j. Field Ornithol., 62(1):117-131

GEOGRAPHIC VARIATION IN THE GRAY

KARENHABERMAN, • D. I. MACKENZIE2 ANDJ. D. RISING3 Departmentof Zoology Universityof Toronto Toronto, Ontario M5S IA I, Canada

Abstract.--We analyzedpatterns of geographicvariation in 11 measurementsof 369 Gray (Tyrannus dominicensis) from 10 localities.In bothsexes, kingbirds from migratory populationshad longerwings and larger bills than thosefrom non-migratorypopulations. In addition,there was significantsexual dimorphism in wing and tail length,as well as in somemeasures of bill size and wing shape.We interpretedthe sexualdifferences in wing and tail lengthand shapeto reflectsexual differences in courtshipdisplays and territorial defense.The differencesin bill size may reflect adaptationsfor niche partitioning,as the bill sizedimorphism tended to be greatestin localitieswhere Gray Kingbirdswere the only Tyrannuspresent.

VARIACION GEOGRJ•FICA EN TYRANNUS DOMINICENSIS Sinopsis.--Analizamosel patron de variaciongeogrfifica de l 1 parfimetrostomados de 369 individuosde Tyrannusdominicensis pertenecientes a 10 localidadesdiferentes. Aves de ambos sexos,de poblacionesmigratorias, mostraron tener las alas y el picomils grande que indi- viduosde poblacionesno migratorias.Ademfis encontramos un dimorfismosexual signifi- cativoen el largo del ala y rabo, comotambi•n en algunasmedidas del tamafiodel pico y forma del ala. Interpretamosque el dimorfismoen el largo y forma de las alas y el rabo, esreflejo de diferencias sexuales en el patr6nde cortejoy la defenzaterritorial. Las diferencias en el tamafiodel picopoddan ser una adaptaci6npara la segregaci6nde nicho,ya que este dimorfismotendi6 a sermayor en localidadesen dondela especieera el finicorepresentativo de su g•nero.

The Gray Kingbird (Tyrannusdominicensis) breeds from the southern Atlantic and Gulf coastsof the United States, south through the West Indies to the islandsoff South America (the Netherlands Antilles east to Trinidad and Tobago), and occasionallyin arid coastalregions of Ven- ezuelaand perhapsColombia. Individuals breeding in the northernparts of this range (United States,Bahamas, and ) migratesouth to northern SouthAmerica in winter, but Gray Kingbirdsare generally residentfrom Hispaniolaeastward (American Ornithologists' Union 1983, Bond 1985, Brodkorb 1950, Hilty and Brown 1986, Voous 1983). Brodkorb(1950) studiedgeographic variation in this .He pre- sortedhis specimensinto five different subspecies,and measuredbirds taken throughoutthe year. The subspeciesdescribed by Brodkorbare not generallyrecognized today (Mayr and Short 1970, Voous 1983), and it is generally agreed that the geographicalvariation, particularly in bill shape,is not well understood(Voous 1983). Here we describepatterns of geographicvariation and sexualdimorphism in the species.

Currentaddress: 312 Durant Way,Mill Valley,California 9d9dI USA. Deceased. Authorto whomcorrespondence should be sent.

117 118] K. Habermanet al. J.Field Ornithol. Winter 1991 Hypotheses Size variation.--Bergmann's Rule, the tendency for individuals of a speciesto be relativelylarge in the coldest,or coldestand dryestparts of the species'range, is the mostcommonly described pattern of geographic variationin sizein (James1970, Zink and Remsen1986). However, we cannotassess size variationin Gray Kingbirdsbecause we have no measure of body size. Weights were not indicated on enough of the specimenlabels for us to use them as a measureof body size; of the featureswe could measureon the museum specimensavailable, tarsus length might reflectbody size,but others(measures of wing, tail and bill size) probablyare little correlatedwith bodysize (Freemanand Jackson 1990, Rising and Somers1989). Ornithologistshave frequently used wing length as sucha measure,but that is precludedin this studybecause T. dominicensisis migratory in part of its range and sedentaryin others. Generally,within a speciesor amongclosely-related species, individuals from non-migratorypopulations have smaller wings, relative to bodysize, than those from migratory populations[e.g., Passerculussandwichensis (Rising, pers. obs.) and in Vireo (Barlow, pers. comm.)], and, indeed, Brodkorb (1950:334) notesthat Gray Kingbirds that migrate have rel- atively longer wings and shortertails than do thosethat do not. Sexualselection.--Commonly, size dimorphismin appearsto have evolvedas a consequenceof intrasexual aggressionor mate choice, that is, of sexual selection.In many birds, males defendterritories, nest sitesand matesfrom othermales, and in suchspecies males are commonly larger than females.Alternatively, in specieswhere maleshave acrobatic aerial displays,such as many shorebirdsand hummingbirds,so-called reversedimorphism (males smaller than females)is common(Jehl and Murray 1986), presumablybecause smaller males are better able to perform suchacrobatic displays. Little has beenwritten on the courtship and territorialityof T. dominicensis.Bent (1942:31)cites Audubon's (1840) accountof the courtship,which indicatesthat both sexesdo aerial displays duringcourtship, and Smith (1966) describesthe "Wing Flutter" display, typical of Tyrannus,for T. dominicensis.Although both sexesperform courtship displays,in kingbirds in general only the males engagein territorial defenseand the "Tumble Flight" display associatedwith it, andtypically male Tyrannusare moreaggressive toward conspecifics than females(Smith 1966). There is no reasonto suspectthat the behaviorof T. dominicensisis substantiallydifferent from that of better-studiedTy- rannus (Smith 1966), and male T. dominicensishave the modified outer primary wing feathersthat are characteristicof adult male kingbirds (Pyle et al. 1987). This probablyreflects a specialrole of thesefeathers during courtshipand territorial displaysof male Tyrannusin general. Assumingthat male T. dominicensisengage in territorial fights more frequentlythan females,we predictthat maleswill be larger, on average, than females,particularly with regard to wing length, both becauseit doubtlessreflects overall body size to someextent as well as flying ability. Alternatively,we couldargue, as did Jehl and Murray (1986) for shore- Vol.62, No. 1 GeographicVariation ofGray Kingbirds [ 119 birds, that acrobaticdisplays select for relatively small males, but this seemsless likely to us becausethe Tumble Flight displaysof Tyrannus do not appear to be as acrobaticas thoseof someshorebirds and hum- mingbirds. Nichepartitioning.--Van Valen (1965) postulatedthat the variability in structuresused for obtainingfood will be greatestin populationswhere the diversityof differenttypes of foodsavailable is greatest.This is the so-called"niche variation hypothesis."The diversityof food availableis influencedboth by the kinds of foodspresent and by the amount of competitionfor those foods.Where the amount of food is potentially limiting, intraspecificcompetition for it is of selectiveimportance. Where sympatricspecies use similar foods,interspecific competition may alsobe important.However, intraspecific competition is probablymore important than interspecificcompetition even in crowdedcommunities because the nicheoverlap of conspecificsmust be greaterthan that of non-conspecifics, and if both individualsof a pair feedon their territory, their overlapping requirementsmay be to their mutual detriment.The magnitudeof this effect dependson resourceavailability and the extent to which the re- quirementsof the male and femaleoverlap, and presumablyselection to reduceintersexual competition would act to exaggeratea differencethat evolvedas a consequenceof sexual selection(Jehl and Murray 1986, Price 1984). Island populationsof melanerpinewoodpeckers, and other populationsthat are allopatricwith otherwoodpecker species of similar size,show significantly more dimorphism in bill sizethan do populations that are sympatricwith congeners(Selander 1966). A similar situation pertainsin southwesternPicoides woodpeckers (Short 1971), and hasbeen describedin other populationsof birds, especiallythose found on islands (Lack 1971). The Gray Kingbird is found throughoutthe Caribbeanregion, and is the commonestWest Indian flycatcher. On many islands, it is the only Tyrannusfound, whereason othersit is sympatricwith one or two other species.The Tropical Kingbird (T. melancholicus)is found on Grenada, Trinidad and in the NetherlandsAntilles; the Fork-tailed Flycatcher(T. savana)is of irregular occurrencein the NetherlandsAntilles; the (T. cubensis)is foundon Cuba andthe Isle of Pines,and formerly on Great Inagua; the LoggerheadKingbird (T. caudifasciatus)is found in ,Cuba, the Isle of Pines, the ,Jamaica, Hispaniola and (but not the Virgin Islands;Raffaele 1989). The Giant Kingbird is decidedlylarger than the Gray Kingbird, is un- commonand tends to be found in woodlandsand pine forests,whereas the Gray Kingbird is generallyfound in more open habitats.It is, thus, probably not in significantcompetition with T. dominicensis.The Log- gerheadKingbird likewise is more of a woodlandspecies than the Gray Kingbird, althoughit canbe seenin opencountry where Gray Kingbirds are found (Bond 1985, Raffaele 1989). In the Netherlands Antilles the Tropical and Gray Kingbirds,and the Fork-tailed Flycatcherhave similar habits, but the Gray Kingbird is far more commonthan the other two 120] t:. Habermanet al. J.Field Ornithol. Winter 1991 species(Voous 1983). The Gray Kingbird is sympatricwith the Eastern Kingbird (T. tyrannus)in Florida. However, in the United Statesthe Gray Kingbird breedsin mangrovesand other bushesor treesin the salt marsh, and the EasternKingbird nestsinland (Brodkorb1950), sothere is virtually no ecologicaloverlap betweenthe species. All kingbirdsforage primarily by sallying from a perch and catching flying insects.The 's bill sizeprobably influences the sizesand types of insectstaken (Hespenheide1964). If interspecificcompetition for food is significantfor Gray Kingbirds, we predict that there will be more dimorphismin bill size in populationsof Gray Kingbirdsthat are allo- patric with other Tyrannusthan in thosethat are sympatricwith other congenericspecies, especially T. melancholicusand T. caudifasciatus.

MATERIALS AND METHODS This researchwas initiatedby the late D. I. MacKenzie, who borrowed specimensin adult plumage from severalof the large North American collections(listed in Acknowledgments).He took the following 11 mea- surements(see Table 1) with dial caliperson 369 specimens:wing length, the chord of the closedwing from the wrist joint to the tip of the longest primary wing feather; wing tip to 4th, wing tip to 5th and wing tip to 10th, the distancesfrom the tip of the wing to the tip of the 4th, 5th and 10th primary feathers,respectively; outer tail lengthand inner tail length, the distancefrom a point of insertionbetween the two middle rectrices and the longestouter and longestinner tail feather, respectively;tarsus length,the diagonaldistance from the mid-point of the posteriorjunction of the tibiotarsusand tarsometatarsusto the distal edge of the last un- dividedscute at the juncture of the tarsometatarsusand the baseof the middle toe; total bill length, the chord of the bill from the point where the integumentof the foreheadmeets the culmen;bill length(from nares), the chordof the bill from the anteriormargin of the nares;bill width, at the anteriormargin of the nares;and bill depth,the distanceperpendicular to the long axis of the bill from the culmento the lower edgeof the ramus at the anterior margin of the nares. The vastmajority of the specimensmeasured from residentpopulations were collectedin November-March whereasthose from migratory pop- ulations(Great Inagua, Bahamas,Florida) were collectedin April-Au- gust.Migratory Gray Kingbirdsmolt their remigesand rectriceson their wintering grounds,and their molt is completeby mid-March. The molt of the sedentaryindividuals is more protracted,and individualsin their first winter may retain somejuvenal feathers.Any such birds were ex- cludedfrom theseanalyses. Thus, feather wear may be more advanced in our migratory samplesthan in our residentones. As well, someof the individualsfrom the residentpopulations may be wintering migratory birds. For analysesof geographicvariation, we dividedthe specimensinto 10 differentpopulations, nine from West Indian islands(or groupsof islands) andone from Florida (Fig. 1). In someinstances, geographically proximal Vol.62, No. I GeographicVariation ofGray Kingbirds [ 121 islandswere pooledto increasesample sizes, and of the total sample,54 specimenswere eliminated from analyseseither because they were missing severalmeasurements or becausethey couldnot be logicallypooled with oneof thesesamples. The West Indian samplesused were as follows:the Netherlands Antilles; Trinidad; Lesser Antilles (including Grenada, Montserrat, St. Lucia, Antigua and Guadeloupe); the Virgin Islands (includingSt. Eustatiusand St. Croix); Great Inagua; Cuba (including the Caymen Islands);Hispaniola; Puerto Rico; and the Bahamas(New ProvidenceIsland, Crooked Island, Cat Island, Grand Bahama and Great Abaco). Sample sizesand averagesare listed in Table 1. We usedtwo way analysesof variance(ANOVA) (sortedby sex and locality) (SPSSX; ANOVA) (SPSSX 1986) to elucidateunivariate pat- terns of geographicvariation. For these analysesonly the largest six sampleswere used.There were at least 11 birds of eachsex in thesesix. To examine multivariate patterns of geographicvariation, we used discriminantfunctions analyses (DFA) (SPSSX 1986). For the DFAs, missingvalues were estimatedusing multiple regression,calculated on the two measurablevariables that bestpredict the missingone; we omitted individualsthat were missingmore than two measurements.We analyzed sexesseparately, and usedonly sampleswhere N > 11. For the DFAs we pooledTrinidad with the LesserAntilles. This restrictedthe DFA to sevensamples of malesand six of females,and a total of 315 individuals (189 males and 126 females). We usedMantel's test to comparematrices of phenotypicdifferences (DF1 and DF2 scoresof males)with a matrix of minimumgeographic distancesand a matrix of the reciprocalsof geographicdistances. These testsindicate whether or notpatterns of phenotypicvariation simply reflect the geographicdistances among the localities;the reciprocalsemphasize patternsof variationamong geographical close localities (Mantel 1967). NTSYS-PC, version1.50, programMXCOMPG (Rohlf 1988) was used for Mantel's tests;1000 randompermutations were donefor eachtest. We assessedsexual dimorphismin each variable in the six largest samplesusing ANOVA, and multivariatelyusing DFA. We did two sets of DFA analyses,one using all 11 variablesand the otherusing just the four bill measurements.

RESULTS Size variation.--The two-way ANOVA (Table 2) showedthat there was highly significantinterpopulational variation in all 11 variablesmea- sured.As well, there was significantsexual dimorphism in wing length, two measuresof wing shape(wing tip to 4th and wing tip to 5th), both measuresof tail length, bill width and bill depth. Specifically,in all samples,males had significantlylonger wings and tails than females. Without a known measureof bodysize, it cannotbe determinedwhether this reflectslarger body size of males,or relativelylonger wing and tail feathersin males,perhaps associated with male flight displays.There was, however,no sexualdimorphism in tarsussize, indicating that there 122] K. Haberrnanetal. J.Field Ornithol. Winter 1991 Vol.62, No. 1 GeographicVariation ofGray Kingbirds [ 123 124] K. Haberraanetal. J.Field Ornithol. Winter 1991

GREAT INAGUA

•• HISPANIOLAPU ERT O RICO VIRGIN IS.

Jamaica • o North 0

t • % % Grenada•St.Lucia• •

FIGURE1. Map of the Caribbeanregion showing the locationsfrom which Gray Kingbirds (T•ra•s do•mce•s•s) were measured.The namesof the largestsamples are in capital letters. is little if any overall differencein body size betweenthe sexes.Females tendedto have wider and deeperbills than males (Tables 1 and 2). The interaction term showsthat there was significantgeographic variation in the amount of sexual dimorphismin total bill length and two measures of wing shape (Table 2). In the DFA of males from the sevenlargest samples,there were four highly significantdiscriminant functions (P < 0.001). However, 63% of the total amonggroup variancewas explainedin the first functionand 23% by the second,and only an additional9% and 4% by the next two, respectively.DF 1 was positivelycorrelated with all measuresof bill size and wing length; DF 2 was also positivelycorrelated with measuresof bill size,especially bill width and the measuresof bill length.Individuals from Great Inagua, the Bahamas and Florida, that is the migratory populations,had large DF1 scoresand thuswere relativelylarge in these measurements(see also Table 1). The individuals from the LesserAntilles alsohad large scoresfor DF1, aswell as large scoresfor DF2. Otherwise, there was no clear pattern of geographicvariation. Overall, 66% of the individualswere correctlyclassified by the discriminantfunctions. These classificationresults (Table 3) show that the individuals fell into two Vol.62, No. 1 GeographicVariation of Gray Kingbirds [ 12 5

TABLE2. Resultsof ANOVA of geographicvariation among Gray Kingbirdsfrom six regions.

Significance Among regions Between sexes Interaction Variable (df = 5) (df = 1) (df = 5) Wing length *** *** ns Wing tip to 4th *** *** * Wing tip to 5th *** *** * Wing tip to 10th *** ns ns Outer tail length *** *** ns Inner tail length *** *** ns Tarsus length *** ns ns Total bill length *** ns * Bill (from nares) *** ns ns Bill width *** * ns Bill depth *** *** ns

* P < 0.05. ** P < 0.01. *** P < 0.001. virtually completelydistinct groups, Antilles-Virgin Islands-Puerto Rico- Hispaniolaand Great Inagua-Bahamas-Florida,that is, into migratory individualsand non-migratoryones. Only two of the 129 non-migratory birds were placedwith the migratoryones, and only three of the 60 migratoryindividuals were classifiedas non-migratory. In the DFA of the six samplesof femalesthere were three highly significant(P < 0.001) discriminantfunctions, and a fourth significant one(P < 0.003). The firstexplained 47% of the variationamong groups, the second 37%, and the third and fourth about 7% each. Unlike the males,only measuresof wing shape,wing tip to 4th and wing tip to 5th were highly correlatedwith the first DF, whereas measuresof bill size and wing lengthwere highly positivelycorrelated with DF2. As with the males, migratory vs. non-migratoryindividuals were separatedon the DF1 axis,with femalesfrom migratorypopulations tending to havemore pointedwings than thosefrom non-migratoryones. Birds from the Lesser Antilles and Puerto Rico tendedto havelarge bills. Overall, 77% of the femaleswere correctlyidentified by the discriminantscores; only two of the 38 migratorybirds from Great Inagua and Florida were incorrectly identifiedas non-migratoryones, and onlythree of the 88 non-migratory oneswere identified as migratory individuals (Table 4). The correlationsamong matrices of phenotypicdifferences (DF1 and DF2 scores)and geographicaldifferences were low, the largest(and only statisticallysignificant) one being -0.47 (t = 1.78; P = 0.04), between DF 1 scoresfor femalesand the reciprocals of geographicdifference. These resultsindicate that patternsof phenotypicvariation are essentiallyin- dependentof geographicdistances, even though they are not independent of geographicposition. 126] K. Habermanetal. J.Field Ornithol. Winter 1991

TABLE3. Actual vs.predicted group membership (number with percentagein parentheses) of male Gray Kingbirds.

Predictedgroup membership Lesser Virgin Puerto Hispan- Baha- Actual group N Antilles Islands Rico iola Inagua mas Florida

Lesser Antilles 18 13 0 3 2 0 0 0 (72) (0) (17) (11) (0) (0) (0) Virgin Islands 19 0 10 9 0 0 0 0 (0) (53) (47) (0) (0) (0) (0) Puerto Rico 49 1 5 34 9 0 0 0 (2) (10) (69) (18) (0) (0) (0) Hispaniola 43 0 0 8 33 1 1 0 (0) (0) (19) (77) (2) (2) (0) Great Inagua 25 0 0 0 0 14 9 2 (0) (0) (0) (0) (56) (36) (8) Bahamas 17 1 0 1 0 5 9 1 (6) (0) (6) (0) (29) (53) (6) Florida 18 0 0 1 0 1 3 15 (0) (0) (6) (0) (6) (17) (67)

Sexualdimorphism.--The wings of maleswere significantly longer than thoseof femalesin all six of the largestsamples (Table 5). This is consistentwith the hypothesisthat malesare selectedto have relatively longerwings either for aerial displaysor for territorial defense.Males alsohad significantlylonger tails (especiallythe lengthof the outer-most tail feathers)than femalesin all samples,save the LesserAntilles (Table 5). This, too, might be of significancein their aerial displays.Perhaps males are simply generallylarger than females,but the absenceof di- morphismin tarsuslength (Table 5) suggeststhat this is not the case; tarsussize in birds that have similar behaviormay be an indicatorof body size (Rising and Somers1989). The resultsalso indicatethat males may tend to have a more forked tail (differencebetween outer and inner tail feather length) than females.Certainly suchdimorphism is pro- nouncedin someother Tyrannus (T. forficatusand T. savana).Wing shape, asreflected by the distancefrom the wing tip to the tip of the 5th primary, was also dimorphicin three of the populations(Table 5), with females tendingto havemore pointed wings. The valuesof thismeasure, however, weresmall (Table 1), anddoubtless subject to relativelyhigh measurement error, and thus this apparent differenceshould be viewed with caution. There was significantdimorphism in bill length only in the Florida population,where males had longer bills than females.In the Great Inaguasample, females had significantly wider and especially deeper bills than males,and femalebill depthwas significantly greater than malesin the Puerto Rico sampleas well (Tables 1 and 5). In the discriminantfunctions analyses using all 11 variables,there was significantsexual dimorphism in four of the five largestsamples (Table Vol.62, No. 1 GeographicVariation of GrayKingbirds [ 127

TABLE4. Actual vs.predicted group membership (number with percentagein parentheses) of female Gray Kingbirds.

Predictedgroup membership Lesser Virgin Puerto Hispan- Actual group N Antilles Islands Rico iola Inagua Florida

Lesser Antilles 22 16 I 4 1 0 0 (73) (5) (18) (5) (0) (0) Virgin Islands 11 0 8 2 1 0 0 (0) (73) (18) (9) (0) (0) Puerto Rico 34 0 6 25 2 0 1 (0) (18) (74) (6) (0) (3) Hispaniola 21 0 0 5 14 1 1 (0) (0) (24) (67) (5) (5) Great Inagua 23 0 0 1 0 20 2 (0) (0) (4) (0) (87) (9) Florida 15 0 1 0 0 0 14 (0) (7) (0) (0) (0) (93)

6). Using only the four bill variables,however, there was significant dimorphismonly in the Great Inagua sample,in which 35% of all of the variation in bill size was due to sexual dimorphism.Fig. 2 showshis- togramsof the DFA scoresbased on the four bill measures.Although the discriminantfunctions are somewhatdifferent among the six analyses,in generalbirds with the largestDF scoreshad the largestbills. On Great

TABLE5. Resultsof ANOVA betweensexes of 11 measuresin the six largestsamples of Gray Kingbirds.

Allopatric with other Tyrannus Sympatricwith other Tyrannus Lesser Virgin Great Puerto Hispan- Variable Antilles Islands Inagua Rico iola Florida Wing length ***+ **+ ***+ ***+ **+ ***+ Wing tip to 4th **+ ns ns ns ns **+ Wing tip to 5th ns ns **- ***- ns **- Wing tip to 10th ns ns ns ns ns ns Outer tail length ns ** + *** + *** + ** + *** + Inner tail length ns ns *** + *** + ns ** + Tarsus length ns ns ns ns ns ns Total bill length ns ns ns ns ns **+ Bill (from nares) ns ns ns ns ns *+ Bill width ns ns *- ns ns ns Bill depth ns ns ***- *- ns ns

* P < 0.05. ** P < 0.01. *** P < 0.001. + Males > females. - Females > males. 128] K. Habermanet al. J.Field WinterOrnithol. 1991

FLORIDA GREAT INAGUA

P=0.11 P < 0.001

! i ! ! , 0 , i i , i , -1 0 I 2 3 4 -3 -2 -1 0 1 2 3

P = 0.39

•.• PUERTORICO

i i i i i [ i i [ , i i i -2 -1 0 1 2 3 4 -3 -2 -1 0 1 2 3 4

VIRGIN ISLANDS LESSER ANTILLES

P = 0.06 ea P = 0.62

0 i i i i i [ i i i i i i i -3 -2 -1 0 1 2 3 4 -3 -2 -1 0 1 2 3 Bill DFA Scores Bill DFA Scores

FIGURE2. Histogramsof the discriminantfunction scores based on fourmeasures of bill sizeof male(open) and female (shaded) Gray Kingbirds(Tyrannus dominicensis) from thesix largest samples. Although the six analyses differ in detail,in generalindividuals to the righthave relatively large (or longand narrow) bills whereas those to the left have relatively small bills. Vol.62, No. I GeographicVariation ofGray Kingbirds [ 129

T^BI.V.6. Summary of DFA betweensexes of Gray Kingbirds usingall 11 variablesand four bill variables(numbers are canonicalcorrelations squared and explain the pro- portionof the total variationthat is betweensexes).

Sample All variables Bill variablesalone

Lesser Antilles 0.62* 0.08 ns Virgin Islands 0.66** 0.26 ns Great Inagua 0.74*** 0.35*** Puerto Rico 0.41'** 0.05 ns Hispaniola 0.26 ns 0.08 ns Florida 0.78 *** 0.23 ns

* P < 0.05. ** P < 0.01. *** P < 0.001.

Inagua, the only locality examinedin which there is significantbill di- morphism, females had larger bills than males. There was also this tendency,albeit not significant,in the samplesfrom Hispaniola and Puerto Rico as well.

DISCUSSION There was significantgeographic variation in Gray Kingbirds. The discriminantfunctions analyses showed that the most striking pattern of geographicvariation was between migratory and non-migratoryGray Kingbirds,the latter having significantlylonger wings, longerbills and more pointed wings than the former. Recall, however, that the non- migratory individuals were generally collectedearlier in the year than the migratoryones, so this result may be an artifactof the dateof collection. This seemsunlikely, however,because the feathersof migratory individ- uals shouldbe, if anything,more worn than thoseof the non-migratory ones(and hencewould be expectedto have shorter,not longer, wings), and seasonaldifferences could not explain the differencesin bill size. Also, the fact that virtually all individualswere correctlyidentified as to migratory status suggeststhat the winter-collectedsamples were made up of residentindividuals not admixedwith substantialnumbers of win- tering migrants. With the data at hand, it is difficult to know whether the migratory individualshave wings that are longer relativeto their body size, or whether they are perhapssimply larger than the non-migratorybirds. If tarsuslength can be taken to be a measureof body size, as it can be in somespecies (Rising and Somers1989), the largestindividuals (from the relativelylarge samples)are from the LesserAntilles, Virgin Islandsand Puerto Rico, and the smallestfrom Hispaniola, the Bahamasand Florida. This indicatesthat it is relative wing length, not absolutewing length, that differs betweenthe migratory and non-migratorybirds. Male Gray Kingbirdswere larger than femaleswith regard to wing length,tail lengthand (in onepopulation) bill length.Females had greater bill depththan males(in two populations).There was no sexualdimor- 130] K. Habermanet al. J.Field Ornithol. Winter 1991 phism in tarsus length, suggestingthat there is little or no dimorphism in "bodysize." Dimorphism in wing lengthand tail lengthperhaps reflects the behavioraldifferences between sexes. The bill size dimorphismwas most striking in the samplefrom Great Inagua, one of the siteswhere Gray Kingbirds are the only tyrannid. As well, bill dimorphismwas nearly significantin Florida and the Virgin Islands,where T. dominicensis is either allopatric,or effectivelyso, to other Tyrannus.On the otherhand, dimorphismwas the least in the samplesfrom Puerto Rico, Hispaniola and the LesserAntilles. T. dominicensisis sympatricwith T. caudifasciatus in the first two localities,but is the only Tyrannusin the LesserAntilles. These results,thus, could be taken to show weak support for the niche variation hypothesis,although such an interpretationis obviouslyten- tative, and it is clearly desirableto have more large samples,especially those of birds from Trinidad and the Netherlands Antilles where T. dominicensisapparently closelyoverlaps with T. melancholicus. Brodkorb(1950) recognizedfive subspeciesof T. dorninicensis,T. d. dominicensis(Hispaniola, Puerto Rico, Virgin Islands), T. cl.fugax (U.S. and Bahamas,including Great Inagua), T. d. sequax(Cuba, Isle of Pines, s. Bahamasand perhapsJamaica), T. d. tenax(Netherlands Antilles) and T. d. vorax (L. Antilles). He useddifferences in size and proportions,for the mostpart, to describethese subspecies, but notedthat T. d. voraxhas darker gray upper parts than the others.On the basisof size and shape, our resultswould seemto justify the recognitionof only two subspecies, migratory and non-migratoryGray Kingbirds.The type specimenof T. d. dorninicensisis from SantoDomingo, and hencethe name dominicensis has priority for the non-migratoryGray Kingbirds.The type associated with the next oldestname, vorax,is from Martinique. Thus, on the basis of size,both dominicensis and voraxrefer to non-migratoryT. dominicensis. Brodkorb(1950) proposedthe namefugax for the Gray Kingbirdsfrom the U.S. and Bahamas,and this name has page priority over the other subspecies(T. cl. sequaxand T. d. tenax) that he proposed.The type specimenof T. d. fugax was collectedon Cedar Key, Florida. Thus, T. d. fugax is an appropriatename for the migratory individuals.Pending examinationof additionalmaterial, we suggestthat birds from Cuba, the Isle of Pines and Jamaica be placedin T. d. fugax, and thosefrom the Netherlands Antilles in T. d. dominicensis, or in T. d. vorax if the color differenceis judged sufficientgrounds for recognitionof that subspecies.

ACKNOWLEDGMENTS

We thankNancy Flood, Ned Johnsonand Richard Snell for their manyhelpful comments. We alsoappreciate the cooperationof the curatorsof the followingcollections who kindly allowed MacKenzie to measuretheir specimens:Field Museum of Natural History, Uni- versityof KansasMuseum of Natural History, University of Michigan Museum of Zoology, and the Royal Ontario Museum.

LITERATURE CITED

AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list of North American birds. 6th ed. AmericanOrnithologists' Union, Lawrence,Kansas. 877 pp. Vol.62, No. 1 GeographicVariation of Gray Kingbirds [ 131

AUDUBON,J.J. 1840. The birds of America, Vol. 1 (Reprint). Dover, New York, New York. 254 pp. BENT,A.C. 1942. Life historiesof North Americanflycatchers, larks, swallows,and their allies.U.S. Nat. Mus., No. 179. 555 pp. BOND,J. 1985. Birds of the West Indies. Collins, London. 256 pp. BRODKORB,P. 1950. Geographicalvariation in the Gray Kingbird, ryrannusdominicensis. Auk 67:333-344. FREEMAN,S., ANDW. M. JACKSON.1990. Univariate metricsare not adequateto measure arian body size. Auk 107:69-74. HES?ENHEIDE,H.A. 1964. Competitionand the Genus ryrannus.Wilson Bull. 76:265- 281. HILTY, S. L., ANDW. L. BROWN. 1986. A guideto the birdsof Colombia.Princeton Univ. Press,Princeton, New Jersey. 836 pp. JAMES,F. 1970. Geographicvariation in birds and its relationshipto climate. Ecology 51:365-390. JEHL, J. R. JR., AND B. G. MURRAY,JR. 1986. The evolutionof normal and reverse sexualsize dimorphismin shorebirdsand other birds. Current Ornith. 3:1-86. LACK,D. 1971. Ecologicalisolation in birds. Blackwell,Oxford. 404 pp. MANTEL, N.A. 1967. The detectionof diseaseclustering and a generalizedregression approach. Cancer Res. 27:209-220. MAYR,E., ANDL. L. SHORT. 1970. Speciestaxa of North Americanbirds. Publ. 9, Nuttall Ornith. Club, Cambridge,Massachusetts. 127 pp. PmCE,T. D. 1984. The evolutionof sexualsize dimorphismin Darwin's Finches.Am. Nat. 123:500-518. PYLE, P., S. N. G. HOWELL, R. P. YAmCK, AND D. F. DESANTE. 1987. Identification guideto North AmericanPasserines. Slate Creek Press,Bolinas, California. 278 pp. RAFFAELE,H.A. 1989. A guideto the birdsof PuertoRico and the Virgin Islands.Revised Edition. PrincetonUniv. Press,Princeton, New Jersey.254 pp. RISING,J. D. 1988. Geographicvariation in sexratios and bodysize in winteringflocks of SavannahSparrows (Passerculus sandwichensis). Wilson Bull. 100:183-203. •, ANDK. M. SOMERS.1989. The measurementof overall body sizein birds.Auk 106:666-674. ROHLF,F.J. 1988. NTSYS-PC NumericalTaxonomy and Multivariate AnalysisSystem. Exeter, Setauket,New York. 162 pp. SELANDER,R. K. 1966. Sexual dimorphismand differential niche utilization in birds. Condor 68:113-151. SHORT,L.L. 1971. Systematicsand behaviorof someNorth Americanwoodpeckers Genus Picoides(Aves). Bull 145 Amer. Mus. Nat. Hist., New York. 118 pp. SMITH, W.J. 1966. Communicationand relationshipsin the Genus Tyrannus.Monogr. No. 6, Nuttall Ornith. Club, Cambridge,Massachusetts. 250 pp. SPSSX. 1986. SPSSX user'sGuide, 2nd edition.SPSS Inc., Chicago,Illinois. 988 pp. VAN VALEN,L. 1965. Morphologicalvariation and width of ecologicalniche. Am. Nat. 99:377-389. Voous, K.H. 1983. Birds of the NetherlandsAntilles, 2nd ed. De Walburg Pers, Utrecht. 327 pp. ZINK, R. M., AND J. V. REMSEN,JR. 1986. Evolutionary processesand patterns of geographicvariation in birds. Current Ornith. 4:1-69. Received13 Mar. 1990; accepted28 Sep. 1990.