EVOLUTIONARY RELATIONSHIPS AMONG THE NORTH AMERICAN MALLARDS

PAUL A. Jo•X•SG.•

This studyis the report of an attemptto understandthe evolutionary relationshipsexisting within a groupof mallardlikeducks native to North America. The groupincludes the CommonMallard, `4nas platyrhynchos platyrhynchosL.; the BlackDuck, `4nas rubripes Brewster; the Florida Duck,`4nas fulvigula fulvigula Ridgway; the Mottled Duck,.4nas fulvi- gula macuIosaSennett; the Mexican Duck, .4nas diazi diaxi Ridgway; and the New Mexican duck, .4nas diazi novimexicana Huber. All but one of these (the CommonMallard) are restrictedto North America, and all theseAmerican forms possessa sexuallynondimorphic plumage. In all other respectsthey are extremelysimilar to the CommonMallard, and a study of their relationshipsto this form was believedpossibly to providean instructiveexample of speciation.

REVIEW OF RANGES,POPULATIONS, AND NOMENCLATURE

The Mexican and New Mexican Ducks In 1886 Ridgway describeda new speciesof mallard (.4nas diazi) from Puebla, Mexico, which, similar to the then recently (1874) de- scribedFlorida Duck, differed from the CommonMallard (`4nas platy- rhynchos)in its lack of sexualdimorphism. Accordingto Ridgway, the new speciesdiffered from the Florida Duck in its more Mallardlike characteristics,namely the distinctband of white on the secondarywing coverts and its less-fulvousover-all coloration. Thirty-six years later, Huber (1920) describedanother species of mallard from New Mexico, which he named `4nas novimexicana. Huber was seeminglyunaware of the similaritiesshown by the New Mexican Duck to the Mexican Duck, but in 1922 Conoverreferred to an extralimital Nebraskanspecimen as .4has dia•i novimexicana,and Phillips (1924) treated the New Mexican Duck in the samemanner in his monograph. Although the closerela- tionshipbetween the New Mexican and the Mexican duckswas there- fore realizedrelatively early, the paucityof specimensand field observa- tions made the exact geographicrelationship between the two forms uncertain. In 1946 Lindsey summarizedthe situationas follows: "The known nesting range of the New Mexican Duck is confinedto a small area of the south-westernUnited States, but the presumptiverange extends southwardsinto Chihuahua, Mexico, where its relation to the northern breedinglimits of the Mexican Duck (.4. diazi diazi) is unde- terrained." 4 JOlt NSGARD, Relationships among Mallards [ Auk ß I. Vol. 78

ARIZONAi 4G : NEWMEXICO ! ' :8 4G TEXAS

...... j .

....,', •, '•, SINALOA)---" ,[ ( ' • '•. /•--- •.-.• ,.•- "-•'-, ..• ,/ '• ,, ) ß( ,' :, '., ,• ZACATECAS,-"' ,

Sight Specimen Range Mex•c•In Duck O ß ,.r-• (inc) New Mexicofi") MottledDuck O ß • __

JALISCO•. • OL So•ulo? O 50 I00 I•0 20<) •0 MICNOACAN

An attempt has been made to map the geographicranges of the Mexican and New Mexican ducks,and thus to help visualizetheir geo- graphicrelationships. This was doneby plotting all availablespecimen recordsand suchsight recordsas appearedjustified from the literature, personalcommunications, and the major United Statescollections. The Auk 1 Jon•-sg^Ru, Relationshipsamong Mallards 5 Vol. 78 J resultingmap (Figure 1) showsfairly closeagreement with thc com- parable maps of Dclacour (1956), Arcllano and Rojas (1956), and Leopold(1959). It followsthe map of Leopoldin indicatingan unbroken range bct;vccnthe New Mexican and Mexican populations,although recordsarc spottyfor the area in question. If birds at the northernparts of the rangearc migratory,for whichthere is someevidence, it is probable that they winter in the heart of the range of diazi; habitat conditionsarc very poor on the Chihuahualakes, and they arc unattractiveto waterfowl (Saundersand Saunders,1949). Of all the racesof North American mallards,perhaps least is known about the Mexican Duck, and this is certainly the case regarding the total populationsize. Since 1947 the U.S. Fish and Wildlife Service has surveyedthe lakes of interior Mexico during its annual January waterfowl inventories. Mr. R. H. Smith, who has flown this route each year since1951, has provided (in litt.) the data obtainedfor the Mexican Duck during theseyears. The sinailesttotal numberof Mexican Ducks recordedwas 780 in 1951, and the largest total was 10,322 in 1958. Althoughthe northernChihuahuan lakes--Toronto, Bustillos,and Mexi- canos--•veresurveyed, it is of interestthat no Mexican Ducks have been recordedthere; Lago Santiaguillo,Durango, is apparentlythe northern wintering limit of the Mexican Duck in Mexico. Its westernlimit would appearto be Lago Sayula, Jalisco,and the easternlimit in the 1natshy region near Oriental, Puebla. These data would indicate that the size of the adult MexicanDuck populationis probablyless than 20,000 birds. New Mexicangame biologists have estimated that the populationin their state is extremely snmll, possiblynot over 100 adult birds.

The Florida and Mottled Ducks In 1874 Ridgwaydescribed a new form of mallard from St. Johns, Florida, whichhe consideredto be a subspeciesof the Black Duck. As the Black Duck was at that time called Anas obscura,the new bird was namedAnas obscuravar. fulvigula. It differedmainly from the Black Duck in that the malar region,chin, and throat were immaculatebuff rather than streakedwith dusky,and the plumagehad a general ochraceous,rather than dusky,tone. At the time it was known only from Florida. Later (1878) Ridgway relegatedit to the rank of a full species,Arias fulvigula. Elevenyears later, Sennerr(1889) described from Texas yet another speciesof mallard, which he called .4has maculosa. Sennett stated that the most marked differences between zt. )•ulvigulaand A. maculosawere the latter's streaked, rather than plain buff, cheeksand a green,rather than purple,speculum in additionto a mottled, rather than streaked,general coloration. The Mottled Duck 6 Joa•s•^RD, Relationshipsamong Mallards [ .•.uk [Vol. 78 was addedto the A.O.U. Check-listin 1890 as Anas Iulvigula •naculosa, and no later argumentsfor retainingthe specificstatus of maculosahave been presented. Although Phillips had vacillatedearlier betweenaccepting and reject- ing the Mottled Duck and Florida Duck as distinct races, in 1924 he synonymizedthe two forms. Later, Peters (1931) accepted•naculosa as being distinctfrom Iulvigula, althoughDelacour and Mayr (1945) as well as Hellmayr and Conover (1948) synonymizedthem. The form •J,aculosa is still acceptedby the A.O.U. Check-list. Regardlessof the questionof the validity of the describedplumage differences,which will be discussedlater, a definite hiatus in the ranges of the populationsseems to exist. The range of the Florida Duck (see Figure 2, which is modifiedfrom Howell, 1932, and Sprunt, 1954) is largely confined to the southern half of peninsular Florida, with its northern limits being in the vicinity of Gainesville. There are appar- ently no specimensof fulvigula or maculosathat have been taken in Mississippi. A single specimenhas been taken in Alabama (Imhof, 1958) and may representeither fulvigula or •nacuIosa. In addition, severalsight recordsalong the coastof Alabamahave beenobtained for Florida or Mottled ducks. Mottled Ducks occur commonlyalmost to the eastern border of Louisiana (Oberholser, 1938). They are found mainly in the coastaltier of parishesof Louisiana(Lowery, 1955) west to the Texas border, and in Texas likewise follow the coastal counties for the entire length of the state (Singleton, 1953). The range of the Mottled Duck alsoextends a longdistance into Mexico (Saunders,1952, 1953) in the statesof Tamaulipasand Veracruz. The range map pre- sentedin Figure 2 is a compositeof informationfrom the sourcescited above,and from additionalsight and specimenrecords available to me. The Florida Game and Fresh Water Fish Commission has taken a censusof the Florida Duck populationannually since 1948 by the aerial transectmethod. During the years 1948 to 1955, the estimatedtotal adult populationvaried from 17,000to 30,000 (Sincock,1957). The 1956and 1957inventory estimates for the entire statewere 9,000-16,000 and 7,000-10,000 birds, respectively,indicating, according to Sincock,a sharpdrop in Florida Duck numbersin recentyears. A reasonableesti- mateof the presenttotal adult Florida Duck populationwould appear to be 10,000. In Louisiana,inventories of the Mottled Duck populationhave been madeyearly since1952. Estimatesin recentyears have averaged about 10,000birds (M. M. Smith,in litt.), whichare spreadout fairly evenly over the marshesof southeastand southwestLouisiana, being especially Vol.Auk78 ] Jo]•r•*s(;^RD,Relationships among Mallards 7

• •IssISSIPPI • •,LABAMA '• FLORIOA i

P, LouiSiANA

Sight Sp•cimetlRange ,_• MottledOuck FloridaDuck BlockDuck V ß Mollord A ß Me•ic•Duck 0 ou• c**., • '•) • Florida(In•)O

Figure 2. Geographicranges of Mottled and Florida ducks. 8 JO}•NSC.;XRD,Relationships among Mallards [ .auk [ Vol. 78 frequentin the salt and brackishmarshes of the coast. In Texas the coastalwaterfowl surveys for the years 1947 to 1952provide an estimate of winter Mottled Duck populationsin that state (Singleton, 1953). Mottled Ducks during these five years averagedover 7,000 birds in October(maximum of 23,000in 1947), decreasingmonthly to an average minimumof from 1,000to 2,000 birdsfrom Novemberthrough February. Estimatedsummer Mottled Duck populationsover a 26,000-square-nfile area were 18,500in March and 21,000 in August 1952. Thesebirds are distributedover the entire coastalprairie region, extending inland from 50 to 70, or possibly100, miles (Singleton,1953). Little can be said regarding the size of the Mexican populationof Mottled Ducks. Saunders(1953) reportsthat "severalhundred" have beenseen during the Januaryinventories along the Mexican coast,and Saundersand Saunders(1949) statethat ground studiesin Tamaulipas and Veracruz indicatedthat severalthousand birds are present. In sum- mary, the total Mottled Duck populationmight be estimatedat 30,000 adult birds in the United Statesplus an additional number. possiblynot over 5,000, in Mexico.

The Black Duck Gmelin described the American Black Duck in 1788 as ,4•ms obscura, basedon Pennant's"Dusky Duck" froin New York. This name was useduntil 1908,when ,•nas obscurawas found to be preoccupied(Allen, ct at., 1908). In 1902, however,Brewster had describeda subspeciesof the Black Duck, which he calledrubripes, and thus this name was taken for the specificepithet to replaceobscura. This left Brewster without a name for his newly describedform, a conditionhe rectifiedin 1909 by proposingtristis. The ensuingarguments for (Brewster, 1910; Phillips, 1920; Townsend, 1912) and against (Dwight, 1919; Pirnie, 1928; Shortt, 1943) the admissionof tristis have filled many pages,and the long-delayedacceptance (1931) of this form by the A.O.U. Check-list committeewas xvithdrawnin 1945 following Shortt's (1943) careful analysis of seasonalvariations in soft-part coloration. The present consensus that tristis is not an admissible form is followed here. In 1945 Delacourand Mayr stated that "it seemsobvious that the Mexican and Black Ducks are only sub-specificallydistinct from the Dusky Duck" (]ulviguIa), and consequentlythe Black Duck was treated in their revisionas .4•ms[ulvigula rubripes. However, in 1956 Delacour modifiedhis opinion regardingthe Mexican and Florida ("Dusky") ducks' affinities,stating that the latter will mate preferentially xvith the Common Mallard rather than the Black Duck, and so restored the name ,4•as rubripes to the Black Duck. Vol.Auk78 ] JOI-INSGARI),Relationships an, o•g Mallards 9

The recenttreatise by Stewart (1958) on the distributionof the Black Duck is a thorough and detailedaccount that can scarcelybe improved. His mapsof the wintering and breedingranges of the Black Duck agree well with the data collectedduring the presentstudy, and are adoptedin simplifiedform (Figure 3) here. Stewart divided the eastern United Statesinto six regionsof approximatelyequal size (indicatedin Figure 5), and determinedthe average wintering Black Duck populationsfor eachregion, based on the annualJanuary U.S. Fish and Wildlife Service ilwentoriesfor the five years 1952-1956. These data, roundedto the nearest 1.000, appear in Table 1, together with the average percentage

I ! I / I

Figure 3. Southern limit of breeding (broken line) and wintering (shaded) ranges of Black Duck. 10 JO>Ir,'SGARB,Relationships among Mallards [ auk [Vol. 78

TABLE 1 REGIONAL DISTRIBUTION OF WINTERING BLACK DUCKS

Total Birds/party Per cent Region BlackDucks hour total

East-Central U.S.F.W.S. counts 128,000 -- 19 Christinas counts -- 76.2 19.3

Atlantic Coastal U.S.F.W.S. counts 417,000 -- 61 Christinas counts -- 239 60.7

Southeastern U.S.F.W.S. counts 62,000 -- 9 Christmas counts 13.9 3.5

Great Lakes U.S.F.W.S. counts 32,000 5 Christmas counts 31.5 8.0

Northern New England U.S.F.W.S. counts 35,000 5 Chrlstinas counts 32.5 8.2

Western U.S.F.W.S. counts 4,000 -- 1 Christmas counts -- 0.9 0.2 of the total U.S. populationobserved in each region. In addition are includedcomparable data compiledfrom the annual Audubon Christmas countsfor the years 1950-1957. Becausethe numbersof observersvary greatly in differentareas, average total Black Ducksrecorded per party hour (for all countson which Black Ducks were observed) are listed, rather than averagetotal numbersof Black Ducks seen. A breakdown of thesedata by statesis also presentedin Figure 4, providinga more detailedcomparison of Black Duck wintering densitiesthroughout the eastern states. These two sets of data, obtained by widely differing means,agree remarkablywell in regard to the relative distributionof the total wintering population. The only area of poor agreementis in the southeasternstates, where a differenceof 5.5 per cent occurs. This differenceis at least partly explainedby the fact that in Florida and Louisiana the Fish and Wildlife inventories include Florida and Mottled duckswith Black Ducks, sincethey cannotbe distinguishedduring aerial counts. This factor tends to exaggerateestimated Black Duck popula- tions in these states. Total numbers of Black Ducks estimated on these annual January inventories provide the best available indication of population size, althoughthese survey data cannotbe consideredto be completelyaccu- rate estimatesof actual numbersof wintering Black Ducks (Stewart, •auk 1 JOR•SOARn,Relationships among Mallards 11 Vol. 78 ]

BLACK DUCKS/ PARTY LessThan I :.'.'....'.•1- 10.9 • 11-49.9 50- 200 Figure 4. Relative densities of wintering Black Ducks, eastern states.

1958). During the five years includedin Stewart'sstudy, this average totalpopulation figure was approximately 678,000. If the 1957and 1958 inventoriesare included,the seven-yearaverage is 647,000 birds, which will 1;eused here as an estimateof the total Black Duck population.

The Commo• Mallard The Common, or Green-headed,Mallard was describedby Linneaus in 1758as .dnasplatyrhynchos in the 10thedition (p. 125) of his Systema Naturac. Previousto 1906,it wasgenerally referred to as .dnasboschas, beforethe page priority of platyrhy•choswas established(L/Snnberg, 1906). The speciesoccurs throughout the northernhemisphere, and has 12 Jon•,•sc,.am),Relationships a,non# Mallards [ Auk [ Vol. 78

one of the widest breeding distributionsof any speciesof waterfowl in the world. Becauseof the rarity of this speciesin the easternstates during the colonial period, the Mallard (unless uncapitalized,the corn,nonname Mallard is used here to mean .4. p. platyrhynchos)was not recognized as a common speciesin the United States, since it was replaced in the East by the Black Duck. Evidently it was not before the late 1800's that the Mallard beganto movenoticeably eastward, as indicatedby the scarcityof referencesto it before 1900. Brewster (1906), referring to the Cambridgeregion, states: "During the earlier years of my shooting experiencethe Mallard was unknownto our local gunners.and every- where throughout New England was considereda rare bird. Within the past decade,however, it has becomea regular and not uncommonvisitor in autumn to several localities in eastern Massachusetts .... " Town- send (1905) indicatesa similar situationin the sameregion. At presentthe Mallard must be considereda fairly numerousnesting speciesthroughout the East, at least as far east as central New York. Although Maine probably may not yet be consideredas within the nestingrange of the Mallard, this specieshas been observedthere with increasingfrequency, and recently(1953) Coulterprovided the first two definitenesting records for that state. At presentMallards and Mallard x Black Duck hybrids compriseapproxhnately 1 per cent of the fall hunter kill in Maine (Coulter, 1954). In Ontario Mallards have in- creasedgreatly aroundthe Toronto area (Goodwin, 1956), and to the north nestas far eastas the JamesBay region (Manning, 1952). For the purposesof comparisonwith the BlackDuck, a map showing the breedingand wintering distributionof the Mallard in the eastern stateshas beenprepared (Figure 5). The breedingrange has been modifiedfrom previousrange maps (Kortright, 1942; Phillips,1924) to showthe easternextension of the breedingrange, and the winter distri- butionis basedmainly on an analysisof the AudubonChristmas count data. Localareas of winter concentrationare indicated. To providean additionalcomparison with the BlackDuck (Figure 4), the averagetotal Mallards recordedper party hour (for all countswhere Mallards were seen) havebeen calculated for eachstate for the period1950-1957, and thesedata are presentedin Figure 6. Becauseonly that portionof the Mallardpopulation that is sympatric with the BlackDuck is of biologicalimportance from the practicalstand- pointof actualinteraction between the two genepools, only an estimate of the Mallardpopulations of theeastern states (Mississippi and Atlantic Vol.Auk 78 ] JOHNSGARD,Relationships amon# Mallards 13 flyways) will be attempted. It shouldbe remembered,however, that the potentialgene pool of the Mallard, if the Central and Pacificflyway (plus Europeanaud Asiau) populationsare considered,is actually immensely greater. Again, the 1hostsatisfactory source of quantitativeinformation on the size of Mallard populationsis the Fish and Wildlife Service January inventories. For the Atlantic flyway, data from thesecounts are showu be!ow for the 10-year-period1949-1958, whereas for the Mississippi flyway they are availablefor the nine-year-period1950-1958.

.4veragetotal •lveragetotal Ratio Mallards Black Ducks Mallard: Black

Atlantic flyway 218,000 457,000 32.2:67.8 Mississippiflyway 3,566,000 190,000 97.5:2.5 Total easternU.S. 3,784,000 647,000 85.4:14.6

A checkon the accuracyof thesedata is providedby a similar calcula- tion of Mallards and Black Ducks observedper party hour during the Audubon Christmascounts over essentiallythe same period of years (1950 1957). Althoughactual numerical comparisons cannot be made, it is possibleto obtain an index to relative densitiesof Mallards and Black Ducksby summingthe averagenumbers of both forms recorded per party hour for all stateswithin each flyway and obtainingratios on the relative abundance of the two forms:

Cumulative Cumulative Mallards Black Ducks Ratio per party hour per party hour Mallard: Black

Atlantic flyway 113.27 297.09 27.6:72.4 Mississippiflyway 2,259.06 97.14 95.9:4.1 Total easternU.S. 2,327.33 394.23 85.5:14.5

These two sets of data show extremely close agreement with one another,particularly in regardto the estimatedratio of Mallards to Black Ducks for the two flywayscombined, where a disagreementof lessthan 1 per cent is found. It seemssafe, therefore, to state that in the eastern United Statesthe Mallard populationoutnumbers the BlackDuck at the presenttime by approximatelysix to one. 14 JOHh-SG^P.D,Relationships among Mallards [ .•uk [ Vol. 78

Figure 5. Breeding (ruled lines) and wintering (shaded) ranges of Common Mallard in eastern states. Blacked-in areas indicate major wintering concentrations.

MORPItOLOGICAL CHARACTERISTICS

The variousplumage and soft-part differencesattributed to eachof the sexually nondimorphicforms can be summarizedas follows: The Mexican and New Pie:ricanducks. Accordingto Ridgway (1886). the Mexican Duck is most like the Florida Duck, differing in its double band of white on the wings, its less-fulvouscoloration, its streaked cheeks,and in its lack of a black spot on the baseof the maxilla. From the Mallard it differs in its obviouslack of sexual dimorphism. When Huber (1920) describedthe New Mexican Duck, no comparisonswith other mallardswere made, other than to say that it is apparently"a memberof the Black Duck group." .auk ] J0XtNSG^RD,Relationships among Mallards 15 Vol. 78 J

MALLARDS / PARTY HR LessThen I "/..':•...... :.•I - 10.9 ..•.• II-49.9 50-200 over200 Figure 6. Relative densities of wintering Common Mallards, eastern state•.

The Florida and Mottled ducks. The Florida ])uck was stated (Ridgway, 1874) to differ froln the Black Duck in the forlner's paler coloration,unmarked cheeks and throat, predominatelyochraceous plum- agetone, and (in males) a blackspot at the baseof the maxilla. The later descriptionof the Mottled Duck, with its streakedcheeks, darker colora- tion, and less-distinctblack 1hark on the maxilla, weakened these distinc- tion• and left no clear-cut difference between these forms and tile Black Duck. The Black Duck. Accordingto Kortright (1942), the Black Duck differs from the Mottled Duck by the former's olive-coloredbill and the 16 Jori NSGARD,Nelatio•ships amon9 Mallards Auk IX ol. 78

absenceof a black spot at its base, by possessinga distinctly streaked throat, and by having narrower buffy edgesto the feathersof the back.

Methods and Materials It was evidentfrom the confusionindicated above that a COlnparative studyof plumageand soft-partdifferences among these forms mustpri- marily includemeasures of variation of the relative darknessof body plumage,the degreeof white barring on the secondarycoverts, and bill coloration. Various supposeddifferences in speculumcoloration (shown by Phillips, 1912, to be highly variable) and the degreeof streakingon the throat, which appearsto be related to the relative darknessof the bodyplumage, were not consideredof major ilnportancefor study. It was decidedat the outsetthat any plumagecomparisons would have to be madewith referenceto a standardthat wouldextend over a range includingboth extremesas well as gradedintermediate conditions. The "Hybrid index"system, which has been so successfully used by Anderson (1949), Sibley(1950), andothers for estilnatingobjectively phenotypical variations,seemed of possibleapplication here. It was discoveredthat the plumagesof femalehybrid combinations between Mallards and Black Ducks correspondedso closelyto thoseof pure Mexican, Florida, and Mottled clucksthat patternsof correspondingfeather regions could be consideredidentical. Therefore, it was decidedto use pure female Mallardsand BlackDucks for plulnageextreme standards, and graded hybrid combinationsfor the intermediate conditions. Applyingthe techniqueof the Hybrid index,the extremerepresented by the lightestplumage condition was givena numericalvalue of "0," and the darkestextrelne was given a value of "4." The intermediate conditionthus had a value of "2," and the 1/4 and 3/4 conditionshad valuesof "1" and "3." In theory, an index number of "2" would result from an F1 hybrid, and scoresof "1" and "3" would be obtainedfrom the two backcross combinations if no dominance were involved. Of course,wild-taken hybrids cannot be definitelyestablished as suchcom- binations,and in most casesprobably would be the result of lnore com- plicatedrecombinations and segregations.However, by directcomparison with specimensin the Museumof ComparativeZoology at Harvard of known-parentagehybrids between Mallards and Black Ducks obtained from experimentalbreedings by Phillips (1915, 1921), it was possible to determine that no dominance xvasinvolved. Feathers from wild-taken hybrids froln the Ithaca region, which closelycorresponded to these known-parentagebirds, were usedas referencetypes. .•uk ] Jo•s6.•m),Relationships amon• Mallards 17 Vol. 78 ]

Six feather areas, which includedthe major points of plumagediffer- encesmentioned above, were selectedfor study. These areas were as follows:

Outer (6th) Rectrices (Row "A," Figs. 7-9): Predominantly white in platy- rh3mchos,solid dusky in rubripes. Scapulars (Row "B," Figs. 7 9): Wide terminal edge and three subterminalbuff bars in platyrhy•whos,narrow terminal edge and no barring in rubripes. I !anks (Row "C," Figs. 7-9): Wide terminal edge and subterminal "V" in platyrhy•chos, narrow terminal edge in rubripes. Secondary Coverts (Ro•v "D," Figs. 7-9): Broad white subterminal bar in platy- rl•yJ,chos, this absent or very faint in rubripcs. Under-tail Coverts (Row "E," Figs. 7-93: Predominantlywhite in platyrhynchos, dusky with narrow buff edge in rubripes. Breast (Row "F," Figs. 7 9): Predominantlywhite with brown sFotsor crescent in platyrhy•whos,dusky, usually with faint "V" and narrow buff edge in rubripcs. Representativefeathers from theseareas are shownin Figure 7, with pure female Mallard ("0") at extremeleft and pure female Black Duck ("4") at extremeright. This figure was usedwhen determiningindex w,lues of all femalesexamined. Sincethe specimensfrom the populations other than platyrhynchosand rubripes•vere obviouslynot hybrids,per- haps "I"umage index" rather than "Itybrid index" would be a more appropriatedesignation for the method as it was used here. A valid comparisonof •nalesfrom all formswas impossible,owing to the sexualdimorphism present in the Mallard. For visualcomparison only, the correspondingfeathers from males of the Mallard, the Black Duck,and hybrid combinations that probably closely approach the typical F• and the two backcrosscombinations are presentedin Figure 8. The blendingtype of inheritance,as well as a curiousappearance of female characteristicsin someof the hybridfeathers (e.g., flanksand rectrices), is of generalinterest. To providea meansof objectiveevaluation of plumagedifferences in males,it was necessaryto excludethe Mallard and devisea plumage indexfor the sexuallynondimorphic forms only. This wasaccomplished by selectingfeathers from malesrepresenting the lightest(diazi) and darkest(rubripes) extremes,and three gradedconditions between them. The featherareas used were the sameas in the females,and indeedthe descriptionsgiven for tl:esefeathers from the femalesapply equally well to the males. The index scoringwas thus as follows:

IndexNo. Plumagecondition Lightest, as in diazi, with wide buff and white areas Mainly as in diazi, but reducedbuff and white areas Intermediatebetween diazl and rubripes Mainly asin rubripes,but with slightbarring present Darkest,as in rubripes,with no barringpresent 18 JOHNSGARD,Relationships among Mallards [Vol.[ Auk 78

Figure 7. Reference chart of feathers from female Mallard (left), Black Duck (right), and hybrid combinations. Reading from top down, rows are A to F (in text). Auk ] Jo•ss(;•,•, Relationshipsamong Mallards 19 Vol. 78 J

.i

Figure 8. Chart of feathers from male Mallard (left), Blaek Duek (right), and hybrid combinations. Heading from top down, rows are A to E (in text). 20 Jo•sa^RI•, Relationships among Mallards [ Auk tVol. ?a

Figure 9. Reference chart of feathers from male Mexican Duck (left), Black Duck (right), and intermediate conditions. Reading from top down, rows are A to F (in text). Auk 1 JOItNSGARD,Relationships among Mallards 21 Vol. 78 J

These conditions are illustrated in Figure 9, the reference feather chart, with the left extremerepresenting "0" and the right extreme rep- resenting"4." In the caseof both malesand females,specimens that scored "0" for all six characters would receive a summed score of "0," whereas those that scored "4" for all characters would receive a summed score of "24."

Results Index scoresobtained for eachfeather region of all malesand females exanfinedare presentedin Tables 2 and 3. By totaling the number of observationsfor each index classin a given form, a combinedestimate of that form's relative darknessmay be obtained,which can be compared statisticallywith the estimateof any other form. These combinedobser- vations and a mean index value for each form are also tabulated in Tables 2 and 3. By constructinga contingencytable and determining the chi squarevalues for any two populationsin question,the probablevalidity of separatingthose populations on a plumagebasis may be determined. Using the index data for females, novimexicana was compared with diazi, and fulvigula was comparedwith •naculosa. Too few male speci- mens of diazi and novimexicanawere availablefor a similar comparison, but male specimensof fulvigula and rnaculosawere sufficientfor testing. In the case of female novimexicanaand diazi, no strong indication of distinctiveness(p >0.05) betweenthe two populationswas found. How- ever, in the case of both male and female specimensof •ulvigula and maculosa,separation of the two was distinctlyindicated (p <0.01). An alternativemeans of presentingindex data,that of summingindices for eachspecimen and plotting the summedtotals for eachform, is pre- sentedin Figures 10-11. In this casedata from novimexicanaand diazi were combined,since they showedno indicationof summedindex differ- encesthat might be significant. These diagramsclearly illustrate the difficultiesof findingclear-cut morphological distinctions between any populationsexcept the extremes. A fairly evident differencebetween fulvigulaand rnaculosamay be seenin this diagram,and it is of interest that those specimens(four) of maculosathat achievedthe highest summedindices were from Texas, primarily the Brownsvillearea, rather than from Louisiana. It would be of interest to learn if the Mexican populationof maculosaalso exhibitsthis very dark coloration,but at presentalmost no specimenshave beencollected from that region. Examinationof soft-partcoloration is very difficultfrom preserved specimens,and as a result no critical studieswere made of thesefeatures. However,since some aspects of bill spottinghave beenstressed by the authoritieson theseforms, an attemptwas madeto obtainsome informa- 22 Jo•xs•,•Rv,Relatio•tships among Mallards I Auk [Vol. 78

MALLARD MEXICAN(8•N. MEX ) DUCK FLORIDADUCK MOTTLEDDUCK ?• N=19

N=27

20'

• :.:.:.:

0 4 B 12 16 20 24 HYBRIO INDEX

Figure 10. Distribution frequenciesof summed indices, female specimens.

N=15MEXICAN (&N MEX.) DUGKIn 30- FLORIDADUCK '•:•...•.• MOTTLED DUCK I N=50

25- BLACK DUCK

PLUMAGEINOEX 24

Figure l]. Distributionfrequences of summedindices, male specimens. Vol.Auk78 ] JOttNSGARD,Relationships among Mallards 23

TABLE 2

SUMMARy OF [FqDEX DATA FOR FEMALES

"2 .... 3.... 4" Total specimens

Rectrices plat.yrhy.nchos 30 5 4 - -- 39 nowmex,cana -- 10 6 l -- 17 diazi 1 7 4 -- -- 12 fulvigula -• 11 13 3 -- 27 m aculosa -- 1 l 0 8 -- 19 rubripes -- - l 8 95 104 Scapulars platyrhynchos l 8 13 8 -- 39 novimexican a 1 4 8 4 -- 17 diazi 2 2 7 1 -- 12 fulvigula 2 16 9 -- -- 27 maculosa -- 1 6 12 19 rubripes -- 1 9 94 104 Flanks plat.yrhy.nchos 29 7 3 -- 39 nowmex,cana 3 5 7 2 -- 17 diazi 1 10 1 -- -- 12 fulvigula 3 19 5 -- -- 27 maculosa -- 2 11 6 -- 19 rubripes .... l 23 80 104 Secondarycoverts plat.yrhy.nchos 37 1 l -- 39 nowmex,cana -- l 5 2 -- 17 diazi 2 10 -- -- 12 fulvigula -- 14 l 1 2 -- 27 maculosa -- 2 11 6 -- 19 rubripes -- -- 2 14 88 104 'Fail coverts platyrhynchos 29 7 3 - 39 novimexicana 1 12 4 -- 17 diazi -- 10 2 12 fulvigula - I 1 12 4 -- 27 m aculosa -- 6 l 3 -- l 9 rubripes -- I l 5 88 104

Breast platyrhynchos 30 6 3 - 39 novimexicana l l 3 3 -- -- l 7 diazi - 10 2 - 12 fulvigula 2 8 16 1 -- 27 maculosa 7 12 - 19 rubripes - - 3 20 8l 104 Total combined character observations Index mean (Six per specimen) platyrhynchos 173 39 22 -- 0.35 novimexicana 6 59 30 7 -- 1.37 diazi 6 49 16 l 1.17 fulvigula 7 79 66 10 1 1.49 maculosa 6 51 57 -- 2.45 rubripes .... 9 89 526 3.83 Jo}•'sc:x,r, l•elationshipsamo•tg Mallards [Vol. 78

TABLE 3 SUMlk{ARY' OF INDEX DATA FOR MALES

"0 .... 1" "2" "3" "4" Total specimens

Rectrlces novimexzcana 3 diazi 5 4 ------9 fulvigula -- 13 26 7 4 50 1 11 15 3 30 rubripes ------1 49 50 Scapulars novimexzcana 2 -- 2 -- -- 4 diazi 2 7 ------9 fulvigula 4 26 18 2 -- 50 maculosa 2 5 16 7 -- 30 rubripes ------1 49 50 Flanks novimexzcana 3 - I -- -- 4 diazi 7 2 ------9 fulvigula 15 31 4 -- -- 50 15 9 6 -- 30 rubripes -- -- 15 35 50 Secondarycoverts novimexicana 3 diazi 6 3 ------9 f ulvigula -- 26 23 I -- 50 5 18 7 -- 30 rubroes -- I 10 17 22 50 Tail coverts novimexw ana 3 diazi 5 4 ------9 fulvigula -- 32 18 -- -- 50 13 17 -- -- 30 rubripes -- -- 2 11 37 50 Breast novimexwana 2 diazi 1 8 ------9 fulvigula -- 17 25 8 -- 50 1 20 9 -- 30 rubripes -- -- 1 20 29 50 Total combined character observations Index mean (Six per specimen) novimexic ana 16 5 3 -- 0.46 diazi 26 28 ------0.52 fulvigula 19 145 114 18 4 1.48 maculosa 2 40 91 44 3 2.03 rubripes -- 1 13 65 221 3.69 tion on these points. The major question concernsthe presenceor absenceof a black spot on the baseof the maxilla of males. Ridg-way (in Ferrari-Perez, 1886) used this character to separate fulvigula (present) from rubripes(absent), and likewiseKortright (1942) indi- catesthat it is absentin rubripes. However, Shortt's plate in Kortright's volume,as well as that in Shortifs (1943) paper of soft-part colorsof Auk ] JOHNSGARD,Relat,io•ships arnorod Mallards 25 Vol. 78 l the IglackDuck, showsthis spotclearly. •[ have notedit on all live male lglackDucks in breedingcondition that •[ have examined,but the olive- slate color of the rhamphothecaof preserved ]glack Duck specimens obscuresthis mark. it was also found to be presenton all 17 of the male specimensof fulvigulaand all 12 maculosaexamined for it. It •vas absentfrom five male specimensof diazi, and thus appearsto be typical of f•dvigula,maculosa, and rubripes. In summary it may be said that none of the describedplumage or soft-partcharacters, aside from the sexualdimorphism of platyrhynchos, were found to be of absolute diagnosticvalue in differentiatingany populationfrom all other populations. •[t was found that the distinctive plumagecharacteristics of the two extremes,platyrhynchos and rubripes, are bridgedby the other populationsand exhibit the characteristicsof thesepopulations in hybrid individuals. The Geneticand Hormoi•al Basis of Sex•ial Di•izorphis,z At this point, where it is apparentthat the sexualdimorphism of the CommonMallard providesthe only real qualitativemorphological basis for distinctionfrom the other populations,it is of prime importanceto investigatethe geneticbasis for such dimorphismin order to ascertain the taxonomic"value" of this character. That is, are the geneticchanges required for converting a highly sexually dimorphic phenotypeto an essentiallynondimorphic one so great that these phenotypesmust be regardedas constitutingtwo species? in all probabilityno one will ever be able to determineby how many genesone distinctspecies differs from another, but if it can be determinedthat the loss of sexual dimorphism is dependentupon only a very few genes,then the taxonomicvalue of such dimorphismmust be consideredslight. it has been discoveredthat male Mallards can be "feminized" by orchidectomy,followed by the implantationof bits of ovarian material in the abdominal cavity (Goodale, 1918). Furthermore, the annual assumptionof the "eclipse"plumage of the Mallard providesadditional evidencethat the male Mallard carriesthe geneticpotentiality of female plumagepatterns. However, suchfemalelike plumages can be obtained only during periodsof little or no testicularactivity, and thus would not providean explanationfor the permanentlynondimorphic plumages of the forms concernedhere. A genetic basis for the loss of sexual dimorphismin theseforms is necessary,possibly similar to that found in somebreeds of DomesticFowl (e.#., Sebrights). In suchbreeds both sexesare normally hen colored,and the factor producinghen coloration in males is determinedby a single,dominant autosomal gene (Hurt, 1949), which, in the presenceof either male or female hormone.acts 26 Jo•t•sc^u•, I?ela•ionshipsamong Mallards [[Vol. Auk 78 on the feather follicle and preventsthe expressionof male plmnage. Gonadectomizedbirds, however,assume a more malelike plumagein the absence of sex hormones. Three female Black Ducks were ovariectomized duringthe presentstudy, and pluckedfeather areas grew backnormal feathers. Therefore it seems unlikely that the same mechanismthat appliesto the DomesticFowl canbe postulatedfor theseducks. How- ever, it is mostfortunate that experimentalstudies on the geneticnature of sexualdimorphism in the Mallard, Mottled Duck, and Black Duck have been conducted(Phillips, 1915, 1921), and these studiesare so crucialto the evaluationof the geneticimportance of sexualdimorphism in waterfowlthat they will be summarizedhere in detail. Phillipshybridized males of "Florida Duck" (actuallyMottled Duck œromLouisiana), Black Duck, Pintail (Anas acuta), and AustralianGray Duck (Anas poecilorhynchasuperciliosa) with female "Call Ducks" (domesticAnas platyrhynchos)of typical•vild-type plumage. All of the crosseswere extendedto the F2 generation,as well as to many of the backcrosscombinations. The resultshe obtained(1921) for the three, mallardlike forms were:

Mallard x Florida Duck: 38 F2 males obtained. Two nearly pure Mallard types, three nearly pure Florida types. Backcrossesto Mallard female bring out many pure Mallards. Mallard x Black Duck: 23 ire males obtained. Two most extreme Mallard variates lack solid green head, white neck-ring, and light-colored, vermiculated abdomen. Backcrossesto Mallard females show no pure Mallard types. Mallard x Australian Duck: 15 F2 males obtained. Very slight segregation. Most extreme Mallard type scarcely differs from Fx individuals. Backcrossto Australian female was the only one made, so not directly comparablewith the other backcrosses.

In the mating betweenthe Mallard and the i'intail, "the smallestpos- sibleamount of segregationwas found, both in the straight F2 generation and in backcrosseswith the Pintail male." Regarding the Mallard x "Florida" Duck crosses,'Phillips (1921) statesthat the hybrids exhibit The most extreme example of segregationthus far discoveredin speciescrosses between either ducks or pheasants. Parental types are almost recovered from among 38 F2 males. Backcrossesto Mallard parent producea more uniform gen- eration of nearly pure Mallard types. The Florida Duck mustbe supposedto repre- sent, from the genetic point of view, substantially a Mallard, with the male sec- ondary sex characters not expressed.

He states further:

It is apparentthen that in this speciescross we have a nearer approachto ortho- dox Mendelismthan in any other thus far reportedamong birds. Out of 38 males two were nearly pure Mallard types, and three almost pure female types, these latter closely resemblingthe Florida Duck parent. Vo•..auk7s ] Jou ssc.Am), Relationships among Mallards 27

Although Phillips did not pursuethis point, it shouldbe stated that, accordingto the principleof multiplefactor inheritance,a matingof two genotypesdiffering in a characterby two pairs of independent,non- dominantgenes •vould produce in an Fs generationof 38 phenotypesan expectedrecovery of roughlyfour pure parentaltypes (Sinnott et al., 1950). Judgingfrom Phillips' description,the Fs generationof this crossclosely approximated these expectedresults. It seemsjustified therefore to state that the difference in male plumages between the CommonMallard and the "Florida" (Mottled) Duck is probablyexplain- able by a geneticdifference of the magnitudeof two gene pairs. As indicatedby Phillips, the Mallard x Black Duck F2 generation exhibited less tendencytoward segregationthan the Mallard x Florida l)uck cross,but more so than did the crossesinvolving the Australian Gray Duck and Pintail with Mallards. Thus we may concludethat the numberof genesdetermining the Black Duck's plumagedifferences from the Mallard is greater than is the casewith the Florida Duck, and fewer than the genetic differencesbetween the plumagesof the Mallard and the Australian Duck or Pintail. Phillips believed,however, that the Black Duck possessesthe geneticcapacity for assumingthe male Mallard plmnage. He states(1915): We are thus led to the belief that the wild Black Duck actually containsMallard charactersnot presentin the Australian Duck, and that these are slightly accen- tuated by associationin the zygote with characters from the Mallard parent. In other words, there is present in the wild Black Duck, latent Mallard characters, which sometimesappear with age. These charactersbecome immediately apparent when a quarter dose of Mallard is combined with a three-quarter dose of Black Duck.

It would seem apparentthat the describedplumage differencesthat havebeen the basisfor the erection,at onetime or another,of six species of North American mallards are scarcelyvalid characterson which to basespecies judgments. This is emphasizednot only by the evidently slight geneticvariations that can effectthese differences, but also by the considerableindividual variation and overlapof charactersattributed to the various forms.

EGG-WHITE PROTEINS

Introduction and Methods In conjunctionwith Dr. C. G. Sibley'sresearch on the electrophoretic characteristicsof arian protein systems,an attemptwas madeto compare the egg-whiteproteins of as many of the forms under investigationas possible. The theoreticalimportance of protein similarities in deter- mining phyletic relationshipsis now widely reco•ized. Comparative 28 JoH•s•^•v, Relationshipsamong Mallards I ,•uk [VoL 78 studiesutilizing electrophoresisare of relatively recent development; the history,methodology, and theoreticalbasis of thesetechniques have been summarizedby Sibley (1960). Comparative electrophoreticegg-white studies on waterfowl were begunby McCabeand Deutsch (1952), who analyzed37 speciesof birds, includingeight of waterfowl. Sibley (1960) has reportedon several additionalspecies of Anatidae,including some material from the Mallard and the Black Duck that has beenused in the presentstudy. Equipment and techniquesutilized have been describedin detail by Sibley and Johnsgard(1959a, 1959b).

Variation with Anas To illustrate the amount of variation that is encountered among diverse speciesof ,'1has (sensu Delacour), samplesfrom speciesrepre- senting the extreme morphologicalvariation have been selectedfor presentation. These are the Common Pintail (,4has acura), Shoveler (,4has clypeata),and Gadwall (,4has strepera). Electrophoreticprofiles of the egg-whiteproteins from thesespecies are presentedin Figure 12. These curves, projected on a centimeter scale, indicate the variability encounteredin the stated sample size. Each of the sampleswas from a differentegg. The uppermostof the three linesindicates the maximum vertical distanceencountered in the combinedprofiles, the lowermostthe minimumvertical distance, and the centralline representsa meanprofile sample,plotted optically, for eachspecies. In all casesvertical variability is of least significance,relative height of the peak is of greater signifi- cance,and the horizontalmobility from the point of application (indi- .catedby arrows) to eachpeak is of the greatestsignificance. Using the meansample profiles of eachspecies and superimposingthem on the same chart providesan easy meansof optical comparison. Thus it may be observedthat the Gadwallpossesses a higher peak in the ovoconalbmnin region than do the other two species. The presumedovomucoid and ovalbuminfractious of the Pintail have somewhatgreater mobilitiesthan have the correspondingfractions of the other species.

Variation within the North ,4roerican Mallards Within the mallard group, sampleswere obtainedfrom the Mallard. the Florida Duck, and the Black Duck. These profilesare also pre- sentedin Figure 12. The Florida Duck profilewas obtainedfroin eight samplesof a single,slightly incubated egg, whereasthe othersrepresent singlesamples from eachof severalfresh eggs. Thereforeless credence canbe placedin the resultsfrom the Florida Duck, and it is only included for the sakeof completeness.The compositemean profilesfrom these Vol.^uk78 ] Jo}tNSG^•D, .Relationships among Mallards 29

Figure12. Eleetrophoreticprofiles of egg-whiteprotein samples (see text for explanation). 30 JOHNSG^RI),Relationships among Mallards [Vol.I' Auk 78 forms,plus domestic "Pekin" Mallard (/tnas platyrhynchos),have been superimposedfor easycomparison. It will be seenthat the meanpro- files of the Mallard and Black Duck are extremely similar, and in fact the Pekin profileexhibits greater difference from the wild form than does the BlackDuck. Consideringthe inferior conditionof the Florida Duck sample,it too showsa great similarity to the typical mallard profile, differing only in the mobility of the ovoconalbuminfraction. In the original profilesthis fraction exhibited,by its indefiuiteand diffused borders,indications of having undergonepartial chemicalbreakdown. It is very probablethat fresh sampleswould show even closer similarity to the others.

SEXUAL BEHAVIOR

Data on the sexual behavior of Mallards and Black Ducks have been presentedelsewhere (Johnsgard, 1960), and will be summarizedbriefly here, togetherwith noteson the behavior of the other forms concerned. The behaviorof the Mallard has been extensivelystudied by Lorenz (1941; 1951-1953) andmore recently by Weidmann(1956) andothers. Thesestudies have all indicatedthat there are three major male "court- ship" displaysthat are primarily concernedwith pair formation. These are, in Lorenz'sterminology, the "Grunt-whistle,"the "Down-up," and the "Head-up-tail-up"with associated"Nod-swimming." The female's courtshipdisplays consist primarily of "Nod-swimming"and "Inciting." Thesesame displays are found,with varying degreesof modification,in all the mallardlikeducks as well as in other closelyrelated groupsof surface-feedingducks, but in theseless closely related species qualitative as well as quantitativedifferences occur in the behaviorof both sexes. A studyof the courtshipdisplays among wild and semiwildMallards and Black Ducks,during whichover 3,000 Mallard and over 1,300 Black Duck displayswere tallied,has failed to revealany qualitativedifferences in the male courtshipdisplays of thesetwo forms. Femalecourtship displayswere not includedin the quantitativestudies owing to their nondiscretenature, but no qualitativedifferences were observedbetween Mallardsand Black Ducksin thesefemale displays. Only a few statis- tically significantquantitative differences were detectedin the male court- shipdisplays of Mallardsand Black Ducks. Of these,the moststriking differencewas that male BlackDucks have a lower displaythreshold; i.e.,in a giveninstance of maledisplay more male Black Ducks are likely to displaytoward a female than would male Mallards in the samesitua- tion. It is suggestedthat this is a resultof the Black Duck'sabsence of specialmale plumagecharacters, which likely intensifythe needfor a promptand sensitiveresponse to femalesignals. Secondly,male Black Vol.auk78 ] JoH•sG^R•, [?elationshipsamong Mallards 31

Duck displaystended to be slightly more predictableand uniform than were 1haleMallard responses. This differencealso was attributed to be the probableresult of selectionfor specificmale responsesin the absence of male plumagedifferentiation. A few observationson the behaviorof Mallard x Black Duck hybrids were obtainedfor wild and captiveindividuals. Hybrids displayedwith the sameapparent intensity as did "pures"of eachform, and alsoshowed no deviationin qualitativefeatures of displayfrom the parental forms. All of the displays of the parental forms, including copulation, were observedin hybrids, but in too small numbers to submit to statistical analyses. Intermediatehybrids mingled with Mallard courting groups as frequentlyas •vith Black Ducks, althoughhybrids that were predomi- nantly sinfilar to one or the other parental form generally were found with the form they mostresembled. Occasionally"pure" Mallards could be observeddisplaying in a group of Black Ducks or vice versa, but normally the courting groups of the two forms remained almost com- pletely segregated. No observations on the sexual behavior of the Mexican Duck were obtained,but captivespecimens of the Florida Duck and Mottled Duck have been observeddisplaying. These observationsare also too few to analyzestatistically, but all the courtshipdisplays of Mallards and Black Ducks have been obse•wedin these races and in exactly the same situa- tions and form as in Mallards and Black Ducks. Except for the slight plumage differencesof all of the forms studied, it would be impossible to recognizewhether a particular bird was a Mallard, a Florida Duck, a Mottled Duck, or a Black Duck on the basisof its sexual behavior.

ISOLATING MECHANISMS

As shownearlier (Johnsgard,1960), no qualitativedifferences in the behaviorof male Mallards and Black Ducks were found that might serve as a basisfor speciesrecognition differences; indeed, no major quantita- tive differenceswere observedthat might indicatea possiblestrengthening or diminutionof any particular displaysunder the impact of selection againsthybridization between the two forms. This meansthat, in a mixedpopulation of unmatedMallards and BlackDucks, selective mating may otherwise be effectedby (1) plumage differencesin the males, (2) slightecological or geographicdistributional differences that would result in reducedprobability of nfixed pairing, or (3) differencesin the temporalsequence of pair formation. The last-mentionedpossibility may be ruled out on the basis of the behavioral studies cited above. The other pointswill be consideredhere. 32 JOHNSC^RD,]x'elationsl•ips among Mallards /Vol.[ Auk 78

Correctmate selectionby femaleson the basisof maleplumage differ- encesis an obviousand simplemeans by which the two populationscould be maintainedas perfectly discreteand separateentities; male Black Duckscan be distinguishedfroin maleMallards at greatdistances by even the unpracticedhuman eye. That sucha discriminationmight exist cannot be argued; that it actuallydoes occur and so resultsin "correct"mate choicesis worth investigation. If sexualselection in mallardsis actually as important as is suggestedby the daborate male plumagepattern of the Common Mallard, it must be concludedthat female Mallards tend to selectfor mate the more brightly coloredbirds. There is someevidence to supportthis theory (Johnsgard,1960). One wondersthen if female Black Ducks, confronted with both dull-colored drake Black Ducks and the comparatively"gaudy" male Mallards, might not tend to "choose" the brighter bird. Lorenz (1941) found that in the sexually nondi- morphic Meller's Duck (Anas melleri) females showedas strong a tendencyto mate with male Mallards as with their own species. This couldpossibly also be true of the Black Duck, which is so similar in out- ward appearanceto the Meller's Duck. Goodwin (1956), in a stimu- lating discussionof this problem, has pointed out the rather obvious effectsof mixed mating. He states: "If the ducks of either speciesare less selectivewith regard to cross-matings,then that specieswill tend to be less successfulif a surplusof drakes exists." That there exists a preponderanceof drakesin adult flocksof both Mallardsand BlackDucks cannotbe questioned. 5,lostavailable data indicatethat between50 and 60 per centof the combinedsex populations of Mallardsand BlackDucks are males. This meansthat not all malescan be successfulin obtaining mates, and the effectivenessof sexual selection is thus enhanced. In comparingresident populations of Mallards and Black Ducksin the Toronto area, Goodwin (1956) noted that since the introductionof a sinall stock of Mallards and Black Ducks in 1931 the two forms have hybridized"fairly freely and yet Mallardshave increased rapidly in pro- portion to Black Ducks." He suggeststhat a selectionagainst Black Duckssuch as postulatedabove might be operatingthrough a tendency towardsnonselective mating or towardscross-mating in the caseof the Black Duck females. It would be most valuable to learn if most mixed pairing in the wild state involvesmale Mallards and female Black Ducks as would be expected,or, on the other hand, if female Mallards often mate with male Black Ducks. I have recorded or found reference to three pairings of the former type and six pairings of the latter kind, indicatingthat both combinationsdo occur,but the sampleis far too sinall to suggestthe favoring of either combination. Auk ] J0}x•SC,^•D,b?elatio•zships amon# Mallards 33 Vol. 78 .I

In spite of the essentiallyidentical nature of the displaysof Mallards and Black Ducks, there is a surprisingamount of separationin courting groupsbetween the two forins when they occuron the samewater area. Whether this is the result of mutual attractionamong like-plumaged birds or a result of very slight differencesin habitat preference is difficult to decide. The marsh-dwellingtendencies of the Mallard, as opposedto the forest- and maritime-inhabitingproclivities of the Black Duck, are certainly the most conspicuousdifferences to be found betweenthe two forms. This was well documentedby Benson(1937) who investigated the breeding ecologyof the Black Duck on some natural water areas near Ithaca, New York. One of these areas, Spencer Marsh, closely approachesthe typical cattail and tule-bordered marshesfound in the westernprairie states,whereas the others,Danby Pond and the Michigan Itollow swamp and Beaver Pond, are typical representativesof the woodlandponds of the eastern states. The more alkaline cattail marsh was actually found to contain many more waterfowl-food plant species than the xvoodedponds. However, in the xvoodedswamp and ponds Black Ducks usually nest every year, xvhereasin SpencerMarsh they had not beenrecorded as nesting. Bensonconcluded that the Black Duck "is apparentlycognizant of the fact that it needsa backgroundof shadows affordedby swamps,woods, and numerouslogs and stumps,"and "is not nearly so much at easeon Spencer Marsh as on the Beaver Pond and Danby Pond." In his book on the Black Duck, Wright (1954) has also pointed out that during the breeding seasonthe bird is essentiallya "forestduck." This contentionis further supportedby a breedingstudy done on Kent Island, Maryland, where 80 per cent of 161 Black Duck nestswere found in woodedareas (Stotts, 1955). Although the Mallard will, under unusual circumstances,nest solnetimesin heavily wooded areas (Bent, 1923), this is a comparativelyrare situation.

ESTIMATION OF GENE POOLS AND HYBRIDIZATION INCIDENCE

From the data presentedearlier, it is possibleto estimatewith a fair degreeof certaintythe approximatesizes of the genepools of the popu- lations involved. By assigningthe Mallard populationsof the eastern United States (Atlantic and Mississippiflyways) an arbitrary value of 100, the relativeestimated total populationsizes of the other forms are as follows: Black Duck 17, Mottled Duck 0.91, Mexican Duck (including New Mexican population)0.52, and Florida Duck 0.26. This clearly illustratesthe comparativelysmall sizes of these southernpopulations and the easeby which they might theoreticallybe swamped. To deter- mine ho•v much, if any, interbreedingoccurs among the sexually non- 34 JO•,•sG^RD,Relationships among Mallards [ Auk [ Vol. 78 dimorphicforms where they are in contactwould be impossible. How- ever, in the two areaswhere the Mallard overlapsin its breedingrange with these nondimorphicforms the presenceof hybridization is easily proved. A relatively small amount of breedingcontact exists betweenthe Mal- lard and the Mexican ("New Mexican") Duck in central and southern New Mexico. In regard to the frequency of hybridization, Lindsey (1946) states: The extreme southern part of the Mallard's breeding range covers most of the known nesting range of the New Mexican Duck; frequent hybridization produces individuals showingwidely varying proportionsof the traits of the parental species. Almost any day in winter and early spring, interesting wild hybrids may be seen at close range, mingling with wild Mallards which also winter in the duck ponds at Rio Grande Park, Albuquerque. Hybrids usually outnumberthe pure New Mexicans here; few of the latter winter this far north. Except for this statement,no other estimateof hybridizationincidence for the area is available. It is somewhateasier to estimatethe incidenceof hybridizationoccur- ring between the Mallard and the Black Duck in the eastern United States, for the area of contact is much broader and a considerable amount of banding and hunter-kill data are available. No one has thus far accumulatedsuch data with the intention of determining the exact hybridizationincidence in any area, and these figures generally result from the occasionaloccurrence of a bird that cannotbe readily classified as either a Mallard or Black Duck. Undoubtedlythe great majority of recombinantindividuals that only showslight deviationsfrom the typical Mallard or Black Duck plumagepatterns are not recordedas hybrids; juveniles,females, and malesin nonbreedingplumage are also likely to be overlooked. However, to provide a very minimum estimate of hybridizationincidence, an attempt was made to accumulatesuch data from a numberof states. These data have been obtainedprimarily by directly contactingpersons engaged in bandingoperations, as practically noneof this kind of informationhas beenpublished, and their assistance in compilingthese figures is gratefullyacknowledged. The most satis- factory data obtainedin this way, from the standpointof samplesize (a minimumof 4,000 total birds), are presentedin Table 4. Nearly all of the personscontacted indicated that their figuresonly represented"obvious," or "pronounced,"hybrids. Thus D. Handley, who providedthe data from Ohio, states:"If minor hybrid character- isticswere consideredthen the Mallard-Blackhybrids would makeup aboutfive to ten per centof the total Mallard and BlackDuck popula- tions." This wouldmean that the hybrid figuresin this caseshould be .•uk 1 JO•tNSG^m),Relationships among Mallards 35 Vol. 781

TABLE 4 MINIMUM MALLARDX BLACK DUCK HYBRIDIZATIONINCIDENCE, VARIOUS STATES

No. No. Per cent Per cent Per cent Mal- No. by- by- Mal- hybrids State Period lards Blacks brids brids lards Blacks

Ohio 1950-1953 2,834 1,118 69 1.7 70.5 6.2 Delaware 1951-1958 2,160 2,489 74 1.5 42.5 2.5 Michigan 1935-1958 3,563 6,955 92 0.9 33.9 1.3 Vermont 1946-1957 335 3,647 19 0.5 8.4 0.5 Pennsylvania 1948-1958 1,957 8,531 25 0.2 18.6 0.3 Maine 1948-1957 423 23,907 39 0.2 1.9 0.2 multipliedby a correctionfactor of from betweenthree to six, and the samecan probablybe safelystated for the other figures. Except for the data from Pennsylvania,there is a direct correlationbetween the per- centageof recordedhybrids and the percentageof Mallards in the com- binedpopulation. This mightbe interpretedto meanthat a femaleBlack Duck is morelikely to matewith a Mallard where Mallardspredominate than a female Mallard is likely to mate with a Black Duck where the latter predominate. It might, however,also simply mean that hybrids are more frequent where the two forms have been in contact longest, that is, in the westernpart of the Black Duck's range. In any event, it can be safelyconcluded that increasedsympatry of range is likely to result in increasedhybridization. Unfortunately, no detailed data from the westermnostedge of the Black Duck's range could be located. Dr. W. H. Elder informed me (pers. comm.) that during his banding operationsin Illinois Mallards comprisedabout 90 per cent of the combinedMallard and Black Duck catch,and that the ratio of Black Ducks to hybrids was approximately ten to one (about 700 to 70). Dividing this figure by two, there still remainsat least 5 per cent of the Black Duck gene pool "tied up" in hybrids in that area. No direct evidencethat hybridspossibly are being selectedagainst, either through reducedfertility or reducedability to obtain mates be- causeof intermediatecharacteristics, was obtainedduring this study or hasbeen indicated by otherstudies. The completespectrum of plumage variationsfrom pure Mallard to pure Black Duck indicatesthat a suffi- cientnumber of hybridmatings occur to allow the phenotypicexpression of everypossible genetic recombination. For examplethe followingnum- bersof wild male hybridswere observedat Howland's Island, New York, during a two-monthperiod. Six almost pure Mallards, six predomi- nantly Mallard, sevenintermediate, eleven predominantlyBlack Duck, and eight almostpure Black Duck. If, however,hybrids actually are 36 Jo•r•-sc•.•D, Relationsh,ipsamong Mallards [ Auk [ Vol. 78 being selectedagainst, the roughly six-to-oneratio of Mallards to Black Ducks in the eastern states cau be interpreted to mean that for every mixed pair formed there resultssix times as much damageto the Black Duck genepool as to that of the Mallard. If the hybridsare not selected against,and they mate randomlywith both Mallards and Black Ducks, then the possihilityof eventualswamping of the Black Duck gene pool is six times greater than the alternativepossibility. In sumnmry,these followiug points place the BlackDuck at a selective disadvantage:(1) Its smallergene pool will be affectedmuch soonerby disadvantageoushybridization or possibleswamping than will that of the Mallard, (2) the male Black Duck'ssexually nondimorphic plumage and lack of specialsignal charactersare, theoreticallyat least, of less sexual selectiouvalue than are thoseof the Mallard in a mating situationwhere sexual selectionis intense,and (3) forest clearing,farin pond construc- tit;n, and more intensive land use in the eastern states have resulted in continuallydecreasing Black Duck, and increasingMallard, habitat.

Q]VOLUTIONARY IMPLICATIONS AND TAXONOM[C CONCLUSIONS The extremely wide Holarctic range of the Common Mallard leaves little doubt that it is the oldest of the forms under discussion, and can prohably be regarded as being closest to the ancestral forin. I also helieve that the sexually dimorphic condition of the Common Mallard is of long standing,for a strikingly similar plunmgepattern is found in the otherwise well-differentiatedand allopatric Australian and New Zealand Chestnut-breastedand Brown Teals (•nas castanea-/tnasauk- landicacomplex), which must have been isolatedfrom the Mallard for a nmch longer period than any of the populationsunder consideration here. Of theselatter forms, I believethat the degreeof geneticdiffer- encefrom the CommonMallard, from least to greatest,is indicatedhy the sequencediazi % maculosa• fulvi•ula % rubripes. Isolationof the Black Duck from the ancestralnmllard stockprobably originatedthrough the establishmentof an easternpopulation that was separatedby the hardwoodforests froin the prairie populationsof Arias. Such isolationwould allow-for the loss of sexual dimorphismthrough reducedpressures for speciesrecognition (Sibley, 1957). Probablythis coastal population gradually moved westward and uorthward into the swampsand waterwaysof the hardwoodand evergreenforests, where selectioufor a darker, concealing,body plumagewould likely obtain. By the time the populationhad extendedfar enoughwestward to estab- lish a secondarycontact with the prairie nmllards, sufficienthabitat adaptationhad presumablybeen evolvedto reducegreatly the actual contact between the two forms. Auk ] JOlq•SG.ai•D,Relationships among Mallards 37 Vol. 781

Much more recently,breeding populations of mallardswere established on the central plateauand easterncoasts of Mexico. This may have occurredduring a periodof glaciation,when the prairie mallardpopula- tion was forced southwardand the ancestralBlack Duck populationwas restricted to the southeasternstates. Lakes in the central plateau of Mexicowere apparentlymuch more extensive during this period (Jaeger, 1925), and suitablebreeding habitat may have likewisebeen much more extensive. Later, when the warmer and drier postglacialclimate resulted in a drying up of many of thesepluvial lakesand causeda northward movementof waterfowl populations,a small, residual populationof mallardsprobably remained on the few areasof suitablehabitat on the plateau. At the sametime, the coastalpopulation, separated from the plateaupopulation by au altitudinalbarrier of 7,000feet, probablybegan to extendnorthward along the Gulf Coast. Both the coastaland plateau populationsprobably lost their sexualdimorphism rapidly, judging from the simple genetic control of this character,in responseto a lack of. other syrupatticbreeding species of Arias. The coastalpopulation also evolveda darker plunmge,presumably in responseto its more humid environment. This •vas achievedby the same mechanical,if not genetic, mechanismas in the Black Duck, that is, by the reduction in width of the buff edging and bars of the contour feathers. The plateau popula- tion, however, remainedessentially unchanged in this respect. Even- tually part of the coastal population reachedpeninsular Florida (after crossingan area of apparently now relatively unfavorablehabitat along the coastsof Mississippi,Alabama, and westernFlorida) and established a semi-isolatedpopulation in the southernpart of that state. This explanationthus hypothesizesthe establishmentof three major isolated populationsfrom parental prairie mallard stock, each of which independentlyevolved sexually nondimorphic plumage and varying de- grees of darker coloration. The same mutations may, however, have been utilized in some casesto achieve these ends. As indicated earlier, the Black Duck may have lost its sexual dimorphismby an inhibition of the genetic basis for this character, whereas in the Mottled Duck at least the genesgoverning it seemto have beeneliminated. I am in firm agreementwith Delacour(1956) that dia•i and •ulvi•ula shouldbe consideredsubspecies of Arias platyrhynchos,and likewise agreethat novi•ne.ricanais inadmissibleand shouldbe synonymizedwith diaci. Resultsof the plumagestudy, however, force me to concludethat •rm•culosais a valid race, a deternfinationin oppositionto Delacour's opinion. 38 JomnSC,^RD, Relationships among Mallards [ Auk [ Vol. 78

An accurateand completelysatisfactory disposition of rubripes can- not, in my opinion,be made. No moderntaxonomist has as yet formally proposedthe conspecificityof rubripesand platyrhynchos,although the possibilityhas been suggested(Baillie, in Goodwin, 1956). Delacour (1956) statesthat only the rather wide area of sympatryin breeding ranges prevents this interpretation. Dr. Milton Trautman, who has studiedthe Black Duck carefullyfor many years,has concluded(in litt.) that "the Black and Mallard probablyare not 'good'species as we nor- mally define speciesin birds, but that they are 'potential' specieswhich as yet have developedno geneticisolation but which display marked (subspecificat least) habitatdifferences and a markedmutation in colora- tion." I completelyagree with this opinion,and feel that thesehabitat differencesare of greater significancethan is the relativelyrecent sym- patry of the forms. The marked separationof the two when they are inhabitingthe samewater area and the relatively infrequentoccurrence of mixed courting groups indicate that recognitioncapacities are well developed. Hybrid frequencyalso indicatesthat pairing is far from randomin wild populations. Arguing againstthe species-distinctnessof the forms are the appar- ently completehybrid fertility, the great morphological(including both plumage and egg-white proteins) similarities, the essentiallyidentical sexualbehavior, and the evidencefor increasinghybridization associated with increasingsympatry. In addition,there is no indicationof any tendencytoward the developmentof any uniquesignal characters. Thus, dependingupon one's viewpoint, evidencecan be marshaledeither to treat the two formsas subspeciesor as species.There is little probability that speciationin the Black Duck will progressany further becauseof the extensivesecondary contact, and indeedit is possiblethat the form will eventuallybecome so geneticallyintermixed with the Mallard that specificrecognition will be impossiblefrom a practical as well as a theoreticalstandpoint. I, therefore,believe that the most satisfactory statusfor rubripesis to considerit a subspeciesof z4nasplatyrhynchos, but one that exhibitsa greaterdegree of differencesin ecologicalprefer- enceand socialsegregation than normallyoccurs in subspecies.Neither a subspecificnor a specificrelegation can be entirely satisfactoryat present,and only time and continuedinvestigations are likely to prove or disprovethe conclusionsreached here.

ACKNOWLEDGMENTS

The original data formed the basisfor a doctoraldissertation, which wassubmitted to the GraduateSchool of CornellUniversity on 15 June Auk 1 JO•NSC^R•, Nelationshipsamon!l Mallards 39 Vol. 78 ]

1959. I would like to expressmy sincereappreciation to my committee chairman,Dr. CharlesG. Sibley,for his continuedassistance and advice. For financial aid in the form of fellowshipsI am indebted to Cornell Universityand the National ScienceFoundation. The State University of New York provideda summerresearch honorarium in 1957. The use of aviary facilitiesfor keepingducks, and the opportunityto utilize the electrophoreticequipment there, is gratefullyacknowledged. My other committeemembers, Drs. L. C. Cole, W. C. Dilger, and A. van Tien- hoven, provided various kinds of assistance;Dr. Dilger's helpfulness in allowing my use of his motion picture equipment was especially appreciated. For the use of unpublisheddata or other assistanceI am most grateful to C. E. Addy, D. Benson,W. Bock, J. F. Cassel,E. B. Chamberlain, P. Claude,M. W. Coulter,W. F. Crisey,W. H. Elder, R. J. Fleetwood, W. L. French, J. Hagar, D. Handley, A. S. Hawkins, O. H. Hewitt, S. Hoyt, J. A. Imhof, R. Latimer, A. S. Leopold, A. A. Lindsey, J. Lynch, D. F. McKinney, A. Merriam, W. R. Miller, R. S. Palmer, M.D. Pirnie, F. A. Pitelka, M. M. Smith, R. I. Smith, R. H. Smith, M. B. Trautman, and D. A. West. Egg-white materials or birds were providedby the New York State Department of Conservation,the Delta Waterfowl ResearchStation, the Wildfowl Trust, J. Martin, J. Kiracofe, B. Dane, and R. E. Phillips. Use of museumfacilities at Cornell Univer- sity, the American Museum of Natural History, and the Museum of ComparativeZoology at Harvard Collegeis acknowledged.

SUMMARY

A comparativemorphological, behavioral, and electrophoreticstudy was undertakenon six populationsof North Americanmallards that are currently consideredby the A.O.U. Check-listto comprisefour species and two additional subspecies.These include ,4has p. platyrhynchos, •q. d. diaxi, •4. d. novimexica.na,.4. f. fulvigulc•,,4. •. maculosa,and .4. rubripcs. These forms are essentiallyallopatric in their breeding distributions,with the major exception of .4. p. platyrhynchosand •. rubripes, which overlap throughouta several-hundred-milezone in east-central United States. Geographicranges and populationsizes are estimated,and evidence is presentedto indicatethat the presentsympatry of rubripesand platy- rhynchosis of recentorigin and is rapidly increasing. Plumage differ- encesin the populationswere investigatedby the "Hybrid index"method, and it was concludedthat, except for the sexual dimorphismof platy- rhynchos,only minor quantitativedifferences in featherpatterns differ- 40 Jolt-•ss^RD,Relationsl•ips among Mallards [ Auk [Vol. 78 entiatethe populations. Evidencefrom experimentalhybridization was presentedto show that the geneticbasis for the sexual dimorphismof platyrhynchosis relatively simple, and should not be consideredan importanttaxonomic criterion. Analysisof the sexualbehavior of male platyrhynchosand rubripesfailed to reveal any important differences otherthan slightdifferences in maleresponse thresholds. Electrophoretic studiesof the egg-•vhiteproteins of pla.tyrhynchos,rubripcs, and •ulvigula likewise failed to show any noticeabledifferences in the proportionsand electrophoreticcharacteristics of theseproteins. A discussionof ecological,geographic, and behavioralisolating mecha- nismsof the populations,and the incidenceof hybridizationwhere they overlap resultedin the conclusionthat none of the forms could reason- ably be consideredspecifically distinct from any or all of the others. It is recommendedthat diazi, )Culvigula,and maculosabe consideredgeo- graphic races of Arias platyrhynchos,and that novimexicanabe con- sidereda synonymof diazi. It is also concludedthat rubripes cannot satisfactorilybe considereda distinct speciesand that increasinghybrid- ization indicatesthat it might eventuallybe necessaryto considerit a subspeciesof Anas platyrhynchos.

LITERATURE CITED

AZZE•, J. A., et al. 1908. Fourteenth supplementto the A.O.U. check-list. Auk, •S: 345-399. AU•RlC^• OR•lx•OI•OS•SxS' U•o•r. 1957. Check-list of North American birds. Fifth ed. 691 pp. A•rDERSO•,E. 1949. Introgressivehybridization. John Wiley & Sons, N.Y. ARE•Z^•O, M. and P. Roj•xs. 1956. Aves acuaticasmigratorias en Mexico. I. Instituto Mexicano de Recursos Naturales Renovables, A.C. Mexico, D.F. 270 pp. BE•so•r, D. 1937. Studies in the ecology of the Black Duck, .•lnas rubripes Brewster. M.S. thesis, Cornell U., Ithaca. B•T, A. C. 1923. Life histories of North American wild fowl. Bull. 126, U.S. Nat. Mus. 244 pp. BR•wsx•R, W. 1902. An undescribedform of the Black Duck (Anas obscura). Auk, 19: 183-188. BREWSTER,W. 1906. The birds of the Cambridgeregion of Massachusetts.Mem. Nutt. Orn. Club, Cambridge. 426 pp. BREWSXER,W. 1909. Something more about Black Ducks. Auk, •6: 175-179. Ba•WSXER,W. 1910. Resurrection of the Red-legged Black Duck. Auk, •?: 323-333. Co>•owa,H. B. 1922. A•ms diaxi novimexicana and Grus ame•,icana in Nebraska. Auk, •9: 412. COt:nTER,M. W. 1953. Mallard nesting in Maine. Auk, 70: 490. COt'nT•R,M. W. 1954. Some observations of Mallards in central Maine. ]gull. Me. Aud. Soc., 10: 20-23. Auk ] IOHNSGARD,]•elatio•ships a•ttong Mallards 41 Vol. 78 / '

DELACOUR,J. 1956. The waterfowl of the world. Vol. II. Country Life, London. 232 pp. D•:L^COUR,J. and E. MAYR. 1945. The family Anatidae. Wilson Bull., $7: 3-55. Dw•c,H%J. 1909. The singularcase of the Black Duck of North America. Auk, I•6:422 426. FE•R^m-P•R•Z, F. 1886. Catalogue of animals collected by the Geographical and Exploring Commissiouof the Republic of Mexico, with descriptionsof five new species,and critical remarks on others of great or less rarity or interest by Robert Ridgway. Proc. U.S. Nat. Mus., 9: 125-199. GOO•)^LF•,H. D. 1918. Femiuized male birds. Genetics, 3: 276-299. Goo•)w•N, C.E. 1956. Black Duck and Mallard populations in the Toronto area. Ont. Field Biologist,10: 7-18. H•.L•AYR, C. E. aud H. B. CONOWR. 1948. Catalogue of the birds of the Americas. Pt. I, No. 2. Pub. Field Mus. Nat. Hist., Zool. Ser., Vol. 13. 434 pp. How•[L, A H. 1932. Florida bird life. Florida Dept. of Game and Fresh Water Fish. 579 pp. Hu•, W. 1920. Descriptiouof a new North Americanduck. Auk, 37: 273•74. Hux% F.B. 1949. Geneticsof the fowl. McGraw Hill, N.Y. 590 pp. IMHOF, T. A. 1958. Receut additions to the avifauua of Alabama. Auk, 75: 354-357. JAEgeR,F. 1925. Untersuchungenfiber das diluviale Klima iu Mexico. Zeitschr. Gesell.f/fir Erdkunde,Berlin. No. 9-10: 366--373. JO•NSG^RD,P. A. 1960. A quantitativestudy of the sexual behaviorof Mallards and Black Ducks. Wilson Bull., 71•: 133-155. KoRrm½•%F. H. 1942. The ducks,geese and swansof North America. Wildl. Mgmt. Inst., Washington. 476 pp. LEOPOLD,A. S. 1959. Wildlife of Mexico--The game birds and mammals. U. of Calif. Press, Berkeley. 568 pp. LINDSEY,A.A. 1946. The nesting of the New Mexican Duck. Auk, 63: 483•92. L;5•q•q•Ro,E. 1906. Einige Nomenklaturfragen. Jour. f/fir Orn., $4: 52•533. LoReNz, K. Z. 1941. VergleichendeBewegungsstudien an Anatinen. Jour. f/fir Orn., 119: 194-294. LoR•sz, K.Z. 1951-1953. Comparative studieson the behaviour of the Anatinae. Avic. Mag., $7: 157-182; $8: 8-17, 61-72, 86-94, 172-184; $9: 24-34, 80-91. LoweRY,G. H., JR. 1955. Louisianabirds. La. State Univ. Press. 556 pp. McC^u•, R. A. and H. F. DEUZSCH. 1952. The relationships of certain birds as ind,icatedby their egg white proteins. Auk, 91: 1-18. MANNING, T. H. 1952. Birds of the West James Bay and southern Hudson Bay coasts. Nat. Mus. Can. Bull. 125. 113 pp. O•RHOLS•R, H. C. 1938. The bird life of Louisiana. Louisiana Dept. Cons. Bull., I•8: 1-834. PEr,RS, J. L. 1931. Check-list of birds of the world. Vol. I. Harvard Univ. Press. 354 pp. PHILLIPS, J. C. 1912. A reconsiderationof the Americau Black Ducks with specialreference to certain variations. Auk, I•9: 295-306. PHILLIPS, J. C. 1915. Experimental studies of hybridization among ducks and pheasants. Jour. Exper. Zool., 18: 69-144. PHiLZIPS,J. C. 1920. Habits of the two Black Ducks, .4has rubripes rubripcs and ,4has rubripes tristis. Auk, 37: 289-291. 42 JOHNSGARb,Relationships among Mallards 1' Auk [Vol. 78

PHILLIPS, J. C. 1921. A further report on speciescrosses in birds. Genetics, 6: 366-383. PHILLIPS, J. C. 1924. A natural history of the ducks. Vol. 11. Longmans, London. 409 pp. PIRNIE, M.D. 1928. The plumage changes of certain waterfowl. Ph.D. thesis, Cornell University, Ithaca. RiI)GW^Y, R. 1874. Notes upon American water birds. Amer. Nat., 8: 108-111. Rm(;w^¾, R. 1878. Descriptionsof several new speciesand geographicalraces of birds contained in the collection of the United States Museum. Proc. U.S. Nat. Mus., 1: 247-255. RmGw^¾, R. 1886. Preliminary descriptions of some new speciesof birds from southern Mexico, in the collection of the Mexican Geographical and Exploring Commission. Auk, 3: 331-333. S,xummRs,G. B. 1952. Waterfowl wintering grounds of Mexico. Trans. 17th North American Wildl. Conf., pp. 89-100. S^ummRs, G. B. 1953. The Tule Goose (.rinser albifrons gambelli), Blue Goose (Chen coerulescens)and Mottled Duck (.4nas fulvigula maculosa) added to the list of the birds of Mexico. Auk, 70: 84-85. S.au•im•s, G. B. and D.C. S^u>•mms. 1948. Report on migratory waterfowl investigationsin Mexico. U.S. Fish and Wildlife Service, 107 pp. mimed. Ss>•>•sTz,G. B. 1889. A new speciesof duck from Texas. Auk, 6: 263-265. SHO•TZ, T. M. 1943. Correlation of bill and foot coloring with age and season in the Black Duck. Wilson Bull., 5li: 1-7. Sm•s¾, C. G. 1950. Species formation in the Red-eyed Towhees of Mexico. Univ. Calif. Pub. Zool., •0: 109-194. SmLSY,C. G. 1957. The evolutionaryand taxonomic significanceof sexual dimorphismand hybridization in birds. Condor, •9: 166-191. S•us¾, C. G. 1960. The electrophoreticpatterns of avian egg-white proteins as taxonomic characters. Ibis, 10'•: 215-284. SIm.S¾, C. G. and P. A. Jo•Ir4Sg•a•u. 1959a. Variability in the electrophoretic patterns of avian serum proteins. Condor. 61: 85-95. Smus¾,C. G. and P. A. JoH>•sg^•u. 1959b. An electrophoreticstudy of egg-white proteins in twenty-three breeds of the Domestic Fowl. Amer. Nat., 93: 107-115. Si>•coc•c,J.L. 1957. Quarterlyprogress report for investigationsproject. Florida Waterfowl Investigations. 18 pp. mimed. SI>•gLszo>•,J. R. 1953. Texas coastalwaterfowl survey. F.A. Report Series Xo. 11. Texas Game and Fish Comm., Austin. 128 pp. SINNOTT,E. W., L. C. DUNN, and T. Do•zH^Nsx•v. 1950. Principlesof genetics. McGraw Hill, N.Y. 505 pp. Sem•>•z,A., J•. 1954. Florida bird life. Cowand-McCannInc., N.Y., and the Nat. Audubon Soc. 527 pp. S•w^•z, R. E. 1958. Distribution of the Black Duck. U.S. Fish and Wildlife Circular 51. 8 pp. S•OTZS,V. 1955. Black Duck breedingstudy ends in the Kent Island areas. Md. Tidewaters News, 1'• (4): 1-4. Xuk 1 Jo•Nsc^RD, Relationshipsamong Mallards 43 Vol. 78 i

Tow•-ssNI), C. W. 1905. The birds of Essex County, Massachusetts. Mem. Nutt. Orn. Club, Cambridge. 352 pp. Tow•sE•i), C.W. 1912. The validity of the red-legged subspeciesof Black Duck. Auk, g9: 176-179. WEmM^N•, U. 1956. Verhaltensstudienau der Stockente(Anas platyrhynchosL.). Das Aktionssystem. Zeitschr. f/Jr Tierpsychologie,13: 208-271. WRm•t, B. $. 1954. High tide and an east wind. Stackpole Co., Harrisburg. 162 pp.

The I47ildfowlTrust, Slimbridge,Glos., England.