The Condor97z792-803 0 The Cooper OrnithologicalSociety 1995 GEOGRAPHIC, ECOLOGICAL AND SUBSPECIFIC VARIATION IN THE SONG OF THE RUFOUS-BROWED PEPPERSHRIKE (CYCLARHIS GUJANENSIS)’ PABLO L. TUBARO AND ENRIQUE T. SEGURA Laboratorio de Biologia de1Comportamiento, Institute de Biologia y Medicina Experimental, Obligado 2490. 1428 - BuenosAires, Argentina Abstract. We describethe patternsof songvariation in the Rufous-browedPeppershrike (Cyclarhis gujanensis).Individual variation was estimatedusing a sampleof 659 songs belongingto 2 1 individuals.This analysisshowed that this speciesuses a repertoireof up to seven songtypes. Geographic, ecological, and subspecificpatterns of variation were estimatedusing four songsfrom eachof 93 birds recordedfrom Mexico to Argentina.On eachsonogram we measured10 temporal,frequency, and structuralfeatures of the song. Principal componentanalysis based on the correlationmatrix of thesedata showedthat subspeciessing similar songs. However, there were significant relationships between principal componentvalues and latitude,indicating that songsfrom equatorialareas are shorterin duration, have highermaximum frequencyand number of syllables,broader bandwidth and are lessrepetitive than thosefrom moretemperate areas. In addition,we classifiedthe recordingsites into threecategories according to the actualvegetation: open, mixed, and closed.A multivariateanalysis of varianceafter removing the effectof latitudeand altitude, showedthat habitat typesdiffer in songstructure. Songs from openand mixed areashave a narrowbandwidth, and a lower maximum, minimum and emphasizedfrequencies than thosefrom closedsites. We analyzedthese patterns according to the currentideas about soundpropagation in naturalenvironments and the geographicalpattern of body sizevari- ation in the Rufous-browedPeppershrike. Key words: Rufous-browedPeppershrihz; Cyclarhis gujanensis; song; subspecies; latitude: altitude; habitat; reverberationhypothesis; Bergmann s’ rule. INTRODUCTION the attenuation of different frequencies (Morton The analysis of bird-song features and their re- 1975, Marten and Mailer 1977, Marten et al. lationship to habitat structure has been a subject 1977). These studiesreported the existenceof an of researchsince the pioneering studies of Chap- “acoustic window” near the forest ground which puis (197 1) and Morton (1975). Those studies favors the use of low frequencies for long-range showedthe existenceof songdifferences between communication in closedhabitats. However, this bird communities living in tropical forests and observation does not explain why speciesliving open fields. In particular, Morton (1975) found one or two meters above the ground or in open that the frequencies used by tropical forest spe- areasdo not use still lower frequencies,since they cies are lower than those from open tropical ar- presumably would propagate even better. Ryan eas. This relationship remained valid even when and Brenowitz (1985) suggestedthat differences differences in body size and phylogenetic effects in spectral distribution of ambient noise could were accounted for (Ryan and Brenowitz 1985). explain such a difference. In addition, Wiley and More recently, Wiley (199 1) found that, on av- Richards (1982) stressedthe effectsof reverber- erage, the maximum frequency of temperate for- ation rather than differencesin frequency atten- est songsis lower than those of marsh and grass- uation among habitats as the main selective fac- land. However, the causes of this relationship tor in songdesign. According to this view (hence- are not clear. forth referred to as the “reverberation hypoth- Early attempts to explain the relationship be- esis”), in addition to frequency, the temporal tween frequency of songsand habitat focusedon structure of the song is important for long-range communication. Slowly modulated signals are favored in forests,because they avoid the acous- tic degradation generated by the accumulation I Received3 January1995. Accepted1 May 1995. of ethos. In open fields, the main sourceof acous- WQI SONG VARIATION IN THE RUFOUS-BROWED PEPPERSHRIKE 793 tic degradation is random amplitude fluctuation and eCOlOgiCa faCtOrS may influence song struc- produced by moving cells of air with different ture. temperature and humidity. This favors signals The song of the Rufous-browed Peppershrike with high rates of repetition (Wiley and Richards has not been describedusing representative sam- 1978, Richards and Wiley 1980). ples of sonograms and cuantitative measure- The predominance of rapidly modulated ments of frequency and temporal variables. This sounds in songs from grass and edge habitats, specieshas a repertoire of two songtypes: a very and the presenceof pure tone, slowly modulated infrequent, slow seriesof four to eight descending songsin forest, was describedby Morton (1975). notes (which is attributed to the female), and a A recent study of the vocalizations from bird musical whistled song, with multiple ascending communities of eastern North America also and descending modulated notes (Skutch 1967). showed that temporal properties of song were We considered the later as the primary song of strongly associatedto habitat (Wiley 199 1). Spe- the species and we will refer exclusively to it cifically, birds of forest avoided the use of notes hereafter. with short repetition periods, as predicted by the reverberation hypothesis. METHODS Another line of research on song design con- Recordings used in this study were derived from cerns the study of single speciesliving in a di- a variety of sources;recording localities and oth- versity of habitats. In spite of the limited gen- er details are provided in Appendix 1. The sam- erality of this approach, it has the advantage of ple includes songsof 93 individuals from within reducing the confounding effects of phylogeny the ranges of seven subspecies(sample sizes in- and morphology in songcomparisons. Studies of dicated in parentheses):C. g. flaviventris (5), ni- the White-throated Sparrow (Zonotrichia albi- caraguae (16) subflavescens(lo), ochrocephala collis, Wasserman 1979, Waas 1988), the Sum- (32), gujanensis (12), pawus (8), and saturatus mer Tanager (Piranga rubra, Shy 1983), the (10). These samplescover almost the entire range Northern Cardinal (Cardinalis cardinalis, An- of latitudes and altitudes over which this species derson and Conner 1985), and the Great Tit is found (Fig. 1). Subspecific identity was as- (Parus major, Hunter and Krebs 1979) indicate signed according to the geographic location of the use of lower frequencies and narrower band- the songrecording and the following bibliograph- width in forested than in more open habitats. ic sources: Chapman (19 17, 1926), Griscom However, the Rufous-collared Sparrow (Zono- (1932), Hellmayr (1935), Dickey and Van Ros- trichia cape&s) appears to show the opposite sem (1938), Pinto (1944) Phelps and Phelps pattern, with higher frequency and broader band- (1963), Olrog (1979), and Howard and Moore width associated with more closed and mesic (199 1). Environmental information (altitude and environments (Nottebohm 1975, Handford and habitat type) was derived from the data sheets Lougheed 199 1). This speciesalso avoids the use accompanying the recordings. Latitude was in- of rapid trills in forested areas, thus conforming ferred in missing cases,using the geographicco- to the predictions ofthe reverberation hypothesis ordinates of the recordingsite. Habitat types were (Nottebohm 1975, Handford 1988, Handfordand classifiedinto three categories:closed (including Lougheed 199 1, Tubaro et al. 1993). tropical humid forestand montane forest),mixed This diversity of phenomena indicates the need (including coniferous forest, mixed forest and de- for other comparative studies. In this paper we ciduous forest), and open (arid scrub, gardens, present an analysis of the pattern of song vari- and savannas). ation in the Rufous-browed Peppershrike (Cy- On each sonogram (made with a Proaudio clarhisgujanensis). This specieslives in a variety Spectrum 16 Sound Blaster [Media Vision] and of habitats ranging from mesic forest to xero- the ADDA 16 software [LIS 19921)we measured morphic woodlands and arid savannas,and from the following variables (see Fig. 2): song length the sealevel to 3,500 m in the Andes Mountains. (SOLEN), note length (NOLEN), intemote in- It also shows a high degree of polytypism, with terval (INI), maximum and minimum frequen- 2 1 subspeciesdescribed in its range from eastern cies (MAX and MIN, respectively), bandwidth Mexico to central Argentina (Howard and Moore (BAND = MAX-MIN), emphasized frequency 199 1). This provides the possibility for clarifying (frequency with the higher amplitude in the song, the extent to which phylogenetic, geographical, ENF), number of notes (NN), number of sylla- 194 PABLO L. TUBARO AND ENRIQUE T. SEGUR4 pat-w.6 gujanensis subflavescens Y safuratus l 1 0 2 FIGURE 1. Locationof songsamples, as well as the approximatedistribution of the subspeciesconsidered. The sizeof the filled circlesindicates the numberof birdsrecorded in eachsite. bles (NS), and a “repetition index” (RI). The RI 15-l 00). Details of the provenance of these re- is the mean number of times a syllable type is cordings are shown in Appendix 2. Song types consecutively repeated in a single song. We de- were defined according to the shape and order fine notes and syllables according to Tubaro of syllables. Since we found that the Rufous- (1992). browed Peppershrike has a repertoire of several Individual variation was