Taxonomic Aspects of Avian Hybridization

TAXONOMIC ASPECTS OF AVIAN HYBRIDIZATION

LESTER L. SHORT

INSTANCESof natural hybridizationfrequently pose difficult taxonomic problems.A decisionregarding the taxonomicstatus of hybridizingforms shouldbe basedupon an evaluationof all availablebiological information concerningtheir relationship.The paucity of availableinformation usually hamperstaxonomists treating hybridizationamong the higher vertebrates. Also it is difficult or impossibleto subjectsuch animals, and particularly animal populations,to experimentalstudy. Neverthelesstaxonomists must renderdecisions that reflect current availableknowledge. I offer the follow- ing discussionand definitionsin the hopeof facilitating suchdecisions by taxonomists,especially those working with terrestrial vertebrates. Ex- amplesforming the frameworkfor the discussionare drawn from the literature of avian hybridization,including some work of my own. The widely divergent treatment of hybridizing forms evident in the recent ornithological literature demonstratesthe need for some guide- lines. At one extremethis divergenttreatment appearsto reflect a hold- over of typologicalthinking (e.g. Godfrey, 1966; Sutton, 1967), while the oppositeextreme (e.g. Phillips et al., 1964; West, 1962) resultsfrom an overly strict interpretationof the biologicalspecies definition in that different speciesdo not interbreed,and henceinterbreeding forms must be conspecific.The resultsof such divergenttreatment are taxonomicover- splittingin the formercase, and overlumpingin the latter. I definehybridization as the interbreedingof individualsof morphologically and presumablygenetically distinct populations,regardless of the taxonomicstatus of such'populations (seen Anderson, 1949: 61; Short, 1965: 360). This definition of hybridization,essentially that of Mayr (1963: 110), does not differ fundamentally from its broader genetic meaning,in which it can even includeinterbreeding of genotypicallydif- ferent individuals; I have simply restrictedits usage. Included are both sympatricand allopatrichybridization; the latter includessecondary intergradation,but not primary intergradation (see Mayr, 1963: 368 369). Secondary intergradation is distinguished,occasionally with difficulty, from primary intergradation by the increasedvariability of populations within comparedwith thoseoutside the area of intergradation,contrasted with the normal variability of populationswithin and outsidean area of primary intergradation. I define "hybrids" as the productsof sympatric and allopatric hybridization (including secondaryintergradation), and "intergrades"as the productsof primary intergradation. The practice of defining hybridization on the basis of the taxa involved should be dis-

84 The Auk, 86: 84-105. January, 1969 Jan., 1969] Taxonomic Aspectsof Avian Hybridization 85 couraged,because it requiresa taxonomicdecision before "hybridization" can even be employed (see Mayr, 1963: 111; Bigelow, 1965), and it unnecessarilylimits the use of this word. I cannot see the need for en- cumberingtaxonomically a term that is fundamentallynontaxonomic. How distinct must populationsbe to warrant consideringtheir inter- breedingas "hybridization"? The populationsought to be sufficiently distinct to permit recognitionof a hybrid as such. I suggestthat their distinctnessought to approach,equal, or exceedthe levelof distinctnessof subspecies.Criteria for subspeciesare a purely taxonomicmatter, ir- relevantto this point, the ideabeing that the populationsare about as distinct, or more so, than subspecies.For example,designation of an area of primary intergradationdepends on the distinctnessof the populations involved.These populations are usuallyconsidered as subspecies,but that dependson the criteria of the taxonomistviewing the situation. I have designatedas intergradesabove the productsof interbreedingthrough an area of primary intergradation,but this does not preclude the possibility of true hybridizationoccurring between individuals of the sametwo populations. Wanderingbirds. from one populationoccasionally may traverse the area of intergradationand mate with birds of the other population, producingtrue hybrids. Of courseone might obtain individualsof the two populationsand mate them under captive conditionsto produceoffspring that are F • hybrids,not intergrades. While the framework for the considerationof hybridization presented herein is novel, most of the basic ideas are drawn, some insensibly,from the vast literature on speciationand populationgenetics. Especially helpful have been the important works of Mayr (1942a, 1963), Anderson (1949, 1953), Dobzhansky(1951), Stebbins(1959), and Sibley (1957, 1959, 1961). The biologicalinformation forming the basis for a taxonomicdecision in casesof hybridizationshould ideally include: 1) knowledgeof the occurrenceand/or lack of occurrenceand the place of occurrenceof hybrids; 2) knowledgeof the distribution and habitats of the parental forms and hybrids; 3) knowledgeof the relative frequenciesof hybrid and parental phenotypesin the area of hybridization;4) knowledgeof the type of cross- ing (F •, F 2, backcross)that is occurring;5) knowledgeof the occurrence and the extent of any introgression(introgression is gene flow resulting from hybridization;for discussionand definitionsee Short, 1965: 360); and 6) knowledgeof the populationdynamics (especiallyof the spatial distributionof individualsand the populationstructure) of the forms involved. Althoughfull data are often unavailablefor someof theseitems, theremay existinferential bases for their considerationpermitting a tenta- 86 LESTERL. SaORT [Auk, Vol. 86 tive taxonomicevaluation. Other (behaviorial,genetic, ecological) data are of coursedesirable; their availability greatly aids in the taxonomic evaluation,and strengthensthe decisionthat is ultimatelyreached. Hybrids producedunder captive conditionsare not discussedin this paper. I agree with Mayr (1963: 112) that "the mere possibilityof hybridizationin captivity provesnothing as far as speciesstatus is con- cerned." Artificially induced hybridization proves only the existenceof considerablegenetic similarity and compatibility. Of coursestudies of hybridizationunder captive experimental conditions are a usefuladjunct to the study of natural hybridizationin thoseanimals that can be maintained and bred successfully,and in sufficient numbers. The severalcategories of naturalhybridization are discussedin the order from rare hybridizationto frequenthybridization and morecomplex interactions.

INFREQUENT AND RARE HYBRXDIZATION Singleinstances of hybridizationare generallyof limited taxonomicuse- fulness,although they may provide clues to relationship. Many intergeneric arian hybrids involving taxonomicallyvalid genera have been recorded.Other purportedcases involve species of overlysplit generathat shouldbe congeneric,for examplehybrids among species of North Amer- icanhummingbird "genera" (Short and Phillips, 1966). Recentlyreported unique intergenerichybrids include a flycatcherContopus sordidulus X Empidonaxtraillii (Short and Burleigh,1965), a woodwarbler Dendroica strfataX Seiurusnoveboracensis (Short and Robbins,1967) and a hum- mingbird Cynanthus latirostrisX Euge'nesfulgens (Short and Phillips, 1966). In the first two of thesecases the occurrenceof the hybrid tends to corroborategenerally accepted generic relationships and the current adjacentplacement (A.O.U., 1957) of the generainvolved. Study of the third caseled to the suggestedclose relationship between Cynanthus and Eugenes,which previously had not beenconsidered closely related. Thus, lone hybrids betweenspecies representing valid genera may corroborate relationshipsdetermined by other studies,or they may suggestcomparisons not previouslyconsidered. The occurrenceof severalunique intergeneric hybridsinvolving different species of the sametwo generais morestrongly suggestiveof genericrelationship. I have elsewhere(Short and Phillips, 1966) noted as inconsistentwith modernphylogenetic and evolutionary conceptsthe suggestion(Banks and Johnson,1961: 26) that intergeneric hybridizationmight occur more frequently than intragenerichybridization. Aside from serving a taxonomicfunction, these hybrids may help to elucidatethe pattern of evolutionof a group,or they may aid in establish- ing the patternsof evolutionof a trait or traits within a group.. Jan., 1969] Taxonomic Aspectso! Avian Hybridization 87

Unique intragenerichybrids vary in their taxonomicsignificance. Gen- erally they are of little significancein groupsin which hybridsare uncommon. Miller (1955) suggestedthe evolutionaryhistory of certain North Americanwoodpeckers of the genusDendrocopos mainly on the basisof a singlehybrid D. scalarisx D. villosus.Recently, I have found (Short, MS) numbersof hybrids of D. scalarisx D. nuttallii, and two hybridsof D. nuttalliix D. pubescens.The newevidence from these specimens necessitates considerable modification of Miller's views. Hence a loneintrageneric hybrid is of limited useas a basisfor inferringrelation- ships. The amountof hybridizationmay suggestor corroboraterelation- shipsin generasuch as the duck genusAnas, that exhibit, for reasons elucidatedby Sibley (1957), a great deal of hybridization(Johnsgard, 1960). Important here is the rare occurrenceor absenceof hybridsbe- tween certain species,contrasted with more frequently occurring wild hybridsamong other speciesof a genus. One must, of course,evaluate theseoccurrences on th'ebasis of the opportunityof the variousspecies to hybridize,i.e. the degreeof sympatryof potentiallyhybridizing species. More distantrelationship is impliedif few, one,or no hybridsoccur, while closerrelationship is suggestedby occurrenceof crossesinvolving several to many hybrids (Johnsgard,1960). Hybridsmore frequent than the uniqueor rare hybrid,yet lessfrequent than sporadic,i.e. up to 10 or so, may assumegreater taxonomicsignificance. If the speciesrepresent different generaclose relationship is suggested,possibly including the mergingof the genera. Recently reported casesin this categoryare th'oseof North Americanhybrid sparrowsJunco byemallsx Zonotrichiaalbicollis (Short and Simon,1965) and humming- birdsArchilochus alexandri x Calyptecostae (Short and Phillips,1966). When consideredwith other kinds of evidencein the light of instancesof intergenerichybridization between certain of theseand otherspecies of the genera mentioned,their h'ybridizationsuggests generic merger. Another interestingcase is the hybridizationbetween the ducksMergus cucullatus and Bucephalaclangula (Cockrum, 1952; in the U.S. National Museum is an additionalsubadult hybrid male). Althoughthese species represent different genera sometimesplaced in different "subfamilies"(A.O.U., 1957), behavioralstudy (Johnsgard,1961a) has indicated their close relationship.The existenceof severalhybrids and a morphologicalcom- parisonof the two species,and especiallycomparison of the females,lead me to concurwith Johnsgard.The divergencein morphologicalspecializa- tions amongthe ducks (subfamily Anatinae) seemsto. belie their funda- mentallyclose relationship and the recentnessof their originand adaptive radiation (Delacour and Mayr, 1945). 88 LtsrER L. SaORT [Auk, Vol. 86

Existence of several intragenerichybrids between two speciesmay be evidencefor their relativelyclose relationship. In groupswhere hybridization is rare, severalhybrids betweensympatric species may suggestthat they formerlycomprised a superspecies,a superspecies being a "groupof entirely or essentiallyallopatric taxa that were once races of a single speciesbut which now have achieved speciesstatus" (Amadon, 1966). This may be obvious,as in the case of the meadowlarksSturnella magna and S. neglecta(Lanyon, 1966), which have recentlycome broadly into sympatry. A few hybrids have occurredat scatteredpoints within the area of sympatry. Sympatricintrageneric hybrids tend to occurwhere one speciesis at th'eborder of its range,encouraged by restrictedmate choice where that speciesis uncommon.Such specialcircumstances usually are involvedbecause sympatric species normally are effectivelyreproductively isolated,which is partly responsiblefor their sympatry. The woodpeckers Dendrocoposnuttallii and D. pubescensare broadly sympatric in Cali- fornia, which comprisesalmost all of nuttallii's range. Two of their three known hybrids (Short, MS) are from the southernextreme of the range of D. pubescensin California; the locality of the third hybrid (Ridgway, 1887) is unknown. South of the range of pubescens,nuttallii replaces pubescensin riparian vegetation like that occupiedby the latter where they are sympatric. These two woodpeckersprobably formerly comprised a superspecies.The formerexistence of a superspeciesmay be lessobvious, as perhapsin the broadly sympatricand sporadicallyhybridizing tanagers Thraupis palmarum and T. virens (Haverschmidt, 1966). Likewise in a groupprone to hybridize like the grouse(Tetraoninae; see Sibley, 1957), even more numeroushybrids in sympatry may suggestthe former existence of superspecies.An exampleis the caseof the Eurasian capercaillie(Tetrao urogallus) and black grouse (T. tetrix), which interact ecologicallyand hybridize occasionallyin their extensivearea of sympatry, althougheach todaycomprises a superspecies of its own with related Asian species (Short, 1967). Existenceof a superspeciesmay be directly inferred from the occurrenceof two to several hybrids in a contact between allopatric con- genericspecies. An exampleis the case of two hybrids of the tanagers Piranga ludovicianaand P. olivacea,respectively, of westernand eastern North American (Tordoff, 1950; Mengel, 1963). When discussingth'e numbersof natural hybrids, one must recognize the vastly greateropportunity to recoverhybrids of waterfowland other gamebirds,compared with otherbirds. Fully half the populationof certain game speciesmay be killed yearly by hunters,and hencea considerable portion of obvioushybrids (mainly males) are recovered.The discussion hereinis mainly concernedwith nongamespecies. Jan., 1969] Taxonomic Aspects of Avian Hybridization 89

ZONES OF OVERLAP AND HYBRIDIZATION The term "hybrid zone" has been looselyused to designateany area of secondarycontact in which h'ybridsoccur. I restrict a "hybrid zone" (see below) to include an area of hybridizationwhere only hybrids occur, as distinguishedfrom a "zoneof overlapand hybridization." I prefer to include both of these categorieswithin the categoryof secondaryintergradation, thus retaining the latter general term for all situations involving hybridization and backcrossingbetween allopatric forms in a secondarycontact. A zone of overlap and hybridization is an area of secondaryintergradation occupied by numeroushybrids and both parental formsas well. The parentalphenotypes must occurin numberssufficient to precludetheir representingextreme hybrid phenotypes(recombinants). I arbitrarilyconsider the occurrenceof both parentalphenotypes in num- bersgreater than 5 per cent, taken together,of the populationin the zone of overlap and hybridization as indicating that at least some parental phenotypesnormally comprise a part of this population.The use of 5 per cent has no function other than to insure the presenceof parental phenotypes;if these otherwisecan be demonstrated,their per cent occurrence is unimportant. A zone of overlap and hybridization has two taxonomically significant features. The existence therein of numerous parental phenotypesshould result from impedimentsto gene flow, pre- sumably the action of partial isolating mechanisms(Short, 1965: 418; Bigelow,1965). Immigrationof parentalphenotypes from outsidethe zone alsomay take placewhen hybrids are at a selectivedisadvantage. Secondly, phenotypesof the parentalforms are sympatricin this zone; their reactions can include primary (F •) hybridization, competition,and reinforcement or breakdownof isolatingmechanisms (Table 1.) Other than the indirectevidence for partial isolatingmechanisms providedby the existenceof a zoneof overlapand hybridization,there appears to be no logical basis for deciding the taxonomic status of the interacting forms. The amount of hybridizationand the relative proportionof hybrids and parental phenotypes.in the zone do not provide useful, nonarbitrary criteria for a taxonomicdecision, for both may be influencedby factors suchas the size of the contact; henceneither necessarilyis correlatedwith the efficacyof partial isolatingmechanisms. The presenceand maintenance of parentalphenotypes in the zoneof overlapand hybridizationappears to constitute a nonarbitrary basis for a taxonomic decision. The taxonomic value of this phenomenonis enhancedby the direct relationshipit seems to havewith functioningreproductive isolating mechanisms. I restrictthe term "semispecies"(Amadon, 1966) to thoseforms actually or potentially capable of forming a zone of overlap and hybridization. 90 LESTER L. SHORT [Auk, Vol. 86

TABLE 1 A CLASSIFICATION OF So!viE HYBRIDIZING FOR!VrS

Forms involved Distribution Interactions

Subspecies,or subspecies contiguouslyor strictly primary intergradation, or groups of polytypic species allopatric potentially capable of so doing subspecies,or subspecies strictly allopatric hybridize in hybrid zone groups of polytypic species (or potentially capable of so doing)

...... TAXONO!V[IC SPECIES BORDER ...... semispecies (allospecies) basically allopatric, but form zone o/ overlap and some sympatry evident hybridization (or potenti- or possible ally capable of so doing); competition; reinforcement of isolating mechanisms allospeciesof a basically allopatric; some rare, inconsequential or no superspecies or no sympatry hybridization; effective isolating mechanisms, rarely breaking down; competition if sympatric related, but not Sympatric or allopatric rare or no hybridization; allospecific species effective isolating mechanisms, very rarely breaking down; competition or not when sympatric

These semispeciestruly combine attributes of populations at both the specieslevel, suchas morphologicaldistinctness and reproductiveisolation, even if only partial, and at the intraspecieslevel, such as the ability to interbreed. On the basis of suggestiveevidence for existenceof partial isolating mechanismsin situations involving semispecies,I recommend that semispeciesbe consideredtaxonomically as species(Table 1). Semi- speciesare one type of allospeciescomprising a superspecies(Amadon, 1966). Other superspeciesare comprisedof allospeciesthat infrequently or never hybridize in areas of secondary contact or limited areas of sympatry, or of allospeciesthat are completely allopatric. All types of allospeciesare geographicallyrepresentative forms taxonomically con- sideredspecies (Areadon, 1966). Zonesof overlap and hybridizationmay be stable, but often they may changedynamically, either on a broad level or locally. The frequencyof suchchanges suggests a need for continuinganalysis and reexaminationof all zonesof overlap and hybridization. In his discussionof the invasion by the woodpeckerDendrocopos syriacus of the European range of its closerelative D. •najor from southeasternEurope earlier in this century, Bauer (1958) showsthat hybridization occurred in a narrow belt along Jan., 1969] TaxonomicAspects of Avian Hybridization 91 the advancingzone of overlapand hybridizationas D. syriacusexpanded its rangeto the northwest.As the front passeda givenarea, hybridization diminished,then ceased,and th'etwo speciescame to existin sympatry. Hybridizationin this caseseems due to restrictedmate choiceamong im- migrantsof D. syriacus;interbreeding diminishes as its populationreaches sufficientsize to make availableenough • conspecific mates. A similar caseinvolves the titmice Parus caeruleusand P. cyanus(Vaurie, 1957). Late in the past centurythe AsianP. cyanusmassively invaded the east Europeanand west Asian range of the largelyEuropean P. caeruleus.Since the 1880sthe rangeof P. cyanushas receded somewhat, but a vastarea of overlapstill exists.Hybridization, apparently widespread during the period of expansion,has diminished and nowtakes place only in th'enorthern part of the zoneof overlap(Vaurie, 1957). In thesetwo casesthe zoneof overlap and hybridizationhas becomegreatly restricted,while the former regionwhere the zoneoccurred is now an area of sympatry,or simplya zone of overlap. The gulls Larus argentatus and L. hyperboreusare sympatric without interbreeding over much of the northern Holarctic region. Icelandwas formerlyoccupied by L. hyp.erboreus,but not by L. argentatus.The recent arrival of tke latter as a breedingbird in Iceland has resultedin widespreadhybridization (A. Ingolfsson,pers. comm.), despitetheir sympatryelsewhere. All theseexamples are specialcases of zonesof overlap and hybridization--they might also be consideredcases of combinationsof situations,as shownbelow. As the formsinvolved are basicallyreproductively isolated, they are species.I would not designate themas semispecies; they are formersemispecies, able to coexistover large areaswithout interbreeding. Large, broad zonesof overlap and hybridizationare not infrequent. (The adjectives"large," "small," and "moderate"herein pertain to the overall extent or area of the zone,while "broad" and "narrow" describe the widthof the zoneat a specifiedpoint, or the averagewidth of the zone if no point is specified.)The warblersVermivora pinus and V. chrysoptera extensively overlap and hybridize in eastern North America. Parental phenotypesapparently are maintainedthroughout the zoneof overlapand hybridization,and intro.gression is taking place (Short, 1963). In the same regionthe ducksAnas platyrhynchos and A. rubripesbroadly overlap and hybridize,with parentalphenotypes remaining more common than hybrids (Johnsgard,1961b, 1967). Likewise,the kingfishersCeyx erithacusand C. rufidorsusinterbreed and overlapover a largearea includingmost of Sumatra,Borneo, and the Malay Peninsula(Sims, 1959). In mostcases of this type the parentalphenotypes and recombinantsclosely resembling themcomprise a majorityof the populationin muchof the area of over- 92 LESTERL. SUORT [Auk, Vol. 86 lap and hybridization.These forms are semispeciesthat I believeshould be accordedthe taxonomicstatus of species. Moderatezones of overlapand hybridizationexist in many areas. In central North America broad zones of overlap and hybridization exist betweenthe grosbeaksPheucticus ludovicianus and P. melanocephalusand the buntingsPasserina cs'anea and P. amoena.Although broad, the zones are limitedin extentbecause these species inhabit only river valleyswithin the vast central Great Plains area the zone encompasses.The maximum width of the grosbeakzone is about 200 miles (West, 1962; ShOrt,un- publisheddata from recently collectedNebraska specimens).Within this zoneboth parentaland hybrid phenotypesare encountered.Slight introgressionis detectableimmediately adjacent to the zone of overlap and hybridization. The buntingsoverlap and hybridize along river valleys in the sameregion (Sibley and Sh'ort,1959a), but the zone is broader, about 400 mileswide. The easternsemispecies, P. cyanea,recently has bred sporadicallyfar within the range of P. amoena in California, Arizona,and Utah (somecitations are in Sibley and Short, 1959a). Sig- nificantly, most pioneer birds were apparently phenotypicallypure P. cyanea, and some individuals securedmates of their own speciesdespite being far beyondthe species'normal range. Examplesof small and narrow zonesof overlap and hybridization are providedby semispeciesof gulls (Larus glaucescensand L. argentatus) and chickadees(Parus atricapillusand P. carolinensis). The gulls meet in southernAlaska where they sometimesoccur in mixed colonies.In at least somecolonies considerable hybridization occurs, altho.ugh parental phenotypesapparently are maintained despite interbreeding (Williamson and Peyton, 1963). The rangesof the chickadeesare sporadicallycon- tiguousfrom the MississippiRiver to the Atlantic Coastof North' America. Known contactsmainly involveno overlap (Tanner, 1952; Brewer, 1963), but slight overlapwith limited hybridizationoccurs in southernIllinois.

HYBRID ZONES A hybrid zone is an area occupiedby a hybrid populationconnecting two parentalgene pools (Short, 1965 and above). The parentalphenotypes togethercomprise less than 5 per cent of the hybrid zone population,i.e. they occurso infrequentlythat they might be due to chancerecombination. Occasionalimmigrants or wanderersfrom the range of one or the other parental form could accountfor someoccurrence of parental phenotypes, which are virtually or actually lacking from most hybrid zones. Parental phenotypesdo not occur togetherin the hybrid zone; rather, each is limited usually to that edgeof the hybrid zone appropriatefor its adjacent Jan., 1969] Taxonomic Aspects of Avian Hybridization 93 distribution. Each border of the zone is marked by an increasein the parental phenotypeof the immediatelyadjacent form, and a decrease in hybrid and introgressantindividuals. A significantfeature of the hybrid zone is that the hybrid population simultaneouslyboth connectsand separatesthe parental populations; the latter henceare completelyallopatric (Table 1). The opportunityfor primary (F l') hybridizationis lacking becauseindividuals of the parental formsare no longerdirectly in contact,as they were when the secondary contact initially was established. Reproductiveisolating mechanismsare nonexistentor ineffectual.Analysis of populationsadjacent to the hybrid zone usually will uncoverevidence of introgression.The geneticcontact betweenthe hybrid population and the parental form insuresoccurrence of introgression,whether or not it is detectable.Forms connectedby a hybrid zone are consideredconspecific because of lack or ineffectiveness of reproductiveisolating mechanisms. One is often reluctantto acceptconspecificity of two formsinterbreeding in a hybrid zone when one or both of the forms are polytypic. As free interbreedingand lack of reproductiveisolation demand conspecificity of the formsinvolved, use of the term "subspeciesgroups" is helpful (Table 1). Subspeciescan vary tremendouslyin their degreeof morphological divergence.Also, they may be entirely allopatric, broadly connected throughareas of primary intergradationinvolving gradual or step-clines, or connectedthrough hybrid zones. A group of slightly or moderately differentiated clinal races may comprisea subspeciesgroup having a hybrid zone with anothersuch' group, as in the flickers (Short, 1965) discussedbelow. Of coursepolytypic specieswith numerousraces may havesubspecies groups. even without hybrid zones. The often considerablemorphological differences between forms interbreedingin a hybrid zone is no reasonfor reluctanceto mergethem, for the morphologicalfeatures by which they differ have no significancewith regardto their reproductiveisolation. Why shouldwe questionthe con- specificityof two morphologicallydivergent forms, when individualsof theseforms accept each other as conspecific?When this acceptanceresults in the formationof a hybrid zone, the hybrid populationbecomes a con- nectinglink betweenthe parental forms,which are therebybuffered from direct contact. The formationof an intermediaryhybrid populationattests to the successof the interbreeding,the lack of reproductiveisolation of the parental forms, and the insignificanceof the morphologicaldifferences betweenthe parentalforms with regardto reproductiveisolation. Because the evolutionof reproductiveisolating mechanisms is related only in- directly to morphologicaldivergence, the level of morphologicaldistinct- 94 LESXE• L. Sao•x [Auk, Vol. 86 nessof interbreedingsubspecies groups can approach, equal, or occasionally exceedthat achievedby semispecies. The best test for the existenceof reproductiveisolating mechanisms is a large contactbetween two populations.When a large contactensues, there is a direct threat of geneticswamping of the parentalgene pools throughhybridization and introgression.Any existingincipient isolating mechanismsshould be enhancedunder these circumstances. Sibley (1961) discussesthe reinforcementof isolating mechanismsthrough selection acting againsthybrids; reinforcementis a part of Brown and Wilson's (1956) "character displacement." Large and broad avian hybrid zonesare evident on most of the con- tinents. The morphologicallydivergent auratus and caJer subspecies groupsof the North Americanflicker (Colaptesauratus) form a vast mid- continentaland northwesternhybrid zone (Short, 1965). Introgression affects parental populationsas far from the zone as the east and west coasts of North America. Likewise two forms of grackle (Quiscalus quiscula)form a large hybrid zoneover much of easternNorth America (Chapman, 1940; Huntington, 1952). The flowerpeckersPardalotus melanocephalusand P. "uropygialis" form such a zone in northern Australia (Salomonsen,1961); for severalother fine examplesof this phenomenonsee Keast (1961). Th'e formsinvolved in thesehybrid zones are obviouslyconspecific, as no isolatingmechanisms exist and the parental gene pools are firmly connected.This does not mean that these forms necessarilywill merge completely. Very often, as in most continental hybrid zones,the large parental populationsoccupy an immensearea, allowingfor isolationby distanceand gradualfiltering (seebelow) of intro- gressedgenes away from the hybrid zones. Just as variable clinesoften are evident in casesof primary intergradation,the differential sorting action of natural selectionwill fashion variable clines for introgressant genesand gene combinations.In the flickers mentionedabove, some featuresby which the hybridizingforms differ are sharplydinal within the hybrid zone (e.g. nuchalpatch; seeShort, 1965), while others (e.g. shaft color) are step-dinalwithin the hybrid zone; theseclines assume a gradual slopeoutside the hybrid zone. Narrownessof a hybrid zone is not related necessarilyto any basic geneticincompatibility of two hybridizing forms. Narrower zonesoccur when differencesbetween the parental forms are few. Genetic gradients for variouscharacters across and beyond the hybrid zone inevitably will vary with the selectiveforces regulating gene flow, as noted above. This differentialaction of natural selectionoperating with respectto introgression out of a hybrid zone,which Bigelow (1965) ignoredin his contrary Jan., 1969] Taxonomlc Aspects of Avian Hybridization 95 view of the narrow hybrid zone between Corvus corone and C. "cornix" (below), has no bearing on reproductiveisolation of the two forms; hence it has no bearing on their taxonomic status. When individuals of two forms can interbreed, and their hybrid offspring and successivehybrid populationsare genetically connectedwith both forms, then these forms patently are not reproductivelyisolated species. There exist a few large or moderately large but narrow hybrid zones, notably the famous case of the Eurasian crows Corvus corone and C. "cornix" (Meise, 1928; Vaurie, 1954). These crows form a narrow hybrid zone acrosscentral and north•vesternEurope; little introgression is apparent beyond the zone. A third crow (C. "orientalis") occupies northeasternAsia and forms a broader hybrid zone with C. "cornix" in West Siberia (Mayr, 1942a; Vaurie, 1954). Previousworkers discussing this problem generally overlookedthe great morphologicalsimilarity of thesecrows in virtually all traits exceptthe one color feature by which they differ strikingly, and upon which we have had to depend for evidenceof introgression. The narrownessof this hybrid zone simply may reflect the dearth of morphologicalcharacters available for analysis. Even with free interbreedingthrough a hybrid zone selectionmay fashion a sharp step-cline(see above) for certain genesand gene combinations,perhaps including those responsiblefor the gray versus black backs of the two crows. Advantageousgenes and genesnot stronglydisadvantageous may pass unimpeded through the hybrid zone. For example, morph•ological analysis of a hybrid zone in lizards (Cnemidophorus)of southwestern North America indicatedvery limited introgression,but use of biochemical techniquesproved that introgressionwas actually extensive (Dessauer et al., 1962). As the crow populationsare connectedby a totally hybrid population, and there is no evidence of isolating mechanismsacting to limit the hybridization,we are obligedto considerthem conspecific. A broad h'ybrid zone of moderate extent is found in the Great Plains of North America between the galbula and bullockii groups of orioles (Icterus galbula) (Sibley and Short, 1964). The zoneis roughly200 miles wide, but as it occursprincipally along river valleys its extent is limited. Introgressionis evidentimmediately adjacent to the hybrid zone. A moderate, narrow hybrid zone is that connectingthe bicolorand atricristatus groupsof titmice (Parus bicolor) in east-centralTexas (Dixon, 1955). A true hybrid zone 25 to 50 miles in width occursin five areas.of contact betweenthem. Marked introgressionis not evident. Hybridization in small areas of secondarycontact inevitably places less stresson the parental populationsthan occursin situations.involving massivecontact. Small hybrid zoneshence pose more difficult problems 96 LESTERL. SHORT [Auk, Vol. 86 for the taxonomist than do large hybrid zones. A small hybrid zone involving extensivehybridization but little introgressionmay result from chancetopographical or habitat conditions,restricted mate choice,or other special local factors. Thus a small contact may not test effectively the parental forms' capacity to interbreed. The existenceof a small hybrid zone demonstratesthat reproductiveisolating mechanismsare not fully effective, but the presenceor absenceof partial isolating mechanisms often remains a moot point. Small hybrid zones may persist indefinitely. This simply may reflect their small size and limited genetic effects. There is a possibility that minor genetic contact may benefit populations,whether or not partial isolatingmechanisms exist (Andersonand Stebbins,1954; Stebbins,1959; Short, 1965). Counter selectionfrom this sourcecould balance the effects of weak isolating mechanismsor even negate them (Stebbins and Daly, 1961), producing a stalemate. The problem of establishingthe presenceor absenceof isolating mechanismsin situationsinvolving small hybrid zonesis a seriousone. Never- theless,when no evidenceof isolatingmechanisms exists in such situations we are obliged tentatively to consider the interbreeding forms as conspecific. Converselythe demonstrationof isolating mechanismsfunction- ing in small contact situationssupports consideration of the interbreeding forms as species. A small, narrow hybrid zone existsin Colombiabetween the tanagers Ramphocelusfiammigerus and R. "icteronotus"(Sibley, 1958). The hybrid zoneis but 10 or 12 mileswide, and a total changefrom one "pure" parental population to the other takes place within about 50 miles. A hybrid zone about 16 miles acrossconnects the wrens Campylorhynchus rufinuchahumilis and C. r. nigricaudatusin Chiapas,Mexico (Selander, 1964, 1965; see also Short, 1966). The Colombiantoucans Pteroglossus torquatus torquatus and P. t. sanguineusform a hybrid zone only about 12 miles across(Haffer, 1967). Finally a very small hybrid zone connects the birds-of-paradiseAstrapia stephaniae and A. "mayeri" (Mayr and Gilliard, 1952). Introgressionwas detectedin a population of the latter about 15 miles from the hybrid zone. Regardless of the size of these hybrid zones, they are indicative of free interbreeding; the forms involved thereforeshould be consideredconspecific.

HYBRID SWARMS Hybrid swarmsare hybrid populationsof small to large size that are nearly or actually geneticallyisolated from populationsof the parental forms. Their lack of contact with the parental stock hampersevaluating these populationstaxonomically. It is usually impossibleto establish Jan., 1969] Taxonomic Aspects o! Avian Hybridization 97 whether the swarm resulted from breakdown of isolating mechanisms under the exigenciesof local selectiveforces, or whether suchmechanisms simply were lacking. On one hand the occurrenceof several hybrid swarmsinvolving the same two forms suggeststhat their isolatingmech- anismsare weak and probably ineffective,or nonexistent. On the other hand,populations which are partly sympatricor occuradjacently without interbreedingmay form a lone hybrid swarm; this would suggestlocal breakdownof extant isolating mechanismsas the cause of the hybrid swarm. Evidence for the occurrenceof introgressionshould be sought in populations of the parental forms nearestthe hybrid swarm. The existence of introgressionmay be usedas a basis for consideringthe forms involved as conspecific,while lack of introgressionwould be inconclusive.Evidence for the stabilizationof a single hybrid swarm also would warrant conspecificstatus for the parental forms. The basis for this conclusionis that stabilizationproves the parental genotypes.are sufficiently com- patible and alike to permit selectionof adaptivelysatisfactory genotypes from their recombinants,at least in oneregion. Isolatedsingle instances of hybrid swarmsnot exhibiting stabilizationremain a problem. A hybrid swarmdoes prove that the formsinvolved can interbreed, at leastin certain circumstances.Lacking other concreteevidence, especially sympatry elsewhere,for the presenceor absenceor reproductiveisolating mechanisms, I suggestthat the lone instanceof a hybrid swarm providesa tentative, albeit weak, basisfor mergingthe formsin question.This action is more justifiablethan assigningeach of them full speciesstatus, because existence of the hybridswarm proves that they havenot achievedcomplete reproductive isolation,or at any rate had not at the time the swarmbecame established. Various insular populationsof the highly polymorphicPachycephala pectoralisof the SouthwestPacific (approximately 70 races) comprise hybrid swarms (Mayr, 1942a; Galbraith, 1956). These apparently re- sultedfrom colonizationof certainislands by individualsrepresenting two races. Most of these hybrid swarmshave retained the high variability typical of hybrid populations.Solomonsen (1950) regardedthe Iceland populationof redpollsas a stabilizedhybrid swarmof Carduelisflammea x C. hornemanni,although the two speciesare sympatricelsewhere. Other examples(Colaptes auratus, Pipilo erythrophthalmus)of hybrid swarms are discussed below.

COMBinATiOnS OF S•TUA•O•S Variouscombinations of situationssometimes occur between populations, includingdifferent types of hybrid reactions. These combinationsoccur 98 LESTERL. SaORT [Auk, Vol. 86 becauseof genetic differencesamong various populationsof two forms, reflectingthe simplefact that evolutiondoes not proceedat the samerate in all populationsof a species. The often difficult taxonomicproblems presentedby these combinationsought to be solvedon the basis of the extent of reproductiveisolation. Thus, existenceof a large hybrid zone would more than offset the presenceelsewhere of a small area of overlap. Converselya large area of sympatrywithout interbreedingwould override existenceelsewhere of a hybrid swarm or a small hybrid zone. Neverthe- less instancesoccur, like those of circular overlap discussedbelow, that involve both a large area of overlapand a large h'ybridzone. Each case must be decidedafter careful analysisof every situation and an evaluation of the sum of the reactions. Combinations of situations are sufficiently complicatedto merit discussingseveral examples. Miller's (1941) intensivestudy of the North Americanjuncos led him to recognizea number of species,notwithstanding hybridization among them. The easternJunco "byemalls" forms a large hybrid zone with the westernJ. "oreganus."Miller attemptedto showthat the hybrid population is a stablehybrid race ("cismontanus")affiliated witk hyemalis. The parental populationsmainly involved in the interbreedingare virtually identical in those mensural characters that Miller used to demonstrate the supposedstability of the hybrid race. A hybrid population should not showincreased variability in featuresnot differing in the parental forms; hencethe supposedstability of "cismontanus"is spurious.Miller's analysis of "cismontanus"was based on only two samplesfrom the vast area ascribed to that form. He admitted (1941: 342-343) that "actual" hybrids cannot in fact be distinguishedfrom individualsrepresenting "cismontanus." Critical examination (Sh'ort, MS) of large numbers of northwesternjuncos, many taken after Miller's study, indicates that "cismontanus"actually representsa vast hybrid zonepopulation, and that considerableintrogression occurs beyond this zone. SinceMiller's (1941) work, Dickinson (1953: 128-131) and Phillips (1961: 372 377) have studiedthe interbreedingof hyemalisand oreganusthrough "cismontanus"; both concludedthat the interbreedingof theseforms shows them to be conspecific.The SouthernRocky Mountain Junco"caniceps" forms isolated hybrid swarmswith the race thurberi of J. "oreganus"in easternCalifornia and Nevada, and a moderatehybrid zone with mearnsiof J. "oreganus" in Wyoming, Idaho, and Utah. Finally, although Junco "aikeni" of the Black Hills regionis geographicallyisolated from otherjunco populations, severalhybrids (aikeni X mearnsi) were obtainedfrom the populationof aikeni geographicallynearest to another junco (J. "oreganus"mearnsi). I concurwith the suggestionsof Mayr (1942b), Dickinson (1953), and Jan., 1969] Taxonomic Aspects of Avian Hybridization 99

Phillips (1961; and in Phillips et al., 1964) that the oreganus,caniceps, aikeni, and hyemalisgroups ( = subspeciesgroups) of juncos shouldbe consideredconspecific (: Junco byemalls). The genus Junco actually appearsto be a superspecies,a view Mayr (1942b: 379) suggested;the componentspecies are J. byemalls,J. phaeonotusand J. vulcani. These are probablycongeneric with Zonotrichia (Short and Simon,1965). Three morphologically divergent forms of North American flickers (Coloptesauratus), consideredby someas separatespecies, hybridize in North America (Short, 1965). The eastern auratus and western cafer subspeciesgroups form a large hybrid zoneacross North America,as mentioned above. The third form (chrysoidesgroup) inhabits the south- westerndesert region, and it formsboth variable and semi-stabilizedhybrid swarmswith the western cafer group in Arizona and Baja California (Short, 1965; Short and Banks, 1965). In addition, it forms a single tenuous narrow hybrid zone with flickers of the ca•er group along one river valley in Arizona. The existenceof the auratus x ca•er hybrid zone, the narrowca•er • chrysoideshybrid zone,and evidencefor somestabiliza- tion in a hybrid swarm of ca•er • chrysoidesjustify conspecificstatus of these forms. The towhee (Pipilo erythrophthalmus)of most of North America and the towhee(P. "ocai") of the Mexican highlandsinteract in theefollowing threeways (Sibley, 1950, 1954; Sibleyand West, 1958; Sibleyand Sibley, 1964): 1) they form semiconnected,variable hybrid swarmsthat almost constitutea hybrid zone in westernMexico; 2) they form two small but moderatelybroad hybrid zonesin south-centraland eastern Mexico; and 3) they are sympatricwith little or no interbreedingin severalparts of Oaxaca and one area in eastern Mexico. Introõressionis evident about the hybrid zonesand is affecting populationsin at least one of the three areas of sympatry. When hybridization and introgressionaffect major populationsof two forms,these cannot be consideredfully reproductively isolated. It is not fitting to designatetwo forms as specieswhen the extent of their interbreedingand introgressionoutweighs the extent of their sympatry. Hence the towheesare best consideredas conspecific. The bulbulsPycnonotus cafer and P. leucogenysbroadly overlapin India and Pakistan. Scatteredhybrids occur in the overlap area near the western end of caJer'srange, while two localhybrid swarmsalso occur in that region (Bannu and Kohat in northwesternPakistan; Sibley and Short, 1959b). These bulbuls are evidently species exhibiting occasionalbreakdown of reproductiveisolation, probably only where one speciesis uncommonand the choice of mates is therefore limited. Chapin (1948) reporteddifferent reactionsbetween the African paradise 100 LESTERL. S•ORT [Auk, Vol. 86 flycatchersTerpsiphone ru/iventer and T. viridis. The subspeciesT. r. nigricepsand T. v. viridis occur sympatricallywithout interbreedingin someareas, but in Gambia and PortugueseGuinea they formed a hybrid zone that apparentlyhas becomea relatively stabilizedhybrid population, designatedas T. r. ru/iventer. Other racesof thesespecies (T. r. somereni and T. v. ferreti) form a zone of overlapand hybridizationin Uganda, where the hybridsappear less numerous than individualsof the parental types. It seemsbest to considerthese flycatchers as separatespecies.. The sparrowsPasser domesticusand P. hispaniolensisare broadly sympatric in Spain, northwesternAfrica, the Balkans, Asia Minor, and south-centralAsia. Yet they have formeda large stabilizedhybrid population in Italy, P. d. italiae, considereda race of P. domesticusbecause of its hybrid zone with that form (Meise, 1936; Vaurie, 1956, 1959), small stabilizedand nonstabilizedhybrid swarmsin centralAlgeria and southern Tunisia, and a zone of overlap and hybridization in northern Algeria (Meise, 1936). Although genetic exchangebetween these forms indicates that they comprisea singleevolutionary unit, the great area over which they coexistwithout interbreedingmerits assigningspecific status to each. Three taxa of sapsuckers,once regarded as separatespecies, were merged in Sphyrapicusvarius followingstudies by Howell (1952). The three formsare: the northeasternS. "varius," the Rocky Mountain S. "nuchalis," and the westernand northwesternS. "ruber." Two races of ruber (ruber, daggetti) hybridize with nuchalis. In one case (daggettix nuchalis) a small,narrow zone of overlapand hybridizationhas resulted,with parental phenotypesas numerousas the hybrids. The other case (ruber x nuchalis) involves a moderate, narrow zone of overlap and hybridization with few hybrids. Sphyrapicus"nuchalis" and S. "varius" make contactin Alberta, but little is known of their reactionsexcept that hybrids are few and interbreeding apparently is limited. Sphyrapicus "ruber" and S. "varius" overlapin British Columbia,where hybrids rarely are produceddespite the considerableextent of the area of overlap (Howell, 1952; Dickinson, 1953). Thus in every situationinvolving these sapsuckers gene exchange is limited; barriers seemto prevent free interbreeding. As no basis for mergingthem exists,I considerthe sapsuckersS. varius, S. nuchalisand S. ruber as species.These semispecies comprise the superspeciesS. varius.

CASES OF CIRCULAR OVERLAP These situationsare very complexand all known casesrequire further study before even tentative taxonomic decisionscan be rendered. It is crucial that all reactions between two forms be documented before under- takinga full taxonomicevaluation. For exampleVaurie and Snow(1957) Jan., 1969] TaxonomicAspects of Avian Hybridization 101 mentionevidence of interbreedingbetween the terminalpopulations of the major and minor groupsof Parus major; the terminal forms involvedin thisfamous case of circularoverlap thus do notoverlap without interbreeding,as previously had been supposed (Mayr, 1942a: 182). Cain (1955) andKeast (1961) discusseda case of circularoverlap in Platycercusparrots of Australia.A terminalform Platycercuselegans elegansintergrades with P. e. adelaidae,which in turnhybridizes with the otherterminal form flaveolus. One hybridexists between flaveo.lus and P. e. elegans;otherwise these forms contact or overlapslightly without interbreeding.The crux of the problemis a need for analysisof the adelaidaex [laveolusinterbreeding. If a zoneof overlapand hybridization exists,Jlaveolus should be considereda separatespecies. This wouldnot involvecircular overlap. Should free interbreeding occur through a hybrid zone,circular overlap would exist. I would then considerthe three forms conspecificon the basisof the geneticexchange among the populations andrelatively small overlap of Jlaveolusand elegans. The circularoverlap involving gulls of the Larusargentatus group is exceedinglycomplex, despite the recentwork of Smith (1966) and others (citedby Mayr, 1963: 508-510). Thereis not evenagreement on which formscomprise the complex.Mayr (1963) includesL. glaucoidesin the argentatuscomplex, but excludesL. glaucescensand L. schistisagus.Re- cent studies(Williamson and Peyton, 1963; Vaurie, 1965: 735) have shownthat the lattertwo formshybridize with andare closelyrelated to L. argentatus.Stresemann and Timof•eff-Ressovsky (1947) demonstrated that gapsexist between Eurasian populations of this complexthat pre- viouslywere thought (Mayr, 1942a) to be continuous.Any taxonomic decisionin this casewill be prematureuntil thereis further documentation of all existingreactions. Othercases of circularoverlap Mayr (1963) citesalso are fraughtwith problems.These especially involve the continuity of linksin thecircle, and theexact reaction of theterminal populations. When true circular overlap is demonstrable,I would consider the terminalforms as conspecificif their geneticconnection is continuousthrough' areas of primary intergradationand hybridzones. Major gapsin the continuityof the chain, includingthe existenceof zonesof overlapand hybridization, would neces- sitatedesignation of severalspecies; the situationwould not be oneof true circular overlap.

ACI•I•OWLEDGMEI•ITS DeanAmadon, Walter Bock,Eugene Eisenmann, and Ernst Mayr kindly read this manuscript,and I am grateful for their suggestionsbenefiting it. Of courseI take full responsibilityfor the views expressedherein. Studies of the author that con- tributedto this report havereceived support from the ChapmanMemorial Fund of 102 LESTERL. SHORT [Auk, Vol. 86 the American Museum of Natural History, the National ScienceFoundation (Grant GB-5891), Adelphi University, and the U.S. Fish and Wildlife Service.

SUMMARY Instancesof natural hybridizationin birds provide the frameworkfor a discussionof guidelinesfor taxonomica]lyappraising various phonomena of hybridization. Hybridization and semispeciesare defined, and the superspeciesconcept is relatedto hybrid situations. The occurrenceof unique hybrids is generally of little taxonomiccon- sequence,but these may corroborate or suggestrelationships. Several hybrids betweenspecies of different genera suggestgeneric relationship, while several intragenerichybrids may suggestthe existenceof super- speciesor former superspecies. Zones of overlap and hybridization occur between semispeciesthat are taxonomicallyspecies. The occurrenceof parentalphenotypes with hybrids indicates the operation of partial isolating mechanismsrestricting gene exchange.In such casesthe parental forms are actually sympatric. On the other hand, hybrid zonescontain only hybrids, and they connectas well as separateconspecific populations (subspecies or subspeciesgroups). The fact that the hybrid populationboth' connectsand separatesthe allopatric parental forms attests to their lack of reproductiveisolation. Unique hybrid swarmspose serious problems that are discussed. Combinationsof relationships,including hybridization, are discussedand examplesare provided. In suchcases the extentof hybridizationand introgressionis weighed against the degree of reproductiveisolation and the amount of sympatry exhibitedby the populationsinvolved. Instancesof circular overlap are complicatedand they seemto require further study before they can be taxonomicallyresolved.

LITERATURE CITED A•ADON, D. 1966. The superspeciesconcept. Syst. Zool., 15: 245-249. A•ERICAN ORNITttOLOGISTS' UNION. 1957. Check-list of North American birds. Fifth Ed. Baltimore, Amer. Ornithol. Union. ANDERSON,E. 1949. Introgressivehybridization. New York, J. •Viley and Sons. ANDERSON,E. 1953. Introgressive hybridization. Biol. ]Rev., 28: 280-307. ANDERSON,E., AND G. L. STEBBINS,JR. 1954. Hybridization as an evolutionary stimulus. Evolution, 8: 378-388. BANKS,]R. C., ANDN. K. JOHNSON. 1961. A review of North American hybrid hum- mingbirds. Condor, 63: 3-28. BAUER,K. 1958. Zur systematischenStellung des Blutspechtes.Pp. 22-25 in J. F. Naumann-Ehrung, Leipzig, Urania-Verl. BraELOW,R. S. 1965. Hybrid zones and reproductive isolation. Evolution, 19: 449-458. BREWER,R. 1963. Ecological and reproductive relationshipsof Black-capped and Carolina Chickadees. Auk, 80: 9-47. Jan., 1969] Taxonomic Aspects of Arian Hybridization 103

BRow•, W. L., JR., ^•D E. O. W•LSO•. 1956. Character displacement. Syst. Zool., 5: 49-64. C^•, A. J. 1955. A revision of Trlchoglossushaematodus and of the Australian Platycercine parrots. Ibis, 97: 432-479. C•^m•, J.P. 1948. Variation and hybridization among the paradise flycatchers of Africa. Evolution, 2: 111-126. CI•-•PMAIff,F.M. 1940. Further studies of the genus Quiscalus. Auk, 57: 225-233. Coc•:Rv•, E. L. 1952. A check-list and bibliography of hybrid birds in North America north of Mexico. Wilson Bull., 64: 140-159. DELACOUR,J., A•D E. MAYR. 1945. The family Anatidae. Wilson Bull., 57: 1-55. DESSA•ER,H. C., W. Fox, A• F. H. Po•½•. 1962. Starch-gel electrophoresisof transferins, esterasesand other plasma proteins of hybrids between two sub- speciesof whiptail lizard (genusCnemidophorus). Copeia: 767-774. D•c•:•so•, J. C., JR. 1953. Report on the McCabe Collection of British Columbian birds. Bull. Mus. Comp. Zool., 109: 123-209. D•xo•, K. 1955. An ecological analysis of the interbreeding of crested titmice in Texas. Univ. California Publ. Zool., 54: 125-206. DOBZ•A•SX¾,T. 1951. Genetics and the origin of species. New York, Columbia Univ. Press. GALBRAITa,I. C.J. 1956. Variation, relationshipsand evolution in the Pachycephala pectoralissuperspecies (Aves, Muscicapidae). Bull. Brit. Mus. (Nat. Hist.), Zool., 4: 133-222. GOdfReY,W.E. 1966. The birds of Canada. Natl. Mus. Canada,Bull. 203: 1-428. HAF•, J. 1967. Speciation in Colombian forest birds west of the Andes. Amer. Mus. Novitates, no. 2294: 1-57. HAV•RSC•m% F. 1966. Hybrids of Thraupis palmarum and Thraupis virens. Bull. Brit. Ornithol. Club, 86: 5-6. How•]•]•, T. H 1952. Natural history and differentiation in the Yellow-bellied Sap- sucker. Condor, 54: 237-282. Hv•I•6•o•, C.E. 1952. Hybridization in the Purple Grackle, Quiscalusquiscula. Syst. Zool., 1: 149-170. Joa•s6AR•,P.A. 1960. Hybridization in the Anatidaeand its taxonomicsignificance. Condor, 62: 25-33. JO•SGAR•, P.A. 1961a. The sexualbehavior and systematicposition of the Hooded Merganser. Wilson Bull., 73: 227-236. JO•SCARD, P. A. 1961b. Evolutionary relationships among the North American Mallards. Auk, 78: 3-43. JOa•,'SGAR•,P.A. 1967. Sympatry changesand hybridization incidencein Mallards and Black Ducks. Amer. Midl. Nat., 77: 51-63. K•.aST, A. 1961. Bird speciationon the Australian continent. Bull. Mus. Comp. Zool., 123: 305-495. L^•¾o•, W.E. 1966. Hybridization in meadowlarks. Bull. Amer. Mus. Nat. Hist., 134: 1-25. MAYR, E. 1942a. Systematicsand the origin of species.New York, Columbia Univ. Press. MAYR,E. 1942b. Speciationin the junco. [A review.] Ecology, 23: 378-379. MAYR, E. 1963. Animal speciesand evolution. Cambridge, Massachusetts,Belknap Press of Harvard Univ. Press. MAYR, E., A• E. T. G•]•AR•. 1952. The Ribbon-tailed Bird of Paradise (Astrapia mayeri) and its allies. Amer. Mus. Novitates, no. 1551: 1-13. 104 LESTZa L. SHOaT [Auk, Vol. 86

Mmse, W. 1928. Die Verbreitung der Aaskrahe (Formenkreis Corvus corone L.). J. Ornithol. 76: 1-203. MEXSE,W. 1936. Zur Systematik und Verbreitungsgeschichtede Haus- und Weiden- sperlings,Passer domesticus (L.) und hispaniolensis(T.). J. Ornithol. 84: 631-672. Me•ceL, R. 1963. A second probable hybrid between the Scarlet and Western tanagers. Wilson Bull., 75: 201-203. MLLLER,A. H. 1941. Speciation in the avian genus Junco. Univ. California Publ. Zool., 44: 173-434. Mn•zeR, A H. 1955. A hybrid woodpeckerand its significancein speciationin the genusDendrocopos. Evolution, 9: 317-321. P•rr•LxPs,A.R. 1961. Notas sistemgticassobre aves Mexicanas. I. Anales Inst. Biol., 32: 333-381. PI•z•Ps, A. R., J. H. M^RS•r^•;L,JR., ^•) G. Mo•so•. 1964. The birds of Arizona. Tucson, Univ. Arizona Press. Rmcw^¾, R. 1887. On a probable hybrid between Dryobates nuttallii (Gamb.) and D. pubescensgairdnerii (Aud.). Proc. U.S. Natl. Mus. (1886), 9: 521-522. S^Lo•ro•s•, F. 1950. The birds of Greenland. Copenhagen,Ejnar Munksgaard. S^LO•rO•SE•, F. 1961. Notes on flowerpeckers (Aves, Dicaeidae) 6. The super- speciesPardalotus striatus. Amer. Mus. Novitates, no. 2068: 1-31. Sez^•l•eR, R. K. 1964. Speciation in wrens of the genus Campylorhynchus. Univ. California Publ. Zool., 74: 1-306. SeL^•I•R, R. K. 1965. Hybridization of Rufous-naped Wrens in Chiapas, Mexico. Auk. 82: 206-214. S•roRz, L. L., JR. 1963. Hybridization in the wood warblers Vermivora pinus and V. chrysoptera. Proc. 13th Intern. Ornithol. Cong. (1962), 1: 147-160. S•oRr, L. L., JR. 1965. Hybridization in the flickers (Coloptes) of North America. Bull. Amer. Mus. Nat. Hist., 129: 307-428. S•roRz, L. L., J•. 1966. Speciation in wrens of the genus Campylorhynchus. [A review.] Wilson Bull., 78: 243-245. S•oR% L. L., JR. 1967. A review of the genera of grouse (Aves, Tetraoninae). Amer. Mus. Novitates, no. 2289: 1-39. S•roRz, L. L., JR., ^•l• R. C. B^•xs. 1965. Notes on birds of northwestern Baja California. Trans. San Diego Soc. Nat. Hist., 14: 41-52 S•rORT,L. L., JR., ^•l• T. D. BvRz•xo•r. 1965. An intergeneric hybrid flycatcher (Contopus X Empidonax) from Idaho. Proc. Biol. Soc. Washington, 78: 33-37. S•roRz,L. L., J•., ^•l• A. R. Pm•zxvs. 1966. More hybrid hummingbirdsfrom the United States. Auk, 83: 253-265. SHORZ,L. L., JR., ^•l• C. S. Ro•s. 1967. An intergeneric hybrid wood warbler (Seiurus X Dendroica). Auk, 84: 534-543. S•roRz,L. L., JR., ^•l• S. W. Sx•o•. 1965. Additional hybrids of the Slate-colored Junco and the White-throated Sparrow. Condor, 67: 438-442. Smze¾ C.G. 1950. Speciesformation in the Red-eyed Towhees of Mexico. Univ. California Publ. Zool., 50: 109-194. S•z•¾ C.G. 1954. Hybridization in the Red-eyed Towhees of Mexico. Evolution, 8: 252-290. Smz•¾ C.G. 1957. The evolutionary and taxonomic significance of sexual selection and hybridization in birds. Condor, 59: 166-191. SmLE¾ C. G. 1958. Hybridization in some Colombian tanagers, avian genus Ramphocelus. Proc. Amer. Phil. Soc. 102: 448-453. Jan., 1969] Taxonomic Aspects oJ Avian Hybridization 105

SmLE¾,C. G. 1959. Hybridization in birds: taxonomic and evolutionary implica- tions. Bull. Brit. Ornithol. Club, 79: 154-158. SmLE¾,C.G. 1961. Hybridization and isolatingmechanisms. Pp. 69-87 in Vertebrate Speciation(W. F. Blair, Ed.). Austin, Univ. Texas Press. SmLE¾,C. G., A•D L. L. S•oRt, JR. 1959a. Hybridization in the buntings (Pas- serina) of the Great Plains. Auk, 76: 443-463. StorEY, C. G., A•D L. L. SHORt, JR. 1959b. Hybridization in some Indian bulbuls PycnonotuscaJer X P. leucogenys. Ibis, 101: 177-182. SmLE¾,C. G., A•D L. L. S•oRt, JR. 1964. Hybridization in the oriolesof the Great Plains. Condor, 66: 130-150. StorEY, C. G., ^•D F. C. SmLE¾. 1964. Hybridization in the Red-eyed Towhees of Mexico: the populationsof the southeasternplateau region. Auk, 81: 479-504. S•ZE¾, C. G., a• D. A. WEst. 1958. Hybridization in the Red-eyed Towhees of Mexico: the eastern plateau populations. Condor, 60: 85-104. S•Ms, R.W. 1959. The Ceyx erithacus and rufidorsus speciesproblem. J. Linnean Soc. London, Zool., 44: 212-221. SM•t•, N. G. 1966. Evolution of some Arctic gulls (Larus): an experimental study of isolating mechanisms. Amer. Ornithol. Union Ornithol. Monogr., no. 4: 1-99. STEam,S, G. L., JR. 1959. The role of hybridization in evolution. Proc. Amer. Phil. Soc., 103: 231-251. StEam,s, G. L., JR., a• K. Daz¾. 1961. Changes in the variation pattern of a hybrid population of Helianthus over an eight-year period. Evolution, 15: 60-71. StRESEM^•q•q,E., ^•q• N. W. T•2VIOV•Evv-REssovsxY.1947. Artenstehung in geo- graphischen Formenkreisen. I. Der Formenkreis Larus argentatus-cachinnans- Juscus. Biol. Zentralbl., 66: 57-76. Svt•'o•, G.M. 1967. Oklahoma birds. Norman, Univ. Oklahoma Press. Ta•ER, J. T. 1952. Black-capped and Carolina Chickadees in the southern Ap- palachian Mountains. Auk, 69: 407-424. Tom•ovv,H.B. 1950. A hybrid tanager from Minnesota. Wilson Bull., 62: 3-4. V^IJRIE,C. 1954. Systematicnotes on Palearctic birds, no. 5, Corvidae. Amer. Mus. Novitates, no. 1668: 1-23. VAIJRIE, C. 1956. Systematic notes on Palearctic birds, No. 24, Ploceidae. Amer. Mus. Novitates, no. 1814: 1-27. VAIJRIE,C. 1957. Systematicnotes on Palearctic birds, no. 26, Paridae: the Parus caeruleuscomplex. Amer. Mus. Novitates, no. 1833: 1-15. VAVR•E, C. 1959. The birds of the Palearctic fauna, vol. 1, Passeriformes. London, H. F. and G. Witherby. VavRm, C. 1965. The birds of the Palearctic fauna, vol. 2, Non-passeriformes. London, H. F. and G. Witherby. VavR•r, C., a• D. S•ow. 1957. Systematic notes on Palearctic birds, no. 27, Paridae: the genera Parus and Sylviparus. Amer. Mus. Novitates, no. 1852: 1-43. WESt, D. A. 1962. Hybridization in grosbeaks (Pheucticus) of the Great Plains. Auk, 79: 399-424. Wrzr•aMso•, F. S. L., A•D L. J. PE¾•'ON. 1963. Interbreeding of Glaucous-winged and Herring Gulls in the Cook Inlet Region, Alaska. Condor, 65: 24-28. The American Museum oJ Natural History, New York, New York 10024.