Geographic Variation in the Song of the Rufous-Collared Sparrow in Eastern Argentina ’

The Condor95:588-595 0 The Cooper Ornithological Society 1993

GEOGRAPHIC VARIATION IN THE SONG OF THE RUFOUS-COLLARED SPARROW IN EASTERN ARGENTINA ’

PABLO L. TUBARO AND ENRIQUE T. SEGURA Laboratorio de Fisiologia de1Comportamiento, Instituto de Biologia y Medicina Experimental. Obligado2490. 1428 BuenosAires, Argentina

PAUL HANDFORD Department of Zoology, Universityof WesternOntario, London, Ontario N6A 5B7, Canada

Abstract. Variation in trill features of Rufous-collared Sparrow (Zonotrichia capensis) song is described in three sites located on the northeast coast of Buenos Aires Province, Argentina. There is a distinctive “talar” dialect in sites supportingnatural woodlands. This dialect is characterizedby slower trills with lower minimum frequenciesthan the one present in the open grassyfields (“steppe” dialect). Talar dialect is absent in a site where natural woodlands have been replaced by grassyfields and open man-made parks, suggestingthat the vocal tradition changedafter habitat modification. Detailed analysis reveals that talar and steppe dialects intergrade in a songcline, following the vegetational ecotone. This song cline has been stable in location and featuresduring the last decade, even though there were no geographicbarriers impairing population or cultural exchangebetween habitats. Key words: Rufous-collaredSparrow: Zonotrichia capensis;trill variation; dialect; song cline.

INTRODUCTION thus giving rise to regional dialects (Nottebohm Bird song dialects, that is, particular types of 1969, 1975; Handford 1981, 1988; Lougheed et songsshared by birds inhabiting the same area, al. 1989). Rufous-collared Sparrow dialects have have been describedin several species(see Mun- been well studied in northwestern Argentina dinger 1982). Two of the most studied are the where there is a strong pattern of habitat varia- White-crowned Sparrow (Zonotrichia leuco- tion with longitude and altitude (Ring 1972; Not- phrys) (see Baker and Cunningham 1985, tebohm 1975; Handford and Nottebohm 1976; Kroodsma et al. 1985) and the Rufous-collared Handford 198 1, 1988; Handford and Lougheed Sparrow or Chingolo (2. capensis) (Nottebohm 1991). These studies have shown the existence 1969,1975; Nottebohm and Selander 1972; Ring of a relationship between trill dialect and vege- 1972; Handford and Nottebohm 1976; Hand- tation, and a high temporal stability in regional ford 198 1, 1988; Handford and Lougheed 199 1; dialect distribution during the last two decades. Lougheed and Handford 1992). In areaswith homogeneousvegetation trill fea- The song of this specieshas two parts: an in- tures seem to be relatively constant over great troduction and a final portion referred to as distances. For example, in the lowlands of Bue- “theme” and “trill,” respectively. The theme is nos Aires, Santa Fe and Cordoba Provinces, Not- usually composed of two to five whistled as- tebohm (1969, 1975) found a relatively homo- cending or descending notes. According to the geneousdialect (hereafter referred to as “steppe” number, shape and ordering of the notes, it is dialect), characterized by a trill interval of about possible to distinguish different themes. Within 60-70 msec. This area is also characterized by a an area, each male usually sings an individually homogeneousherbaceous steppe (the “Argentine distinctive theme. On the other hand, the main pampas,” Cabrera 1976). pattern of trill variation is among populations. Here, we describe the geographic pattern of In particular, trill interval (time between con- trill variation of the Rufous-collaredSparrow over secutive notes in the trill) is relatively constant a small geographicarea located near the north- within an area, but it changesamong life zones, east coast of the Buenos Aires Province, Argen- tina. The original habitats present in this area included marsh, open woodland locally termed ’ ’ Received9 October1992. Accepted22 February “talar” for the dominant tree, “tala” (Celtis t&z), 1993. and steppe. However, at certain sites within this SONG VARIATION IN A SPARROW 589 area present-day vegetation differs from the orig- all the natural habitats are in a relatively good inal one, becausehuman activities have replaced state of conservation. the talar by grassyfields, gardensand parks. This Site B-Punta Indio: an area with some tourist situation opens the possibility for making song activity, where marked modifications in the orig- comparisons among sites where the original veg- inal vegetation have been made, mainly through etation still remains and those in which this veg- the elimination of the talar woodland and its etation has been replaced during the last few de- replacement by grassy fields, gardens and open cades. In particular, the aim of this study is: (1) parks. The size of the sampled area was about 3 to provide new evidence about the association km2, and comprises the marsh and steppe hab- between dialects and vegetation, (2) to study the itats as well as the place occupied in the past by shape and temporal stability of the dialect a well developed talar. boundaries, and (3) to assessthe possible effect Site C-Estancia Luis Chico: an area of about 1 of habitat change upon dialect distribution. km*, including only the talar and the steppehab- itat. This area has been partially altered because of the past exploitation of calcareous deposits, MATERIALS AND METHODS agricultural practices, and the introduction of The study area was located in the Partido of cattle. In this site the talar exists as a secondary Magdalena, on the northeast coastof Buenos Ai- growth and its physical structure is different from res Province, Argentina (Fig. 1). From the Rio that in Site A, mainly becauseof the small stature de la Plata inland, three different habitats can be and the dispersed distribution of trees. distinguished: Song sample. The songsof 897 adult individ- (1) Marsh: a seasonally flooding lowland and uals were recordedusing a UHER 4000 Report-L higher ground with grasscover, near the coast at a speed of 9.5 cm/set, and a directional car- of the Rio de la Plata. There are three dom- dioid Let 980 LEEA microphone. Wild 2. ca- inant communities of plants: Salicornia am- pensissingers are always males (King 1972), but bigua-Paspalum vaginatum; Scirpusolmery- the possibility that a small fraction of females Paspalum vaginatum-Distichlis spp.; and sometimes sing cannot be ruled out. Eight hun- Mentha pulegium-Pamphalea bupleurifolia. dred and forty-five individuals were recorded at (2) Talar: a xeromorphic woodland comprising Site A between October and January (59 in 1984- several tree species:Celtis tala (fam. Ulma- 1985; 24 in 1985-1986; 149 in 1986-1987; 215 ceae); Scutia buxzfilia (fam. Rhamnaceae); in 1987-1988; 72 in 1988-1989; 127 in 1989- Jodina rhombifolia (fam. Santalaceae) and 1990; and 199 during October 1991); 25 at Site Acacia caven (fam. Leguminosae), among B during January (16 in 1988-1989; and 9 in others. 1989-1990); and 27 at Site C, on 13 January (3) Steppe: an herbaceous open grassland, es- 1989. sentially without trees (with the exception of On each sonogram (made with a Kay Electric some planted rows of Eucalyptus sp., and a Sonagraph 7029-A, with the wide band setting few isolated C. tala. The main herbaceous and filtered to pass the 80-8,000 Hz frequency communities present in this area are: Stipa range) five trill featureswere measured: total du- charruana-Cynara cardunculus-Diodia das- ration, number of notes (NN), trill interval (TI) ycephala; S. charruana-Danthonia monte- (calculated as trill duration/[NN - l]), maxi- vidensis-Eryngium ebracteatum; Stipa pap- mum and minimum frequencies (FMAX and posa-Stenotaphrum secundatum-Distichlis FMIN, respectively). spp. Detailed descriptions of these commu- Statisticalanalysis. We made a two-way multi- nities can be found in Cabrera (1949) Parodi variate analysis of variance (MANOVA) using (1940), Vervoorst (1967) and Leon et al. three variables (IT, FMAX and FMIN), and the (1979). SYSTAT statistical package (Wilkinson 1987). In this analysis we included the 1987-l 988 song Within this area, we studied three sites (Fig. data from Site A, permitted by the completeness 1): of this sample. We also included the data from Site A-Estancia El Destino: an area of about 7 Site C, and the pooled data of 1988-1989 and km2. Part of this area is a private reserve, where 1989-1990 from Site B. In this case, the possi- 590 P. L. TUBARO, E. T. SEGURA AND P. HANDFORD

BUENOS AIRES

BUENOS AIRES PROVINCE

BAHIA SAMBOROMBON

FIGURE 1. Map showingthe locationof the studyarea: Site A (EstanciaEl Destino);Site B (Punta Indio); and Site C (EstanciaLuis Chico). bility of recording the same bird twice was neg- to the fact that talar and steppe songsdiffered at ligible for three reasons.First, the density of sing- Sites A and C, but not at Site B (Table 1). ers found at this site is high. Second, the song For convenience, we introduce the term “talar recordingswere made in different neighborhoods dialect” to refer to the slower trilled songs(higher for each period. Third, estimates of the disap- TI values) heard in talar, and use the term “steppe pearancerate for adult Rufous-collared Sparrows dialect” to refer to the faster TI songsheard on are high, ranging from 30 to 77.6% (Miller and steppe (Figs. 2, 4). At Site A, talar dialect also Miller 1968, Handford 1980). showed a lower FMIN than the steppe dialect Becauseof the small number of birds recorded (Fig. 3). Although there are clear songdifferences at marsh habitat in Sites A and B, we excluded in relation to the vegetation type, the pattern of these data from the formal analysis. However, song change is clinal and accompanied the en- they are plotted in Figures 2 and 3 for compar- vironmental ecotone. Figure 2 shows that this ison. TI cline at Site A has been stable during at least the last six years. Becauseof the small size and RESULTS uneven spatial distribution of the song samples Two-way MANOVA showed a significant effect taken during 1984-1985 and 1985-1986, they of site (Wilks’ Lambda = 0.607; F6,496= 23.40; are not presented,although they show essentially P < O.OOl), habitat (Wilks’ Lambda = 0.880; the same pattern of variation described above. F 3.248= 11.27; P < O.OOl), and site-habitat interaction (Wilks’ Lambda = 0.791; F6,496= 10.28; DISCUSSION P < 0.001). Student t-test (Bonferroni-adjusted Our results show the existence of a tight rela- probabilities) showed that this interaction is due tionship between trill features and vegetation SONG VARIATION IN A SPARROW 591

-“I‘ . I 1 2 3 140C L I - 120- -_ :: E i y loo- __ - a. 3 .- q.‘ : 5 80 :?.‘i. : . * . d .-1.: . .. ’ ’ 60 l . . . : + -E

20-

0 o 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7

DISTANCE 1km) DISTANCE (km)

0' 1 0 1 2 3 4 5 6 7 1 2 3 4 5 6 7 DISTANCE 1km) DISTANCE (km1 FIGURE 2. Individual trill interval values in Site A, from the coast of the Rio de la Plata (marked as zero on the X axis) to the inland. The year in which the song sample was taken is indicated at top right of each graph. Along the top of the 1987 graph, the lines 1, 2, and 3 indicate the area occupied by marsh, talar and steppe habitats, respectively. type, similar to those reported in other areasfrom areas, which showed faster trills (lower TI) in Argentina. The songs found in talar woodland open scrub and grasslandsthan in more forested have longer TI and lower FMIN than thosefound areas (Nottebohm 1975; Handford 198 1, 1988). on the steppe. This syndrome of song structure This trend is also compatible with current ideas agreeswith trends reported from other argentine on the acoustic properties of natural environ- 592 P. L. TUBARO, E. T. SEGURA AND P. HANDFORD

5000 I 2 3 I 4500

._.____---.__-___- ___.-_.--.-___---.___--~.. 4000 I- . I

3500

3000

2500

I I I 1 I I I 2ooo0 1 2 3 4 5 6 7 DISTANCE (km) FIGURE 3. Diagram of the minimum frequency variation in Site A, during the season 1987-1988. Lines at the top of the graph indicate the approximate extent of each habitat type, as in Figure 2.

TABLE 1. Descriptive statisticsof songsfrom Sites A, B, and C.

Original habltat l-alar steppe Student’sf test” n= 164 n = 43 f (205) P Site A (Ea. El Destino) Trill interval (milliseconds) 80 i 13., 57 + 7 10.935 o.ooo* Minimum frequency (kHz) 3.3 t 0.2 3.6 f 0.1 12.350 o.ooo* Maximum frequency (kHz) 6.9 & 0.3 6.8 f 0.2 1.550 0.123

Original habltat Student’st testl’ T&II y=ppge n= 13 f (20) P Site B (Punta Indio) Trill interval (milliseconds) 61 +4 62 i 6 0.591 0.561 Minimum frequency (kHz) 3.6 * 0.2 3.6 i 0.3 0.149 0.883 Maximum frequency (kHz) 6.9 ? 0.4 1.4 i 0.5 2.742 0.013

Or&al habitat T&r steppe Student’st test n= 16 n= 11 f (25) P Site C (Ea. Luis Chico) Trill interval (milliseconds) 69 + 8 61 *5 3.425 0.002* Minimum frequency (kHz) 3.6 + 0.1 3.6 & 0.1 0.918 0.367 Maximum frequency (kHz) 7.4 + 0.4 7.6 ? 0.4 1.008 0.323 dR k SD. ” Two tailed. ( TheseP-values are not adjustedfor multiple comparisons.We assumethat thereare nine multlple comparisonsto be adjusted.Those comparisons with a Bonferroni-adjustedP-value of ~0.05 are indicatedwith an asterisk. SONG VARIATION IN A SPARROW 593

MARSH TALAR STEPPE

80 z....:.

.I. .. . . ’ (3) ..,.. :...?..: ..,.,. ..i :.. = 1...: : .

0 200 400 600 800 1000 1200

DISTANCE ( meters 1

FIGURE 4. Diagram of the trill interval variation in Site C, during the season 1988-1989. ments: slow modulated signals are favored in 1984-1985 to 1991 show the temporal stability forests, because they avoid the acoustic degra- of the TI cline and, from songs recorded and dation generated by the accumulation of ethos. heard in the talar in 1978, it is possible to back- In open fields, the main source of acoustic deg- track the existence of the “talar” dialect for at radation is random amplitude fluctuation pro- least 14 years. This result at a local scale also duced by nonstationary heterogeneities.This fa- agreeswith evidence of regional dialect stability vors signals with high rates of repetition (Wiley in northwestern Argentina, obtained during the and Richards 1978, Richards and Wiley 1980). last 20 years (Nottebohm and Selander 1972; At Site A, the transition from 60 msec TI in Nottebohm 1975; Handford and Nottebohm steppe to loo-120 msec TI in talar woodland 1976; Handford 1981, 1988). occupies only 2 km, in an area where there are The presenceof a differentiated trill dialect in no physical barriers to population or cultural ex- Sites A and C, where natural vegetation has been change between dialectal populations. Several relatively preserved, contrasts with the absence stepped song transitions have been reported in of the talar dialect at Site B, an area where his- northwestern Argentina associated with strong torical information indicates the existenceofwell environmental gradients (Nottebohm 1975, developed talar woodland during the first half of Handford and Nottebohm 1976, Handford 1988, this century (Parodi 1940). These facts suggest Lougheed et al. 1989), pointing out again the that the talar dialect could have gone extinct in importance of habitat features, particularly the Site B after the habitat changed. This scenario vegetation type, in the processof dialect differ- differs with that developed by Handford (1988), entiation. about the persistenceof original song dialects in Talar habitat occupies, even in the undis- northwestern argentine areas,where natural veg- turbed Site A, only a small area, but that seems etation has been drastically modified by human sufficient to affect the cultural evolution of Ru- activities. However, these apparent contradic- fous-collared Sparrow song. Recordings from tory scenariosmay be the result of the different 594 P. L. TUBARO, E. T. SEGURA AND P. HANDFORD geographic scales in which trill variation was HANDFORD,P. 1980. Return rates among highland studied. The area occupied by dialects analyzed Rufous-collared Sparrows. J. Field Omithol. 5 1: 176-177. by Handford (1988) was at least ten times larger HANDFORD,P. 1981. Vegetational correlatesof vari- than that reported here for talar dialect. Because ation in the songof Zonotrichiacapensis, in north- of the small area occupied by talar dialect and western Argentina. Behav. Ecol. Sociobiol. 8:203- its great perimeter of contact with steppedialect, 206. it is possible that immigration movements of HANDFORD,P. 1988. Trill rate dialectsin the Rufous- collared Sparrow, Zonotrichia capensis,in north- birds from surrounding open areas is responsible western Argentina. Can. J. Zool. 66:2658-2670. for the rapid talar dialect disappearance at Site HANDFORD,P., ANDF. NOTTEBOHM.1976. Allozymic B. and morphological variation in population sam- Song replacement could occur not only as a ples of Rufous-collared Sparrow, Zonotrichia capensis,in relation to vocal dialects. Evolution 30: result of a demographic process, but also as a 802-817. selective learning phenomenon, as proposed by HANDFORD,P., ANDS. C. LQUGHEED.199 1. Variation the “Ranging hypothesis” (Morton 1986) and the in duration and frequency charactersin the song “Acoustic adaptation hypothesis” (seeRothstein of the Rufous-collared Sparrow, Zonotrichia ca- and Fleischer 1987). Accordingly, the song tra- pensis,with respect to habitat, trill dialects and body size. Condor 93:644-658. dition that is better adapted to propagate at dis- HANSEN,P. 1979. Vocal learning: its role adapting tance in the environment would be preferentially soundstructures to long distancepropagation, and copied by birds (Hansen 1979, Gish and Morton a hypothesis on its evolution. Anim. Behav. 27: 1981, Nottebohm 1985, Morton 1986, Morton 1270-1271. KING, J. R. 1972. Variation in the songofthe Rufous- et al. 1986). This processcould shift the dialect collared Sparrow, Zonotrichia capensis,in north- frontier after the habitat changed, expanding western Argentina. Z. Tierpsychol. 30:344-373. steppedialect over talar altered areas.Again, this IOIOODSMA,D. E., M. C. BAKER,L. F. BAPTISTA,AND dialect replacement could be very rapid (about L. PETRINOVICH.1985. Vocal “dialects” in Nut- a decade) at Site B because of the great frontier tall’s White-crowned Sparrow, p. 103-134. In R. F. Johnston [ed.], Current Ornithology. Vol. 2. with steppe dialect and the small size of talar Plenum Press, New York. dialect population. If this interpretation is cor- LEON,R.J.C., S. E. BURKART,AND C. MOVIA. 1979. rect, changes in a regional scale could require Serie fitogeogrlfica N” 17. La vegetation de la Re- hundreds or thousands of years. piblica Argentina. Relevamiento fitosociologico de1 pastizal de1 norte de la depresmn de1 salado. ACKNOWLEDGMENTS Secretariade Estado de Agricultura y Ganaderia, INTA. Instituto de Botanica Aaricola. We thank A. Kacelnick, A. Lemoine, S. Lougheed, F. LOUGHEED,S. C., AND P. HANDFORD. 1992. Vocal Nottebohm, J. Reboreda, and S. Talia for their com- dialects and the structure of geographicvariation ments on several versions of the manuscript. E. Mor- in morphological and allozymic charactersin the ton, G. Walsberg, and an anonymous reviewer made Rufous-collared Sparrow, Zonotrichia capensis. valuable criticisms. F. Gabelli and G. Simonetti heloed Evolution 46:1443-1456. us to tape record songsduring 1986/87 and 1989790 LOUGHEED,S. C., A. J. LOUGHEED,M. RAE, AND P. breeding seasons,respectively. We also thank the Elsa HANDFORD. 1989. Analysis of a dialect boundary Shaw de Pearson Foundation for the permission and in chaco vegetation in the Rufous-collared Spar- facilities for field work at Site A. The sonogramswere row. Condor 91:1002-1005. made at the Laboratorio de Investigaciones Sensori- MILLER, A. H., AND V. D. MILLER. 1968. The be- ales.This work was supportedby the ConsejoNational havioral ecology and breeding biology of the An- de InvestigacionesCientificas y Tecnicas,and the Uni- dean Sparrow, Zonotrichia capensis.Caldasia 10: versity of Buenos Aires, Argentina. 83-l 54. MORTON,E. S. 1986. Predictions from the ranging LITERATURE CITED hypothesis for the evolution of long distance sig- BAKER,M. C., AND M. A. CUNNINGHAM. 1985. The nals in birds. Behaviour 99:64-86. biology of bird-songdialects. Behavioral and Brain MORTON,E. S., S. L. GISH, AND M. VAN DER VOORT. Sciences8:85-133. 1986. On the learning of degraded and unde- CABREM,A. L. 1949. Las comunidadesvegetales de graded songsin the Carolina Wren. Anim. Behav. 10salrededores de La Plata. Lilloa 20:269-347. 34:8 15-820. CABRERA,A. L. 1976. Regiones fitogeograficasAr- MUNDINGER,P. C. 1982. Microgeographicand mac- gentinas.Enciclopedia Argentina de Agricultura y rogeographicvariation in acqui~ed~vocalizations Jardineria. 2nd ed. Vol. 2. Sect 1. Editorial Acme of birds. D. 147-208. In D. E. Kroodsma and E. S.A.C.I., BuenosAires. H. Miller‘[eds.], Acousticcommunication in birds. GISH, S. L., AND E. S. MORTON. 1981. Structural Vol. 2. Academic Press, New York. adaptations to local habitat acousticsin Carolina NOTTEBOHM,F. 1969. The songof the Chingolo, Zo- Wren songs.Z. Tierpsychol. 56:74-84. notrichia capensis,in Argentina: description and SONG VARIATION IN A SPARROW 595

evaluation ofa systemofdialects. Condor 7 1:299- ROTHSTEIN,S. I., AND R. C. FLEISCHER.1987. Vocal 315. dialectsand their possiblerelation to honest status NOTTEBOHM,F. 1975. Continental patterns of song signallingin the Brown-headed Cowbird. Condor variability in Zonotrichia capensis: some possible 89: l-23. ecologicalcorrelates. Am. Nat. 109:605-624. VERVOORST,F. B. 1967. Serie fitogeogrkficaN” 7. La NOTTEBOHM,F. 1985. Sound transmission,signal sa- vegetaci6n de la Reptiblica Argentina. Las co- lience, and songdialects. Behav. Brain Sci. 8: 112- munidades vegetalesde la Depresi6n de1 Salado 113. (Provincia de BuenosAires). Secretariade Estado NOTTEBOHM, F., AND R. K. SELANDER.1972. Vocal de Agricultura y Ganaderia de la Nacibn, INTA. dialectsand genefrequencies in the Chingolo spar- Instituto de Botanica Agricola. row (Zonotrichia capensis). Condor 741137-143. WILEY,H., AND D. G. RICHARDS.1978. Physicalcon- PARODI,L. R. 1940. La distribucibn geograficade 10s straints on acousticcommunication in the atmo- talares de la Provincia de Buenos Aires. Darwin- sphere. Implications for the evolution of animal iana 4~33-56. vocalizations. Behav. Ecol. Sociobiol. 3:69-94. RICHARDS,D. G., AND H. WILEY. 1980. Reverbera- WILKINSON,L. 1987. SYSTAT: the system for sta- tions and amplitude fluctuations in the propaga- tistics. SYSTAT, Inc., Evanston, IL. tion of sound in a forest: implications for animal communication. Am. Nat. 115:381-393.