The Auk 111(4):970-978, 1994

MITOCHONDRIAL ANALYSIS OF GENE FLOW BETWEEN NEW ZEALAND ( PLATYRHYNCHOS) AND GREY (A. SUPERCILIOSA)

JUDITH M. RHYMER,1'3 MURRAY J. WILLIAMS,2 AND MICHAEL J. BRAUN1 XLaboratoryof MolecularSystematics, Smithsonian Institution, MRC 534, Washington,D.C. 20560, USA; and 2Directorateof Scienceand Research, Department of Conservation, P.O. Box 10-420, Wellington,New Zealand

ABSTP,•CT.--Oneof the more well-known examplesof hybridization in is the fre- quently documentedoccurrence between sexually dimorphic Mallards (Anasplatyrhynchos) and severalclosely related nondimorphic speciesin the complex.In New Zealand, the Grey (Anas superciliosasuperciliosa) is the indigenous, nondimorphic Mallardlike ,and extensive hybridization with introduced Mallards has been implicated in the populationdecline of Grey Ducks.Individuals from throughoutthe countrywere classified phenotypically as parentals or hybrids based on variation in , bill color, and leg color.We confirmedspecies-specific mitochondrial DNA haplotypesby comparingrestriction- enzymefragment patterns in Grey Ducksand New ZealandMallards to thoseof PacificBlack Ducks (A. superciliosarogersi) from and Mallards from North America, respectively. Our data indicatethat hybridizationhas led not only to introgressionof Grey Duck mtDNA into Mallard populations(the predicteddirection of gene flow), but also to significantin- trogressionof Mallard mtDNA into Grey Duck populations.Thus, the contentionthat hy- bridization between Mallards and nondimorphic speciesinvolves primarily Mallard males with femalesof the other speciesis not upheld for this examplefrom New Zealand. The speciationprocess appears to be undergoingreversal. Received I April 1993,accepted 2 July 1993.

THE INCIDENCEOF interspecific and interge- Mallard/Mottled Duck (A. fulvigula)hybrids also neric hybridization in the order are being reportedin someareas of Florida (Ma- is higher than in any other orderof birds,reach- zourek and Gray 1994). In fact, the AOU (1983) ing 30 to 40% by some estimates(Grant and declared the Mexican Duck to be conspecific Grant 1992). In addition, a substantial propor- with the Mallard becauseof extensivehybrid- tion of interspecifichybrids (20%)in this order ization between them. have been reported to be fertile (Scherer and In New Zealand, the Grey Duck (A. superci- Hilsberg 1982),so there is potential for exten- liosa superciliosa)is the indigenous, nondi- sive gene flow and introgressionbetween some morphic, Mallardlike species.Grey Ducks are species. thoughtto have colonizedfrom Australia,like Amongthe morewell-known examples is the much of New Zealand's avifauna (Baker 1991), frequentlydocumented incidence of hybridiza- and the PacificBlack Duck (A. superciliosarogersi) tion betweensexually dimorphic Mallards (Anas in Australiais virtually identicalphenotypical- platyrhynchos)and several closelyrelated, non- ly to the New ZealandGrey Duck (Frith 1982). dimorphic species. For instance, in North Mallards were introduced by the Otago Accli- America,hybridization with Mallards hasbeen matization Societyinto the southern region of implicated as one factor in the population de- South Island, New Zealand in the mid-1800s cline of American Black Ducks (A. rubripes; from Europeangame-farm stock and into North Johnsgard1967, Heusmann 1974, Ankney et al. Island by the Auckland AcclimatizationSociety 1987), Hawaiian Ducks (A. wyvilliana;Griffin et in the 1930s from North America (Williams al. 1989), and Mexican Ducks (A. platyrhynchos 1981). diazi; Hubbard 1977). Increasing numbers of Over the last few decades, Mallard popula- tions have increaseddramatically, while Grey 3 Presentaddress: Department of BiologicalSci- Duck populationshave declined. Extensivehy- ences,Clemson University, Clemson,South Carolina bridization with Mallards, facilitatedby the loss 29634, USA. of natural habitatsto agriculture,has been im- 970 October1994] Mallardand Grey Duck Gene Flow 971

plicatedin the populationdecline of Grey Ducks this study,we soughtspecies-specific markers in a situationanalogous to thosein North Amer- for Mallards and Grey Ducksusing restriction- ica (Williams 1981,Gillespie 1985).By the early enzymeanalysis of mtDNA, a techniquethat is 1980s, levels of hybridization estimated from generallymore sensitivethan allozymeelectro- plumage variation had increasedto over 50%in phoresis.Our study focusedon the analysisof populations on South Island near Dunedin, pure Grey Ducksand Mallards togetherwith while the proportion of pure Grey Ducks had several individuals that had been classified as declined to lessthan 5% (Gillespie 1985). This hybrids on the basisof their morphological hasled to concernthat the Grey Duckmay even- characteristics. tually disappearas a separatespecies from New We also used the results to infer the direc- Zealand (Weller 1980,Gillespie 1985),as it has tionality of hybridization.If hybrid matingsoc- from the Marianas Islands. The so-called Mar- cur primarily between colorful Mallard males iana Mallard (A. oustaletiSalvadori) is generally and Grey Duck females,then the majority of consideredto be a hybrid betweenstray Mal- hybrid individuals should have a Grey Duck lard Ducksand A. superciliosa(Delacour and Mayr mitochondrial haplotype, since avian mtDNA 1945, Yamashina 1948; but see Reichel and is maternally inherited (Giles et al. 1980, Wa- Lemke 1994). tanabe et al. 1985). Although assortativemating appearsto pre- vail between Mallards and American Black METHODS Ducks in North America (E. Morton unpubl. data),and betweenMallards and A. superciliosa Samplecollection.--We sampled Mallard, Grey Duck, in Australia (Braithwaite and Miller 1975) and and hybridindividuals from North andSouth Islands New Zealand (Hitchmough et al. 1990), there of New Zealand. We also analyzed Mallards from is some experimental evidence that more col- North America(as part of a largerstudy of the mallard orful Mallard males outcompetedull-colored complexof waterfowl) and PacificBlack Ducks from males for mates(Brodsky et al. 1988). Ankney Australiato ensurethat we couldunequivocally iden- et al. (1987)also suggested that the majorityof tify the species-specificmtDNA haplotypesof the pa- hybrid matingsin North America may result rental species. Birdsshot by hunterswere collectedin May 1991 from forcedcopulations between Mallard males from localitiesthroughout the country, as part of a and BlackDuck femalesduring renesting. largerstudy by M.J.W. to assessthe extentof hybrid- Several studies have tried to estimate the ex- ization in New Zealand. Samplesfrom six areas on tent of hybridization and introgressionbe- North Island(Lake Whangape, Cambridge, Ohakune, tween Mallards and one or another of the non- Wairoa, Wanganui, Lake Wairarapa)and one area on dimorphicspecies by qualitativeassessments of South Island (Lake Ellesmere) were selected for anal- phenotypic variation (Braithwaite and Miller ysisin thisinitial study(Fig. 1).Samples were selected 1975, Gillespie 1985, Heusmann 1974; Kirby, on the basisof phenotypiccharacters to include 12 U.S. Fishand Wildlife Serviceunpubl. report), Grey Ducks,9 Mallards,8 Grey Duck-like hybrids, but it is difficult to distinguishhybrids mor- and 14 Mallardlike hybrids.It hasbeen observedthat hybridstend to be Grey Duck-like or Mallardlike in phologicallyafter more than one generationof appearance(Yamashina 1948, Weller 1980). backcrossingto a parentalspecies (Williams and Heart tissue from these birds was stored in 70% Roderick1973, Rhymer unpubl. data). In ad- ethanol and shipped at room temperature. We also dition to the use of phenotypiccharacters to analyzedheart and liver tissuefrom additionalbirds estimatelevels of gene flow, somebiochemical that had been collected between 1986 and 1991 and techniqueshave been applied to the Mallard storedat -80øC. Thesesamples included seven Grey group. However, no species-specificmarkers Ducks(Pohangina River; Apiti, Orua River), six Mal- have been found to distinguishMallards from lards (Apiti), and two Grey Duck-like hybrids (Ron- American BlackDucks using allozyme electro- gotea;Apiti), all from North Island.We alsoanalyzed phoresis(Ankney et al. 1986) or restriction-en- tissueand blood samplesof North AmericanMallards and blood samplesof 10 Pacific Black Ducks. The zyme analysisof mitochondrialDNA (mtDNA; Pacific Black Ducks were chosen from an area near Avise et al. 1990). Canberra in eastern Australia (where no Mallards are Allozyme studiesalso have been unsuccess- found) to comparewith our pure Grey Duck samples. ful in separatingMallards from A. superciliosa Phenotypicscoring.--Variation in some phenotypic both in Australia and New Zealand (Braithwaite charactersof each was quantified, using criteria and Miller 1975, Hitchmough et al. 1990). In similar to those outlined by Gillespie (1985) and 972 R•M•, WILLIAMS,.•_ND BP,•UN [Auk, Vol. 111

TASLE1. Phenotypiccriteria used to scorehead, wing and leg characteristicsof Grey Ducksand Mallards. NORTH ISLAND Face (0) Two black stripes,the upper superciliarystripe of uniform width and extending well beyond eye; lower stripe tapering from gape to below eye and Cambridge giving way to mottled cheek;clean creamstripe be- tween eye and crown, throat and face cream. Dhakune (1) Two blackstripes, upper stripe as described above, Wanganui lower stripe merging with mottled face midway be- tween bill and eye, clean cream stripe between eye Rongotea -- Apiti and crown; face and cheek mottled cream but with Pohangina R. obviouscream patch at baseof bill between the two black stripes. (2) Superciliarystripe obviousand extending from e Wairarapa SOUTH bill to well beyondeye, lower stripeabsent but small ISLAND patch of black feathersat bill baseat gape; mottled creamstripe between eye and crown, faceand cheek entirely mottled black on cream background,throat cream. Lake Eriesmere (3) Entire face mottled black on fawn background, superciliarystripe indistinct and narrow,small patch at gape more heavily mottled to appear darker than 0 300 rest of face, small fawn patch at bill base between km gapeand bill top,throat clean to lightly mottledfawn. (4) Dark mottled face and throat, black on fawn backgroundin female, blackish-greenon dark fawn in male; no obviouseye stripe, no fawn patch at bill Fig. 1. Collection sites on North and South Is- base. lands, New Zealand. (5) Predominantlydark green head,face, and throat (males only). Legs Braithwaite and Miller (1975). These included plum- (l) Dark olive greenish-brown;(2) khaki; (3) yel- agevariation on the headand wing, and differences low-orange to dull orange;(4) bright orange;(5) red in legand bill color(details in Table1). Eachcharacter orange. wasassigned a scoreranging from 0-1 up to 4-6, with Bill Grey Duck charactersscored at the low end of the (0) Uniformly black;(1) blackwith very dark green scale(lowest overall cumulative scorepossible was 1) particularly at baseand along edge of upper mandi- and Mallard charactersat the upper end (highest pos- ble;(2) predominantlyblack / darkgreen, some yellow sible cumulative score was 25 for a Mallard male). or brown at tip; (3) black and brown/yellow; (4) yel- Cumulativescores were designedsuch that Mallard low-green; (5) green or a bluish shade (much more malestended to have the highestscores. Intermediate commonin New Zealand than yellow or orangecolor scoresare indicative of hybrid characteristics. found in North American Mallards). Laboratorymethods.--Total genomic DNA was iso- Speculum and upper alar bar latedfrom eachsample. Proteinase K (200/•g//•1)and (0) Green, no discernible bar; (l) green, obvious SDS (0.5%) were added to 50/•1 of blood in TNE ex- thin and narrow whitish/brown line; (2) green, bar traction buffer (10 mM Tris, 100 mM NaC1, 100 mM distinct but narrow and mottled fawn color; (3) pur- EDTA pH 8.0) or 0.5 to 0.7 g of tissuehomogenized ple, bar distinctbut narrow and mottled fawn color; in ice-cold TNE. Following an overnight incubation (4) purple, bar distinct,wide, mottled fawn; (5) pur- at 55øC, the NaC1 concentration was raised to 200 mM ple, bar narrow, pure white; (6) purple, bar wide, pure and RNase was added to a concentrationof 100/•g/ white. mi. Samples were incubated at 37øCfor ! h and ex- Posterior border to speculum tracted with equal volumesof phenol:chloroform: (0) Black only; (l) black followed by thin (! ram) isoamylalcohol (25:24:1) and chloroform:isoamyl al- white line; (2) black followed by narrow (1-2 ram) white line; (3) black followed by conspicuous(3-4 cohol (24:1) for 20 rain each at 55øC. DNA in each ram) white bar; (4) black followed by wide (>4 ram) samplewas precipitatedin ice-coldethanol and re- white bar. suspendedin TLE (10 mM Tris, 0.1 mM EDTA pH 8.0) to a concentrationof 0.5 to 1.0/•g//•l. The con- centrationand purity of each DNA samplewere es- timated spectrophotometrically. DNA from each individual was digestedwith 16 October1994] Mallardand Grey Duck Gene Flow 973

TAnLE2. MitochondrialDNA haplotypesfor PacificBlack Ducks, New ZealandGrey Ducks, and Mallards. Sequenceof restrictionenzymes is ApaI, AvaI, BamHI, BclI, BglI, BstEII,DraI, HincII, HindIII, Hinfi, PstI, PvuII, RsaI,Sau3AI, StyI, and TaqI. Number of parentalsand hybridswith each haplotypelisted.

Hybrids Haplotypes Parentals Gray Duck-like Mallardlike 1 AGEGLDAECAECBCDE 4 2 ...... G .... B-C 3 $ B ...... G .... B-½ 3 New Zealand Grey Duck 1 B ...... AA-B 2 2 -D .... B- -B ..... ½ 8 a 3B ...... 0 .... B-C 8

Mallard i -D-DFC- I-CF-CDCC 7 2 - D - DFC - - - CF- CDCC 6 • 3 - D- DFC - K- CF- CDCC 1 Onewas "•ure" Mallard. One was "pure" Grey Duck. different restriction endonucleases(2 •g per digest) (unmapped) fragment observedfor all 16 enzymes. under conditionsrecommended by the enzyme sup- This methodhas the advantagethat it includesall of plier. The enzymesincluded 4 enzymesthat recog- our data, but suffers from the fact that some characters nize four-basesequences, 2 that recognize five-base (fragments)are not independentof one another.Sec- sequences,and 10 that recognizesix-base sequences. ond, we inferred the presenceor absenceof restric- Fragmentsin digestedsamples were separatedon 0.7 tion sitesfor all enzymeswhere this was possibleand to 1.2% agarosegels and transferred to nylon mem- deleted thoseenzymes where it was not (Hinfi, StyI, branes(MSI Magnagraph)via Southern(1975) blot- RsaI, Sau3AI, and TaqI). ting. Purified mtDNA, isolated from fresh Mallard heart tissue in a cesium chloride gradient, was la- belled with 32pby random priming (Feinberg and RESULTS Vogelstein1983) and usedto probethe total genomic DNA on each blot. A molecular size standard (•bX174 We scoredeach individual for an average of Hae III plus lambda/HindIII) was usedto determine 120 restriction fragments.In the circular mt- fragmentsizes on autoradiograms.The fragmentpro- DNA molecule, this correspondsto 120 restric- file for eachendonuclease was assigneda letter using tion sites,representing about 600 basepairs of the letter designationsof Kesslerand Avise (1984), recognitionsequence. To ensurethat we had Avise et al. (1990) and Avise (pers. comm.) as a base- identified pure Grey Duck and Mallard haplo- line for consistencyamong studies. Thus, each indi- typesbefore attemptingto characterizehybrid vidual's mtDNA haplotypeis codedas a seriesof 16 individuals, we compared New Zealand Grey letters.Fragment sizes for eachprofile are available from the senior author upon request. Duck and New Zealand Mallard mtDNA hap- Statisticalanalysis.--Genetic distances were esti- lotype profilesto thoseof PacificBlack Ducks mated for eachhaplotype pair accordingto the Nei and North American Mallards, respectively. and Li (1979) method for unmapped fragment data Nine mtDNA haplotypeswere observedalto- (see Nei 1987:106-107) using the analysis package gether,three each in New ZealandGrey Ducks, RESTSITE(version 1.1; Nei and Miller 1990). The re- Pacific Black Ducks, and New Zealand Mallards sulting distancematrix was then pheneticallyclus- (Table 2). We found no fixed differences be- tered by the unweighted pair-group method with tween Pacific Black Ducks and New Zealand arithmeticaverages (UPGMA; Sneathand Sokal 1973), Grey Ducksor betweenNew ZealandMallards using mtDNA haplotypesas OTUs. Net divergence and North American Mallards. Grey Duck hap- between subspeciesand specieswas calculated,cor- rectingfor within-taxonhaplotype variation (Nei and lotype 3 and Pacific Black Duck haplotype 3 Miller 1990). We used PAUP 3.1.1 (Swofford 1991) to were identical, and New Zealand Mallard hap- estimatea phylogeny for the haplotypesusing dif- lotype 2 was identical to that of one North ferent codingschemes for the restriction-digestdata. American Mallard haplotype(Rhymer unpubl. First, we evaluated the presenceor absenceof each data, Kessler and Avise 1984). 974 RHYMER,WILLIAMS, AND BRAUN [Auk, Vol. 111

TABLE3. Nucleotidedivergence among mtDNA haplotypesof PacificBlack Ducks from Australia (AU), New Zealand(NZ) Grey Ducks,and Mallardsestimated from restriction-fragmentdata (method of Nei 1987).

Haplotype 1 2 3 4 5 6 7 8 ! AU Black Duck 1 2 AU Black Duck 2 0.0026 3 AU Black Duck 3 0.0040 0.0014 4 NZ Grey Duck ! 0.0047 0.0047 0.0027 5 NZ Grey Duck 2 0.0048 0.0066 0.0086 0.0090 6 NZ Grey Duck 3 0.0040 0.0014 0.0000 0.0027 0.0086 7 Mallard 1 0.0!!9 0.016! 0.0179 0.0180 0.0108 0.0179 8 Mallard 2 0.0!08 0.0160 0.0168 0.0!70 0.0108 0.0168 0.0013 9 Mallard 3 0.0!20 0.0175 0.0180 0.0184 0.013! 0.0180 0.0013 0.0018

The mtDNA of Grey Ducks and PacificBlack lotype 2. Thirteen of 14 Mallardlike hybrids had Ducksare highly similar with an estimatednu- Mallard haplotypes(five had haplotype 1, six cleotide sequencedivergence (after correction had 2, two had 3), and one had Grey Duck 3, for within speciesvariation) of only 0.09%,a while 5 of 10 Grey Duck-like hybrids had Grey reflection of their close historical affiliation. We Duck haplotypes(one had haplotype2 and four found nine species-specificmarkers between A. had 3) (Table 2, Fig. 3). Of the five Grey Duck- platyrhynchosand A. superciliosa,however. Grey like hybrids that had Mallard haplotypes,two Ducks and Pacific Black Ducks have diverged had Mallard haplotype 1 and three had Mal- from Mallards by 1.40 and 1.43%,respectively. lard 2. All A. superciliosahaplotypes (New Zealandand The proportionof hybridswith Mallard and Australia combined)were very similar (genetic Grey Duck haplotypes was compared to eval- divergenceranging from 0.00 to 0.90%),as were uate whether hybrid matings are primarily be- the three haplotypeswithin New ZealandMal- tween Mallard males and Grey Duck females. lards (0.13 to 0.18%; Table 3). Thishypothesis predicts a predominanceof Grey Figure 2 shows the strict consensusof five Duck haplotypes in hybrid individuals. There equally-parsimonioustrees derived from par- was, however, a significant tendency for hy- simonyanalysis of the unmappedfragment data. brids to have Mallard (n = 19) rather than Grey Key featuresof this tree includeseparation of Duck (n = 7) haplotypes(X 2 = 6.54, P < 0.025). A. platyrhynchosand A. superciliosahaplotypes into two distinctgroups, clustering of GreyDuck DISCUSSION and PacificBlack Duck haplotypesinto a single group that doesnot subdividealong Species-specificmitochondrial DNA haplo- lines, and the basal branching of Grey Duck typeswere distinguishedfor Mallardsand Grey haplotype2 from the Grey Duck/Black Duck Ducks in New Zealand. The genetic distance group.All of thesefeatures also were apparent (correctedfor within speciesvariation) is 1.4%, in the UPGMA analysis,in the parsimonyanal- considerablymore than the 0.7%divergence be- ysisusing only inferred restrictionsite changes, tween the two mtDNA lineagesof North Amer- and in a high proportionof treesresulting from ican Mallards (one of which containsAmerican bootstrappingof either parsimonydata set. Black Ducks; Avise et al. 1990) and between Phenotypicscores ranged from 1 to 4 for Grey Mallards and Mottled Ducks (Kessler and Avise Ducks and 18 to 24 for Mallards, with Grey 1984).It is also higher than that for the most Duck-like hybrids scoring between 5 and 10, divergent pair of A. superciliosahaplotypes and Mallardlike hybridsscoring between 12 and (0.9%). 16. The boundarybetween Grey Duck-like and Hybridization and introgressionappear to be Mallardlike hybridsis somewhatarbitrary. Hy- extensivein New Zealand, despitethe relative- brids were found at virtually all collection sites. ly large genetic distancebetween Grey Ducks One of the 19 individuals characterized phe- and Mallards (for the Mallard complexof spe- notypicallyas a "pure"Grey Duck had the Mal- cies;Rhymer unpubl. data). Individuals char- lard haplotype2 and one of the 15 phenotyp- acterized by hybrid phenotypic traits are pres- ically "pure" Mallardshad the Grey Duck hap- ent throughout the country on both islands. October1994] Mallardand Grey Duck Gene Flow 975

1 25 -- NZ Mallard 1 ß ß Mallard mtDNA haplotype O Grey DuckmtDNA haplotype 20 ß NZ Mallard 3 o ß

15 ß ß NZ Mallard 2 10

NA Mallard 5

NZ Grey Duck 1 0 F M F M F M 24 (100%) Grey Duck Mallardlike 16 (100%) NZ Grey Duck 3 Grey Ducks -like Mallards 12 (100%) Hybrids AU Black Duck 1 5 (100%) Fig. 3. Comparison of phenotypic scores and AU Black Duck 2 mtDNA haplotypesfor birds categorizedas pure Grey Ducks,Mallards, as Grey Duck-like hybrids,or Mal- lardlike hybrids. AU Black Duck 3

1983,Carr et al. 1986,Tegelstrom 1987, Lehman NZ Grey Duck 2 et al. 1991], amphibians [Spolskyand Uzzell 1984, Lamb and Avise 1986], and fish [Avise and Fig. 2. Strict consensusof five most-parsimonious Saunders 1984]). trees relating mtDNA haplotypesof Mallards, New The contention that hybridization between Zealand (NZ) Grey Ducks and Pacific Black Ducks Mallards and closelyrelated nondimorphic spe- from Australia(AU) basedon unmappedrestriction- cies involves primarily Mallard males with fe- fragment data for 16 enzymes (one North American [NA] Mallard haplotype included for comparison). malesof the other speciesis not upheld in New Each fully resolvedtree was 69 steps long (con- Zealand.The higher proportionof hybridshav- sistencyindex excludinguninformative characters = ing Mallard mtDNA as opposedto Grey Duck 0.920, retention index = 0.971). Same topology ob- mtDNA indicates that crosses between Mallard tained from analysisof reduced data set consistingof females and Grey Duck males must occur fre- inferred restrictionsites for 11 enzymes.In latter case, quently. Hybrid femalesapparently mate suc- there were six most-parsimonioustrees, each of length cessfully with either Mallard or Grey Duck 34 (consistencyindex excluding uninformative char- males.The fate of hybrid males is lesspredict- acters = 0.912, retention index = 0.955). Branch-and- able. If male plumage quality is an important bound algorithm employed for all tree searches. factor in their selection as mates, the femalelike Branch lengths that were invariant in all equally- parsimonioustrees given both as number of inferred plumageof many hybrid malesmay reducetheir fragment changes(above) and as number of inferred chancesof successfullyattracting (and back- site changes(below). Numbers in parenthesesindi- crossingwith) females of either species.They cate percentageof times that node appearedin anal- could, presumably,be successfulin obtaining ysesof 1,000 bootstrappedpseudoreplicate data sets. forcedcopulations with any renestingfemales. Our study provides only a conservativees- timate of the degree of introgressivehybrid- Our geneticdata for theseindividuals indicate ization becauseit is basedon phenotypic char- that hybridizationhas led not only to introgres- acters and uniparentally inherited genetic sion of Grey Duck mtDNA into Mallard pop- markers.Analysis of phenotypictraits of birds ulations(the predicteddirectionality of hybrid- from controlledbreeding crosses between Mal- ization), but also to significantintrogression of lards and Grey Ducks has shown that pheno- Mallard mtDNA into Grey Duck populations. typic charactersbecome less reliable indicators This addsto the increasingnumber of examples of hybridization, as the level of introgression of mitochondrial gene exchangebetween spe- increases (Williams and Roderick 1973). Al- cies in other taxa (e.g. mammals[Ferris et al. though we were able to identify two "hidden" 976 RHYming,WILLIAMS, AND BP.•UN [Auk, Vol. 111

hybrids,complete characterization of all hybrid tegrity of A. superciliosasuperciliosa as a separate individuals will require species-specificnuclear speciesis doubtful (Williams 1981). There are DNA markers, which are biparentally inherit- few other examplesof suchgeographically ex- ed. tensive introgression following the purposeful The genetic similarity of Mallards in New introduction of one speciesinto the range of Zealand to those in North America is not sur- another (see also Echelle and Connor 1989). prising given their intentional introduction Nondimorphic species in the Mallard com- from North America in this century. Despite plex often are presumedto have evolved from the geographic isolation of Australia and New the Mallard (Johnsgard1961), although it is pos- Zealand,however, there is no phylogeneticdis- sible that they evolved from a nondimorphic continuity of mtDNA haplotypes between A. "proto-Mallard." Regardlessof the evolution- superciliosarogersi in Australia and A. superciliosa ary history of the group, it appears that the superciliosain New Zealand. Baker (1991) sug- speciationprocess is undergoing reversalin New gestedthat post-Gondwanalanddispersal across Zealand. Anas platyrhynchosand A. superciliosa the Tasman Sea via west wind drift was the couldhave diverged as long agoas 700,000 years principal route of colonization of avifauna from (based on clock estimation), but it is clear that Australia to New Zealand. He discussed the idea no pre- or postzygotic mechanisms have arisen that New Zealand subspeciesof Australian spe- to prevent extensive introgression. Evidence cies are about 20,000 years old. One might at- from captive studiesindicates that hybrids sur- tempt to estimatethe time of divergenceusing vive and are as reproductively successfulas the a molecular-clockapproach. Based on our data, parentalspecies (Haddon 1984). Our geneticdata the sequencedivergence between A. superciliosa now lend supportto the phenotypicevidence subspecies(correcting for within subspecies that points to the eventual loss of identity of variation) was 0.09%.Using a clock calibration the Grey Duck as a separate speciesin New of 2% mtDNA sequenceevolution per million Zealand, and the subsequentdominance of a years (Shields and Wilson 1987), we estimate hybrid swarm akin to the "." that the two subspeciesdiverged about 45,000 Thus, the loss of natural Grey Duck to years ago. Given the expectederror rate of such agriculturehas been compoundedby successful estimates (Sheldon and Bledsoe 1993), this date introductions of the highly adaptable Mallard does not conflict with Baker's suggestion of and the subsequenteffects of interspecifichy- 20,000 years ago. bridization. However, this molecular-clock estimate is probablyoverly simplistic.Limited samplesizes and limited geographicsampling of PacificBlack ACKNOWLEDGMENTS Ducksmake our sequence-divergenceestimates We thank C. Krajewski and G. Shields for helpful somewhat uncertain. Also, the mtDNA• haplo- commentson a previous draft of the manuscript,D. types of the two subspeciesdo not resolve into Swofford for analytic advice, J. Avise for providing separate clades. Some haplotype pairs (both mtDNA fragment-size profiles from previous studies amongand within subspecies)are about10 times of ducks, and P. Fullagar and C. Davey for sending as divergent as the weighted estimate (Table 3). PacificBlack Duck blood samplesfrom Australia. As- Using the sameclock calibration, we would be sistanceto M.J.W. for the collectionof specimenswas forced to conclude that these haplotypes di- provided by the New Zealand Lottery ScienceBoard. Support for J.M.R. was provided by a Smithsonian verged 330,000 to 450,000 years ago. Yet, one Institution Molecular Evolution Postdoctoral Fellow- of the haplotypes was identical between sub- ship. species,yielding an estimated time of diver- gence of zero. This shared haplotype might be L•ORE CITED due to more recentcolonization or continuing AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list gene exchangebetween subspecies.In fact,birds of North American birds, 6th ed. Am. Ornithol. may be moving in both directionsbetween pop- Union, Washington, D.C. ulations;a Grey Duck banded in New Zealand ANKNEY, C. D., D. G. DENNIS, L. N. WISHARD, AND J. in the 1950s was recovered in Australia (Wil- E.S•. 1986. Low genicvariation between Black liams 1981). Ducks and Mallards. Auk 103:701-709. Given the extensivegene flow between Mal- ANKNEY, C. D., D. G. DENNIS, AND R. C. BAILEY. 1987. lards and Grey Ducks in New Zealand, the in- IncreasingMallards, decreasingBlack Ducks: Co- October1994] Mallardand Grey Duck Gene Flow 977

incidenceor causeand effect.J. Wildl. Manage. SavannahRiver Ecology Laboratory, Aiken, South 51:523-529. Carolina. AvIsœ, J. C., C. D. ANKNEY, AND W. S. NELSON. 1990. HADDON,M. 1984. A re-analysisof hybridization Mitochondrialgene treesand the evolutionary betweenMallards and Grey Ducksin New Zea- relationship of Mallard and Black Ducks. Evo- land. Auk 101:190-191. lution 44:1109-1119. HEUSMANN, H. W. 1974. Mallard-Black Duck rela- Awsœ,J. C., ANDN. A. SAVNDERS.1984. Hybridiza- tionshipsin the Northeast.Wildl. Soc.Bull. 2:171- tion and introgressionamong species of sunfish 177. (Lepomis):Analysis by mitochondrialDNA and HITCHMOUGH, R. A., M. WILLIAMS, AND C. H. allozyme markers.Genetics 108:237-255. DAUGHERTY.1990. A genetic analysisof Mal- BAKER,A. J. 1991. A review of New Zealand orni- lards,Grey Ducks,and their hybridsin New Zea- thology. Curr. Ornithol. 8:1-67. land. N. Z. J. Zool. 17:467-472. BRAITHWmTE, L. W., AND B. MILLER. 1975. The Mal- HUnnARD,J.P. 1977. The biologicaland taxonomic lard,Anas platyrhynchos, and Mallard-Black Duck, statusof the Mexican Duck. New Mexico Dep. A. superciliosarogersi, hybridization. Aust. Wildl. Game and Fish Bull. 16. Res. 2:47-61. JOHNSGARD,P. A. 1961. Evolutionaryrelationships BRODSKY,L. M., C. D. ANKNEY, AND D. G. DENNIS. among the North American Mallards. Auk 78:3- 1988. The influence of male dominance on social 43. interactions in Black Ducks and Mallards. Anim. JOHNSCARD,P.A. 1967. Sympatrychanges and hy- Behav. 36:1371-1378. bridization incidence in Mallards and Black CARR, S. M., S. W. BALLINGER,J. N. DERR, L. H. Ducks. Am. Midi. Nat. 77:51-63. BLANKENSHIP,AND J. W. BICKHAM. 1986. Mito- KESSLER,L. G., ANDJ. C. AWSE. 1984. Systematicre- chondrial DNA analysisof hybridization be- lationshipsamong waterfowl ()inferred tween sympatricwhite-tailed deer and mule deer from restrictionendonuclease analysis of mito- in West Texas. Proc. Natl. Acad. Sci. USA 83:9576= chondrialDNA. Syst.Zool. 33:370-380. 9580. LAMB, T., AND J. C. AvISE. 1986. Directional in- DEI•COUR,J., ANDE. MAYR. 1945. The family Anat- trogressionof mitochondrial DNA in a hybrid idae. Wilson Bull. 57:3-55. populationof tree frogs:The influenceof mating ECHELLE,A. A., ANDP. J. CONNOR.1989. Rapid, geo- behavior. Proc. Natl. Acad. Sci. USA 83:2526-2530. graphicallyextensive genetic introgression after LEHMAN, N., A. EISENHAWER,K. HANSEN, L. D. MECH, secondarycontact between two pupfishspecies R. O. PETERSON,P. J.P. COGAN, AND R. K. WAYNE. (Cyprinodon,Cyprinodontidae). Evolution 43:717- 1991. Introgressionof coyotemitochondrial DNA 727. into sympatric North American gray wolf pop- FFJ_NBERG,A. P., AND B. VOGELSTEIN. 1983. A tech- ulations. Evolution 45:104-119. nique for radiolabeling DNA restriction endo- MAZOUREK,J. C., AND P. N. GRAY. 1994. The Florida nucleasefragments to high specificactivity. Anal. duck or the Mallard. Fla. Wildl. 48:29-31. Blochem. 132:6. NEI, M., AND W.-H. LI. 1979. Mathematical model FERRIS,S. D., R. D. SAGœ,E.-M. HUANG, J. T. NIELSON, for studyinggenetic variation in termsof restric- U. RITTE, AND A. C. WILSON. 1983. Flow of mi- tion endonucleases. Proc. Natl. Acad. Sci. USA tochondrialDNA acrossa speciesboundary. Proc. 76:5269-5273. Natl. Acad. Sci. USA 80:2290-2294. NEI, M. 1987. Molecular evolutionary genetics.Co- FRn'H,H.J. 1982. Waterfowlin Australia.Angus and lumbia Univ. Press, New York. RobertsonPublishers, Sydney. N•, M., AND J. C. M•LLER. 1990. A simple method GILES, R. E., H. BLANC, H. M. CAN'N, AND D.C. WAL- for estimatingaverage number of nucleotidesub- LACE. 1980. Maternal inheritance of human mi- stitutionswithin and between populationsfrom tochondrial DNA. Proc. Natl. Acad. Sci. USA 77: restriction data. Genetics 125:873-879. 6715-6719. RF,ICHEL, J. D., ANDT. O. LEMKE.1994. Ecologyand GILLESPIE,G. D. 1985. Hybridization, introgression, of the Mariana Mallard. J. Wildl. Man- and morphometricdifferentiation between Mal- age. 58:199-205. lard (Anas platyrhynchos)and Grey Duck (Anas SCHERER,S., ANDT. HItSBERG.1982. Hybridisierung superciliosa)in Otago,New Zealand.Auk 102:459- und Verwandtschaftsgradeinnerhalb der Anat- 469. idae-eine systematischeund evolutionstheore- GRANT,P. R., ANDB. R. GRANT.1992. Hybridization tischeBetrachtung. J. Ornithol. 123:357-380. of bird species.Science 256:193-197. SHELDON, F. H., AND A. H. BLEt}SOE. 1993. Avian GRIFFIN, C. R., R. J. SHALLENBERGER,AND S. I. FEFER. molecularsystematics, 1970s to 1990s.Annu. Rev. 1989. 's endangered waterbirds: A re- Ecol. Syst. 24:243-278. source management challenge. Pages 1165-1175 SHIELDS,G. F., AND A. C. WILSON. 1987. Calibration in Proceedingsof Freshwaterand Wildlife Sym- of mitochondrialDNA evolution in geese.J. Mol. posium (R. R. Sharitz and J. W. Gibbons, Eds.). Evol. 24:212-217. 978 marMrs,WILLIAMS, aND BI•,UN [Auk, Vol. 111

SNEATH, P. H. A., aND R. R. SOKAI,. 1973. Numerical WATANABE,T., M. MIZUTaNI, S. WAK•NA, aND T. TOMI- . W. H. Freeman, San Francisco. TA. 1985. Demonstration of the maternal in- SO•JTHERN,E.M. 1975. Detectionof specificsequenc- heritance of avian mitochondrial DNA in chick- es amongDNA fragmentsseparated by gel elec- en-quailhybrids. J. Exp. Zool. 236:245-247. trophoresis.J. Mol. Biol. 98:281-283. WELLrs, M. W. 1980. The island waterfowl. Iowa SPOLSKY,C., ANDT. UZZELL.1984. Natural inter-spe- State Univ. Press, Ames. cies transfer of mitochondrial DNA in amphib- WILLL•MS,M.J. 1981. The duckshooter'sbag. Wet- ians. Proc. Natl. Acad. Sci. USA 81:5802-5805. land Press,Wellington, New Zealand. SWOFrORD,D. 1991. PAUP: Phylogenetic analysis WILLIAMS,M. J., AND C. RODERICK.1973. Breeding using parsimony, version 3.1. Illinois Natural performance of Grey Duck (Anas superciliosa), History Survey,Champaign, Illinois. Mallard (Anasplatyrhynchos) and their hybrids in TEGEI.STROM, H. 1987. Transfer of mitochondrial DNA captivity. Int. Zoo Yearb. 13:62-69. from the northernred-backed vole (Chlethrionmys YAM^SmNA, Y. 1948. Notes on the Marianas Mal- rutilius)to the bank vole (C.glareolus). J. Mol. Evol. lard. Pac. Sci. 2:121-124. 24:218-227.