MICROGEOGRAPHICSONG DIALECTSIN THEORANGE-TUFTEDSUNBIRD (NECT ARINIAOSEA )

by

NOAM LEADER1,2), JONATHAN WRIGHT 3) and YORAM YOM-TOV 1,4) (1 ZoologyDepartment, Faculty of Life Sciences, T el-AvivUniversity, Tel-A viv69978, Israel; 3 Schoolof BiologicalSciences, University of Wales, Bangor, UK)

(Acc.3-VIII-2000)

Summary Inastudyof male song in theorange-tufted sunbird ( Nectariniaosea )inan urbanneighbor- hoodin Ramat-A viv,Israel, we discovered the occurrence of songvariation on a microgeo- graphicscale in the form of twodistinct dialects with a sharpboundary between them. The maindistinction between the two song dialects is the frequency of thetrill, which comprises theterminal part of the song. A largedifference of 2-3kHz in thepeak frequency of the trill wasdiscovered between the two dialects, which could be easilydistinguished by ear.Thirty- sevenmales were recorded singing the ‘ low’dialect and 21 birdssang the ‘ high’dialect. Four otherbirds sang both dialects or ‘ hybrid’songs. Along the boundary that separated the two dialectpopulations, neighboring birds sang different dialect songs, although they were only 20-30meters apart. All four ‘ bilingual’birds occupied territories near the dialect boundary. Thehistorical processes leading to the formation of this dialect system may result from the patternof humansettlement at the time of the establishment of this neighborhood in the early 1950’s.The spatial distribution of the two sunbird dialect populations, and the apparent low dispersalrates of birdsfrom their natal dialect area, suggest the existence of a mechanism, whichcurrently maintains these dialects at the current boundaries.

2) Correspondingauthor; e-mail address: [email protected] 4) Wethank A. Elkayamfor assistance in the Ž eldand R. Zilbermanfor valuable information andhelp. We are grateful to L.F .Baptista,C.K. Catchpoleand P .K.McGregor for providing helpfulcomments on earlier drafts of this manuscript. This study was supported by a grant fromthe Israeli Academy of Sciences and Humanities, Tel-A vivUniversity and the Israel Cohenchair for Environmental Zoology to YY . c KoninklijkeBrill NV ,Leiden,2000 Behaviour137, 1613-1627 ® 1614 LEADER,WRIGHT&YOM-TOV

Introduction

Geographicalvariation invocalizations has beendocumented in a wideva- riety ofavianspecies (Mundiger,1982). Some species showa geographical mosaic oflocal populations,each characterized bydistinctive variants ofa givenvocalization. These different groupsof individualsconstitute ‘dialects’ whichmay be deŽ ned as separate, adjacent aggregationsof birds,with well- deŽned boundaries, and with different vocalizations (Rothstein &Fleischer, 1987).The variations inbird song may even occur on a microgeographic scale, betweenneighboring groups of birds, which might, at least poten- tially,interact orinterbreed with eachother (reviewed in Mundinger,1982). Suchmicrogeographic variation is synonymouswith the term ‘local dialects’ (McGregor,1980). The most striking exampleof microgeographic variation insong is wherebirds haveonly one song type (or a verysmall repertoire) andthere is close sharingbetween neighbors. In such cases notonly may the songspresent differ betweenlocalities, butit is also frequentlyfound that the changefrom place toplace is notgradual, rather asuddenswitch fromone dialect to anotheroccurs at aparticular place (Catchpole& Slater, 1995). Thebest knownexample of sharp dialect boundariesis in the sedentaryrace ofthe white-crownedsparrow ( Zonotrichialeucophrys nuttalli ),originally studied byMarler &Tamura(1964 )andby Baptista (1975). Explanationsfor these patterns ofvariation are historical accidents, i.e. non-adaptivecultural foundereffects (Thielcke,1969, 1973; King, 1972; Baker,1975; Baptista, 1975,1977; Mundiger, 1980; Payne, 1981 ),genetic differences betweenpopulations ( e.g. Nottebohm,1972; Baker & Cunning- ham,1985 ),physical circumstances, i.e. different songtypes beingmore or less suitable indifferent habitats becauseof their acoustical properties ( e.g. Morton,1975; Wiley & Richards,1978; Gish &Morton,1981), and Ž nally, social behavior, i.e. interactions betweenneighboring males (Payne,1981, 1982,1983; Rothstein &Fleischer, 1987;Payne et al., 1988; Bell et al., 1998). Theorange-tufted sunbird ( Nectarinia osea )is asmall (6-8g ),socially monogamouspasserine, showingsubstantial sexual dimorphismin plumage colors andbody size. It is territorial andsedentary throughout most ofits range.At the beginningof the 20thcentury this sunbirdwas considered extremely rare in Israel, andwas restricted mainly to the southernparts ofthe Jordanvalley anddesert oases nearthe shores ofthe DeadSea at ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1615

Jericho andEn-Gedi (Paz, 1986 ).Since the 1930’s,it has expandedits range dramatically andtoday is averycommon resident in gardensnear human dwellings,throughout Israel. This expansionis attributed to the abundance ofornamentalgardens, plentiful with cultivated nectar-secreting owersyear round. Cramp& Perrins (1993)describe orange-tuftedsunbird song as asweet sibilant warble,composed of a fewintroductory notes followedby a rapid hardmusical trill. However,apart fromthis short description,the songof the orange-tuftedsunbird has neverbeen studied thoroughly.Virtually nothing is knownof dialects, ofother spatial songvariations orof inter-individual variation. Theaims ofthis studyare (1)to describe the songof the orange-tufted sunbirdand (2 )to examinepatterns ofgeographic variation in the songof this species.

Methods

Thestudy was carried out at Ramat-A viv,a suburbanneighborhood of T el-Aviv,Israel (32°05¢ N, 34°47¢ E)during1997-1999. The study site is about1.5 km 2,situatedin an urban area,with a mixtureof private gardens, lawn and park areas between residential buildings and privatehomes. Most vegetation consists of many introduced nectar-secreting plants, which attractsunbirds ( e.g.Hibiscus spp., Malvaviscusarboreus , Tecomariacapensis , Aloe spp., Callistemon spp., Erythrina spp.). Thisparticular sunbird population has been the subject of continuousbehavioral research formore than 10 years(Goldstein et al.,1986,1987; Goldstein & Yom-Tov,1988; Markman et al.,1995,1996; Zilberman et al.,1999).Therefore, most birds recorded were color-banded allowingindividual recognition. The sunbird’s tolerance to human presence in urban areas, allowedfor relatively easy observations and song recordings without any apparent disruption ofsunbirdbehavior. Vocalizationsof 63 territorial male orange-tufted sunbirds were recorded extensively duringthe entire length of thethree breeding seasons ( February-September1997-1999). This included51 males that were present on their territories for the entire three year duration and 12malesthat were present for at least one breeding season ( 5ofwhichreplaced males that haddisappeared). Allrecordings were made within close proximity to singing birds (5-10 m) using a SennheiserME-67 ‘ shotgun’microphone and a SonyWM-D6 cassette recorder. Songs weredownloaded to a computerfrom a SonyTC-KE500S 3-headstereo cassette deck and weredigitized at 22050 Hz with 16 bit precision, band-pass Ž lteredfrom 2 to10 kHz to eliminateurban background noise and stored as individual computer Ž les.Spectrograms usedfor measurements were produced with a 512-pt.FFT (frequencyresolution 55 Hz, timeresolution 2.90 ms, Hamming window and 87.5% overlap) using A visoftSASLab Pro c forWindows V ersion3.4 ( Avisoft ® 1998).All measurements were taken using A visofts’ 1616 LEADER,WRIGHT&YOM-TOV

Fig.1. Spectrogram of a typicalmale orange-tufted sunbird song. Detailed are the four phrasesin the song as well as the method of calculation of acoustical variables noted in Table 1. built-inmeasurement cursors. 711 songs from 63 territorial males were recorded and used intheanalyses. Between Ž veand 41 songswere recorded from each individual ( x¯ SE = 11.85 1.34songs/ bird, N = 63). § Inaddition § to songrecordings in theRamat-A vivstudy site, we made a fewrecordings of malesunbird song in the surrounding neighborhoods as wellas at afewlocalities throughout Israel.

Methodsof analyses

Todescribeacoustic variation in thesongs from different individuals, we divided each song intodifferent phrases as shown in Fig. 1. Theterms we used to describesong are illustrated in Fig.1 andfollow the terminology of Catchpole& Slater(1995). An element is acontinuous traceon a spectrogram,separated by atleast 3 msfrom other traces. A syllableis arepeated unitof one or moreelement. Wemeasured24 acoustical variables on each song (Fig. 1, Table1). After examining the correlationmatrix, we entered the variables into a principalcomponents analysis. A principal componentanalysis creates a smallset of summary variables that efŽ ciently summarize the variationpresent in the original set of 24variables.

Results Songcharacteristics Inthe Ramat-Avivstudy site, onceterritories are established, male orange- tufted sunbirdsstart singinga single stereotyped‘ full song’delivered from ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1617

TABLE 1. Acoustical variables measuredin spectrograms of the ‘high’and ‘low’ dialects in the Ramat-Avivsunbird population

Variables Methodof calculation SFLFREQ Min.frequency of ‘First’element (kHz) SSLFREQ Min.frequency of ‘Second’element ( kHz) SMLFREQ Min.frequency of ‘Middle’element ( kHz) STRLFREQ Min.frequency of ‘Trill’element (kHz) SFHFREQ Max.frequency of ‘First’element ( kHz) SSHFREQ Max.frequency of ‘Second’element (kHz) SMHFREQ Max.frequency of ‘ Middle’element (kHz) STRHFREQ Max.frequency of ‘Trill’element (kHz) SFBAND Frequencybandwidth of ‘ First’element ( kHz) SSBAND Frequencybandwidth of ‘ Second’element (kHz) SMBAND Frequencybandwidth of ‘ Middle’element (kHz) STRBAND Frequencybandwidth of ‘Trill’element (kHz) SFMODUL Frequencymodulation of ‘First’element ( D kHz /D s) SSMODUL Frequencymodulation of ‘Second’element ( D kHz /D s) SMMODUL Frequencymodulation of ‘Middle’element ( D kHz /D s) STRMODUL Frequencymodulation of ‘ Trill’element ( D kHz /D s) STOTALTIME Totalsong duration ( s) STRILLTIME Durationof ‘ Trill’element (s) SFIRST No.of ‘First’notes SMIDDLE No.of ‘Second’notes SSECOND No.of ‘Middle’notes STRILLS No.of ‘Trill’notes SNNOTES Totalno. of notes TRILLRATE No.of ‘Trill’notes / durationof ‘ Trill’element highvantage points such as tree-tops oroverhead wires. Males continueto singthroughout the breedingseason (February-September).Mean rate of songdelivery of males was 2 .56 0.23songs/ min.(range 1-15, N = 60 males). Singingis most pronounced§ during the nest buildingstage, during whichtime males singfrom a highsong-post just abovethe nest, while the female is busycollecting nesting material. Males are aggressive in defense oftheir territory at this time andlaunch attacks fromthese song-postsat approachingmale rivals, as territories are especially proneto trespass by males seekingextra-pair fertilizations afewdays before the resident female starts laying(Goldstein &Yom-Tov,1988). Twodistinct variations ordialects ofthis songwere discovered in the studyarea, which were maintained duringthree breedingseasons (see below; Fig. 2). 1618 LEADER,WRIGHT&YOM-TOV

Fig.2. Spectrograms of orange-tufted sunbird song dialects in Ramat-A viv:( A-D)‘ low’ dialectsong; (E-H) ‘high’dialect song; (I) exampleof a ‘bilingual’male. This male was recordedsinging ‘ low’(I-1) and ‘ high’dialects ( I-2)and mixed dialect songs (I-3, I-4) in one continuoussong-bout. ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1619

Fourdifferent elements havebeen identiŽ ed, appearing in most individu- als’ song(Figs 1and2). These elements appearin well-deŽ ned syllables and syntaxis retained inall songs.Most songsstart witha Žrst syllable consist- ingof 1-8 frequency-modulated elements (Felement). IndividualF elements are oftengiven throughout the yearas calls, howevermales start producing them in anorganized syllable formjust beforethe beginningof the breed- ingseason, as aprecursorto full song.The second syllable typeconsists of1-4less frequencymodulated whistle elements (Selement). This syllable is sometimes absent especially in the ‘high’dialect song.The third syllable typeconsists ofahigh-pitchedelement precedingthe trill (Melement). Inthe Ramat-Avivpopulation this syllable is sometimes absent in the ‘high’dialect song,and consists ofonly one element in the ‘low’dialect song(only one birdwas recordedgiving 2-3 M elements). Inother locations, this syllable mayconsist of1-3M elements. Thelast part ofthe songalways consists ofa slow trill madeup of1-12elements andis veryvariable in quantitybetween males. Themain distinction betweenthe twodialects foundin the studypopula- tion is the frequencyof the trill. Trill elements ofboth dialects start at about 3.6-3.8kHz, yet the ‘low’dialect’s peaktrill frequencyreaches ameanof 5.1 kHz(4.7-5.4 ),whereas that ofthe ‘high’dialect reaches ameanof 7.3 kHz (6.9-7.9).There is thus nooverlapin peak frequencies and all songscould beassigned unequivocallyto oneof the twogroups. This large frequency differenceof 2-3 kHz between the twodialects canbe distinguishedby ear while in the Želd. Thirty-sevenmales wererecorded singing the ‘low’dialect and21 birds sangthe ‘high’dialect. Fourbirds sangboth dialects or‘ hybrid’songs (Fig.2). Song types didnot vary within males overthe breedingseason or fromyear to year.In all the cases wherea territorial male disappearedeither duringor after the breedingseason, the newmale that occupiedthe territory sangan identical songtype.

Quantitative analysis ofacoustic variation

Ina principal components(PCA )analysis the Žrst three principal compo- nents extracted explained30.1%, 14.5% and 10.7% of the total variation in the data set, whereasthe remainingfour explained less than10% each(8.2%, 7.5%,6.0% and 4.5% ).Together, these sevencomponents account for 81.6% 1620 LEADER,WRIGHT&YOM-TOV

TABLE 2. Details ofprincipalcomponents analysis performedon 24 acousti- cal measures of sunbirdsong

PC Eigenvalue %totalvariance Cumul. % 1 7.23 30.1 30.1 2 3.49 14.5 44.7 3 2.58 10.7 55.4 4 1.98 8.2 63.6 5 1.80 7.5 71.1 6 1.43 6.0 77.1 7 1.07 4.5 81.6

Sevenprincipal components were extracted. Eigenvalues greater than one show principal componentsthat explain more variance than a singleoriginal variable. ofthe total variancein these data.Although the Žrst principal component mayaccount for the most variation,it maynot be the axis ofmost biologi- cal interest (McGregor,1992 ).Table 2showsall sevenprincipal components extracted havingEigenvalues in excess ofone. Seventeenof the 24variables hadcorrelations greater than0.7 with the sevenprincipal components(Table 3). PC1 corresponds to anaxis primarily ofincreasing bandwidthof the trill (mainly dueto rising maximumtrill frequency),decreasing trill rate andfewer S elements. PC2corresponds to an axis primarily ofdecreasingtrill lengthand fewer trill elements. Increasing values onPC3correspond to increasing total songlength, increasing number ofF elements anddecrease ofminimum frequency of F elements. PC4 showsincreasing bandwidthof type F elements anddecreasing frequency modulationof type S elements. PC5shows decreasing values ofminimum andmaximum frequencies of S elements. PC6corresponds to anaxis of decreasingvalues ofbandwidth of F elements, their maximumfrequency , anddecreasing frequency modulation of Felements. PC7shows increasing values ofthe maximumfrequency of M elements. t-tests betweendialects (Table 4)showedsigniŽ cant differences forPC1, PC5,PC6 and PC7. No differencebetween dialects was foundfor PCs 2,3, and 4. Tosummarize, the major distinctions betweenthe dialects are that the highdialect has amuchhigher peak trill frequency,larger trill bandwidth anddecreased trill rate. Thehigh dialect also has alowerpeak frequency ofthe Felement, as well as decreasedbandwidth and frequency modulation ofthe Felement, ahigherpeak frequency of the Melement, andfewer S ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1621

TABLE 3. Correlation coefŽcients betweenacoustic variables andthe seven principal components

Factor1 Factor2 Factor3 Factor4 Factor5 Factor6 Factor7

SFLFREQ ­ 0.08 ­ 0.04 ­ 0.79 0.23 ­ 0.070.04 0.13 SSLFREQ 0.19 0.05 ­ 0.09 ­ 0.30 ­ 0.90 0.06 0.10 SMLFREQ 0.36 ­ 0.05 0.09 ­ 0.26 ­ 0.21 0.34 0.71 STRLFREQ 0.070.41 0.05 ­ 0.06 ­ 0.600.14 0.39 SFHFREQ ­ 0.37 0.16 ­ 0.310.11 0.04 ­ 0.78 ­ 0.06 SSHFREQ 0.18 0.01 ­ 0.04 0.14 ­ 0.89 0.08 0.12 SMHFREQ 0.140.04 0.03 0.16 ­ 0.27 ­ 0.12 0.85 STRHFREQ 0.84 0.01 0.06 ­ 0.05 ­ 0.290.20 0.33 SFBAND ­ 0.310.22 0.29 ­ 0.09 0.08 ­ 0.82 ­ 0.13 SSBAND 0.14 ­ 0.12 0.02 0.83 0.14 0.11 ­ 0.11 SMBAND ­ 0.18 0.09 ­ 0.17 0.62 ­ 0.02 ­ 0.43 ­ 0.14 STRBAND 0.89 ­ 0.09 0.05 ­ 0.04 ­ 0.180.19 0.27 SFMODUL ­ 0.13 0.03 ­ 0.03 ­ 0.04 0.11 ­ 0.91 ­ 0.04 SSMODUL ­ 0.310.16 0.02 0.72 ­ 0.010.03 0.33 SMMODUL ­ 0.27 ­ 0.09 ­ 0.130.05 0.06 ­ 0.35 0.08 STRMODUL 0.59 ­ 0.09 ­ 0.20 ­ 0.22 ­ 0.400.15 0.30 STOTALTIME 0.11 ­ 0.34 0.81 0.160.04 0.12 0.08 STRILLTIME 0.41 ­ 0.85 ­ 0.010.02 0.04 0.20 ­ 0.07 SFIRST 0.19 0.28 0.74 0.020.03 0.32 0.39 SMIDDLE ­ 0.230.38 0.41 0.13 ­ 0.09 ­ 0.31 ­ 0.34 SSECOND ­ 0.77 0.01 ­ 0.220.16 0.01 ­ 0.38 ­ 0.09 STRILLS ­ 0.01 ­ 0.97 ­ 0.01 ­ 0.070.07 0.05 ­ 0.03 SNNOTES ­ 0.39 ­ 0.680.54 0.07 0.08 0.00 0.14 TRILLRATE ­ 0.88 0.09 ­ 0.02 ­ 0.06 0.06 ­ 0.24 0.09 Expl. var 4.352.81 2.64 2.03 2.44 3.14 2.16 Prp. totl 0.180.12 0.11 0.08 0.10 0.13 0.09

Variableswith r > 0.70are in italics. elements inthe song,which if theyexist, havea higherpeak frequency and also beginat ahigherfrequency .

Geographicalvariation

TheRamat-A vivpopulation is separated into twodistinct areas, eachwith a distinct dialect anda sharpboundary (Fig. 3). The ‘ high-low’dialect border appearsto runin astraight line fromNW to SE.Theborder comprises arecreational park(composed mainly of Tipuanatipu trees), high-school groundsand a shoppingcenter, areas whichare lackingsuitable nectar- 1622 LEADER,WRIGHT&YOM-TOV

TABLE 4. t-test of principal factors between‘ high’and ‘ low’ dialects

PC Mean low Mean high t p PC1 ­ 0.691 0.994 ­ 10.61 < 0.000001 PC2 ­ 0.051 ­ 0.151 0.36 0.716 PC3 ­ 0.067 0.053 ­ 0.42 0.675 PC4 0.046 ­ 0.107 0.55 0.581 PC5 0.254 ­ 0.354 2.24 0.029 PC6 ­ 0.187 0.393 ­ 2.12 0.039 PC7 ­ 0.229 0.540 ­ 3.02 0.004

Nhigh = 21, Nlow = 35.

Fig.3. Map of the Ramat-A vivstudy showing the territorial boundaries of orange-tufted sunbirdsin 1998. ( L)‘low’dialect; (H) ‘high’dialect; (BL) bilingualmale. ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1623 secreting plants forsunbirds. At the NWas well as the SEcorners,however, bothsides ofthe bordercontain suitable territories andneighboring birds canbe heard singing different dialect songs,although they are only20- 30meters apart.All four‘ bilingual’birds singingboth dialects or‘ hybrid’ songs,occupied territories nearthe dialect border. Thetwo dialects seem tobe conŽ ned to the studyarea anddo not continueinto neighboringareas. Recordingsmade adjacent to the NEcorner ofthe studysite (Tel-AvivUniversity campusgrounds and the Afeka neighborhood)reveal different patterns ofsong.Preliminary recordingsfrom anumberof locations inIsrael reveal the occurrenceof distinct dialects on amacrogeographicscale, some varyingin tempo,syntax, frequency and the shapeof elements. This Žndingis less surprising,as most populationsare isolated fromone another by large areas ofunsuitable habitat. Wefoundvery low rates ofdispersal outside the dialect areas. Of21 markedbirds singingthe lowdialect, 16were banded as adults between 1989and 1998 in the studyarea, three werebanded as Žrst yearbirds and twoas nestlings. Ofthese onlythree birds (14%,one adult, one Ž rst year birdand one nestling )werebanded outside the currentdialect boundaries. Ofninemarked birds singingthe highdialect, sevenwere banded as adults between1994 and 1998, and 2 werebanded as Žrst yearbirds. None of these birds werebanded outside the currentdialect boundaries.One male (OAWY) bandedin 1997as anestling in the highdialect area,occupied a territory at the boundaryin 1998and sang a mixeddialect song.

Discussion

Themain Žndingsof our study are that the songsof the orange-tufted sunbirdshow marked microgeographical variation within asmall area. Twoadjacent anddistinctive local dialect populationsare present with asharplydeŽ ned boundary between the different dialect areas. Bilingual males, singingboth dialects, occupyterritories onthe boundary.This close geographicassociation ofbirds singingthe same basic songand the lack ofvariability withingroups is characteristic oflocal dialects (McGregor& Thompson,1988). Dialects andgeographical song variation has also beendiscovered in anotherspecies ofsunbird, the splendidsunbird, Nectarinia coccinigaster 1624 LEADER,WRIGHT&YOM-TOV

(Grimes, 1974;Payne, 1978 )andprobably exist in at least twoother sunbirds species, Nectarinia kilimensis and Nectarinia chalybeus ,as well (Grimes, 1974). Colonizationof new areas bysongbirds is thoughtto beone process throughwhich regional differences insong may arise (Baker& Cunningham, 1985).Sunbird settlement in manyparts ofIsrael appearsto be dependenton nectar producingornamental bushes and trees plantedby peoplein gardens. Onecan speculate onthe historical processes leadingto the formationof the Ramat-Avivdialect system. This neighborhoodwas established inthe early 1950’sandhas slowlyexpanded throughout the years.Old aerial photographsreveal that the neighborhoodgrew out of twoconcentrations of homes,separated byvacant, barren areas. Apparently,at least twofounding populationsof sunbirds settled nearthese homesand gardens in different parts ofRamat-Aviv. Each population could have sung a different variant ofthe song,based on the place theyoriginated from. Y oungbirds would havesettled in vacantterritories neartheir natal territories andwould start singingthe local song,learned from their fathers andimmediate neighbors. This wouldpersist, resulting in small clusters ofsunbird territories around private homesand gardens existing at the time, eachcluster ofbirds singing adifferent dialect. Humanexpansion through the years will haveslowly connectedthe twoneighborhoods and thus connectedthe twosunbird concentrations,leading to a seemingly homogenouspopulation today ,yet with twodialects. Interestingly, aerial photographsclearly showthe neighborhoodwas fully built andinhabited by people,with well-developedgardens, as early as 1963. Thus,it seems oddthat whilst it tookprobably less than10 years forthese sharpdialect boundariesto arise dueto ageographicalbarrier, an additional 35years wouldhave gone by after the isolating barrier hadbeen lifted, withoutthem diffusing. Theissue ofhow local dialects are maintained andwhether they have signiŽcant evolutionaryconsequences has beenthe subject ofintense interest andcontroversy in recent years (summarized inKroodsma et al.,1984;Baker &Cunningham,1985; Rothstein &Fleischer, 1987). Catchpole& Rowell (1993)have shown that inthe case oflocal dialects ina populationof the Europeanwren ( Troglodytestroglodytes ),a small yet signiŽcant natural boundary,suchas a200meter widelake, is all that is neededfor separate songtraditions to develop.According to them,such ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1625 boundariesare clearly insigniŽcant in terms ofdispersal, butmay form aneffective social barrier betweenboth populations (Catchpole & Rowell, 1993). Inour study ,the area alongthe boundarywhich is unoccupiedby sunbirds (arecreational park,school grounds and a shoppingcenter; Fig.3), is not onlysmaller, it also doesnot provide a complete separation betweenthe two dialect populations.In our case, however,it seems that dispersal outside the dialect area is uncommon.This is consistent withearlier Žndingsof ahigh degreeof relatedness betweenneighboring individuals in ourstudy area, as evidencedfrom high proportions of band sharing discovered in genetic analysis ofthis sunbirdpopulation (Zilberman et al., 1999). Thespatial distribution ofthe twosunbird dialect populationsin Ramat- Aviv,togetherwith the observeddispersal patterns stronglysuggest the existence ofa mechanism,which is notmerely the consequencesof a nonadaptivecultural foundereffect 50years ago,that currentlymaintains these dialects at the currentboundaries. Moreover, if sucha mechanism doesindeed exist, it is natural to assume that oneof its main characteristics is the ability ofbirds to discriminate betweentheir ‘home’dialect and foreigndialects. Indeed,preliminary results fromplayback experiments on male sunbirdshave revealed a signiŽcant differential behavioralresponse by males to playbacksof lowand high dialects. Suchbehavioral discrimination implies the capability torecognize differences betweenthe twodialects by the birds. Thepronounced differences wehavediscovered between these twoadja- cent sunbirddialects in the peaktrill frequency(2 to 3kHz),seem substantial comparedto those reportedin manypublished song dialect studies onother songbirdspecies. Thesunbird dialect system in Ramat-Avivis also unique in that it is recent andbased mainly onhuman impact oncolonization.We are currentlyresearching the stability andfunction of these dialects in two avenues:(1 )investigating the genetic compositionof males ofthe twodi- alects, whichwill helpus to establish the rate ofgene  owbetween the two dialect areas, andto determine if andhow the songdialects reportedhere affect gene owin this species; and(2) conducting playback experiments in orderto investigate discrimination capability anddifferential responses of males andfemales to different songdialects. 1626 LEADER,WRIGHT&YOM-TOV

References

Baker,M.C. (1975). Song dialects and genetic differences in white-crowned sparrows (Zonotrichialeucophrys ).— Evolution29, p. 226-241. ——&Cunningham,M.A. (1985). The biology of bird-song dialects. — Behav.Brain Sci. 8,p. 85-133. Baptista,L.F .(1975).Song dialects and demes in sedentarypopulations of the white-crowned sparrow (Zonotrichialeucophrys nuttalli ).— Publ.Zool. Univ. Calif. 105, p. 1-52. ——(1977).Geographical variation in songand dialects of the Puget Sound white-crowned sparrow.— Condor79, p. 356-370. Bell,D.A., Trail, P .W.&Baptista,L.F .(1998).Song learning and voacl tradition in Nuttall’ s white-crownedsparrows. — Anim.Behav. 55, p. 939-956. Catchpole,C.K. &Rowell,A. (1993).Song sharing and local dialects in apopulationof the Europeanwren Troglodytestroglodytes .—Behaviour125, p. 67-78. ——&Slater,P .J.B.( 1995).Bird song: biological themes and variations. — Cambridge UniversityPress, Cambridge, 248 pp. Cramp,S. &Perrins,C.M. ( 1993).The birds of the western Palearctic. V ol6. — Oxford UniversityPress, Oxford. Gish,S.L. &Morton,E.S. (1981).Structural adaptations to local habitat acoustics in Carolina wrensongs. — Z.Tierpsychol.56, p. 74-84. Goldstein,H., Eisikovitz, D. &Yom-Tov,Y .(1986).Infanticide in thePalestine sunbird. — Condor88, p. 528-529. ——,Verbeek,N.A.M., Eisikovitz, D. &Yom-Tov,Y .(1987).Sunbirds prefer to feedin the sun.— Ardea75, p. 293-295. ——&Yom-Tov,Y .(1988).Breeding biology of the orange-tufted sunbird in Israel. — Ardea76, p. 169-174. Grimes,L.G. (1974).Dialects and geographical variation in thesong of the splendid sunbird Nectariniacoccinigaster .—Ibis116, p. 314-329. King,J.R. ( 1972).V ariationin the song of therufous collared sparrow, Zonotrichiacapensis , innorthwestern Argentina. — Z.Tierpsychol.30, p. 344-373. Kroodsma,D.E., Baker, M.C., Baptista, L.F .&Petrinovich,L. (1984).V ocal‘ dialects’in Nuttall’swhite-crownedsparrow. — Curr.Ornithol. 2, p.103-134. Markman,S., Y om-Tov,Y .&Wright,J. (1995).Male parental care in the orange-tufted sunbird:behavioral adjustments in provisioning and nest guarding effort. — Anim. Behav.50, p. 655-669. ——,Yom-Tov,Y .&Wright,J. (1996).The effect of maleremoval on female parental care inthe orange-tufted sunbird. — Anim.Behav. 52, p. 437-444. Marler,P .&Tamura,M. (1964).Song dialects in three populations of white-crowned sparrows.— Science146, p. 1483-1486. McGregor,P .K.( 1980).Song dialects in the corn bunting ( Emberizacalandra ). — Z. Tierpsychol.54, p. 285-297. ——(1992).Quantifying responses to playback: one, many, or composite multivariate measures.— In:Playback and studies of animal communication (P .K.McGregor,ed.). PlenumPress, New York,p. 79-96. ——&Thompson,D.B.A. (1988).Constancy and change in local dialects of the corn bunting.— Ornis.Scand. 19, p. 153-159. ORANGE-TUFTEDSUNBIRDSONGDIALECTS 1627

Morton,E.S. (1975).Ecological sources of selection on avian sounds. — Am. Nat.109, p. 17-34. Mundiger,P .C.(1980).Animal cultures and a generaltheory of cultural evolution. — Ethol. Sociobiol.1, p. 183-223. ——(1982).Microgeographic and macrogeographic variation in the acquired vocalizations ofbirds. — In:Acoustic communication in birds (D.E. Kroodsma & E.H. Miller,eds). AcademicPress, New York,p. 147-208. Nottebohm,F .(1972).The origins of vocallearning. — Am. Nat.106, p. 116-140. Payne,R.B. (1978).Microgeographic variation in songsof thesplendid sunbird Nectarinia coccinigaster :populationphenetics, habitats, and song dialects. — Behaviour65, p. 282-308. ——(1981).Population structure and social behavior: models for testing the ecological signiŽcance of songdialects. — In:Natural selection and behavior: recent research and newtheory ( R.D. Alexander& D.W.Tinkle,eds). Chiron Press, New York,p. 108-120. ——(1982).Ecological consequences of song matching: breeding success and intraspeciŽ c songmimicry in indigobuntings. — Ecology63, p. 401-411. ——(1983).The social context of songmimicry: Song-matching dialects in indigobuntings (Passerinacyanea ).—Anim.Behav. 31, p. 788-805. ——,Payne,L.L. &Doehlert,S.M. (1988).Biological and cultural success of song memes inindigo buntings. — Ecology69, p. 104-117. Paz,U. (1986).Birds. — In:The fauna and  oraof Israel. V ol.6 (A.Alon,ed.). Ministry of DefensePublishing / TheSociety for Protection of Nature, Israel, T el-Aviv,p. 409-410. (inHebrew). Rothstein,S.I. & Fleischer,R.C. (1987).V ocaldialects and their possible relation to honest statussignaling in thebrown-headed cowbird. — Condor89, p. 1-23. Thielcke,G. (1969).Geographical variation in birdvocalizations. — In:Bird vocalizations (R.A.Hinde, ed.). Cambridge University Press, Cambridge, p. 310-339. ——(1973).On the origin of divergence of learnedsignals ( songs)in isolatedpopulations. —Ibis115, p. 511-516. Wiley,R.H. &Richards,D.G. (1978).Physiological constraints on acousticcommunication inthe atmosphere: implications for the evolution of animal vocalizations. — Behav. Ecol.Sociobiol. 3, p. 69-74. Zilberman,R., Moav,B. &Yom-Tov,Y .(1999).Extra-pair paternity in the socially monoga- mousorange-tufted sunbird ( Nectariniaosea osea ).— IsraelJ. Zool.45, p. 407-421.