July2000] ShortCommunications 787 The Auk 117(3):787-791, 2000 Molecular Evidencefor the SystematicPosition of Urocynchramuspylzowi JEFFG. Gr•oTIq• Departmentof Ornithology,American Museum of NaturalHistory New York,New York10024, USA Urocynchramuspylzowi, known variously as Prze- Groth1998). Paynter (in Peters1968) doubted the in- walski's Rosefinch,Pink-tailed Rosefinch, and Pink- clusionof Urocynchramuswith carduelinesbecause tailed Bunting,is a finch-likebird of uncertainsys- of its ten-primariedcondition and becausethe red tematic affinities that is endemic to the mountains of colorationin its tail setit apartfrom rosefinches.The westernChina. In many linear classifications(e.g. ten-primariedstate prompted at least one author Sharpe1909, Vaurie 1959, Peters 1968, Morony et al. (Domaniewski1918) to arguefor classifyingthe bird 1975),it is listed next to the Long-tailedRosefinch in its own family,the Urocynchramidae. (Uragussibiricus) in the Carduelinae,to which it is PlacingUrocynchramus among other oscinefami- similar in size and severalplumage characters. Uro- lies hasbeen frustrated by the absenceof both skel- cynchramushas a muchthinner bill than Uragus,but etal and fluid-preservedspecimens of this bird. In- both specieshave disproportionately long tails com- formation on basic life-history attributes such as paredwith the restof thecardueline finches, and the mating behavior,nest structure,and vocalizations malesare bright pink on thethroat, breast, and belly. generallyis lacking,adding to thedifficulties of pro- Femalesof both speciesare streaked,sparrow-like, ducing convincingarguments for its taxonomicas- and lackthe brightpink coloration.Uragus and Uro- signment.Nevertheless, it is now possibleto extract cynchramusalso are exclusivelyAsiatic and live at DNA from dried museum skins that are decades old; high elevationsnear the centerof speciesdiversity of thesetechniques allow acquisition of numerouschar- rosefinchesin the genusCarpodacus, to which a close actersthat are useful in analyzingsystematic rela- relationshipwith Uragushas never been questioned. tionshipsof problematictaxa when freshbiological The few availableaccounts of its ecologyand behav- materialis not available.A prerequisitefor suchan ior (Przewalski 1876 [recounted in Vaurie 1956], analysisis comparableDNA sequenceinformation Sch•fer1938, B. Kingpers. comm.) state that the hab- from an appropriateset of othertaxa. itat associations,patterns of flight, and posturesof Phylogeneticanalyses of completemitochondrial Urocynchramusare similar to thoseof Uragus. cytochrome-b(cyt b) genesshow a well-supported Despiteits resemblanceto Uragus,Ur'ocynchramus separation between cardueline finches and emberi- frequentlyhas beenclassified as an emberizidbun- zid sparrows(Groth 1998). If Urocynchramusis either ting (Sharpe1888, Dresser 1902, Sibley and Monroe a carduelineor an emberizid,then its cyt-bsequence 1990).Przewalski (1876), who first collectedand de- shouldallow discriminationbetween these two pos- scribedthe bird, notedthe similarityof its bill struc- sibilities.My analysis(Groth 1998) also included a ture to that of buntingsand regarded its songas sim- wide rangeof otheroscine groups, thus providing a ilar to that of the buntingCynchramus (=Emberiza) broadphylogenetic context for placementof Urocyn- schoeniclus.Sushkin (1927) consideredthe external chramus if it is neither a car.dueline nor an emberizid. anatomyand structureof the hornypalette of Uro- Methods.--I extracted DNA from a 1-mm 2 flake of cynchramusto be most like that of buntings.Zusi toe skin from an adult male (American Museum of (1978) found that featuresof its interorbitalseptum Natural History 782951;collected by Ernst Shfiferat were unlike those of cardueline finches, but he could Kham, westernChina, 18 October1934) usingthe neitherconfirm nor rejectits possiblerelationship to methodsof Mundyet al. (1997).The entire cyt-b gene buntings. was amplifiedin five fragmentsusing a combination The most perplexingaspect of Urocynchramusis of 10primers (Table 1) designedfor oscinebirds. Five the condition of its outer primary (P10), which is of the primershad beenused in otherstudies, but I abouttwo-thirds the lengthof P9. Althougha well- obtainedfresh stocksfrom the manufacturer(Op- developedP10 suchas this is characteristicof many eronTechnologies, Inc.) asa precautionagainst con- oscinefamilies, all fringillids and emberizidsare tamination.Further precautionsincluded perform- "nine-primaried"in that P10 is vestigial.Although ing DNA extractionand assemblyof reactiontubes the nine-primariedcondition has been derived often in an isolatedroom away from the main laboratory in oscines,no taxonwithin a nine-primariedgroup using dedicated"ancient DNA" pipettorsand aero- is knownto haveundergone a characterreversal to sol-barrierpipet tips.I usedAmplitaq Gold kits (Ap- secondarilyderive the "ten-primaried"state (see plied Biosystems)according to manufacturers'in- structionsfor amplificationreactions, but theywere scaled down to total volumes of 10 •tL. All kit com- E-mail:[email protected] ponents (except enzyme), water, and disposable 788 ShortCommunications [Auk, Vol. 117 TABLE1. List of primersused for Urocynchramuscy- replicates,with eachreplicate containing five sepa- tochrome-bsequencing. rate heuristic searches with random-taxon addition. I depicteddistance relationships with a neighbor- Primer• Sequenceb joining tree (Saitou and Nei 1987) generatedusing L14851 c CCTACTTAGGATCATTCGCCCT modified genetic distances(third-position transi- L15068 CTAGCCATACACTACACAGCAGA tionseliminated). Trees were rootedusing data from L15410 c TGAGGAGGATTCTCAGTAGACAA four membersof the Corvida(sensu Sibley and Mon- L15656 c CCAAACCTACTACTAGGAGACCCAGA roe 1990). L15848 CCAAACTACGATCAATAACCTTCCG Resultsand discussion.--Severalobservations sug- H15103 TGCCGAGACGTACAATTCGGCTGA H15460 ATCGTAGGCCTCACACTAGTCCAC gestthat the resultingsequence was authenticUro- H15710 • GGCATAGGCGAATAGAAAGTATC cynchramuscyt b and not a contaminant.In experi- H15934 GGCTAGTTGGCCGATGATGATGAA mentsusing DNA from museumskins, all negative H16141 c AACATACCAGCTTTGGGAG control reactions were blank when visualized on EtBr-stainedagarose gels, whereas reactionswith aL andH referto extensionproducts on the light andheavy strands, respectively.Numbers indicate the positionof the 3' baseof theprim- UrocynchramusDNA were positive. Nevertheless, er accordingto the systemused for Gallus(Desjardins and Morais even thoughthe negativecontrol reactions showed 1990). no amplificationproducts, it is possiblethat the ex- b Primers are listed in the 5' to 3' direction. tract from Urocynchramuswas contaminatedwith cPreviously cited by Groth (1998). DNA from another species.If this were the case, "double" sequences,containing ambiguous base plasticswere irradiated with UV light before use. callsfrom a mixture of both Urocynchramusand the Thermocyclingwas donein a Perkin-ElmerAB19600 contaminant,might have resulted for some frag- machine with an initial hold at 94øC for 10 min fol- ments. However, double sequencesdid not occur. lowed by 40 cyclesconsisting of 95øCfor 20 s, 52øC Furthersupport for the authenticityof the sequence for 20 s, and 72øC for 30 s. To test for contamination, was that the five separatetarget DNA fragments, I also assemblednegative control reactionscontain- ranging in size from 277 to 391 bases (excluding ing all componentsexcept Urocynchramus DNA. Re- primer sites),showed no sitesin disagreementwith- actionproducts were then processedfurther and se- in areasof overlap,suggesting that only onesource quencedon an AB1377automated sequencer accord- was responsiblefor all amplification.It would have ing to methodsin Groth (1998). beenhighly unlikelyfor the sameputative contami- I assembledand edited the sequencesusing Se- nant to havebeen amplified cleanly for all five frag- quenchersoftware (Gene Codes).After scoringall mentsto the exclusionof any amplificationfrom the bases,which included the entirecyt-b gene, I added UrocynchramusDNA, which certainlypredominated it to the cyt-b data of 53 other oscines(1,143 base in the extract.Similarly, I doubt that the sequence pairs each;see Groth 1998). This data set included was that of a nuclearpseudogene because it is un- four putatively divergent lineages (genera) of car- likely that all five pairsof primerspreferentially am- dueline finches,one Hawaiian honeycreeper(Hima- plifieda nuclearproduct to theexclusion of themuch tione),five emberizidgenera of which one(Melophus) more abundantmitochondrial cyt b in the DNA ex- is Asiatic,and nine other membersof the "larger" tract.When comparedwith all othercyt-b sequences Emberizidae(as formerly recognized by AOU [1983], I had previouslygenerated, the final presumptive which I refer to as the superfamilyEmberizoidea). Urocynchramussequence was uniqueand highly di- One additionalentire cyt-bsequence, that from fro- vergentfrom all others. zen liver tissue of Uragussibiricus (Russia, Republic Parsimonyanalysis found five equally shorttrees of Buryatia, 55 km W and 30 km S of Ulan-Ude, 16 (length = 2,290; not shown),all of which showed June1993; University of WashingtonBurke Museum monophylyfor both the Fringillidaeand the "em- CDS 4888), was generated using the laboratory berizoids." Not one of the trees linked Urocynchra- methodsof Groth (1998) and added to the data set muswithin or asthe sistergroup of thesetwo clades. for a total of 55 taxa compared.I analyzedthe data Theseresults clearly show that no phylogeneticre- usingboth parsimony-and distance-basedmethods