Speciation in Golden-Plovers, Pluvialis Dominica and P

Home , American golden plover, Charadriidae, Pluvialis

The Auk 110(1):9-20, 1993

SPECIATION IN GOLDEN-PLOVERS, PLUVIALIS DOMINICA AND P. FULVA: EVIDENCE FROM THE BREEDING GROUNDS

PETERG. CONNORS,• BRIAN J. MCCAFFERY,2 AND JOHN L. MARON• •Universityof California,Davis, Bodega Marine Laboratory, P.O. Box247, BodegaBay, California 94923, USA; and 2YukonDelta NationalWildlife Refuge, P.O. Box346, Bethel,Alaska 99559, USA

ABSTRACT.Two forms of golden-plover have long been consideredsubspecies, Pluvialis dominicadominica and P. d. fulva. Prior studies have shown differencesbetween forms in breedingdistributions, wintering distributions,plumage, morphology, molt, and maturation schedules.We report clear and consistentdifferences in breedingvocalizations and nesting habitat, and strict assortativemating in areas of sympatry in western Alaska. These results indicate a greater degree of differentiation between the forms than was previously appre- ciated. They are appropriately treated as separatespecies and should be referred to under the namesPluvialis dominica, for the AmericanGolden-Plover, and Pluvialis fulva, for the Pacific Golden-Plover.Received 24 May 1991,accepted 18 February1992.

Two FORMSof golden-plover traditionally showing assortativemating in regionsof sym- have been treated as subspecies,Pluvialis dom- patry and clear differencesin molecular genet- inicadominica and P. d. fulva (e.g. Peters 1934, ics, plumage and morphology, biogeography, Bock1958, Vaurie 1964,Mayr and Short 1970, display behavior and vocalizations,and ecolo- AOU 1983). The form dominicabreeds on arctic gy. In practice, decisions often must be based and subarctic tundra from Baffin Island in Can- on a subset of these data. Differences between ada west to westernAlaska; fulva breedsfrom the dominicaand fulva forms of golden-plover the Yamal Peninsula in Siberia eastward to have alreadybeen documentedwith respectto western Alaska (AOU 1983). However, based plumage and morphology(Connors 1983), mi- on an analysisof almost 400 museumspeci- gration routesand wintering areas(AOU 1983), mens, Connors (1983) concluded that there was and molt and maturation schedules (Strese- no evidenceof interbreeding in western Alaska mann and Stresemann1966, Kinsky and Yald- where both forms had been collectedduring wyn 1981,Johnson and Johnson1983, Johnson the breedingseason. Employing a multivariate 1985). We now provide additional evidence statisticalanalysis to characterizespecimens, he documentingextensive breeding sympatry in detectedno increasein intermediate pheno- western Alaska, consistent habitat differences, typeswithin the regionof potentialsympatry species-specificbreeding vocalizations,and as- from Nunivak Island to Point Barrow, and he sortativemating in the two forms.These data, recommendedfull speciesstatus for the forms. bolsteredby moleculargenetic evidenceto be The museumdata, however,could not prove presented elsewhere (F. B. Gill, P. G. Connors, widespreadbreeding sympatry of dominicaand J. L. Maron in prep.), unambiguously indicate fulva,because of the absenceof breedinginfor- that separatespecies status is appropriate for mationaccompanying most museum specimens Pluvialisdominica and Pluvialisfulva. and the possibilityof distributionalchanges during the centuryof collections.Furthermore, STUDY AREA AND METHODS data identifying any potential differences in ecology, breeding behavior, or vocalizations Our observationsand data were collectedat many were not available. The AOU (1983), therefore, sites on the Seward Peninsula and Yukon Delta Na- has retained subspecificstatus for theseforms, tional Wildlife Refuge (YDNWR) in northwestern although other authorities now considerthem Alaska from 1985 through 1990 (Fig. 1). In 1988 (31 May-21 June),Connors and Maron surveyedslopes separatespecies (BOU RecordsCommittee 1986, and ridgesat many siteson the southernSeward Pen- Hayman et al. 1986). insula accessiblealong the gravel highway system Ideally, taxonomic decisions to separate emanating from Nome. The three principal roads-- closelyrelated forms should be supportedby a to Teller, Kougerok, and Council--cover more than broad spectrum of evidence, including data 300 km, repeatedlypassing through or near suitable 10 CONNORS,MCCAFFERY, ANDMARON [Auk,Vol. 110

or groups identified as present in a quadrat (i.e. Loise- leuriaprocurnbens, Ernpetrurn nigrurn, Betula nana, Ledurn palustre,Vacciniurn spp., moss,lichen, and sedge).Data from 16 quadrats were averaged for each nest. We measuredelevation change from the nestto each15-m transect point with a sighting level, totalling these for an index of slope. We measuredelevation at the nest with an altimeter and plotted nest locations on topographic maps.Using the statisticalcomputer pro- gram SYSTAT (Wilkinson 1988), we entered these data (vegetation cover, vegetation height, vegetation richness,slope, elevation) in a discriminant-function analysisto separatethe two populationsstatistically. During both 1988 and 1989 we recorded golden- plover vocalizationswith a Sony WM-D6C cassette recorder and an Audio-Technica AT815a directional microphone on TDK MA metallic tapes. Sonagrams were produced on a Kay ElemetricsModel 7800 dig- ital sonagraphusing the wide-band filter. On YDNWR, McCafiery observedbreeding golden- Fig. 1. Map of Seward Peninsula and Yukon Delta plovers at the following locations(see Fig. 1): Bethel NationalWildlife Refuge(YDNWR) studyareas. Prin- (April-May 1987), Nelson Island (May-June 1985), cipal study sitesdenoted by circles. AndreafskyHills and vicinity (June 1985,May-June 1986,May-July 1987),Ingakslugwat Hills (June1988), golden-plover breeding habitat. We selected three Pikmiktalik River headwaters (June 1985), and Cur- principal study sitesnear: Nugget Creek, mile 31 on lew Lake (April-July 1988-1990).In late May and the Nome-Kougerok Road (64ø54'N, 165ø14'W);Hor- early June 1987,McCafiery, Connors, and SarahGrif- ton Creek, mile 58 on the Nome-Council Road fin observedbreeding golden-ploversat siteseast of (64ø44'N, 164ø01'W); and Feather River, mile 37 on Curlew Lake. In addition, golden-ploverswere ob- the Nome-Teller Road (64ø51'N, 166ø05'W;Fig. 1). served by YDNWR personnel at several sites in the Other sites mentioned by name are: Cripple River, Andreafsky River watershed(June 1988, May-June mile 19 on the Nome-Teller Road; and Crete Creek, 1989). on the road to Woolley Lagoon from mile 38 of the Nome-Teller Road. RESULTS In 1989 we attemptedto revisit the three principal study sites earlier in the season(22 May-13 June). Of areas surveyed to date, we have located However, that spring had exceptionallyheavy snow- large numbers of both forms on breeding ter- fall andlate snowmelt. Highwaycrews were delayed ritories only on the Seward Peninsula. Seward in opening highwaysbecause of the extensivedrifted Peninsula sites, therefore, hav• presented the snow and the need to repair subsequentwashouts most useful and revealing comparisons, and during the snow melt. We were able to reach only provide the data we emphasize here. Obser- one principal study site, Feather River, before leaving the area on 13 June. Most 1989 observations, there- vations from other sites are presentedmainly fore, occurredat a variety of locationswithin 30 km as comparisonswith the Seward Peninsula situation. of Nome. We spent three to six days at each of the principal Identification.--With practice in the field, we study sitesin 1988, locating golden-plover pairs and found that both sexesoffulva and dominicawere nests,characterizing nest habitats, and recording vo- easierto identify on the basisof breeding-plum- calizations. We sampled vegetation characteristicsat age differencesthan indicated in previous re- each nest within a 50 cm x 50 cm square quadrat ports (Conover 1945, Gabrielson and Lincoln placed at four points along each of four transects,for 1959), or by examinationof museumspecimens a total of 16 quadrat points per nest. Transectswere (Connors 1983). Most males in breeding plum- set by compass,running in the four principal direc- age are easily separableby the pattern of white tions from the nest. Quadrat samplingpoints were at on the undersides. In dominica,white at the fore- 2.5 m, 5 m, 10 m, and 15 m from the nest on each transect. We recorded: vegetative cover (as estimated head and sidesof the face continuesonly as far percent of total quadrat area); maximum vegetation asthe upper breast,where it terminates,usually height within the quadrat; and vegetation richness, in a widened, bulbouspatch on eachside of the as indicated by the number of the eight plant species breast. Flanks and undertail coverts are black, January1993] Speciationin Golden-Plovers

FEATHER RIVER NUGGET CREEK HORTON CREEK

Fig. 2. Golden-ploverbreeding distributions at SewardPeninsula study sites. Map dimensionsare 3.6 km x 3.6 km (F = fulva;D = dominica).Circles are schematic,and do not indicateterritory-boundary locations; circle diameter is 350 m. Dark circle centers indicate nest locations.Open circles indicate areas of nesting activity with probablenests. Location of gravel road at eachsite shownas a bold line. Peripheralareas of each site map were not searched. rarely with some white mixed in undertail co- Peninsula study sites (n = 165 in 1988, n = 140 verts. In fulva, the white shows little or no in 1989). Additional differences in color, size, broadening at the upper breast, and almost al- and distribution of spots on wings and mantle ways continues in an irregular streak past the may assist in identification of both sexes in breastand along the flanks to the undertail co- breeding plumage (Dunn et al. 1986, Kevin J. verts, which are predominantly white (see il- Zimmer pers. comm.). Ingvar Byrkjedal (pers. lustrations in Hayman et al. 1986:100-101). Fe- comm.)also has noted differencesin body shape, male dominicashow a pattern of white on black with dominicabeing more slender-bodied and or dark brown underparts similar to that of having a shorter naked portion of tibia. males, but with additional white flecks scat- Breedingsympatry.--Our 1988observations on tered throughout the black area. Terminal white Seward Peninsula demonstrated close and con- bulbous patches are usually evident at the up- sistent breeding sympatry throughout an area per breast, as in male dominica.Many female in which both speciesare common and widely dominicaare almost as uniformly dark on face distributed. We found adjacentfulva and dom- and breast as are males. Female fulva, in con- inica territories repeatedly in suitable habitat trast, are much lighter than males throughout throughout the areas surveyed. At the three the face and breast,and are almostalways light- principal study sites,both forms nested in close er than female dominicain full breeding plum- contactat relatively high densities(Fig. 2). With age.Many female fulva have underpartsmainly only moderate searching of these sites, we lo- white, with only scattereddark feathers on the cated29 nests(12 fulva, 17 dominica)and 15 ad- faceand breast,often forming an irregular, dark, ditional territories probably with nests(5 fulva, triangular patch near each eye. When postnup- 10 dominica);we also found nestsof both species tial molt begins in late June or July, pale feath- elsewhere in areas accessiblealong the roads. ers appear on the face and breast of all adults, Territory sizes and display areas, which were and distinctions between forms and sexes blur. very roughly estimated by observing display- Both sexesin fulva and dominicacan also be ing birds, were sufficiently large in comparison separatedby a difference in number of prima- with the closestinterspecific distances that both ries that extend past the tertials of birds stand- sexesof both speciescertainly had opportuni- ing with wings folded; dominicahas a longer ties for interaction throughout the nesting sea- primary extension, with four or five primary son at these and other sites. Because of the wide- tips exposed,whereas fulva showsonly three or ranging aerial displays and aerial chases,the fewer primary tips (Dunn et al. 1986). Use of infrequencyof territorial defenseon the ground, all these characteristicspermitted us to assign and the frequent absence of non-incubating an identity to every bird seen clearly at Seward birds from the territory, determining the loca- 12 CONNORS,MCCAFFERY, AND MARON [Auk,Vol. 110

-3 -1.8 1.8 3 Fig.3. Breedingdistributions of dominicaand fulva BREEDING-HABITAT DISCRIMINANT SCORE in Alaskaand adjacentregions. Fig. 4. Discriminant-functionanalysis of nesthab- itatsof dominica(n = 16;solid bars) and fulva(n = 12; tion of territory boundaries would have re- striped bars). Increasing(positive) scoresindicate in- quired more time than we could commit to that creasingslope and elevation,along with decreasing task. Therefore, we are unable to state whether vegetation cover, height and richness. golden-plovers of either form were more strict- ly territorial intraspecificallythan interspecifi- Breeding distributions away from the areas of cally. We did observefrequent interspecificin- sympatry are approximate, and are taken pri- teractions, some of which are discussed below. marily from Gabrielsonand Lincoln(1959), AOU We believe that the area of breeding sym- (1983), and Godfrey (1986). patry extends far beyond the study sites and Habitat differences.--Figure2 also documents highway system of the southern Seward Pen- a nonrandom topographic distribution of the insula. Both forms probably breed at mostSew- two forms,with fulva usuallyoccurring at lower ard Peninsula sites where suitable habitats oc- elevations than dominica at each of the sites. cur. During June 1988, U.S. Fish and Wildlife Results of the discriminant-function analysis Service personnel observed displaying or call- indicate significant differencesin nesting hab- ing birds of both speciesat nine widely sepa- itat betweenfulva and dominica(Fig. 4; P < 0.01). rated areas north and east of our study sites Reclassification of each nest based on its dis- (Robert E. Gill and Shelli Vacca pers. comm.). criminant score, a measure of the extent of sta- On YDNWR, breeding fulva were common at tistical separation between groups, correctly all study sites;breeding dominicawere limited identified 81% of dominicanests (13 of 16 cor- to Nelson Island, the Pikmiktalik River, and the rectly reclassified)and 92% of fulva nests(11 of Andreafsky River watershed. North of the Sew- 12). The dominica nests occurred more often in ard Peninsula,there may be additional areasof areasof higher elevation and slope, with spars- breeding sympatry. In that region, dominicais er and shorter vegetation, and more rocks, the common form, but specimensof fulva have whereas fulva nestswere usually at lower ele- been collected as far north as Barrow during vations in denser and taller vegetative cover. early June, and both McCafiery and Frank A. There is someoverlap of habitat, however, with Pitelka (pets. comm.) have seendisplaying male both forms using relatively dry, upland tundra. fulva at Barrow. The northernmost known These results are consistent with our observa- breeding recordof fulva, however, is from Cape tions of the nesting habitats of both golden- Krusenstern (67ø08'N, 163ø4YW), where dom- plovers at YDNWR, and consistentwith the ob- inica is the more common form (P. G. Connors servations of J. L. Dunn (in Kessel 1989) of an unpubl. data). altitudinal differencein nesting habitatsof the These records lead us to construct the distri- two forms on the Seward Peninsula. bution map in Figure 3. The areas of sympatry Vocalizationdifferences.--On the breeding shown necessarilyinvolve assumptionand in- grounds, the two forms share a behavioral rep- terpolation, but we believe that both forms ertoire that generally is similar acrossall Plu- probably breed at least occasionallythroughout vialis species.The vocalizations accompanying these areas wherever suitable habitat occurs. these behaviors,however, differ strikingly be- January1993] Speciationin Golden-Plovers 13

I SECOND I SECOND

Fig. 5. Sonagramsof vocalizationsaccompanying Fig. 6. Sonagramsof vocalizationsaccompanying territorial butterfly-flightdisplay by three male dom- territorial butterfly-flight display by three male fulva: inica,Feather River: top, 3 June 1988; middle, 7 June top, FeatherRiver, 7 June1989; middle, Cripple River, 1989; bottom, 7 June 1989. 31 May 1989;bottom, Crete Creek, 8 June 1989. tween dominicaand fulva. The differencebe- but subsequentnotes vary, especiallyin dura- tween forms in the vocalization accompanying tion. In spite of this minor variation within the male territorial display flight is especially forms, however, differences between forms are emphatic.During this displayflight, malesgive clear and unambiguous.After listening to bouts repeatedvocalizations while flying over their of butterfly-flight-displaycalls of more than 50 territories in a characteristic"butterfly flight" males of each form in the area of widespread (Drury 1961),using slow, measured wing beats, sympatry, we have not heard any mixed dis- with their wings almostvertical at the top of plays,nor intermediatevocalizations, nor have the stroke. The flight occursfrom 10 to 100 m we heard any individual giving a flight display above ground, and frequently rangeswidely, vocalization inappropriate to its plumage. crossingterritories of other plovers.Durations We have observed another notable difference of mostflights of both speciesrange from 30 s between flight displaysof the two forms. The to 3 rain, with callsgiven at ratesof 50 to 130 fulva males frequently begin butterfly flights per minute by dominicaand 20 to 40 per minute with a fluttering ascent, using a rapid, shallow by fulva. wing beat that is distinct from both a normal Sonagramsof the butterfly-flightcalls of three wing beat and the exaggeratedbutterfly-flight individuals of each form are shown in Figures wing beat. This flutter ascent may cover 50 to 5 and 6. The dominicacall is an abrupt "tlifik" 300m beforethe fulvamale begins the butterfly or "tdlifik" (comparewith "ktoodle•" of Drury flight. It is accompaniedby an abruptly termi- ! 96!), strikinglydifferent from the plaintiveand nated version of the butterfly-flight call (Fig. melodic"pe•-er-w•e" (J.T. Nicholsin Bent! 929), 7A-B). The bird also may use the fluttering as- or "pe•-chew-e•" of fulva. Some individual cent and flutter call briefly during an interrup- variation occurs within each form. In dominica, tion of a butterfly flight, sometimesin response calls of different individuals, and sometimes to the flight of another golden-plover through consecutivecalls of a single individual, may its territory. We have not observeda fluttering differ in the number of separateelements that flight in dominica. comprisethe call (Fig. 5). The first "twisted- There is also a common vocalization in dom- rope" note occursin all calls,but a note follow- inica for which we have not discovered a clear ing this is sometimesabsent, and the shapeand equivalent in fulva. This vocalization (Fig. 7C- duration of subsequentnotes varies. To our ears, D) is given by the pursuer in aerial chases,and however,this variation is barely noticeable.The is a frequent aspect of dominicabreeding be- sonagramsof fulva(Fig. 6) beginwith a similar havior. The sonagramshows an obvious simi- initial note having a twisted-rope appearance, larity with the butterfly-flight call of dominica, 14 CONNORS,McCAFFERY, AND MARON [Auk,Vol. I I0

.D I SECOND

Fig. 8. Sonagramsof complex whistles of three male dominica,Feather River: top, 7 June1989; middle, 3 June 1988; bottom, 3 June 1988. Arrows denote re-

I SECOND peating unit.

Fig. 7. (A and B) Sonagramsof vocalizationsaccompanying flutter flight by two male fulva, Crete of Sauer 1962), which occurs in several contexts. Creek (A, 7 June 1989; B, 8 June 1989); (C and D) In both dominicaandfulva, most butterfly flights sonagramsof chasecalls of two maledominica, Feather end with the male descendingin a glide with River (C, 3 June 1988; D, 7 June 1988). wings held over the back in a "V." Upon land- ing, the bird gives a short seriesof consecutive and is usually repeatedin a sequencewithout complex whistles (one to three, usually two) pauses.The call can be mimickedas "tdlifikit- while bobbing its head. This call may be an- tdlifikit" (this study) or "toodle•ka-toodle•ka" swered by a similar complex whistle from the (Drury 1961). female. The complex-whistle call is also given There may be some overlap in contextsbe- in other contextsinvolving pair communication tween the flutter call of fulva (Fig. 7A-B) and on the ground, in chases or other aggressive the chasecall of dominica(Fig. 7C-D). Bothforms interactionsboth on the ground and while fly- usethese respective calls in responseto the flight ing, and during butterfly-flight displays.Com- of an intruder through their territory, although plex whistles of three individuals of each form this occursmuch less often in the caseof fulva. are shown in Figures8 and 9. In eachsonagram In dominica,the call is given while in pursuit of the basic unit is identified between arrows; the the intruder, whereas fulva gives it while in full vocalization in each case consisted of this display over the territory. There also is an in- unit given twice consecutively.These calls are teresting correspondencein compositionbe- similar, but can be distinguished with practice. tween the two calls, in spite of their obvious The dominicacall (Fig. 8) can be mimicked by difference overall. Both calls begin as the but- "wit-weeyoo-wit," repeated.The fulva call (Fig. terfly-flight call of the form, but end with a 9) contains similar elements, but is longer and similar, sharplyjuxtaposed couplet. In dominica has a more bubbly, warbling quality. this is a separaterepetition of the lasttwo notes Alarm calls given at the nest also differ be- of the butterfly-flight call. In fulva the first part tween forms. Sonagramsof three males of each of the couplet is continuouswith a shortened form recorded near their nests are shown in and wavering version of the final note of the Figures 10 and 11. In each figure, sonagramsof butterfly-flight call. Yet, the result is that both low-intensity and high-intensity calls from the calls end similarly. We have heard both calls sameindividuals are shown. The shift to high- usedin interspecific,as well as intraspecificin- intensity alarm was elicited each time by the teractions. observer moving forward toward the calling The most similar vocalization frequently used golden-plover.The dominicamales (Fig. 10) first by both formsis a complexwhistle ("long call" reacted to the presence of the observer with a January1993] SpeciationinGolden-Plovers 15

w 0

I SECOND I SECOND

Fig. 9. Sonagramsof complexwhistles of three Fiõ. 10. Nest alarm ca]Is of three male d0mi,ica, male fulva: top, Crete Creek, 7 June 1989;middle, Crete Creek, 7 June 1989; bottom, Cripple River, 31 bottom, 18 June 1988. For each individual low-inten- May 1989.Arrows denote repeatingunit. sity call shown on left with high-intensity call on right. two-syllable call, "klee-yee•." As the bird be- camemore agitatedat the observer'sapproach, may have been a fulva, or an unusually pale the call shifted abruptly to a variable but mainly dominicafemale, or a more typical dominicathat four-syllable call, "killik-killik" (Drury 1961). had started to molt out of alternate plumage. The fulvamales began calling "peee" (this study) Our observationson the timing of molt at other or "pfeeb" (Sauer 1962) while the observerwas sites are consistent with this latter interpreta- far from the nest, and shifted suddenly to tion. "deedleek" (Sauer 1962) at the observer'sap- Two males on adjacentterritories at Curlew proach. Both forms used additional alarm calls Lake alsowere problematlc.Both exhibited the and other calls occasionally,but the male re- bulbous white patch on the upper breast char- sponsesto a human intruder (Figs. 10 and 11) acteristic of dominica. One of the two also had are fairly consistentwithin forms and quite distinctive between forms. Female alarm calls seemed similar, but more variable than male 4 alarm calls.

Assortativemating.--These differences in vo- 2 calizations provide a sufficient mechanism to maintain assortativeme, ting in areas of sympatry. In 1988and 1989,we identified both sexes in 96 pairs of golden-ploverson the southern Seward Peninsula (1988, 27 fulva, 29 dominica; õ 1989,21 fulva, 19 dominica)and, in all cases,strict ,?,o assortativemating prevailed; we did not dis- cover any mixed pairs. J. L. Dunn (in Kessel 1989) also found no mixed pairs on the Seward Peninsula. At sites on YDNWR, all pairs in which both members were identified exhibited assortative I SECOND mating (24 fulva, 9 dominica,1985-1990), al- Fig. 11. Nest alarm calls of three male fulva. For though we detected three possibleexceptions eachindividual, low-intensity call shown on left with to this pattern. At the Pikmiktalik River in late high-intensity call on right: top, Nugget Creek, 12 June 1985,a relatively pale female golden-plo- June 1988; middle, Horton Creek, 5 June 1988; bot- ver was paired with a dominicamale. The female tom, Feather River, 18 June 1988. 16 CONNORS,MCCAFFERY, AND MARON [Auk, Vol. 110 black, rather than white, undertail coverts. We two low marshy areasalong the Teller Road on did not determine the number of primary feath- 23 May, we found 18 dominica,9 fulva, and 6 ers extending past the tertials (sensuDunn et al. golden-ploversnot seenwell enough for iden- 1986) in either bird. However, both males gave tification, but no evidence of a single pair. These the fulvaflight call during butterflyflights, and two siteswere not characteristicbreeding hab- both were paired with fulva females. During itat for either form and, subsequently,were not three years of work at Curlew Lake, no un- occupiedby breeding pairs. equivocal dominicawere seen. We are inclined We recorded the first butterfly flights by both to view thesetwo malesas examples of variation specieson 24 May 1989. These occurredover in fulva, rather than as unusualdominica males open patches of typical nesting habitat. Over giving fulva vocalizations,involved in nonas- the next five days we observedseveral pairs of sortative pairings. both speciesengaged in equivalent stagesof Pairformation.--To what extentmight pairing pair formation. Displays seen in both species during migration, or differences in timing of were: nestscraping by the male;"torpedo" runs arrival on the breeding grounds, explain as- by the male near the female, with the male's sortativemating and the habitat differenceswe head and body level and back feathers ruffled; observed? In view of the distinct migration "tipping" displaysby the male in front of a nest routes and wintering areas, it is reasonable to cup, with the male facing away from the nest considereither of theseas possible mechanisms. cup, head down and tail up, and the female in Sauer (1962) stated that the fulva he observed the nest cup looking up at the male's undertail on St. Lawrence Island arrived on his study site coverts;and frequent brief copulations.Some already paired. If all fulva are paired during of thesepairing displaysoccurred in two early- migration south of Alaska before they can en- opening patchesof habitat not typical of our counter any dominica,assortative mating would previous year's habitat measurementsfor either be assured.Alternatively, if the speciesarrive species.At both of these sites, no golden-plo- on the breeding grounds at different times dur- vers were present 12 dayslater, when extensive ing the spring melt, with the earlier arriving areas of typical golden-plover habitat were species immediately pairing in the earliest snow-free elsewhere. When we first reached our available habitat, both assortative mating and FeatherRiver site on 5 June 1989,golden-plover the differencesin breeding habitat might result. densities and habitat use were similar to those Our observations on the Seward Peninsula, es- measured in 1988. At this site and elsewhere pecially in 1989, lead us to reject both the we did not find any large difference in early- prearrival-pairing hypothesis and the differ- seasonavailability of typicaldominica and fulva ential-timing hypothesisas having any impor- breeding habitat; areas of both habitats were tance in determining assortative mating and opening on a similar schedule. habitat use in these species. From these observations we believe that both In 1989, unusually heavy spring snows de- forms of golden-plover arrive in northwestern layed the thaw on the Seward Peninsula. The Alaskaunpaired, and on a similarschedule. Pair tundra near Nome was almost 100% snow-cov- formation may begin on the breeding territo- ered on 22 May, when we arrived. Golden-plo- ries, or it may begin at other sites within the vers of both forms had been among the earliest breeding distribution, especially if the breed- arriving shorebirds,with a few individuals of ing-territory habitat remainssnow-covered un- fulvaand dominicaseen over the previousthree til late spring. This may explain Sauer's (1962) days (R. E. Gill pers. comm.). We began observ- observations.When he first sawfulvaon a study ing golden-plovers in the earliest opening site on St. Lawrence Island (8 June), they were patchesof tundra along all three roadswithin already paired. He observednone of the pairing about 30 km of Nome on 23 and 24 May. On displayswe describehere, and even butterfly these two days we found both sexesand both flights were infrequent. There is no evidence, speciesat scatteredsites, with golden-plovers however, to suggestthat the birds were paired present in almost every open patch of non- south of Alaska, and we considerit likely that brushyhabitat. Our total count for two dayswas they had already paired at an early-openingsite 36 dominicaand 22 fulva. Most of thesewere in elsewhere on St. Lawrence Island or adjacent groups with no suggestionof pairs, no aggres- Siberia or Alaska. sion, and no displaysobserved. For example,in Interspecificinteractions.--Despite the evi- January1993] SpeciationinGolden-Plovers 17 dencepresented here of differencesin behavior with their larger relatives.On the SewardPen- and of assortativemating in dominicaand fulva, insula, we noted simultaneousbutterfly flights individuals do interact interspecifically quite and chasesbetween fulva and squatarola,and frequently on the breeding grounds. On the complex whistles given by fulva in apparent Seward Peninsula, interspecific aerial chases reaction to squatarolaflights over a fulva terri- between fulva and dominicaare common, but tory. Ingvar Byrkjedal (pers. comm.) has ob- these chase-and-flee interactions do not nec- servedaggressive interactions involving chases essarily indicate mutual communication.The and complex whistles between fulva and the two forms are very similar behaviorally, how- Greater (or Eurasian) Golden-Plover (Pluvialis ever, sharing many displays that recall their apricaria)on the Yamal Peninsula. Apparently, closeevolutionary relationship,and theseprob- interspecificaggression is not rare in the Cha- ably facilitate more involved interspecificcom- radrii. Pluvialisapricaria also engages in fierce munication. We frequently heard complex territorial battles with the Lapwing (Vanellus whistles (and occasionallyother vocalizations) vanellus) in northeastern Scotland (Parr 1979). given by both forms in reactionto the presence In Siberia,P. squatarolaregularly challengesand of the other form. We also observed interspe- expelsP. apricariafrom its territories (Flint and cific, coordinatedinteractions that might be in- Kondratiev 1977).Interspecific territoriality has terpreted as boundary disputes,entailing par- even been documentedbetween fulva and the allel marching, aggressivepostures and contact Ruddy Turnstone(Arenaria interpres) on St. Law- fights; these were seen as frequently as similar rence Island (Sauer 1962). intraspecific interactions. Despite the relative ease with which most One such interaction between a male dominica breedinggolden-plovers can be classified,birds and a male fulva on Seward Peninsula, 2 June that cannot be classifiedby plumage alone are 1989, lasted for 1 h and 45 min. It began with encounteredat a low frequency, especially on an aerial chase, but most interaction occurred YDNWR. Arguing that such "aberrant" birds on the ground, ranging over an area roughly are not unexpected, McKitrick and Zink (1988) 150 m in diameter. An initial period of parallel offered hybridization as only one of severalhy- marching, mainly 1 to 3 m apart, but over a potheses that can account for intermediate meandering route rather than a stable bound- plumage phenotypes.We cannot rule out the ary, escalatedinto repeated chargeson foot or possibilitythat theseintermediate forms are hy- wing by both individuals, with occasionalcon- brids, but we have discovered no conclusive tact. At times, dominicasat in a nest cup while evidence to support this hypothesis during six fulva circled on foot within 0.5 m; later, fulva field seasons.In fact, we know of only one def- performed scrapingdisplays with dominicaob- inite case in which the male of one form ap- serving. No femalesof either form were present proached a female of the other form in what at any time. Vocalizations were frequent and may have been a pairing context. On 28 May often given by one bird in consecutiveor si- 1986, an unpaired dominicamale within an es- multaneous responseto a call by the other bird. tablished territory in the Andreafsky Hills ap- We recorded 38 complex whistles by fulva and proachedan intruding fulva pair five consecu- 21 by dominica. tive times. The male dominica repeatedly This example indicates a remarkable amount approachedin the aggressivehunchbacked pos- of communication, and suggestsinterspecific ture described by Drury (1961). On one occa- territoriality, but should not be viewed as in- sion, the dominicamale got between the fulva validating our contention that these are sepa- male and female, abandoned the aggressive rate species.We have seen comparable inter- posture, and slowly approached the female. actions between dominica and the Black-bellied When he closedto within 1 m of her, the fulva Plover (P. squatarola)and between fulva and male attacked.The dominicamale may have been squatarola.On both Nelson Island and at Frank- attempting to court the fulva female. Whether lin Bluffson the North Slope of Alaska, we have this courtship would have proceededfurther in observedsquatarola and dominicamales engaged the absenceof the fulva male is unknown. How- in ritualized border displaysincluding parallel ever, we have never documented a confirmed marchesand squabbleswith physicalcontact (B. mixed pair; if these exist at all, they are rare J. McCafiery unpubl. data). At both sites, dom- occurrenceswithin a consistentand widespread inicamales did not yield ground in the fights pattern of assortativemating. 18 CONNORS,MCCAFFERY, AND MARON [Auk, Vol. 110

DISCUSSION istic habitat selection at the level of the entire territory. One factor that may influence selec- Assortativemating.--Previous analysis sug- tion of the particular nest site within the ter- gested sympatric breeding in the two forms ritory is cryptic concealmentfor the incubating (Connors 1983), but the evidence from museum bird. Byrkjedal (1989) has shown that cryptic specimens was not definitive. Field observa- concealment of dominica nests enhances nest tions reported here eliminate any potential am- success.Because dominica and fulva are extreme- biguity. We found dominicaand fulva to breed ly similar in appearancewhile incubating, they in closeand frequent sympatryover large areas both may selectsimilar nest microhabitatswith- of the Seward Peninsula, and in at least occa- in the range of habitats available on their ter- sional sympatry in other areas as far south as ritories. Nelson Island. Frequencyof sympatry north of Our data also may understate the degree of Seward Peninsula is unknown, but sympatry differencebetween averagehabitat typesof the probably occurs,at least occasionally.At the forms, becauseour sites were selectedto pro- heart of their sympatricrange on the Seward vide comparisonsof the formson closelyneigh- Peninsula,the two forms mate assortativelyin boring territories. Many nestswere in or near areas where contact between forms is common. the transition zone between habitat types;nests Vocalizations.--Our field data indicate strik- in areasof moreextreme habitat differences may ing differences in breeding vocalizations be- have been underrepresented. The habitat dis- tween fulva and dominica,and alsosome poten- tinctions are sufficiently clear that we quickly tial mechanismsto maintain assortativemating. learned to accuratelypredict speciesoccurrence We have not recorded and compared the full based on a cursory, intuitive assessmentof the vocalization repertoire of the forms, but com- habitat. In addition to these major differences mon breeding vocalizations used in territorial in behavior and ecology reported here, several advertisement and defense, in intrapair com- other lines of evidenceindicating evolutionary munication, and in nest defense differ mark- differences between the forms have been noted edly. The aerial-display-flightvocalizations are from previous studies. so distinctive that they show little relationship Distribution.--Theforms differ not only in past the first brief note, and geographicvaria- breeding distributions, which overlap in north- tion of these vocalizationsacross the breeding western Alaska (Fig. 3), but in wintering dis- rangesof the two forms appearsto be slight (E. tributions, which do not overlap. The dominica H. Miller and P. G. Connors in prep.). At least form winters in South America, while fulva one common vocalization of each form appears winters in southern Asia and on many Pacific to have no clear, contextual counterpart in the islands south to Australia and New Zealand other's repertoire, and fulva has a distinctive (AOU 1983).South of Alaska, migration ranges flutter flight that we have never observed in are essentiallynonoverlapping, except for very dominica.In the area of widespread sympatry, small numbers of birds of both forms that mi- we have never heard intermediate calls, mixed grate(and in the caseoffulva, winter) along the calls, or "wrong" calls in relation to any bird's Pacific coast of North America. plumage. These vocalization results indicate Morphology.--The two forms differ signifi- both independent evolution of the forms and cantly in morphology (Connors 1983), but with well-developed mechanisms that maintain as- some overlap (less than 10%) even after multi- sortative mating. variate statistical analysis based on measure- Breedinghabitat.--The forms also differ in ments of wing, tarsus and culmen. The form breeding habitat. The limited measurements dominicahas a longer wing, but shorter tarsus presented here define a statistically significant and culmen than fulva. difference between nest habitats. The measured Plumage.--The same specimen study also habitat overlap (Fig. 4) may actually understate identified plumage characteristicsuseful in sep- the degree of habitat distinction between ter- arating males in breeding plumage. Our field ritories of the two forms, because these data work, relying also on differencesidentified by refer only to habitat within 15 m of each nest, Dunn et al. (1986), indicated that field identi- a small area comparedwith the entire territory. ficationof breeding plumagebirds of both sexes Many factorsmay determine species-character- is possible, and is easier than identification of January1993] SpeciationinGolden-Plovers 19 museum skins. Juvenile and adult basic plum- inica(American Golden-Plover) and Pluvialisful- ages also are clearly separable (Hayman et al. va (Pacific Golden-Plover) as previously pro- 1986). posed (Connors 1983). Molt and maturation.--Molt and maturation schedulesalso differ. In dominica,first-year birds ACKNOWLEDGMENTS typically migrate northward to breeding grounds when less than one year old (Strese- Research support was provided by the National mann and Stresemann 1966). They apparently GeographicSociety to Connors (Grant Nos. 3537-$7, undergo wing molt prior to this migration, ar- 3765-88,4049-89). We thank Ron Perry, Yukon Delta riving with fresh primaries (Johnson 1985). In National Wildlife RefugeManager (USFWS), for sup- fulva, some,but not all, first-year birds migrate porting Connors'field work near Curlew Lake in 1987, to breeding grounds when less than one year and Robert E. Gill, Jr., SheHi Vacca, and their col- leagues (Alaska Fish and Wildlife ResearchCenter, old (Johnson and Johnson 1983, Connors 1983). USFWS) for logistic support and field observations These birds do not, however, acquire fresh pri- on the Seward Peninsula in 1988 and 1989. We thank maries until after the breeding season. Most Sarah Griffin for field assistance near Curlew Lake, first-year fulva remain on insular Pacific win- and OscarW. Johnsonand Philip Brunet for help on tering areas during their first potential breed- the Seward Peninsula. Brian Colter, Rick Ernst, Chris- ing season(Kinsky and Yaldwyn 1981, Johnson tine McCafiery, Gene Peltola, Jr., and Peter Zahler and Johnson 1983). supplemented McCaffery's golden-plover observa- Molecular genetics.--The one remaining line tionson YDNWR. We thank Peter Marlet and Doug- of evidence that could bear on the taxonomic lasNelson for assistancein producingsonagrams, and status of the two forms is molecular-genetic Frank A. Pitelka, J. v. Remsen,J.P. Myers, Ned K. Johnson,Kimberly G. Smith, Ingvar Byrkjedal,E. H. information. Specimenswe collectedfrom Sew- Miller, Kimball L. Garrett and D. P. Whitfield for ard Peninsula in 1989 have recently been ana- adviceand encouragementor for suggestionsfor im- lyzed for differences in mitochondrial DNA proving the manuscript. (Frank B. Gill, P. G. Connors, J. L. Maron in prep.). Detailed results will be reported sepa- rately, but the preliminary resultsare consistent LITERATURE CITED with all other lines of evidence:fulva and dom- AMERICAhi ORNITHOLOGISTS' UNION. 1983. Check-list inicaspecimens are geneticallydistinct, with no of North American birds, 6th ed. Am. Ornithol. indication of interbreeding. Union, Washington, D.C. Conclusion.--Thediversity and consistencyof BENT, A.C. 1929. Life histories of North American evidence make the conclusionof full speciation shore birds. U.S. Natl. Mus. Bull. 146. inescapable.This decision is reached whether BOCK,W. J. 1958. A generic review of the plovers one applies the biological-speciesconcept (Mayr (Charadriinae,Aves). Bull Mus. Comp. Zool. 118: 1969, 1970) or the phylogenetic-speciesconcept 25-97. BPa•sn OP,ui?noLcx;is?s' UNION RECORDS COMMITTEE (Cracraft 1983, McKitrick and Zink 1988). These (BOURC). 1986. RecordsCommittee, twelfth re- are independently evolving forms that breed port. Ibis 128:601-603. assortativelyin areas of sympatry. They have BYRKJEDAL,I. 1989. Nest habitat and nesting success been consideredsubspecies only becausetheir of Lesser Golden-Plovers. Wilson Bull. 101:93- plumagesare sufficiently similar that ornithol- 96. ogistshad found them difficult to separate.It is CONNORS,P. G. 1983. Taxonomy, distribution, and now apparent that even this aspectof the two evolution of golden plovers (Pluvialisdominica and forms is more distinctive than previously ap- Pluvialisfulva ). Auk 100:607-620. preciated. The golden-plover situation is com- CONOVER,B. 1945. The breeding Golden Plover of parable to the past subspeciestreatment of the Alaska. Auk 62:568-574. Long-billed and Short-billed dowitchers (Lim- CRACRA•,J. 1983. Speciesconcepts and speciation analysis.Curr. Ornithol. 1:159-187. nodromusscolopaceus and L. griseus),whose tax- DRURY,W. H., JR. 1961. The breeding biology of onomic statuswas resolvedonly after sufficient shorebirdson Bylot Island, Northwest Territo- attention was given to the differencesbetween ries, Canada. Auk 78:176-219. the forms (Pitelka 1950). The golden-plovers DUNN, J. L., J. MORLAN,AND C. P. WILDS. 1986. Field differ as dramatically,and shouldbe treatedas identification of forms of Lesser Golden-Plover. separate speciesunder the names Pluvialisdom- Pages71-73 in International bird identification. 20 CONNORS,MCCAFFERY, AND MARON [Auk,Vol. 110

Proceedingsof the 4th International Identifica- MAYR, E. 1970. Populations, speciesand evolution. tion Meeting, Eilat, Israel. Belknap Press of Harvard Univ. Press, Cam- FLINT, V. E., AND A. J. KONDRATIEV. 1977. Materials bridge. on the biologyof the Gray PloverPluvialis squata- MAYR, E., AND L. L. SHORT. 1970. Species taxa of rola. Beitr. Vogelkd. 23:265-277. North American birds: A contribution to com- GABRIELSON,I. N., AND F. C. LINCOLN. 1959. Birds of parativesystematics. Nuttall Ornithol. Club Publ. Alaska. Wildlife Management Institute and No. 9. StackpoleCo., Washington,D.C. McKrrRIcK, M. C., AND R. M. ZINK. 1988. Species GODFREY,W.E. 1986. Birds of Canada. National Mu- conceptsin ornithology. Condor 90:1-14. seums of Canada, Ottawa. PARR,R. 1979. Sequential breeding by golden plo- HAYMAN, P., J. MARCHANT, AND T. PRATER. 1986. vers. Br. Birds 72:499-503. Shorebirds:An identificationguide to the waders PETERS,J.L. 1934. Check-list of birds of the world, of the world. Houghton Mifflin, Boston. vol. 2. Harvard Univ. Press,Cambridge. JOHNSON,O. W. 1985. Timing of primary molt in PITELKA,F. A. 1950. Geographicvariation and the first-yeargolden-plovers and someevolutionary speciesproblem in the shore-birdgenus Limnod- implications.Wilson Bull. 97:237-239. romus. Univ. Calif. Publ. Zool. 50:1-108. JOHNSON,O. W., ANDP. J. JOHNSON.1983. Plumage- SAUER,E. G.F. 1962. Ethologyand ecologyof golden molt-agerelationships in "over-summering"and ploverson St. LawrenceIsland, BeringSea. Psy- migratoryLesser Golden-Plovers. Condor 85:406- chol. Forsch. 26:399-470. 419. STRESEMANN,E., AND V. STRESEMANN. 1966. Die Mau- KESSEL,B. 1989. Birds of the Seward Peninsula, Alas- ser der Vogel. J. Ornithol. 107:1-448. ka. Univ. Alaska Press, Fairbanks. VAURIE,C. 1964. Systematicnotes on Palearcticbirds. KINSKY, F. C., AND J. C. YALDWYN. 1981. The bird No. 53. Charadriidae:The generaCharadrius and fauna of Niue Island, SouthwestPacific, with spe- Pluvialis. Am. Mus. Novit. 2177. cial notes on the White-tailed Tropic Bird and WILKINSON,L. 1988. SYSTAT: The systemfor statis- Golden Plover. Natl. Mus. N. Z. Misc. Ser. tics. SYSTAT, Evanston, Illinois. No. 2. MAYR,E. 1969. Principlesof systematiczoology. Mc- Graw-Hill, New York.