fig. 2C) suggest evant to our study. It is based on cyt-b (only) a burst of simultaneousspeciation among four divergences of small songbirds (Hawaiian Passerinalineages, with the P. amoena-G.caeru- Honeycreepers,Drepanididae), with multiple, leaand P. versicolor-P.ciris lines splitting again well founded calibration dates based on emer- at some later time. Zink et al. (1998) note that gencetimes of three different islands.Thus, we "star phylogenies"such as this might alsobe have two evolutionary rate estimatesthat pro- arrived at via bursts of extinction. Alternative- vide a reasonablerange of potential divergence ly, additional data may change that interpre- times, a 2% rate with uncorrected p distances tation entirely by strengtheningthe nodes in and, following the method of Fleischer et al. question.We are currently investigatingthat (1998), a 1.6% rate using corrected(K2-P) dis- possibilityby sequencingadditional gene re- tances (Table 4). gions (ND6 and control region) for all Cyano- These rates suggest that Cyanocompsaand compsaand Passerinataxa. A thoroughbiogeo- Passerinabuntings diverged from a common graphic assessmentrequires the teasing apart ancestorapproximately 4.1 to 7.3 Mya. This co- of these alternatives. incideswith a Late Miocene period of acceler- Hybridizationas a measureof relationship.Be- ated cooling and drying during which forests causea well-documentedand extensivehybrid and woodlands gave way to an increasingva- zone exists where they are sympatric on the riety of grassland habitats in western North Great Plains (Sibley and Short 1959, Emlen et America (Riddle 1995 and references therein). al. 1975, Kroodsma 1975), P. amoenaand P. cy- A reductionin size and an increasedability to aneahave been treated as conspecificsby some exploit grasslandfood resourcesmay played a (e.g. Phillips et al. 1964) and considereda "su- role in a Passerinaradiation (diversification and perspecies" of very recent origin by others widespreaddistribution) at that time.This time (Mayr and Short 1970). A single hybrid speci- frame is supported by a Passerinafossil (Stead- men is known for another partly sympatric man and McKitrick 1982) from Chihuahua, bunting sisterpair, P.ciris and P.versicolor (Sto- Mexico,dated to -4 Mya. The fossilfragments rer 1961). The inference drawn from that ob- identify an extinctPasserina species that was in- servationis that P. amoenaand P. cyaneaare the termediate between P. amoena and G. caerulea more closely related of those pairs (e.g. Storer both qualitativelyand in size, and from a re- 1961). The results of this study challengethat gion where both presentlyoccur. interpretation. Passerinaamoena and P. cyanea Among the other well-resolved nodes, the are not sister species,whereas P.ciris and P.ver- calibrations indicate that G. caerulea-P. amoena sicolorhave the most recent origins of any ex- diverged approximately2.4 to 3.7 Mya and the tant membersof the genus.It is becomingin- most recent split within the group, P. ciris-P. creasinglyclear that for birds, the ability to versicolor,occurred 1.5 to 2.1 Mya The bunting hybridize and the degreeto which hybridiza- "pair" P. amoena-P.cyanea has figured promi- tion occursare not necessarilygood measures nently (seeSibley and Short 1959) in the devel- of relatedness(Prager and Wilson 1975). We opment of evolutionary models of songbird add P. amoena-P.cyanea to the growing list of evolution. Typically, those (e.g. Mengel 1970, nonsistertaxa that not only retain the ancestral Hubbard 1973) invoke recent glacial advances capacityto hybridize, but also to form an ap- April 2001] PasserinaBunting Paraphyly 621 parent hybrid swarm (e.g. Colaptes,Moore et al. ELLEGREN, H. 1992. Polymerase-chain-reaction 1991; Icterus, Freeman and Zink 1995). (PCR) analysis of microsatellites--A new ap- proach to studies of genetic relationshipsin birds. Auk 109:886-895. ACKNOWLEDGMENTS EMLEN, S. T., J. D. RISING, AND W. THOMPSON. 1975. This paperis dedicatedto the life and memoryof A behavioraland morphologicalstudy of sym- Allan Galante who loved Mexico, its land, people, patry in the Indigo and Lazuli buntings of the culture,music, and natural history,including its bird Great Plains. Wilson Bulletin 87:145-179. life. C.W.T.is gratefulfor financialand logisticalsup- FELSENSTEIN,J. 1981. Evolutionary trees from DNA port from (1) the Eddy Endowmentfor Excellencein sequences:A maximum likelihood approach. Ornithology,(2) C. Wood, C. Spaw,and the Rohwer Journal of Molecular Evolution 17:368-376. Lab Group at the Burke Museum, University of FELSENSTEIN,J. 1985. Confidence limits on phyloge- Washington,(3) two ChicagoZoological Society So- nies: An approachutilizing the bootstrap.Evo- ciety Conservationand ResearchFund awards, (4) a lution 39:783-791. Center for the Study of Tropical Birds award, (5) FLEISCHER,R. C., C. E. MCINTOSH, AND C. L. TARR. Adolfo Navarro and Octavio Rojas at the Museo de 1998. Evolution on a volcanicconveyor belt: Us- Zoologia at the UniversidadNacional Autonomade ing phylogeographicreconstructions and K-Ar Mexico,(6) Eduardo Moralesand EduardoEspinosa based agesof the Hawaiian Islands to estimate of the Instituto de Historia Natural, Gobierno de molecularevolutionary rates. Molecular Ecology Chiapas,Mexico, and (7) the WesternFoundation of 7:533-545. VertebrateZoology. We thank the curatorsand staff FORSYTHE,D. M. 1974. Song characteristicsof sym- at the Louisiana State University Museum of Zoolo- patricand allopatricIndigo and Paintedbunting gy and the University of WashingtonBurke Museum populationsin the southeasternUnited States. for acquiringand loaningtissue specimens. A. Phil- Ph.D. dissertation, Clemson University, Clem- lips provided valuable field assistanceas did T. son, South Carolina. Brashear.J.K. acknowledgesfunding support from FREEMAN,S., AND R. M. ZINK. 1995. A phylogenetic the Dayton and Wilkie Fundsof the J.E BellMuseum study of the blackbirdsbased on variationin mi- of Natural History. Funding for laboratorywork was tochondrialrestriction sites. Systematic Biology provided by R.M.Z.'s NSF grant DEB-9317945. 44:409-420. GOLDMAN, N. 1993. Statistical tests of models of LITERATURE CITED DNA substitution. Journal of Molecular Evolu- tion. 36:182-198. AMERICAN ORNITHOLOGISTS' UNION. 1998. Check- GRIFFITHS, C. S. 1997. Correlation of functional do- list of North American birds, 7th ed. American mains and rates of nucleotidesubstitution in cy- Ornithologists' Union, Washington,D.C. tochromeb. Molecular Phylogeneticsand Evo- BLAKE,C. H. 1969. Notes on the Indigo Bunting. lution 7:352-365. Bird-Banding40:133-139. HACKETT,S. J. 1996. Molecular phylogeneticsand BREMER,K. 1994. Branchsupport and tree stability. biogeographyof tanagersin the genusRampho- Cladistics 10:295-304. celus(Aves). Molecular Phylogeneticsand Evo- BROOKS,D. R., AND D. A. MCLENNAN.1991. Phylog- lution 5:368-382. eny,Ecology, and Behavior:A ResearchProgram HARSHMAN,J. 1996. Phylogeny,evolutionary rates, in ComparativeBiology. University of Chicago and ducks.Ph.D. dissertation,University of Chi- Press,Chicago. cago, Chicago. BROWER,A. V. Z. 1994. Rapid morphologicalradia- HASEGAWA, M., M. KISHINO, AND T. YANO. 1985. tion and convergenceamong racesof the butter- Dating of the human-ape splitting by a molec- fly Heliconiuserato inferred from patternsof mi- ular clock of mitochondrial DNA. Journal of Mo- tochondrialDNA evolution.Proceedings of the lecular Evolution 22:160-174. National Academy of SciencesUSA 91:6491- HELLACK,J. J., AND G. SCHNELL.1977. Phenetic anal- 6495. ysis of the subfamily Cardinalinae using exter- CUNNINGHAM,C. 1997. Is congruencebetween data nal and skeletal characters. Wilson Bulletin 89: partitions a reliable predictor of phylogenetic 130-148. accuracy?Empirically testing an iterative pro- HILLIS, D. M., B. K. MABLE, A. LARSON, S. K. DAVIS, cedure for choosingamong phylogeneticmeth- AND E. A. ZIMMER. 1996. Nucleic Acids IV: Se- ods. SystematicBiology 46:464-478. quencingand cloning.Pages 321-381 in Molec- DESJARDINS,P., AND R. MORAIS.1990. Sequenceand ular Systematics(D. M. Hillis, C. Moritz, and B. gene organizationof the chickenmitochondrial K. Mabel, Eds.). Sinauer Assocociates,Sunder- genome: A novel gene order in higher verte- land, Massachusetts. brates. Journal of Molecular Biology 212:599- HOWELL,N. 1989. Evolutionaryconservation of pro- 634. tein regionsin the protonmotivecytochrome b 622 KLICKAET AL. [Auk, Vol. 118

and their possibleroles in redox catalysis.Jour- evolution, 3.04. Sinauer Associates, Sunderland, nal of Molecular Evolution 29:157-169. Massachusetts. HUBBARD,J.P. 1973. Avian evolution in the aridlands MARTIN,A. P., AND S. R. PALUMBI.1993. Body size, of North America. Living Bird 12:155-196. metabolic rate, generationtime, and the molec- HUELSENBECK,J.P. AND B. RANNALA.1997. Phylo- ular clock.Proceedings of theNational Academy geneticmethods come of age:Testing hypothe- of Sciences USA 90:4087-4091. sesin an evolutionarycontext. Science276:227- MAYR, E., AND L. L. SHORT.1970. Speciestaxa of 232. North American birds, A contribution to com- JOHNSON,K. P. AND M.D. SORENSON.1998. Compar- parative systematics.Publications of the Nuttall ing molecular evolution in two mitochondrial OrnithologicalClub, no. 9. proteincoding genes (cytochrome b and ND2) in MENGEL, R. M. 1970. The North American central the dabbling ducks (Tribe: Anatini). Molecular plains as an isolating agent in bird speciation. Phylogeneticsand Evolution10:82-94. Pages279-340 in Pleistoceneand recentenviron- KIMURA,M. 1980. A simple method for estimating ments of the central Great Plains (W. Dort, Jr., evolutionaryrates of base substitutionthrough and J. Knox Jones,Jr., Eds.). University Press of comparative studies of nucleotide sequences. Kansas, Lawrence. Journalof MolecularEcology 16:11-120. MOORE, W. S. 1995. Inferring phylogenies from KISHINO, H., AND M. HASEGAWA. 1989. Evaluation of mtDNA variation: Mitochondrial-genetrees ver- the maximum likelihood estimate of the evolu- sus nuclear-genetrees. Evolution 49:718-726. tionary tree topologies from DNA sequence MOORE, W. S., AND V. R. DE FILIPPIS. 1997. The win- data, and the branching order in Hominoidea. dow of taxonomicresolution for phylogenies Journal of Molecular Evolution 29:170-179. basedon mitochondrialcytochrome b. Pages 83- KLICKA, J., J. c. BARLOW,W. B. MCGILLIVRAY, T. J. 119 in Avian Molecular Evolution and System- DOYLE,AND R. M. ZINK. 1999. Evidence sup- atics (D. P. Mindell, Ed.). Academic Press, New York. porting the recent origin and speciesstatus of the TimberlineSparrow. Condor 101:577-588. MOORE, W. S., J. H. GRAHAM, AND J. T. PRICE. 1991. Mitochondrial DNA variation in the Northern KLICKA,J., ANDR. M. ZINK. 1997.The importanceof recentice agesin speciation:A failed paradigm. Flicker (Colaptesauratus). Molecular Biology and Evolution 8:327-344. Science 277:1666-1669. NUNN, G. B., J. COOPER,P. JOUVENTIN,C. J. R. ROB- KLICKA,J., AND R. M. ZINK. 1998.Pleistocene speci- ERTSON, AND G. G. ROBINSON. 1996. Evolution- ation and the mitochondrial DNA clock: A re- ary relationshipsamong extant albatrosses (Pro- sponseto Arbogastand Slowinski.Science 282: 1955. cellariformes: Diomedeidae) established from complete cytochrome-bgene sequences.Auk KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. ED- 113:784-801. WARDS,S. PAABO, E X. VILLABLANCA, AND A. C. PAYNE,R. B. 1992.Indigo Bunting (Passerinacyanea). WILSON.1989. Dynamics of mitochondrialDNA In The Birds of North America, no. 4 (A. Poole, evolution in animals: Amplification and se- P. Stettenheim,and E Gill, Eds.). Academyof quencingwith conservedprimers. Proceedings Natural Sciences,Philadelphia, and American of the National Academy of SciencesUSA 86: Ornithologists'Union, Washington,D.C. 6196-6200. PAYNTER,R. A. 1970.Subfamily Cardinalinae. Pages KORNEGAY,J. R., T. D. KOCHER,L. A. WILLIAMS, AND 216-245 In Check-list of Birds of the World, vol. A. C. WILSON.1993. Pathways of lysozymeevo- 13. (R. A. Paynter and R. W. Storer, Eds.). Mu- lution inferred from the sequencesof cyto- seum of ComparativeZoology, Heffernan Press, chrome b in birds. Journal of Molecular Evolu- Worcester, Massachusetts. tion 37:367-379. PHILLIPS,A. R., J. T. MARSHALL,Jr., AND G. MONSON. KRAJEWSKI,C., AND D. G. KING. 1996. Molecular di- 1964. The Birds of Arizona. Universityof Ari- vergenceand phylogeny:Rates and patternsof zona Press, Tucson. cytochrome-bevolution in cranes.Molecular Bi- PRAGER, E. M., AND A. C. WILSON. 1975. Slow evo- ology and Evolution13:21-30. lutionary loss of the potential for interspecific KROODSMA,R. L. 1975. Hybridization in buntings hybridization in birds: A manifestationof slow (Passerina)in North Dakota and eastern Mon- regulatoryevolution. Proceedings of the Nation- tana. Auk 92:66-80. al Academy of SciencesUSA 72:200-204. LANYON,S. M. 1993. Phylogeneticframeworks: To- PRICE, J., S. DROEGE,AND A. PRICE. 1995. The Sum- wards a firmer foundationfor the comparative mer Atlas of North American Birds. Academic approach. BiologicalJournal of the Linnean So- Press,San Diego, California. ciety 49:45-61. QUINN,T. W., ANDA. C. WILSON.1993. Sequence evo- MADDISON, W. P., AND D. R. MADDISON. 1992. lution in and around the mitochondrial control MacClade:Analysis of phylogenyand character region.Journal of MolecularEvolution 37:417-425. April 2001] PasserinaBunting Paraphyly 623

RIDDLE,B. R. 1995. Molecular biogeographyin the and chimpanzees.Molecular Biology and Evo- pocket mice (Perognathusand Chaetodipus)and lution 10:512-526. grasshoppermice (¸nychomys):The late ceno- TAYLOR,W. K. 1974.A new hybrid bunting (Passerina zoic developmentof a North American arid- cyaneax Passerinaciris). Auk 91:485-487. lands rodent guild. Journalof Mammalogy 76: THOMPSON,C. W. AND M. LEU. 1994. Determining 283-301. homology of molts and plumages to address RIDGWAY, R. 1901. The birds of North and Middle evolutionary questions:A rejoinder regarding America. Bulletin of the United States National Emberizid finches. Condor 96:769-782. Museum, no. 50, Part 1. THOMPSON, C. W., AND M. LEU. 1995. Molts and SCHLUTER, D., T. D. PRICE, AND P. R. GRANT. 1985. plumagesof Orange-breastedBuntings (Passer- Ecologicalcharacter displacement in Darwin's ini leclancherii):Implications for theoriesof de- finches. Science 227:1056-1059. layed plumagematuration. Auk 112:1-19. SIBLEY,C. G., AND B. L. MONROE. 1990. Distribution THOMPSON,W. L. 1968. The songsof five speciesof and Taxonomyof Birds of the World. Yale Uni- Passerina. Behaviour 31:261-287. versity Press.New Haven, Connecticut. VOELKER,G., AND S. V. EDWARDS.1998. Can weight- SIBLEY,C. G., AND L. SHORT,Jr. 1959. Hybridization ing improve bushytrees? Models of cytochrome- in the buntings (Passerina)of the Great Plains. b evolution and the molecular systematics of Auk 76:443-463. pipits and wagtails (Aves: Motacillidae). Sys- SMITH,A. B. 1994.Rooting moleculartrees: Problems tematic Biology47:589-603. and strategies.Biological Journal of the Linnean WEST-EBERHARD,M. J. 1983. Sexual selection, social Society51:279-292. competition, and speciation.Quarterly Review SMITHE, E B. 1975. Naturalist's Color Guide. The of Biology58:155-183. AmericanMuseum of Natural History,New York. WHEELER,W. C. 1990. Nucleic acid sequencephylog- SORENSON,M.D. 1996. "Treerot." University of eny and random outgroups. Cladistics 6:363- Michigan, Ann Arbor. 368. STEADMAN,D. W., AND M. C. MCKITRICK. 1982. A Pli- WILLIAMS,P. L., AND W. M. FITCH.1989. Finding the ocenebunting from Chihuahua, Mexico.Condor minimal change in a given tree. Pages453-470 84:240-241. in The Heirarchyof Life (B. Fernholm,K. Bremer, STORER,R. W. 1961. A hybrid betweenthe Painted and H. Jornvall, Eds.). Elsevier Press,New York. and Varied buntings.Wilson Bulletin 73:209. WILSON, A. C., R. L. CAAN, S. M. CARR, M. GEORGE, STORER, R. W., AND D. A. ZIMMERMAN. 1959. Varia- Jr., U. B. GYLLENSTEN,K. M. HELM-BYCHOWSKI, tion in the Blue Grosbeak (Guiracacaerulea) with R. G. HIGUCHI, S. R. PALUMBI, E. M. PRAGER,R. special reference to the Mexican populations. D. SAGE, AND M. STONEKING. 1985. Mitochon- OccasionalPapers of the Museum of Zoology drial DNA and two perspectiveson evolutionary University of Michigan 609:1-13. genetics.Biological Journal of the Linnean So- STURMBAUER, C., AND A. MEYER. 1992. Genetic di- ciety 26:375-400. vergence,speciation and morphologicalstasis in YANG,Z. 1996.Among-site rate variation and its im- a lineage of African cichlid fishes.Nature 358: pact on phylogeneticanalyses. Trends in Ecol- 578-581. ogy and Evolution 11:367-372. SULLIVAN, J., K. E. HOLSINGER,AND C. SIMON. 1995. YODER, A.D., R. VILGALYS, AND M. RUVOLO. 1996. Among-site rate variation and phylogenetic Molecular evolutionarydynamics of cytochrome analysis of 125 rRNA in sigmodontinerodents. b in Strepsirrhineprimates: The phylogenetic Molecular Biologyand Evolution 12:988-1001. significanceof third position transversions.Mo- SWOFFORD,D. L. 1999. PAUP:Phylogenetic analysis lecular Biology and Evolution 13:1339-1350. using parsimony (* and other methods), ver- ZHANG, D.-X., AND G. M., HEWITT. 1996. Nuclear in- sion 4.062. Sinauer Associates, Sunderland, tegrations:Challenges for mitochondrial DNA Massachusetts. markers. Trends in Ecology and Evolution 11: SWOFFORD,D. L., G. J. OLSEN, P. J. WADELL, AND D. 247-251. M. HILLIS. 1996. Phylogeneticinference. Pages ZINK,R. M. 1996.Species concepts, speciation, and sex- 407-514 in Molecular Systematics(D. M. Hillis, ual selection.Journal of Avian Biology27:1-6. C. Moritz, and B. K. Mabel, Eds.). Sinauer As- ZINK, R. M., AND M. C. MCKITRICK. 1995. The debate socociates, Sunderland, Massachusetts. over speciesconcepts and its implicationsfor or- TAMPLIN, J. W., J. w. DEMASTES,AND J. V. REMSEN,Jr. nithology.Auk 112:701-719. 1993. Biochemical and morphometric relation- ZINK, R. M., S. J. WELLER, AND R. C. BLACKWELL. shipsamong some members of the Cardinalinae. 1998. Molecular phylogeneticsof the avian ge- Wilson Bulletin 105:93-113. nusPipilo and a biogeographicargument for tax- TAMURA, K., AND M. NEI. 1993. Estimation of the onomic uncertainty. Molecular Phylogenetics and Evolution 10:191-201. number of nucleotide substitutions in the con- trol region of mitochondrial DNA in humans Associate Editor: R. Prum