ORIGINAL PAPER

W. Hodl Æ A. Amezquita Æ P. M. Narins

The role of call frequency and the auditory papillae in phonotacticbehavior in male Dart-poison frogs Epipedobates femoralis(Dendrobatidae)

Received: 31 December 2003 / Revised: 27 May 2004 / Accepted: 28 May 2004 / Published online: 29 July 2004� Springer-Verlag 2004

Abstract Territorial males of the pan-Amazonian Dart-poison frog, Epipedobates femoralis, are known topresent stereotypic phonotactic responses to the play-back of conspecific and synthetic calls. Fixed siteattachment and a long calling period within an envi-ronment of little temperature change render this terres-trial and diurnal pan-Amazonian frog a rewardingspecies for field bioacoustics. In experiments at the fieldstation Arataı, French Guiana, we tested whether theprominent frequency modulation of the advertisement-call notes is critical for eliciting phonotactic responses.Substitution of the natural upward sweep by either apure tone within the species frequency range or a reversesweep did not alter the males’ phonotactic behavior.Playbacks with artificial advertisement calls embeddedin high levels of either low-pass or high-pass maskingnoise designed to saturate nerve fibers from either theamphibian papilla or basilar papilla showed that malephonotactic behavior in this species is subserved byactivation of the basilar papilla of the inner ear.

Keywords Acoustic playback experiment Æ Anura ÆCall frequency modulation Æ Phonotaxis Æ Territorialbehavior

Abbreviations FM: Frequency modulated ÆAP: Amphibian papilla Æ BP: Basilar papilla

Introduction

Male anuran vocalizations serve to attract conspecificfemales and to regulate male spacing (Wells 1977; Rand1988). They are often periodic, discrete signals, stereo-typed in both the frequency and time domains (however,for variation in advertisement call complexity see Sch-wartz and Wells 1984; Ryan 1985; Wells 1988; Narinset al. 1998, 2000; Feng et al. 2002). Not all of the ste-reotyped spectral and temporal properties are critical forintraspecific communication (Gerhardt 1988) nor is itlikely that a single call character is responsible for callrecognition, but rather that a combination of charactersis involved (Schwartz 1986, 1987; Gerhardt and Doherty1988; Wells 1988; Gerhardt and Huber 2002).

The frog inner ear contains three organs that respondto airborne sound—the amphibian papilla (AP), thebasilar papilla (BP) and the sacculus (S), althoughonly thefirst two are known to subserve long-distance soundreception. Acute frequency resolution occurs only forenergy falling within the range of sensitivity of the AP(Narins andCapranica 1980; Zakon andWilczynski 1988;Lewis andNarins 1999). Nerve fibers innervating AP haircells exhibit V-shaped tuning curves with distinct bestfrequencies ranging from ca. 0.1 kHz to 1.0–1.4 kHz. Incontrast, BP fibers are broadly tuned and exhibit a narrowrange of best frequencies that typically falls above 1.0 kHz(Zakon andWilczynski 1988). It is currently thought thatthe peripheral auditory system of anurans is generallyspecialized to process species-specific communicationcalls but not exclusively dedicated to them (Feng et al.1990). This is also consistent with the idea that sensitivityto environmental sounds other than the species-specificadvertisement call (e.g., prey-generated sounds, hetero-specific calls, abiotic noise)would clearly confer a selectiveadvantage to the individual possessing it.

Typically, males alter their calling/behavioral patternwhen stimulated by playbacks of conspecific calls at su-prathreshold levels (Zelick and Narins 1982; Ryan 1986;Wells 1988; Grafe 1996; Lewis et al. 2001). Overt behav-

W. Hodl (&)Institute of Zoology, University of Vienna,Althanstrasse 14, 1090 Wien, AustriaE-mail: walter.hoedl@univie.ac.atTel.: +43-1-427754495Fax: +43-1-42779544

A. AmezquitaDepartamento de Ciencias Biologicas,Universidad de Los Andes, Bogota, Colombia

P. M. NarinsDepartments of Physiological Science and Organismic Biology,Ecology and Evolution, University of California Los Angeles,Los Angeles, CA 90095, USA

J Comp Physiol A (2004) 190: 823–829DOI 10.1007/s00359-004-0536-1

ioral responses of males in response to playbacks of eithersynthetic or natural calls provide evidence of perceptionand recognition of the acoustic signal and thus is consid-ered the least invasivemethod to test the hearing ability ina free-living amphibian (Narins and Zelick 1988). TheDart-poison frog Epipedobates (=Allobates)1 femoralis(Fig. 1) is known to exhibit conspicuous behavioral re-sponses to the playback of conspecific calls. Intensity-dependent thresholds for antiphonal calling and forphonotactic approach could be determined in playbackexperiments with natural and synthetic calls (Hodl 1982,1983). In aPeruvian population ofE. femoralis, playbacksof synthetic calls with intensities between 56 and 68 dBSPL, measured at a conspecific focal male, generallyevoked complete body orientation toward the broad-casting speaker and subsequent antiphonal calling. Play-backs of advertisement calls with intensities above 68 dBSPL resulted in a sudden cessation of calling activity, headand body reorientation toward the sound source, and azig-zag approach to the broadcasting speaker (Hodl 1982,1987; Narins et al. 2003). Similar reactions are known forother male dendrobatid frogs such as Colostethus palma-tus (Luddecke 1974) and C. nubicola (Gerhardt andRheinlaender 1980).

The advertisement call of E. femoralis consists of ei-ther two (Catuaba, Acre, Brazil; W. Hodl and A. Am-ezquita, personal observations), three (Panguana, Peru;Schluter 1980) or four (all other presently knownAmazonian populations; Hodl 1987) notes each sweep-ing up in frequency within a range of 2.5–4.1 kHz(Fig. 3a; Narins et al. 2003). These frequency-modulated(FM) calls are produced by territorial males from ex-posed positions on the forest floor during the day andcharacteristically given in bouts of up to 41 calls (Narinset al. 2003). Inter-male distances vary between 4 and

30 m and territories may comprise up to 26 m2 whichare occupied and defended for as long as 103 days(Roithmair 1992).

Since males of E. femoralis are so highly reactivewhen presented with a supra-threshold conspecificadvertisement call from within their territory, we deci-ded to exploit this relatively unique robust behavior in aseries of stimulus manipulation playback experiments inthe animal’s natural habitat. These experiments weredesigned to test two specific hypotheses concerning theadvertisement-call frequency of E. femoralis. Hypothesis1: the prominent upward sweep of the call notes is criticalfor eliciting the phonotactic response, which was directlytested in two playback experiments in which the species-specific upward frequency sweep was replaced by (1) adownward-frequency sweep of the same duration, or (2)a pure-tone burst with a single frequency in the rangefrom 2.6 to 4.1 kHz; and hypothesis 2: because of thehigh frequencies comprising the advertisement call, thephonotactic response in this species is mediated by the BP.This was directly tested by broadcasting the species’advertisement call embedded in high levels of either low-pass or high-pass masking noise designed to saturate theafferent nerve fibers originating from either the AP orBP in the inner ear, respectively. If the phonotactic re-sponse to the call persists while the AP fibers are satu-rated by the noise masker, one may conclude that theAP does not subserve phonotactic behavior. Likewise, ifthe phonotactic response to the call persists while the BPfibers are saturated by the noise masker, it is likely thatthe BP does not subserve phonotactic behavior.

Materials and methods

Study animals and study area

Playback experiments with synthetic sound generatedin 1998 on the basis of the temporal and spectral mean

Fig. 1 Calling male of the Dart-poison frog Epipedobatesfemoralis (Dendrobatidae),Arataı, French Guiana. InsertMale E. femoralis afterphonotactic approach in frontof the loudspeaker emittingsynthetic calls

1 Zimmermann and Zimmermann (1988) and Vences et al. (2000)suggest placing Epipedobates femoralis in its own monotypic genus,Allobates femoralis

824

of the Central and Eastern Amazonian populationsknown at the time were undertaken with territorialmales calling in primary forest close to the field stationat Arataı, French Guiana (3�59¢N, 52�35¢W) at theonset of the rainy seasons during December 1998, 1999and January 2000, 2003. The study site is in a lowland(23 m.a.s.l.) wet tropical forest with a climate charac-terized by a long wet season (mid-November to mid-August), irregularly interrupted by a short dry periodaround March, and a longer dry season during theremaining months (Feer 1999). Mean annual rainfalland temperature are 3,000–3,250 mm and 26�C,respectively. Although 34 species representing eightfamilies of anuran amphibians have been identified atArataı (B. Gottsberger and E. Gruber 2004; P. Gau-cher personal communication), only males of Epi-pedobates hahneli and to a lesser extent Adenomerahylaedactyla exhibit partial overlap of their calling-activity periods with E. femoralis and call from similarsites. Males of these species produce calls with energyconcentrated between 4.5 and 6.2 kHz (E. hahneli) and2.1–2.6 kHz (A. hylaedactyla). Calling activity inE. femoralis is greatest in the morning (0700–0900 hours) and late afternoon (1500–1730 hours).Analyses of the acoustic environment at Arataı show ahigh signal-to-noise ratio for E. femoralis throughoutits calling period (Fig. 2).

Synthesis of experimental signals

To determine the role of the frequency modulation(FM) of the call in eliciting the phonotactic response, aset of periodic synthetic calls (Fig. 3b–d) was preparedusing SoundEdit software (version 2.0.7, Macromedia)

and a laptop computer (Macintosh PowerBook G3).The temporal and spectral parameters of the syntheticsignals were derived from the mean values for malesfrom the populations studied up to the beginning of thetest series in 1998 (Schluter 1980; Hodl 1983, 1987). Intests with FM calls, each of the four notes consisted ofeither a constant upward (2.6–3.5 kHz, Fig. 3b top) ordownward (3.5–2.6 kHz, Fig. 3b middle) sweep. In the1998 and 1999/2000 tests with pure-tone calls, eachnote was synthesized as a short tone-burst with fre-quencies of 2.6, 2.9 (Fig. 3b bottom), 3.2 or 3.5 kHz,respectively. Since analyses of the calls produced bymales from the Arataı population revealed frequenciesup to 4.1 kHz (Narins et al. 2003) additional tests withpure-tone signals of 3.8 kHz and 4.1 kHz were under-taken in the 2003 field season. A synthetic advertise-ment call consisted of four notes of 72.5 ms each. Twoconsecutive notes within the same call were separatedby silent intervals of either 40 ms (between 1st and 2ndnote as well as between 3rd and 4th note) or 80 ms(between 2nd and 3rd note), resulting in a call durationequal to 450 ms.

To determine the role of the BP in the phonotacticresponse in this species, we prepared another set ofstimuli in which high levels of either low-pass (0–1.25 kHz) or high-pass (1.5–10 kHz) filtered maskingnoise were added to synthetic natural-call mimics(Fig. 3c, d). Broadband noise was first synthesized withSoundEdit at an SPL equal to that of the synthetic call.Then, to produce low-pass or high-pass masking noise,the unwanted portion of the spectral range was filteredout with the Equalizer procedure in SoundEdit. The twosynthesized noise bands were expected to selectively andtemporarily saturate the 8th nerve fibers of either the APor BP (Narins 1983).

Fig. 2 a Power spectrum andb oscillogram of arepresentative 3-min recordingof the acoustic environment inwhich the playback experimentswith male E. femoralis wereperformed. Intensity isexpressed in relative (non-calibrated) units. Recordingtaken on 14 January 2003 at1510 hours during rainlessconditions. Power spectrumparameters: FFT size, 1,024points; filter bandwidth,266 Hz; frame length, 23.2 ms;window function, Blackman;amplitude, logarithmic. Peaksin the power spectrum from leftto right: 1 E. femoralis, 2 E.hahneli, 3 unidentified insect

825

Playback procedure

Trials were conducted between 1415 and 1800 hoursunder rainless conditions and at temperatures between23.5 and 26.0�C. Territorial males were easily located bytheir vocal activity and exposed calling positions. Weidentified individual frogs by the territories they occu-pied. Acoustic stimuli were broadcast exclusively tovocally-active territorial males. Frogs were not handledprior to, during or after any trial. Once a calling malewas located, a loudspeaker was placed on the forest floorat a distance of 2 m from him and facing the male undertest. All signals were broadcast monaurally through aspeaker-amplifier (Philips SBC BA 130 or BA 170) dri-ven by stored waveforms from the laptop computer(Macintosh PowerBook G3, test series in 1998, 1999/2000), or from CD players (Silva Schneider CDP 286AS, test series in 2003). All playback stimuli werebroadcast at sound pressure levels between 80.8 and84.1 dB SPL (impulse time constant), measured in after-trial controls at a distance of 2 m, approximatelyresembling the level of naturally-calling conspecifics atan equal distance (Hodl 1987). In order to saturate thefibers of the AP, the noise level of the low-pass filterednoise was adjusted to be 12 dB higher than the calls. Inthe experiments with high-pass filtered noise the noiselevel equaled that of the synthetic calls. The sound levelmeter (CEL-493, CEL Instruments) was calibrated witha pistophone (Rion NC-73/M91) several timesthroughout the experimental periods.

During a single 174-s trial, stimuli were broadcast inten consecutive bouts (17.4 s per series), each consistingof ten calls (8.2 s, at a rate of 1.17 calls s)1) and a silentinterval (9.2 s). This temporal pattern approximated that

of the natural call pattern of actively-calling males at thestudy site. For each trial, we noted whether males ori-ented and/or jumped towards the loudspeaker. A pho-notactic approach was only considered successful (andthe playback consequently stopped) when the male undertest came within 30 cm of the sound source during thetrial period. If the frog reached the loudspeaker beforeten consecutive bouts were broadcast, the test was ter-minated. For each successful trial, we calculated thelinear-approach speed by dividing the distance traveled(1.7 m) by the time required to cover that distance. Thistime was estimated by noting the bout number (B30) orinter-bout interval (IBI30) during which the focal malereached a 30-cm perimeter around the loudspeaker. Thetravel time was then set equal to the time elapsed sincethe beginning of the trial plus one-half of B30 or IBI30,whichever was appropriate. Tests in which males failed toapproach the speaker within a single trial period werecounted as negative only if in the subsequent trial (inwhich we usually presented the control stimulus ‘‘normalsweep,’’ Fig. 3b, top), the individual exhibited a positivephonotactic approach. To minimize possible residualeffects of previous tests, individuals were exposed to amaximum of three consecutive tests (including the con-trol stimulus). Following the trials, a rest period of atleast 2 h elapsed prior to the next test. Each FM- or tone-burst stimulus was broadcast until positive and/or neg-ative results of at least 14 individuals were obtained.

Results

We observed head and body orientation, antiphonalcalling (with nearby neighbors), zig-zagging toward and

Fig. 3a–d Sound spectrogramsof playback stimuli used inphonotaxic experiments.a A representative, natural(four-note) call of E. femoralisfrom Arataı, French Guiana,26�C. Bandwidth of analysisfilter: 266 Hz. b Synthesizedcalls used in playback trials:normal upwardly-sweeping FMsignal mimicking the naturalcall (top), reverse sweep(middle), and AM (pure-tone)call at 2.9 kHz (bottom).c Normal upwardly-sweepingFM signal accompanied by low-pass and d embedded in high-pass noise

826

contact with the broadcasting speaker in 46 stimulatedmales. Out of a total of 125 playback trials with 46individuals [17 (1998), 14 (1999/2000), and 15 (2003)]without additional filtered noise, 123 (98.4%) of thetested males head-oriented and subsequently jumpedtoward the loudspeaker broadcasting either FM or pure-tone stimuli (Fig. 4a). Individuals approached theloudspeaker within 30 cm during all playback trials withthe normal and the reversed sweep. Experiments withpure-tone stimuli were 0–6% less effective than FMstimuli in evoking orientation and/or positive phono-tactic approach and 7–27% less effective in attractingthe males to within 30 cm of the sound source (Fig. 4a).However, chi-square analyses revealed non-significantassociations between the acoustic treatment and thenumber of males head-orienting (df=4 in all cases,P=0.507), jumping (P=0.507), or approaching theloudspeaker within 30 cm (P=0.137). Furthermore,linear-approach speed did not differ significantly be-tween individuals with successful approaches in FMtrials and pure-tone trials (Fig. 5, Friedmann’s test:n=7, X2=1.19, p=0.752).

In experiments with call notes combined with low-pass filtered noise, all individuals tested (n=14) orientedand jumped toward the sound source, but only six (43%)made a successful approach within 30 cm during theexperimental trial period (Fig. 4b). The linear-approachspeed for males in these trials did not differ significantlyfrom frogs in control trials (Wilcoxon’s test: n=6,Z=)0.67, P=0.500), even though the signal was 12 dBlower then the level of the low-pass filtered noise. Incontrast, during experiments with call notes combinedwith an equal level of high-pass filtered noise only two

out of 16 individuals (12.5%) oriented and jumped in thedirection of the loudspeaker and none of the testedindividuals made a successful approach.

Discussion

In the present study, we performed a series of acousticplayback experiments using masked and unmaskedstimuli with actively-calling males of E. femoralis in theirnatural habitat. We confirmed that substitution of theupward-sweeping FM-call notes either by pure toneswithin the FM-sweep range or by downward-sweepingnotes did not alter the male’s phonotactic behavior.Specifically, there were no significant differences in theproportion of individuals orienting toward, jumpingtoward or reaching the sound source. In addition, thelinear-approach speed as an indicator of signal attrac-tiveness and localization efficiency did not differ amongthe unmasked stimulus trials. Thus, we conclude that thenatural upward FM sweep in the advertisement call isnot an information-bearing element (sensu Suga 1972)for species recognition or localization in males of E.femoralis. These findings complement earlier behavioraldiscrimination experiments performed with females inwhich synthetic signals lacking FM modulation wereshown to be equally attractive to FM signals in frogs ofthe families Hylidae (Doherty and Gerhardt 1984;Gerhardt 1978) and Leptodactylidae (Wilczynski et al.1995).

Our results support the observation that if anacoustic stimulus has the correct species-specific peri-odicity, and energy falling within the natural call notefrequency range, it evokes phonotaxis in vocally-activemales. So why produce an FM call at all? Several sug-gestions for the utility of FM calls have been put forth.In contrast to a wideband, noise-like stimulus, an

Fig. 4 a Proportion of vocally-active E. femoralis males thatresponded to the playback of synthetic signals, including FM calls(normal and reverse sweeps) and AM calls (pure-tone signals). bSame as a for normal sweeps mixed with continuous low- or high-pass noise. Open bars orientation toward the sound source, stippledbars jump toward the sound source, and filled bars successfulapproach to the sound source. The number of animals tested foreach stimulus is indicated above the bars

Fig. 5 Box plots showing linear-approach speed of E. femoralismales during playback experiments. The boxes indicate the middle50% of the data (interquartile range), the dark horizontal linesrepresent the median values and the vertical lines extend to thesmallest and largest values of the distribution, excluding outliers.Other details same as Fig. 4

827

organism producing a narrowband FM sweep across theidentical frequency range can apply the full effort of itssound production system to one frequency at a time. Inaddition, in any broadband signal (such as an FM-sweep), at least part of the call spectrum has a highprobability of reaching distant individuals even if somefrequencies are strongly absorbed, reflected or masked ina highly-structured environment. Thus, a constant-amplitude FM sweep may confer a selective advantageon the sender by providing multiple parallel pathwaysfor acoustic transmission. In this way, the sweep mayserve to enhance long-distance information transmissionin the species’ natural habitat (for a discussion of rain-forest acoustics see Ellinger and Hodl 2003). Addition-ally, residents may be able to assess the distance tovocalizing intruders by exploiting environmentalacoustic dispersion, i.e., sensing the frequency-depen-dent amplitude changes within the sweep (for insects, seeGerhardt and Huber 2002). Determination of a sender‘sdistance may be especially important for the highlyterritorial E. femoralis, which engage in prolonged fightsonce a territorial limit is transgressed by an intruder(personal observations).

Moreover, our selective masking experiments providestrong evidence that male phonotactic behavior in thisspecies is subserved by activation of the BP rather thanthe AP in the inner ear. Males stimulated by the syn-thetic mimic of the natural call accompanied by low-passnoise oriented toward the sound source and initiated thephonotactic response, but 57% of the males tested failedto approach within 30 cm of the sound source. Duringthese trials the unsuccessful individuals often circled thespeaker without directly approaching it. In these cases, itappeared as if the noise was interfering with the normalsound localization mechanisms. In an earlier acousticplayback study with Eleutherodactylus coqui, individualmales exhibited one of two separate classes of maskingfunctions (Narins 1982): 59% of the males tested re-sponded to playbacks of synthetic calls with a mono-tonically decreasing function of masking noise, whereas41% responded with a peaked masking function. Thus,ambient (or experimentally-supplied) noise may affectdifferent individuals in a population differently, and isworth additional study.

In another acoustic playback study carried out ineastern Puerto Rico, synthetic call notes were mixedwith high-level masking noise before being presentedto individual calling males of E.coqui (Narins 1983). Incontrol trials, male frogs responded to playbacks ofunmasked synthetic calls by producing a short-latency,one-note, antiphonal call. Synthetic calls mixed withhigh-level, low-pass masking noise evoked significantlyfewer antiphonal calls than did control stimuli, yetsynthetic calls mixed with high-level, high-pass mask-ing noise were as effective as control stimuli in evok-ing antiphonal calls. This suggests that saturating thehigh-frequency fibers from the BP (with the high-level,high-pass masking noise) has relatively little effecton the male’s calling behavior. Thus, antiphonal

calling in this species is likely mediated by the AP(Narins 1983).

In contrast, phonotaxis in males of E. femoralis ap-pears to be mediated by the basilar papilla. Using asimilar protocol as Narins (1983), we have now dem-onstrated that saturating the high-frequency fibers fromthe BP with high-pass masking noise significantly re-duced the number of successful approaches made bymales under test. In contrast, temporarily acoustically‘‘lesioning’’ the AP with this method resulted in pho-notaxis behavior that did not differ significantly fromunmasked control trials. Thus, it is becoming clear thatdifferent male behaviors may be subserved by differentinner ear organs in different species. Nevertheless, aunifying principle that appears to be emerging from thiswork is that the inner ear organ subserving the observedbehavior (antiphonal calling in E. coqui, phonotaxis inE. femoralis) corresponds to the organ maximally sen-sitive to the species-specific call mediating that behavior.We believe that this technique of ‘‘acoustic lesioning’’holds promise as a non-invasive field method to tem-porarily and selectively nullify a particular inner organ(or selected frequency range) while testing the efficacy ofthe other (or the remaining frequencies).

Acknowledgements We thank Vanessa Hequet, Muriel Nugent, theAssociation Arataı and especially Philippe Gaucher for theirunfailing and generous assistance with this project. Brigitte Got-tsberger, Daniela Grabul, Edith Gruber, and Christian Proy as-sisted during the field experiments. Norbert Ellinger, HerbertGasser and Daniela Grabul synthesized the playback signals. Thestudy was supported by grants from the Austrian Science Foun-dation (FWF Projects P 11565, P 15345) to W.H. and by NIHGrant no. DC00222 and Academic Senate grant 3501 to P.M.N.

References

Doherty JA, Gerhardt HC (1984) Evolutionary and neurobiologi-cal implications of selective phonotaxis in the spring peeper(Hyla crucifer). Anim Behav 51:981–989

Ellinger N, Hodl W (2003) Habitat acoustics of a neotropicallowland forest. Bioacoustics 13:297–321

Feer F (1999) Effects of dung beetles (Scarabaeidae) on seeds dis-persed by howler monkeys (Alouatta seniculus) in the FrenchGuianan rain forest. J Trop Ecol 15:129–142

Feng AS, Hall JC, Gooler DM (1990) Neural basis of sound pat-tern recognition in anurans. Prog Neurobiol 34:313–329

Feng AS, Narins PM, Xu C-H (2002) Vocal acrobatics in a Chinesefrog, Amolops tormotus. Naturwissenschaften 89:352–356.

Gerhardt HC (1978) Temperature coupling in the vocal commu-nication system of the gray treefrog Hyla versicolor. Science199:992–994

Gerhardt HC (1988) Acoustic properties used in call recognition byfrogs and toads. In: Fritzsch B, Ryan MJ, Wilczynski W,Hetherington TE, Walkowiak W (eds) The evolution of theamphibian auditory system. Wiley, New York, pp 455–483

Gerhardt HC, Doherty JA (1988) Acoustic communication in thegray treefrog, Hyla versicolor: evolutionary and neurobiologicalimplications. J Comp Physiol A 162:261–278

Gerhardt HC, Huber F (2002) Acoustic communication in insectsand anurans. University of Chicago Press, Chicago

Gerhardt HC, Rheinlaender J (1980) Accuracy of sound localiza-tion in a miniature dendrobatid frog. Naturwissenschaften67:362–363

828

Gottsberger B, Gruber E (2004) Temporal partitioning of repro-ductive activity in a Neotropical anuran community. J Trop Ecol20:271–280

Grafe TU (1996) The function of call alternation in the Africanreed frog (Hyperolius marmarotus): precise call timing preventsauditory masking. Behav Ecol Sociobiol 38:149–158

Hodl W (1982) Phyllobates femoralis (Dendrobatidae): Rufverhal-ten und akustische Orientierung der Mannchen (Freilandauf-nahmen). Scientific film Ctf 1788 BHWK Wien

Hodl W (1983) Phyllobates femoralis (Dendrobatidae): Rufverhal-ten und akustische Orientierung der Mannchen (Freilandauf-nahmen). Acompanying paper to scientific film Ctf 1788BHWK Wien; Wiss Film 30:12–19

Hodl W (1987) Dendrobates femoralis (Dendrobatidae): a handyfellow for frog bioacoustics. In: Proceedings of 4th Ord GenMtg Soc Eur Herpetol, pp 201–204

Lewis ER, Narins PM (1999) The acoustic periphery of amphibi-ans: anatomy and physiology. In: Fay RR, Popper AN (eds)Comparative hearing: fish and amphibians. Springer, BerlinHeidelberg New York, pp 101–154

Lewis ER, Narins PM, Cortopassi KA, Yamada WM, Poinar EH,Moore SW,YuXL (2001)Domalewhite-lipped frogs use seismicsignals for intraspecific communication? Am Zool 41:1185–1199

Luddecke H (1974) Ethologische Untersuchungen zur Fort-pflanzung von Phyllobates palmatus (Amphibia, Ranidae). PhDThesis, Johannes Gutenberg University

Narins PM (1982) Effects of masking noise on evoked calling in thePuerto Rican coqui (Anura: Leptodactylidae). J Comp Physiol147:439–446

Narins PM (1983) Synchronous vocal response mediated by theamphibian papilla in a neotropical treefrog: behavioral evi-dence. J Exp Biol 105:95–105

Narins PM, Capranica RR (1980) Neural adaptations for pro-cessing the two-note call of the Puerto Rican treefrog, Eleut-herodactylus coqui. Brain Behav Evol 17:48–66

Narins PM, Feng AS, Yong H-S, Christensen-Dalsgaard J (1998)Morphological, behavioral and genetic divergence of sympatricmorphotypes of the treefrog Polypedates leucomystax in Pen-insular Malaysia. Herpetologica 54:129–142

Narins PM, Hodl W, Grabul DS (2003) Bimodal signal requisitefor agonistic behavior in a Dart-poison frog, Epipedobates fe-moralis. Proc Natl Acad Sci USA 100:577–580

Narins PM, Lewis ER, McClelland BE (2000) Hyperextended callrepertoire of the endemic Madagascar treefrog, Boophis mad-agascariensis (Rhacophoridae). J Zool 250:283–298

Narins PM, Zelick R (1988) The effects of noise on auditory pro-cessing and behavior in amphibians. In: Fritzsch B, Ryan MJ,Wilczynski W, Hetherington TE, Walkowiak W (eds) Theevolution of the amphibian auditory system. Wiley, New York,pp 511–536

Roithmair M (1992) Territoriality and male mating success in theDart-poison frog, Epipedobates femoralis (Dendrobatidae, An-ura). Ethology 92:331–343

Rand S (1988) An overview of anuran acoustic communication. In:Fritzsch B, Ryan MJ, Wilczynski W, Hetherington TE, Walk-owiak W (eds) The evolution of the amphibian auditory system.Wiley, New York, pp 415–431

Ryan MJ (1985) The tungara frog. Chicago University Press,Chicago

Ryan MJ (1986) Synchronized calling in a treefrog (Smilisca sila).Brain Behav Evol 29:196–206

Schluter A (1980) Bioakustische Untersuchungen an Dendrobat-iden in einem begrenzten Gebiet des tropischen Regenwaldesvon Peru (Amphibia: Salientia: Dendrobatidae). Salamandra16:149–161

Schwartz JJ (1986) Male calling behavior and female choice in theneotropical treefrog Hyla microcephala. Ethology 73:116–127

Schwartz JJ (1987) The importance of spectral and temporalproperties in species and call recognition in a neotropical tree-frog with a complex vocal repertoire. Anim Behav 35:340–347

Schwartz JJ, Wells KD (1984) Interspecific acoustic interactions ofthe neotropical treefrog Hyla ebraccata. J Comp Physiol 14:21–224

Suga N (1972) Analysis of information-bearing elements in com-plex sounds by auditory neurons of bats. Audiology 11:58–72

Vences M, Kosuch J, Lotters S, Widmer A, Jungfer KH, Kohler J,Veith M (2000) Phylogeny and classification of poison frogs(Amphibia: Dendrobatidae), based on mitochondrial 16S and12S ribosomal RNA gene sequences. Mol Phylogenet Evol15:34–40

Wells KD (1977) The social behavior of anuran amphibians. AnimBehav 25:666–693

Wells KD (1988) The effect of social interactions on anuran vocalbehavior. In: Fritzsch B, Ryan MJ, Wilczynski W, Hethering-ton TE, Walkowiak W (eds) The evolution of the amphibianauditory system. Wiley, New York, pp 433–454

Wilczynski W, Rand AS, Ryan MJ (1995) The processing ofspectral cues by the call analysis system of the tungara frog,Physalaemus pustulosus. Anim Behav 49:911–929

Zakon HH, Wilczynski W (1988) The physiology of the anuraneighth nerve. In: Fritzsch B, Ryan MJ, Wilczynski W, Hethe-rington TE, Walkowiak W (eds) The evolution of theamphibian auditory system. Wiley, New York, pp 125–155

Zelick RD, Narins PM (1982) Analysis of acoustically-evoked callsuppression behavior in a neotropical treefrog. Anim Behav30:728–733

Zimmermann H, Zimmermann E (1988) Ethotaxonomie und zoo-geographische Artengruppenbildung bei Pfeilgiftfroschen (An-ura: Dendrobatidae). Salamandra 24:125–160

829