An Experimental Investigation of the Bioacoustics of Cowbird Song

Behav Ecol Sociobiol (1981) 9:211-217 Behavioral Ecology and Sociobiology 9 Springer-Verlag 1981 An Experimental Investigation of the Bioacoustics of Cowbird Song Andrew P. King 1, Meredith J. West 2, David H. Eastzer 2, and J.E.R. Staddon 1 1 Department of Psychology, Duke University, Durham, North Carolina 27706, USA 2 Department of Psychology, University of North Carolina, Chapel Hill, North Carolina 27514, USA Received March 10, 1981 / Accepted July 3, 1981 Summary. Female cowbirds (Molothrus ater ater), maintained in isolation from males during the breeding season, respond to the playback of male song with copulatory postures. They respond to some songs more than to others. Cowbird song potency can thus be operationally defined by the proportion of copulatory postures a song elicits across multiple playbacks. The purpose of the present study was to explore whether song potency changes with distance in the field. No field recordings elicited high levels of responding by the females. When songs of known high potency are systematically degraded, the results indicate that female cowbirds are sensitive to small changes in signal to noise ratio and to atmospheric attenuation. The data suggest that cowbird song potency degrades very rapidly with transmission distance in the field. Introduction In the past decade, significant progress has been made in understanding the role of the environmental degradation of sound in the evolution of acoustic signals (e.g. Morton 1975). Most of the work has focused on measuring the attenuation of sound as a function of frequency in a particular habitat by re-recording pure tones from a standardized source and distance. These studies indicate how factors such as habitat type, atmospheric conditions, positions of the signaller in the environment, time of day, and background noise may influence frequency dependent attenuation and constrain the evolution of acoustic communication signals (Chappius 1971 ; Marten and Marler 1977; Marten et al. 1977; Morton 1975; Waser and Waser 1977). Other investigations have emphasized the importance of other forms of environmental degradation such as reverberation and amplitude fluctation (Michelson 1978; Richards and Wiley 1980; Wiley and Richards 1978). These studies emphasize that recipients not only need to detect that the signal has occured but must also be able to discriminate among variant signals. Finally, other investigators have considered the functional consequences of signal degradation (Bremond 1968 ; Richards 1978 ; Schubert 1971 ; Shiovitz 1975). Richards, for example, played back artificially degraded songs of the rufous sided towhee (Pipilo crythrophthalmus) and measured the effect upon individual recognition. All of these studies used the male's territorial response to playback as a measure of song function, and consequently, the signal degradation investigated concerned relatively long distances. Cowbirds offer an additional perspective to the study of sound transmission of bird song for two reasons. First, the male's song is used at short distances from the recipient. As part of the male's courtship display to the female, the song is sung at distances of less than one foot from the female. Second, female cowbirds respond with a copulatory posture to the playback of male song, thus providing a quantitative measure of the short distance function of cowbird songs (West et al. 1979). Some cowbird songs, which are particularly effective in eliciting copulatory postures from female cowbirds, can produce aggressive attacks from other male cowbirds if they are close to the courting pair (West and King 1980). Thus, the short distance 'message' for the female may have evolved to degrade very rapidly so as to avoid alerting other males. The purpose of this report is to present data regarding the effects of recording conditions upon song potency in cowbirds. The measure of potency was determined by playback to captive isolated M.a. ater females: the more responses a song elicited, the more potent it was considered. We used the highest quality 0340-5443/81/0009/0211/$01.40

212 songs we could locate from two subspecies (M. a. ater and M.a. obscurus) and from six geographic locations. Further, we tested the relative potency of the same song re-recorded under different conditions. Would females respond differentially to the same song when it was recorded at greater distances or under different environmental conditions? In addition, conditions were created in which the signal to noise ratio of the song was manipulated independently from the recording distance. Materials and Methods S~tbjects For both experiments, ten different M.a. ater females were tested. All were obtained from the Patuxent Wildlife Station in Laurel, MD, and had been introduced into sound attenuation chambers in late fall. Housing The sound attenuation chambers each consisted of two concentric boxes constructed of plywood and sheetrock. Wood and acoustic tile baffles between the boxes were designed to be most effective between 2 and 16 khz. Suppression was greater than 39 db at 1,000 Hertz, and it increased with higher frequencies to greater than 50 db between 8 and 16 khz. The interior box was a 1.1 m cube, fabric-lined to reduce sound reflection, lighted by two 40-watt Vita Lite tubes and continuously ventilated. Procedure Field Playback Songs. Eleven songs of wild cowbirds recorded at different distances in the field were used (Table 1). Eight songs were selected from the Cornell Library of Natural Sounds and three were recorded in Tompkins County, New York. Ten of the songs were recorded during the breeding season and one (No. 1) was recorded just prior to it (March 17). It was included because it was recorded at 1.5 m. Playback of Same Songs Under Different Conditions9 Two songs were used both of which were originally recorded from individually 85% t* IP 74% p. 36% f, 12- I0-- 8 6 4 2 0 L D k,~im~ ~....5S il I 36% Fig. 1. Sonograms of four versions of song S1 and their potencies in % (for explanation see Materials and Methods). Units on the ordinate are khz. The original recording (A) was rerecorded inside a reverberant room at 0.3 m (B), 3 m (C), and 10 m (D). Note the increased amount of reverberation (as evidenced by the echoes, especially on the introductory notes) and the lowered amplitude on the short, high frequency note (IPU) as a function of increasing distance

213 96% E IF t._~. 62% 70%, J"~r 57% 57% I0 I ~,,r.5s Fig. 2. Sonograms of the six versions of song $2 and their potencies. Re-recording of the original song (E) at 0.3 m inside a reverberant room (F) produced no change in potency. Re-recording at 3 m (G and H) and 10 m (I and 3) produced a significant drop in potency relative to the original version of the song (E). Although there is much more reverberation in song versions re-recorded inside the room (G and /) than those re-recorded outside (H and J), there is no significant difference in potency as a function of the degree of reverberation (see Table 2 for more information) housed males maintained in auditory and visual isolation from other male cowbirds from the winter through the breeding season. High potency isolate songs were used in order to expand the range for manipulation. The recordings were made within 0.3 m with a Uher 4000L recorder and a Sennheiser 435 microphone. Both of these songs had been previously played back to the females and found to be especially potent. Song 1 was played back through the J.B.L. 2420 driver and 2340 horn and re-recorded using a Uher 4000L recorder and Sennheiser 435 microphone. The song was played back in a reverberant room measuring 10 x 11 x 2.3 m. The sound pressure level (SPL) of the playback song was held constant at 83 db A weighted impulse reading at 3 m from the speaker using a B & K 2209 sound pressure meter. The song was played back and re-recorded at three distances: 0.3 m, 3 m and l0 m. For all three recordings, the "record" level of the Uher was adjusted to produce a maximum deflection of -6 db on the VU meter. Thus, for the three re-recordings, the playback level was held constant while the record level was increased as the distance between the playback speaker and microphone was increased (Fig. 1). A fourth recording in the series was also generated. For this recording~ the playback amplitude was 83 db A weighted impulse reading at 0.3 m, was apposed to the three previous recordings for which the playback amplitude was 83 db A weighted impulse at 3 m. This produced a song with a slightly lowered signal to noise ratio relative to the other recordingmade at 0.3 m. The same equipment and general produres were used to produce a second series (Song 2). Five songs were produced by rerecording the original song inside the same reverberant 10 x 11 x 2.5 m room at 0.3 m, 3 m and 10 m and outside in an open field in full sun at 3 m and 10 m. For all of the re-recorded songs, the same record level setting was used. In order to equate signal to noise ratios, the volume of the playback speaker was adjusted at the different distances until it produced an SPL of 90 db peak reading at the face of the recording microphone. This thus produced a series of songs with equivalent signal to noise ratios (i.e., the record level and signal level was held constant) but varying degrees of degradation and reverberation. Both of the songs that were re-recorded inside at 3 and 10 m had much more reverberation than any of the outside recordings, and the 10 m song had more reverberation than the 3 m song (Fig. 2), Signal to noise ratios were measured peak to peak on a Tektronix 5113 oscilloscope. Playback Procedure. The songs were played back at 80_+1.5 db impulse measured with a B & K 2209 sound pressure meter. The playback level was measured at 0.6 m from the speaker, the maximum distance a female could be from the speaker. The ambient SPL in the chambers during playbacks was 41 _+2 db slow reading at 500 Hz and declined to 18_+2 db at 12 khz. The lowest frequencies in cowbird song are at 500 Hz and the highest at 12 khz. The songs were tested during the female's reproductive periods as indicated by egg laying. Each day for approximately six weeks, seven songs were played back to the females in their sound attenuation chambers. The time interval between presentation of each song was 90 rain. The order of song presentation varied each day but was the same for all females. The response measure was the presence or absence of a copulatory posture by each female within one second after the song's onset. The copulatory response is described as follows: approximately 400 msec after the song begins, a female spreads her wings, arches her neck and body and separates

214 Table 1. Mean percent and range of copulatory responses to the songs of wild cowbirds recorded at different distances. The proportion of playbacks of a given song which elicit the response is calculated for each female, and these proportions are averaged across the ten females to derive the mean percent potency of each song. The range provides the highest and lowest individual female response rates Song and Date Geographic Subrecording distance origin species Mean percent and range of copulatory responses Song No. 1 at 1.5 m 3/17 Tulson, obscurus 3 AZ (0-33) No. 2 at 13 m 4/28 Tioga Co., ater 25 NY (0-100) No. 3 at 13 m 5/19 Freeville, ater 39 NY (0 83) No. 4 at 13 m 5/19 Freeville, ater 20 NY (0-75) No. 5 at 13 m 5/19 Freeville, ater 11 NY (0-50) No. 6 at 17 m 4/25 Ohio ater 35 (0~5) No. 7 at 23 m 5/19 Glendora, obscurus 0 CA No. 8 at 23 m 5/19 Glendora, obseurus 5 CA (0-33) No. 9 at 33 m 5/19 Glendora, obscurus 5 CA (0-33) No. 10 at 33 m 5/10 West Virginia ater 3 (0-33) No. 11 at 50 m 5/5 Tompkins Co., ater 5 NY (0 25) Table 2. Mean percent and range of copulatory responses to songs S1 and $2 recorded under different acoustic conditions. The signal to noise ratios were measured peak from the face of a Tektronix 5113 oscilloscope Figure Signal to Mean percent designation noise and range ratio of copulatory responses Playback song S1 : S/N varied A Original recording 48 85 (60-100) B Inside at 0.3 m 43 74 (14-100) Inside at 0.3 m 40 40 (reduced S : N) (0-89) C Inside at 3 m 37 36 (0-54) D Inside at 10 m 27 36 (0-63) Playback song $2: S/N held constant in attempt to determine effect of reververation alone E Original recording 45 95 (60-100) F Inside at 0.3 m 43 90 (67-100) G Inside at 3 m 42 62 (0-100) H Outside at 3 m 43 70 (14-100) I Inside at 10 m 42 57 (0-100) J Outside at 10 m 43 57 (14-100) the feathers surrounding the cloacal region; she then holds this posture for one to seven seconds after the song terminates and then often preens herself. The female cowbird does not respond to the songs of other species and shows a high degree of selectivity with respect to the cowbird songs that elicit the most responses (King and West 1977). If she hears only a few songs each day, the response can be reliably elicited every day for six weeks. Each song was played back to each female an average of five times (range 4-7). The data were then converted to mean percentages to account for the different numbers of playbacks to each female (see note, Table 1). Partial responses such as approach or head orientation do not occur in this species and thus were not considered. Other cowbird songs, not relevant to this experiment, were also played back. Results 1. Female's Response to Field Recordings The wild songs were relatively ineffective releasers of the female's response (Table 1). The differences among song effectiveness for the 11 songs appeared to relate most directly to recording distance. The least effective songs (Less than six percent) were all recorded at distances of 20 m or more. The only other very ineffective song was No. 1, which although recorded at a short distance, was obtained before the breeding season began. The songs recorded at less than 20 m were more potent. Of these, two (No. 3 and No. 6) were responded to by all the females and received response rates of 39% and 35% respectively, which are equivalent to the potencies of songs recorded from groups of males in our aviaries (West et al. 1981). 2. Signal to Noise Ratio and Reverberation Re-recording of the Song 1 slightly reduced its potency (Table 2). With increasing distance beyond 0.3 m between the speaker and the microphone, how-

215 ever, the song's potency declined from 74% to 36%. The difference in potency between the original song and the song at 0.3 m (43 db S :N) was not significant as tested by Wilcoxon test for matched pairs, but the differences between the original and the 3 m song and the original and the 10 m song were significant (P<0.01). The songs recorded at 0.3 m but with a reduced signal to noise ratio (40 db) also produced reliably different levels of responding (P < 0.01) with the song having the higher ratio receiving more responses. The results for the second series indicate that reverberation apparently plays little role in the female's response (Table 2). The same song with the same signal to noise ratio re-recorded inside and outside received the same percentages of responses at both 3 and 10 m. The songs recorded at 3 and 10 m, however, were reliably less effective than either the original song or the same song at 0.3 m (P<0.01). Thus it appears that relatively high levels of reverberation (see Fig. 2 caption) do not affect song potency but that absolute distance independent of the signal to noise ratio does, presumably as a result of atmospheric degradation. Discussion Cowbird songs recorded in the field elicited only minimal responding from captive female cowbirds: the greater the recording distance, the fewer the number of responses indicating the importance of proximity to the preservation of song potency in cowbirds. Moreover, when the same song, recorded at different distances inside and outside, was compared, the same finding was obtained: the females responded most often to the perceptually 'best' version of male's song. The eleven songs tested in the first part of the experiment resulted in an average response rate of 13%. This figure is in significant contrast to female cowbirds' responding to the songs of normally reared males recorded under our laboratory conditions. In six years of testing over 150 females with over 40 normal male cowbird songs, the average response rate has been 35%. Moreover, in no case have any songs received average rates of 10% or below which was the case for six of the songs reported here. The extremely low level of response observed here is also important to ruling out the possibility that the results can be explained by the geographic variation in the origin of the songs. We have previously tested M.a. ater and M.a. obscurus songs recorded within 1 m of the male with females of both subspecies (King et al. 1980). The results of that study showed that although the females respond most often to songs of their own subspecies, they also respond to songs of the other subspecies. The range of responding found in that study was between 7 and 42% to songs of the other subspecies with a mean of 27%. This is far higher than the percentage of responses reported for obscurus songs (Table 1), indicating that the low level of responsiveness to field recordings in this study is due to the distance at which the songs were recorded and not the geographic location. Moreover, as with M.a. ater songs, in no case have any M.a. obscurus songs recorded in the laboratory ever received rates of response less than 10%. Given this line of reasoning, one might have expected song No. 1, which was recorded at a very close distance, to have elicited more responses from the females even though it was from an M.a. obscurus male. It was however recorded in March which is prior to the cowbird's mating season. We have collected data in our laboratory that indicate that such early spring songs are less effective than later ones at releasing the female's response. We recorded songs from five males in March and then again in April. For each male, there was a substantial increase in song potency for the same song types as judged by the female bioassay. Thus, song No. 1 in the present experiment was probably not potent because of the season rather than because of its geographic origin. What are the structural characteristics of cowbird song that could account for changes in potency with distance? Cowbird song is composed of three phrases: an introductory series of notes between 500 and 4,000 Hz, followed by a brief note between 8 and 12 khz with a mean duration of 50 msec, and a final whistled phrase lasting approximately 450 msec and modulated between 5 and 12 khz. The peak amplitude on the introductory notes is typically sung between 8 and 20 db below the peak amplitude of the whistle phrase. Cowbird song is also characterized by discrete amplitude envelopes and rapid amplitude modulation (AM) in all three phrases (Greenewalt 1968). Examination of oscillograms of the wild cowbird songs revealed a lack of both discrete amplitude envelopes and rapid AM in these songs; this type of degradation of a signal has been shown to occur at distances as short as 7 m (Richards 1978). Also, the introductory notes in these songs were barely audible. We know this phrase to be especially important to the female's response (West et al. 1979). In the field, it has frequently been observed that only the whistle phrase of the song is heard at a distance (e.g. Darley 1968). This fact alone might account for the ineffectiveness of these songs. Finally, all of the long distance songs had signal to noise ratios less than 40 db (range 26-40), measured peak to peak. This led us to test the importance of signal to noise ratio for female responsiveness, as all the laboratory re-

216 cordings of potent song used in previous studies had higher S :N ratios. All of these factors probably interacted to reduce song effectiveness. The other series of playbacks was designed to examine the effects of signal to noise ratio, reverberation and atmospheric degradation. The results demonstrated that female cowbirds are sensitive to differences as small as 3 db (43 db vs. 40 db) in a song's signal to noise ratio and to low levels of atmospheric degradation in re-recorded versions of the same song. This does not mean, however, that such small differences are as important when the females compare different songs from different males. These experiments only show that the females can detect such small acoustic differences and that, for a given song, they respond most to the perceptually 'best' version. Thus, both sets of data indicate that song potency as perceived by female cowbirds has evolved in such a way so as to minimize the range of detection. Previous work on cowbird song structure also supports this idea. Our acoustic analyses have shown that the first two phrases of the song are pivotal to eliciting the female's response (West et al. 1979). These two phrases are frequently characterized by low relative amplitude and rapid AM and FM of 500-1,200 CPS. Neither the low amplitude of the introductory notes nor the rapid AM would appear to be designed to travel any great distance. A possible mechanism for the decreased potency of the short distance signal might be found in evidence regarding cowbird hearing, indicating a sharp drop in sensitivity above 8 khz (Hienz et al. 1977). This is the range of the second phrase in cowbird song that is critical to releasing the female's copulatory response. These data are also consistent with a previous finding that relative amplitude can be critical to establishing song potency (West et al. 1979); by increasing the relative amplitude of the second phrase the potency of a song can be increased. It appears that the structural elements in song that code potency as well as the auditory system work in harmony to limit severely the range of detection of potency information. One apparent function of the limited transmission distance of potency may be that it decreases intraspecific aggression among males. Males that sing high potency songs in the presence of other males are frequently attacked and killed. It is important to note that the decrease in potency observed in these experiments involves relatively small distances. Consider the decrease in potency from 90 to 62% when a song is re-recorded at 0.3 and 3 m and the signal to noise ratio is held constant (songs F and G Fig. 2). Many field recordings in this and other species are typically made at 30 or more m. It is thus not surprising that the responses to the field recordings were generally quite low. These facts indicate the impracticality of doing playback experiments in the field in this species where the short distance message is the primary concern. Not only is it difficult to obtain accurate recordings of the short distance message in the field, but it would also be very difficult to play back 'accurate' recordings to the female. A recent study of cowbirds' responses to field playback strengthens these speculations. Dufty (1979) reported that neither male nor female cowbirds responded to playback of male vocalizations at distances of approximately 20 m. These experiments also suggest that reverberation in a song may be of secondary importance. This is interesting because reverberation produced the most extensive alteration of the signal when compared to either small changes in the signal to noise ratio or to the low level of atmospheric degradation produced at short distances. The explanation for this effect might be that reverberation does not necessarily give information about distance from the sound source. By contrast, both signal to noise ratio and the extent of atmospheric degradation will correlate with distance. Thus, it may be that the females are especially sensitive to songs that indicate that the sound source is very close. We and others have observed that a male is very close to the female when he sings a song that leads to copulation (Darley 1978; Laskey 1950; Payne 1973). In summary, the data indicate that the information in the song that codes potency degrades over very short distances. Furthermore, the data show that the female's response to song is a relativistic one: she responds most to the song that sounds closest. These findings complement current theories that stress the importance of understanding the adaptive relationship between sound transmission and reception and the ecological and social environment of the organism (Morton 1975; Marten and Marler 1977). The data also stand as an example of the difficulty of interpreting playback responses to recordings made at some distance. In the cowbird, lack of response to a far recording could mean that the song lacks a critical property, e.g., potency or that it is perceptually inadequate. Thus, only comparison tests using close recordings allow correct interpretation. Acknowledgements. This work was supported by N.S.F. grant BNS 78-07223 and a grant from the University Research Council. We thank R. Haven Wiley for the use of his equipment, W. Stickel and the Patuxent Wildlife Station for assistance in obtaining birds.

217 References Bremond JC (1968) Recherches sur la semantique et les elements vecteurs d'information dans les signaux acoustiques du rougegorge (Erithacus rubecula L.). Terre Vie 22:109~20 Chappuis C (1971) Un example de l'influence du milieu sur les 6mmissions vocales des oiseaux: l'6volution des chantes en for~t 6quatoriale. Terre Vie 25:183-202 Darley JA (1968) The social organization of breeding brownheaded cowbirds. PhD thesis, University of Western Ontario, London Darley JA (1978) Pairing in captive brown-headed cowbirds (Molothrus ater). Can J Zool 56:2249-2252 Dufty AM (1979) Response of brown-headed cowbirds to playbacks of cowbird vocalizations. Paper No. 137 presented at American Ornithologists' Union meeting, College Station, Texas Greenwalt C (1968) Bird song: Acoustics and physiology. Smithsonian Institution, Washington, DC Heinz RD, Sinnott JM, Sachs MB (1977) Auditory sensitivity of the redwing blackbird (Agelaius phoeniceus) and brown-headed cowbird (Molothrus ater). J Comp Physiol Psychol 91:1365-1376 King AP, West MJ (1977) Species identification in the N.A. cowbird: Appropriate responses to abnormal seng. Science 195 : 1002-1004 King AP, West MJ, Eastzer DH (1980) Song structure and song development as potential contributors to reproductive isolation in cowbirds (Molothrus ater). J Comp Physiol Psycho194:1028 1039 Laskey A (1950) Cowbird behavior. Wilson Bull 62:157-174 Marten K, Marler P (1977) Sound transmission and its significance for animal vocalization : I. Temperate habitats. Behav Ecol Sociobiol 2: 271-290 Marten K, Quine D, Marler P (1977) Sound transmission and its significance for animal vocalization : II. Tropical forest habitats. Behav Ecol Sociobiol 2:291-302 Michelsen A (1978) Sound reception in different environments. In: Ali MA (ed) Sensory ecology review and perspectives. Plenum, New York Morton E (1975) Ecological sources of selection on avian sounds. Am Nat 108:17-34 Payne RB (1973) The breeding season of a parasitic bird, the brown-headed cowbird, in central California. Condor 75 : 80-99 Richards DG (1978) Environmental acoustics and song communication in passerine birds. PhD thesis, University of North Carolina, Chapel Hill Richards DG, Wiley RH (1980) Reverberations and amplitude fluctuations in the propagation of sound in a forest: Implications for animal communication. Am Nat 115:381-399 Schubert M (1971) Experimentelte Untersuchungen fiber die reaktionsausl6senden Signalstrukturen des Fitis-Gesanges, Phylloscopus t. trochilus. L., und das Verhalten gegenfiber arteigenen Rufen. Behaviour 38 : 250-288 Shiovitz KA (1975) The process of species-specific song recognition by the indigo bunting, Passerina cyanea, and its relationship to the organization of avian acoustical behaviour. Behaviour 55 : 128-179 Waser PM, Waser MS (1977) Experimental studies of primate vocalizations for long-distance propagation. Z 'Tierpsychol 43 : 239-263 West M J, King AP (1980) Enriching cowbird song by social deprivation. J Comp Physiol Psychiol 94:263-270 West M J, King AP, Eastzer DH, Staddon JER (1979) A bioassay of isolate cowbird song. J Comp Physiol Psychol 93:124-133 West MJ, King AP, Eastzer DH (1981) Validating the female bioassay of cowbird song: Relating differences in song potency to mating success. Anim Behav 29:490-501 Wiley RH, Richards DR (1978) Physical constraints on acoustic communication in the atmosphere: Implications for the evolution of animal vocalizations. Behav Ecol Sociobiol 3 : 69-94