Behavioral Ecology Advance Access originally published online on July 17, 2007
Behavioral Ecology 2007 18(5):944-951; doi:10.1093/beheco/arm063
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A mutual understanding? Interspecific responses by birds to each other's aerial alarm calls
School of Botany and Zoology, Australian National University, Canberra 0200, Australia
Address correspondence to R.D. Magrath. E-mail: robert.magrath{at}anu.edu.au.
Received 18 January 2007; revised 13 May 2007; accepted 13 June 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
Individuals are likely to benefit from responding to the alarm signals of other species with similar predators, and mutual interspecific responses to aerial (hawk) alarms are thought to be common in birds, in part because similarity in alarm call structure among species might facilitate detection or interpretation. However, there has been no test of whether interspecific responses to aerial alarm calls can involve mutual responses between species and only incomplete tests of the response of any species to such heterospecific alarms. We describe the aerial alarm calls of white-browed scrubwrens (Sericornis frontalis) and superb fairy-wrens (Malurus cyaneus) and use a playback experiment to test for mutual responses to each other's aerial alarm calls. The 2 species occur in similar habitats and can co-occur in mixed-species flocks during the nonbreeding season. The aerial alarm calls of both species are high pitched (
7 kHz) and rapidly frequency-modulated calls but are distinct in frequency measures and only the scrubwren's call had 2 parallel sounds. Both species fled to cover after playback of either their own or the other species' alarm calls but never to control sounds. The response to either species' alarm was almost invariant in both species in an experiment at high natural amplitude, but there was a slightly lower response to heterospecific compared with conspecific alarms when playbacks were at the mean natural amplitude. Our results demonstrate, after at least 50 years of interest in the subject, that there can be mutual responses to aerial alarm calls between species. Key words: acoustic communication, aerial alarm calls, interspecific communication, passerines, predation.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
Many species of birds have "aerial" or "hawk" alarm calls that are primarily given to predatory birds in flight (Marler 1955
). These calls have become a classic textbook example of adaptive signal design because many species have calls that are of high constant pitch, with gradual onset and termination, acoustic features that are likely to make it difficult for hawks to hear or locate the call (Marler 1955; Klump and Shalter 1984; Klump 2000). Aerial alarm calls appear to provoke conspecifics to freeze or flee for cover—hence the term "flee" alarm—in contrast to "mobbing" calls, which prompt others to approach and harass a predator (Klump and Shalter 1984; Bradbury and Vehrencamp 1998). Playback experiments to 2 species have confirmed that aerial alarms can alone prompt conspecifics to freeze or flee for cover (Evans et al. 1993; Leavesley and Magrath 2005; see also Griesser and Ekman 2004). Similarly, playbacks of mobbing calls prompt others to approach (e.g., Forsman and Mönkkönen 2001).
Interspecific responses to aerial alarm calls are believed to be common (Marler 1957; Haftorn 2000; Caro 2005), and mixed-species flocks or colonies might even form so that some species can take advantage of alarm calls of others (e.g., Burger 1984; Goodale and Kotagama 2005). However, most claims of interspecific responses among birds rely on observations rather than experiments. For example, Hindwood (1937) observed a mixed flock of insectivorous passerines fleeing for cover after the alarm call of a gray fantail, Rhipidura flabellifera; Marler (1957) saw chaffinches, Fringilla coelebs, fleeing after he heard aerial "seeet" calls of Parus species; Gaddis (1980) reported that members of mixed species appeared to respond to the aerial alarm calls of Carolina chickadees, Parus carolinensis, and tufted titmice, Parus bicolor; and Haftorn (2000) noted that willow tits, Parus montanus, responded to the alarm calls of great tits, Parus major, robins, Erithacus rubecula, redstarts, Phoenicurus phoenicurus, and reed buntings, Emberiza schoeniclus. Such observations are suggestive of interspecific responses, but it is necessary to carry out playback experiments to ensure that the call itself prompts escape. Without playbacks, the apparent "response" of the other species could be due to seeing the predator independently or observing the evasive actions of callers.
Five playback experiments have aimed at testing whether birds respond to the flee alarm calls of other species. These important studies were carried out before the problem of pseudoreplication in playback experiments was widely appreciated (Kroodsma 1989), and so the evidence is incomplete because a lack of replication of playback sounds or lack of controls (Table 1). Members of mixed-species Amazonian flocks responded to playback of an aerial alarm call of a bluish-slate antshrike that acts as a sentinel species (Munn 1986). Downy woodpeckers responded with freezing and vigilance to playback of aerial alarms of 2 parids (Sullivan 1984). Western grebes depart their nests after playback of a Forster's tern, Sterna forsteri, alarm call (Nuechterlein 1981). Nestling great tits, just before fledging, responded to the seeet call of a blackbird (Rydén 1978). Unfortunately, in each case, there is pseudoreplication in experimental design, with only a single example of an alarm call broadcast. This means that one can draw conclusions only about the specific alarm call broadcast and not about interspecific responses (Kroodsma et al. 2001). In one experiment in which alarm call playbacks were replicated, Møller (1988) found that sparrows (Passer spp.) fled after great tit alarms, but the focus of the study was a comparison of real and false alarm calls—to which the birds also fled—so there was no control playback. Overall, there is little evidence that birds respond to the flee calls of other species, in contrast to good evidence of interspecific responses to mobbing calls (e.g., Forsman and Mönkkönen 2001; Rainey et al. 2004) and interspecific responses to alarm calls by mammals (reviewed in Caro 2005).
|
It is important to test whether aerial alarm calls produce interspecific responses, rather than to assume that they do so based on experiments using mobbing calls, because the 2 types of alarm call are quite different structurally and functionally. Aerial alarm calls are designed to be difficult for predators to hear or locate and provoke immediate flight to cover or other cryptic behavior, whereas mobbing alarm calls are designed to be conspicuous and incite approach and sometimes harassment of a predator (above). Individuals giving mobbing calls are likely often to benefit from attracting individuals of other species in addition to their own and so may be selected to give calls conspicuous to other species. By contrast, there may be no general benefit of warning other species to flee from aerial predators in flight, especially if they expose the caller to risk, and so calls may not be selected to communicate with other species.
In addition to the general belief that aerial alarms prompt heterospecific responses, it has been suggested or stated that responses are mutual—that different species respond to each other's calls (Marler 1957, 2004; Matsuoka 1980; Haftorn 2000; Caro 2005). Marler's (1957) suggestion was based on the astonishing similarity of the seeet aerial alarm calls among several European passerines, which might facilitate communication among species. However, all published playback experiments have involved responses by single species to the calls of others (Table 1), and observations of most mixed-species flocks suggest that there are usually one or more calling species to which others respond (e.g., Munn 1986). Such responses may be largely "eavesdropping" on calling species (Goodale and Kotagama 2005). The only exception appears to be some Sri Lankan flocks in which several species can give aerial alarm calls, although even there the majority of calls were given by only 2 species (orange-billed babblers, Turdoides rufescens, or greater racket-tailed drongos, Dicrurus paradiseus; Goodale and Kotagama 2005). In contrast to birds, there is playback evidence of mutual interspecific responses to flee alarm calls in rodents (Shriner 1998) and mutual responses to referential alarm calls in primates (e.g., Zuberbühler 2000, 2001; Fichtel 2004).
The issue of whether responses to aerial alarm calls are solely eavesdropping of single species on others is relevant to the evolution of interspecific communication. Kostan (2002) suggests that evolution of communication among species is likely to proceed from 1) responses of one species to the signals of another to 2) responses of each species to the signals of the other, then 3) asymmetric communication, in which both species respond to each other's signals but only one species deliberately signals to another, and finally 4) mutualism, in which both respond and provide signals. Kostan (2002) illustrates all but the third stage with avian alarm calls, including aerial calls, which might therefore prove a useful model for the evolution of mutualistic communication. Downy woodpecker's response to parid alarms appears to be an example of one species responding to others (Sullivan 1984, 1985), mobbing often appears to entail species responding to each other's alarm calls, and the interactions between dwarf mongooses, Helogale undulata, and 2 species of hornbills, Tokus spp., might be a true mutualism, including hornbills giving alarm calls to predators that are a threat to mongooses but not themselves (Rasa 1983). However, before such examples are used to support models of evolution, it is necessary to establish when interspecific responses really occur and whether they involve species responding to each other's alarms.
Here we report a test of the hypothesis of mutual interspecific responses to aerial alarm calls between 2 species of passerine, the white-browed scrubwren, Sericornis frontalis (Acanthizidae), and superb fairy-wren, Malurus cyaneus (Maluridae). Scrubwrens have high-pitched (ca. 7 kHz) aerial "trill" alarm calls given to predators in flight, and playback experiments have shown that conspecifics respond to these calls by fleeing for cover (Leavesley and Magrath 2005). The fairy-wren is similar in size, overlaps in range and habitat, also usually feeds on the ground, and in winter can form mixed-species flocks including scrubwrens and other insectivores. Its aerial alarm call has not been described but sounds similar to the scrubwren's trill aerial alarm (Magrath RD, Pitcher BJ, Gardner JL, personal observations). Here we describe the aerial alarm call of the fairy-wren, compare it with the scrubwren's call, and perform fully replicated playback experiments to test for mutual responses to each other's calls.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
Study site and species
We studied scrubwrens and fairy-wrens in Canberra, Australia, from March to June 2006 (autumn to winter) in the Australian National Botanic Gardens and in parks around Lake Burley Griffin. Both species are abundant residents at all sites and are accustomed to people. The 40-ha Botanic Gardens contains native woodland, areas planted with Australian plants, and some lawn (Magrath 2001
); parks around the Lake contain lawn, native and exotic trees, and shrubs. The study sites were distributed within a 5-km radius from the Australian National University campus, and all locations contained trees and a mixture of open areas and dense cover.
Scrubwrens are small (12–14 g) passerines that breed in pairs or cooperative groups, lay eggs from August to December, and hold territories throughout the year (Magrath et al. 2000; Magrath 2001; Higgins and Peter 2002). Similarly, fairy-wrens are small (9–10 g) passerines that breed in pairs or cooperative groups and lay eggs from September to January (Rowley 1965; Higgins et al. 2001). They defend territories while breeding, but from March to August often join larger groups, which can wander over several territories (Rowley 1965). The 2 species have overlapping territories during the breeding season and both occur, including together, in mixed-species flocks of small insectivores during the nonbreeding season (Bell 1980; Higgins and Peter 2002; Magrath RD, Pitcher BJ, Gardner JL, personal observations). There has been no detailed study of the roles of different species within these mixed-species flocks or their use of alarm calls, but one local study found that group-living species, particularly superb fairy-wrens and thornbills (Acanthiza species), are core species in the formation on these flocks (Bell 1980). As another group-living species, scrubwrens might also form the core of such flocks where the species is common.
The study sites have resident populations of raptors and large avian omnivores (Taylor and Canberra Ornithologists Group 1992) that prompt aerial alarm calls in scrubwrens and fairy-wrens (Leavesley and Magrath 2005; Magrath RD, Pitcher BJ, Gardner JL, personal observations). Raptors include the collared sparrowhawk, Accipiter cirrhocephalus, that nests within the city and preys on small passerines (Marchant and Higgins 1993). The pied currawong, Strepera graculina, is a common omnivorous predator that is known to take nestlings, fledglings, and occasionally adults of both species (Higgins et al. 2006). Laughing kookaburras, Dacelo novaeguineae, which are large kingfishers, are also common at all sites and prompt aerial alarm calls from both species.
Recording and call analysis
We used life-sized, gliding model predators to prompt aerial alarm calls from each species in the nonbreeding season, from March to May, in 2006. Models were made of extruded polystyrene foam body and wings, with a balsa wood tail, shaped and painted to resemble either a collared sparrowhawk or pied currawong. Drawing plans ensured models of one type were the same size and shape. The wings were cut with a hot wire using airfoil templates, so models could glide 15–25 m when thrown by hand. The sparrowhawk model required a fake tail to ensure flight stability; this consisted of a carbon fiber rod that projected from the tail and with colorless, transparent vanes glued to the end. Both model types had clear vertical tails for lateral stability. We matched the paint color by eye against study skins, and spectrophotometry showed that neither plumage nor painted models reflected in the UV.
We prompted aerial alarm calls from scrubwrens and fairy-wrens in the same way. One of us stood about 10 m from a focal bird or group and threw the model so that it glided past the group, 1–6 m above the birds and at a horizontal distance of 2–10 m. The other person stood between 4 and 7 m of the birds, which were on the ground or in low vegetation (<3 m), and recorded any calls using a Marantz PMD670 digital recorder and Sennheiser ME66 or ME67 directional microphone. We noted the distance of the microphone from the bird that called and the Marantz recording level and microphone used so that we could calibrate recordings against sounds of known amplitude (Charif et al. 1995; below). After recording from a group, we moved to a different site to ensure that we did not record the same individual twice. To check whether the act of throwing prompted calls, we mimed 10 "throws" per species, under the same conditions, while simultaneously recording any sounds. Only one bird, a scrubwren, called immediately after a throw, and that was a "chip" contact call, not an alarm call (Higgins and Peter 2002).
To test the validity of using models to elicit alarm calls, we compared scrubwren calls prompted by models with those given to avian predators in flight, primarily currawongs. Natural recordings were made in 2003 in the Botanic Gardens using Sony TCD-D100 Digital recorders, also sampling at 44.1 kHz, using the same microphones as the prompted recordings (Magrath et al. 2006). We then analyzed the prompted and natural recordings in the same way. We did not have an adequate sample of naturally prompted fairy-wren recordings to carry out the same analyses for that species, but prompted aerial alarms sounded similar to natural calls were similar in structure to the few natural recordings, and we assumed that if the models were adequate for scrubwrens they were likely to be adequate for fairy-wrens. Furthermore, we tested the response of fairy-wrens to their own model-prompted alarms.
We analyzed the calls in Canary 1.2.4, using identical settings on the same computer, for both species and from prompted and natural recordings. We used only high-quality recordings that did not overlap with other calls or loud sounds. We first calibrated the files in Canary against a sound file of known amplitude, measured with a Radio Shack 33–2050 sound level meter, following methods in Charif et al. (1995). Separate calibration files were prepared for each set of recording equipment (individual microphone, cable, and recorder), for every recording level used. The spectrogram settings used a temporal grid resolution of 0.1814 ms with a 96.88% overlap, a frequency grid resolution of 43.07 Hz with a fast fourier transform (FFT) size of 1024 points, a Blackman window function, a –10 dB clipping level, and smooth display style. We set the screen display of spectrograms to the scale of 5 ms cm–1 and 0.8 kHz cm–1 and then measured the following for each call: 1) duration (ms), 2) average amplitude (dB re 1 pWm–2), 3) lowest frequency (kHz), 4) highest frequency (kHz), 5) frequency range (kHz), 6) peak frequency (frequency at maximum amplitude, kHz), 7) lower peak frequency (peak frequency of lower parallel sound, kHz), and 8) rate of frequency modulation (Hz; below). Parallel sounds were present only in scrubwrens and were nonharmonic sounds apparently produced by each side of the syrinx; in most cases, the greatest amplitude occurred in the lower parallel sound (details below). The rate of frequency modulation was measured as the number of complete frequency cycles per unit time visible on the spectrogram, excluding partial cycles at the beginning or end of the element.
Playback experiment
We broadcast both species' aerial alarm calls and 3 controls to 15 groups of each species. We identified separate groups by spatial separation and, in the Botanic Gardens, by color-banded individuals of both species. Each group within a species received a unique set of the 5 playback sounds: 1) scrubwren alarm, 2) fairy-wren alarm, 3) background sound recorded at the same time as the scrubwren alarm, 4) similar background sound for the fairy-wren recording, and 5) bell call of crimson rosella (Playcercus elegans), a harmless parrot. Sounds were recorded from "strangers" that lived at least 3 territories away, thereby avoiding any effect of familiarity. The 2 alarm playbacks used on a group were matched for the type of model (sparrowhawk or currawong) that prompted the call. We used SoundEdit 16 and Canary to compose playbacks. We chose the alarm element with the best signal-to-noise ratio, without overlapping sounds, from a given recording and pasted the element at natural intervals to compose 4-element alarm calls, following methods similar to Leavesley and Magrath (2005). In scrubwrens, a 4-element call signals that a predator is nearby, and playback prompts flight to cover (Leavesley and Magrath 2005). There have been no studies of fairy-wren aerial alarms, but we found that both species gave multi-element calls to model prompts and that 4 element calls were within the natural range for each (fairy-wren: mean 4.9, range 2–10; scrubwren mean: 2.9, range 1–5; n = 15 individuals in 15 different groups for each species). Sound below 4 kHz was filtered out, and the call was amplified on computer. The background sound controls were composed, filtered, and amplified in the same way as the alarm calls and tested whether any responses were to alarm calls rather than background sound or any equipment or protocol effects. The calls of rosellas were used to control for effects of amplitude itself, were composed to last for a similar time to the alarm calls, and were the same amplitude when broadcast.
Playback sounds were burnt to CD and broadcast using a Sony CD Walkman D-EJ751 connected via an amplifier to a Response Dome Tweeter speaker (1.5–20 kHz). All equipment was mounted around the observer's waist with the speaker facing forward. To ensure that calls were broadcast at the same amplitude, and at a natural amplitude, we rerecorded test playbacks in the field 5 m from the speaker, approximately the distance from the birds during the original recordings (fairy-wren: mean 5.1, range 3.5–7 m; scrubwren: mean 4.3, range 3–6 m). We amplitude calibrated the rerecordings in Canary, following methods given above, and measured their broadcast amplitude. We then adjusted amplitude in Canary to keep the mean broadcast amplitude of alarm and rosella elements within approximately 1 dB given constant conditions and burnt the final CDs used in playbacks to birds. The absolute broadcast amplitudes were within the natural range for each species but differed between experiments as explained below.
When carrying out a playback in the field, a single observer slowly followed a group of birds at a distance of about 8 m and awaited times when a focal individual was out of cover, the observer was between 5 and 10 m of a focal (closest) bird, there had been no natural alarm calls in the preceding 5 min, the birds were feeding on the ground, and there was no obstruction between the speaker and focal bird. We also avoided playbacks when the other species (scrubwren or fairy-wren) was within 10 m of the focal bird or any other bird species was in clear view within 10 m. All treatments were carried out on each group, in a random order, with at least 5 min between playbacks. If there was a disturbance during a playback, such as by a predator or a natural alarm call, we repeated that treatment after all others had been completed. In practice, only 6 of 240 playbacks needed to be repeated. The behavior of the focal individual was quantified as follows: 0, none; 1, look up; or 2, flee. If the bird fled, we recorded whether it flew straight to cover (1) or landed out of cover (0).
We carried out 2 experiments that differed in amplitude of playbacks and the number of controls. Different amplitudes were used because initially we wanted to ensure that all focal individuals heard both alarm calls, so broadcast at maximal natural amplitude; but we wanted also to ensure that interspecific responses could occur at mean amplitude because maximal amplitude itself might signal an extreme situation. In the first experiment, we broadcast alarm calls such that the mean amplitude of elements was 64.6 dB at 5 m, which is at the upper limit of amplitudes of the alarm elements originally recorded from fairy-wrens at that distance (mean 56.7 dB; range 49.6–65.7; n = 15) and at the 75th percentile for scrubwren elements at that distance (mean 58.7 dB; range 49.5–71.9; n = 15). In the second experiment, we reduced the broadcast amplitude to a mean of 58.4 dB so that it lay between the means for the 2 species. Furthermore, because there had been no responses to the background sound controls and the second experiment was broadcast at lower amplitude, we used only the rosella (sound) control.
Model predators and playback experiments appeared to cause only short-term disturbance. Birds were usually in or near cover when we launched the model predator and resumed normal activity within seconds or minutes of the model's disappearance. After prompting alarm calls, we moved to another group, and so any one individual saw the model on one occasion. Given that sparrowhawks were seen regularly in the study site and currawongs and kookaburras were common, our model presentation probably had a negligible effect on time budgets. Playback experiments usually provoked immediate flight to cover (below), but again birds quickly resumed normal activity. Overall, 36% of birds that fled to alarms came out of cover within 10 s of the alarm playback, and most others probably reappeared within a minute. Furthermore, any one group received only 2 alarm calls in an experiment, and the 2 experiments were carried out over an 8-week period during the nonbreeding season, from May to June.
Statistical analyses
We used t-tests to compare element structure between species and prompts as residuals were normally distributed, if necessary after log transformation. Analyses of playback experiments used a binary response variable because individuals of both species almost invariably either fled to cover or did not respond at all. There was too little variation in response within a playback type to fit mixed models that included the group as a random term. However, group membership had no effect on the response to playback, so we dropped the random term and used generalized linear models (GLM), with logit link functions, to analyze results. Analyses of call structure were conducted in SPSS 13 (Kinnear and Gray 2006; SPSS Inc., Chicago, IL), and the GLM function in the R statistical package was used for playbacks (Crawley 2002; R Development Core Team 2006).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
Aerial alarm call structure
The aerial alarm elements of both species were high pitched and rapidly frequency modulated, but they differed in frequency and only scrubwren alarms had a dual structure, with 2 parallel sounds (Figure 1). Specifically, scrubwren alarms had a lower low and peak frequency, greater frequency range, but similar high frequency, amplitude, duration, and rate of frequency modulation (Table 2). In all but one case, the scrubwren's element peak frequency was within the lower parallel sound; in the one exception, the peak frequency of the higher sound was 7.54 kHz and the lower one 6.68 kHz. Low frequency and peak frequency were correlated (scrubwren, r = 0.89, P < 0.001; fairy-wren, r = 0.52, P = 0.046; n = 15 in each species), but neither was significantly correlated with frequency range (scrubwren: low r = –0.33, P = 0.36; peak, r = –0.23, P = 0.42; fairy-wren: low r = –0.26, P = 0.36; peak r = –0.07, P = 0.81). Overall, calls were broadly similar in design, but peak frequency (or low frequency) and frequency range together defined nonoverlapping, species-specific aerial alarm calls (Figure 2), and scrubwrens alone had a dual structure.
|
|
|
Scrubwren aerial alarms prompted by models were similar to those given to real predators (Table 2). There was no significant difference in any of 6 measures of element structure between elements provoked by models compared with predators, and calls prompted by models covered most of the natural range. We did not detect any significant difference in the alarms provoked by different predator models in any measure, despite the difference between model types in shape, color, and color pattern (t-tests: P > 0.2 for all measures except [log] duration, where P = 0.094; n = 5 currawong, 10 sparrowhawk prompts). However, the methods were not designed to test for such a difference, and such a test would require appropriate replication of model types and larger sample sizes.
Playback experiments
Each species fled after playback of both its own species' alarm call and that of the other species. In the first experiment, at maximum natural amplitude for fairy-wrens, all but one individual fled to cover after playback of either species' alarm call but none fled to cover after playback of any control (Figure 3a; GLM: 
42= 192.2, P < 0.001). The one exception was a fairy-wren that flew 2 m away from the speaker and toward cover immediately after a scrubwren alarm playback but landed 1 m short of cover. No individual looked up but did not flee. The consistent response means that individuals did not respond differently to conspecific compared with heterospecific calls (GLM interaction: 42= 0).
|
The second experiment, with playback at reduced amplitude, produced a similar result to the first, except that a few individuals did not flee to cover, and there was a small effect of species responding more to conspecific compared with heterospecific alarm calls (Figure 3b). Birds fled to alarm calls but never to the control (GLM, playback type:
22 = 76.0, P < 0.001). Overall, 23/30 individuals fled to cover after heterospecific alarms, whereas 29/30 did so to conspecific alarms (GLM, interaction of playback type and species: 22 = 6.8, P = 0.034). This interaction was confirmed in a simple tabulation of group responses: the focal individuals of 4 groups of fairy-wrens and 3 groups of scrubwrens fled to conspecific but not heterospecific alarms, but none did the opposite (binomial 2-tailed probability = 2 x 0.57 = 0.016). Of those fairy-wrens that did not flee to cover after scrubwren alarms, 2 showed no response, 1 looked up, and 1 flew but not into cover. Of the scrubwrens that did not flee to cover after fairy-wren alarms, 2 showed no response and 1 flew but not to cover. One scrubwren looked rather than fled to playback of a conspecific alarm. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
White-browed scrubwrens and superb fairy-wrens fled to playback of each other's aerial alarm calls, just as they did to conspecific alarms. To our knowledge, this is the first adequately replicated experiment in birds to show responses to heterospecific aerial alarm calls and the first experiment to examine mutual responses to aerial alarm calls despite a long history of observation and discussion (Table 1; Marler 1957
). The results therefore demonstrate mutual interspecific responses, but further work is needed to assess if these species are merely eavesdropping on each other or whether one or both could be signaling to the other (Kostan 2002). Both species gave high-frequency aerial alarm calls of similar duration, and frequency modulated at a similar rate. Nonetheless, the calls were distinct because they differed in frequency measures, and only scrubwren calls had a dual acoustic structure.
The striking feature of the playback results is the strength and uniformity of responses across individuals and species, which we suggest is adaptive given that aerial (flee) alarm calls signal immediate danger (Marler 1955). In the first experiment, every individual fled to each alarm call (from 60 playbacks to 30 groups), and all but one flew directly to cover, with the one exception flying closer to cover. Not one individual merely became vigilant. In the second experiment, we reduced the broadcast amplitude to the mean natural level for each species rather than to the highest natural level for fairy-wrens. Despite the reduction in amplitude, every fairy-wren fled to cover after fairy-wren alarms, as did 12/15 scrubwrens, and again most birds of both species fled to cover after broadcast of scrubwren alarms. Only one individual of each species became vigilant but did not flee after an alarm playback. The immediate flight to these alarms presumably reflects signaling of imminent danger; it makes little sense merely to become vigilant if an aerial predator is near. At least in scrubwrens, a 4-element alarm signals that a predator is flying close and so is a strong signal (Leavesley and Magrath 2005). The responses of other species to playback of aerial alarms can include fleeing, freezing, and increased vigilance or a mixture (Table 1), presumably depending on exposure, type of playbacks, and species and habitat differences in evasive strategy.
There was a slightly greater response to conspecific than heterospecific alarms in the second experiment, in which broadcast amplitude was reduced. In each species, there were 2 individuals that showed no response to playback of the heterospecific call, although 3 other individuals did respond but did not flee to cover, suggesting that both detection and discrimination could be better for conspecific calls, which may be a general avian feature (Dooling 2004). Reduced responses to heterospecific calls may be particularly important in more difficult perceptual tasks, such as at lower amplitude or with greater background sound. Nonetheless, given the small effect size, a larger sample is necessary to confirm whether there is in general a greater response to conspecific alarms. In addition to difficulties with detection or discrimination, it is possible that lower amplitude heterospecific calls were interpreted as implying less threat than comparable conspecific calls.
Models of aerial predators appeared to prompt normal aerial alarm calls in scrubwrens. Alarms prompted by gliding models did not differ significantly in any measure from those to real predators. Given the relative infrequency and temporal unpredictability of naturally provoked aerial alarm calls, models are essential in sampling aerial alarms from multiple species and permitting replicated interspecific playback experiments. Previous studies of responses to aerial threat in birds and mammals have used a variety of "models" including video images, hats, sticks, frisbees, model planes, flapping model birds, and models or skins gliding along a line. The variety of prompts suggest that model details could be unimportant, but there are few quantitative comparisons of prompted and natural alarm calls to confirm that this is true. We believe that such comparisons are important, given avian alarm calls can differ in subtly different contexts (e.g., Klump and Curio 1983; Bayly and Evans 2003; Leavesley and Magrath 2005; Templeton et al. 2005), and birds can respond differently to subtly different alarm calls (e.g., Rainey et al. 2004).
Scrubwrens and fairy-wrens had aerial alarm calls that were similar to each other but differed in structure from aerial alarm calls described for most other species. Their alarms are high pitched but in other respects differ from the classic seeet aerial alarm call described for several European and North American passerines (Marler 1955; Ficken and Witkin 1977). Seeet calls are narrow-band tones of constant or slowly changing frequency, whereas both scrubwrens and fairy-wrens have rapidly frequency-modulated calls, and scrubwrens have relatively broadband calls because of parallel sounds and often a descending initial frequency sweep. The probable aerial alarms of several Sri Lankan species are variable in structure, including broad bandwidth and major frequency sweeps (Goodale and Kotagama 2005), and the aerial alarms of several Australian birds are of relatively low pitch and can include frequency sweeps (Jurisevic and Sanderson 1994). Aerial alarm calls are clearly variable among species, and many differ from the "classic" seeet alarm of several Northern Hemisphere species (Marler 1957).
In addition to structural differences from seeet calls, scrubwrens (and possibly fairy-wrens) encode greater danger with a greater number of elements in the alarm call (Leavesley and Magrath 2005), whereas seeet calls are usually given with little repetition of notes (Marler 2004). An apparent paradox of the scrubwren's encoding scheme is that it may provide predators with greater location cues in riskier situations. However, the high frequency and rapid delivery of scrubwren and fairy-wren alarms probably minimizes risk of eavesdropping, and the opposite encoding scheme would seem open to cheating and would mean that alarms signaling greater danger could be less easy for intended receivers to detect (Leavesley and Magrath 2005).
We do not know if the similarities in call structure of scrubwrens and fairy-wrens facilitated mutual responses to each other's alarm calls, and in general, there is nothing known about the relative importance of structural similarity and learning in affecting responses to aerial alarms, despite the remarkable similarities among the aerial alarm calls of some species (Marler 1957, 2004). Work on mobbing alarm calls suggests that both similarity and learning are likely to be important. For example, nestlings of 3 species of passerine responded to conspecific but not heterospecific mobbing alarm calls, even if raised by the other species, suggesting that birds can have neural templates for call recognition (Davies et al. 2004, 2006; Madden et al. 2005), although other species are known to respond to mobbing alarms of quite different structure to their own, which suggests learning (e.g., Rainey et al. 2004).
|
Funding |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
Australian Research Council (Discovery Project grant DP0665481) to R.D.M.
|
ACKNOWLEDGEMENTS |
|---|
We thank Andy Bennett, Jim Forge, Andy Gosler, Wayne Kotzur, Peter Marsack, Alan Muir, Bob Phillips, Dean Soccol, and Tess Ward for advice or help developing the predator models; Jim Forge and Bob Phillips for electronics; Anastasia Dalziell and Adam Leavesley for advice and help in the field; Andy Horn for advice of acoustics; Leo Joseph and the Commonwealth Scientific and Industrial Research Organization (CSIRO) Australian National Wildlife Collection for loan of skins; Suhel Quader and Kavita Isvaran for statistical advice; Ana Dalziell, Nick Davies, and Mike Double for comments on the manuscript. We obtained permits from the Australian National Botanic Gardens, Environment ACT, and the Ethics Committee of the Australian National University. R.D.M. thanks Nick Davies and Malcolm Burrows for welcoming him to the Zoology Department, University of Cambridge, where the paper was written.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFundingREFERENCES |
|---|
Bayly KL, Evans CS. Dynamic changes in alarm call structure: a strategy for reducing conspicuousness to avian predators? Behaviour (2003) 140:353–369.[CrossRef]
Bell HL. Composition and seasonality of mixed-species feeding flocks of insectivorous birds in the Australian Capital Territory. Emu (1980) 80:227–232.[Web of Science]
Bradbury JW, Vehrencamp SL. Principles of animal communication. (1998) Sunderland (MA): Sinauer Associates.
Burger J. Grebes nesting in gull colonies: protective associations and early warning. Am Nat (1984) 123:327–337.[CrossRef][Web of Science]
Caro T. Antipredator defences in birds and mammals. (2005) Chicago (IL): Chicago University Press.
Charif RA, Mitchell S, Clark CW. Canary 1.2 user's manual. (1995) Ithaca (NY): Cornell Laboratory of Ornithology.
Crawley MJ. Statistical computing: an introduction to data analysis using S-Plus. (2002) Chichester (UK): Wiley.
Davies NB, Madden JR, Butchart SHM. Learning fine-tunes a specific response of nestlings to the parental alarm calls of their own species. Proc R Soc Lond B (2004) 271:2297–2304.[Medline]
Davies NB, Madden JR, Butchart SHM, Rutila J. A host-race of the cuckoo Cuculus canorus with nestlings attuned to the parental alarm calls of the host species. Proc R Soc Lond B (2006) 273:693–699.[Medline]
Dooling R. Audition: can birds hear everything they sing? In: Nature's music: the science of birdsong—Marler P, Slabbekoorn H, eds. (2004) San Diego (CA): Elsevier. 206–225.
Evans CS, Evans L, Marler P. On the meaning of alarm calls: functional reference in an avian vocal system. Anim Behav (1993) 46:23–38.[CrossRef][Web of Science]
Fichtel C. Reciprocal recognition of sifaka (Propithecus verreauxi verreauxi) and redfronted lemur (Eulemur fulvus rufus) alarm calls. Anim Cogn (2004) 7:45–52.[CrossRef][Web of Science][Medline]
Ficken MS, Witkin SR. Responses of black-capped chickadee flocks to predators. Auk (1977) 94:156–157.[Web of Science]
Forsman JT, Mönkkönen M. Responses by breeding birds to heterospecific song and mobbing call playbacks under varying predation risk. Anim Behav (2001) 62:1067–1073.[CrossRef][Web of Science]
Gaddis P. Mixed flocks, accipiters, and antipredator behavior. Condor (1980) 82:348–349.
Goodale E, Kotagama SW. Alarm calling in Sri Lankan mixed-species bird flocks. Auk (2005) 122:108–120.[CrossRef][Web of Science]
Griesser M, Ekman J. Nepotistic alarm calling in the Siberian jay, Perisoreus infaustus. Anim Behav (2004) 67:933–939.[CrossRef][Web of Science]
Haftorn S. Contexts and possible functions of alarm calling in the willow tit, Parus montanus; the principle of ‘better safe than sorry’. Behaviour (2000) 137:437–449.[CrossRef]
Higgins PJ, Peter JM. Handbook of Australian, New Zealand and Antarctic birds. In: pardalotes to shrike-thrushes (2002) Volume 6. Melbourne (Australia): Oxford University Press.
Higgins PJ, Peter JM, Cowling SJ. Handbook of Australian, New Zealand and Antarctic birds. In: boatbills to starlings (2006) Volume 7. Melbourne (Australia): Oxford University Press.
Higgins PJ, Peter JM, Steele WK. Handbook of Australian, New Zealand and Antarctic Birds. In: tyrant-flycatchers to chats (2001) Volume 5. Melbourne (Australia): Oxford University Press.
Hindwood KA. The flocking of birds with particular reference to the associations of small insectivorous birds. Emu (1937) 36:254–261.
Jurisevic MA, Sanderson KJ. Alarm vocalisations in Australian birds: convergent characteristics and phylogenetic differences. Emu (1994) 94:69–77.
Kinnear PR, Gray C. SPSS 14 made simple. (2006) New York: Psychology Press.
Klump GM. Sound localization in birds. In: Comparative hearing: birds and reptiles—Dooling RJ, Fay RR, Popper AN, eds. (2000) New York: Springer. 249–307.
Klump GM, Curio E. Reactions of blue tits Parus caeruleus to hawk models of different sizes. Bird Behav (1983) 4:78–81.
Klump GM, Shalter MD. Acoustic behaviour of birds and mammals in the predator context I. Factors affecting the structure of alarm calls. II. The functional significance and evolution of alarm signals. In: Z Tierpsychol (1984) 66:189–226.[Web of Science]
Kostan KM. The evolution of mutualistic interspecific communication: assessment and management across species. J Comp Psychol (2002) 116:206–209.[CrossRef][Web of Science][Medline]
Kroodsma DE. Suggested experimental designs for song playbacks. Anim Behav (1989) 37:600–609.[CrossRef][Web of Science]
Kroodsma DE, Bayers BE, Goodale E, Johnson S, Liu W. Pseudoreplication in playback experiments, revisited a decade later. Anim Behav (2001) 61:1029–1033.[CrossRef][Web of Science]
Leavesley A, Magrath RD. Communicating about danger: urgency alarm calling in a bird. Anim Behav (2005) 70:365–373.[CrossRef][Web of Science]
Madden JR, Kilner RM, Davies NB. Nestling responses to adult food and alarm calls: 1. Species specific responses in two cowbird hosts. Anim Behav (2005) 70:619–627.[CrossRef][Web of Science]
Magrath RD. Group breeding dramatically increases reproductive success of yearling but not older female scrubwrens: a model for cooperatively breeding birds? J Anim Ecol (2001) 70:370–385.[CrossRef]
Magrath RD, Leedman AW, Gardner JL, Giannasca A, Nathan AC, Yezerinac SM, Nicholls JA. Life in the slow lane: reproductive life history of the white-browed scrubwren, an Australian endemic. Auk (2000) 117:479–489.[CrossRef][Web of Science]
Magrath RD, Platzen D, Kondo J. From nestling calls to fledgling silence: adaptive timing of change in response to aerial alarm calls. Proc R Soc Lond B (2006) 273:2335–2341.[Medline]
Marchant S, Higgins PJ. Handbook of Australian, New Zealand and Antarctic Birds. In: raptors to lapwings (1993) Volume 2. Melbourne (Australia): Oxford University Press.
Marler P. Characteristics of some animal calls. Nature (1955) 176:6–8.[CrossRef]
Marler P. Specific distinctiveness in the communication calls of birds. Behaviour (1957) 11:13–39.[Medline]
Marler P. Bird calls: a cornucopia for communication. In: Nature's music: the science of birdsong—Marler P, Slabbekoorn H, eds. (2004) San Diego (CA): Elsevier. 132–177.
Matsuoka S. Pseudo warning call in titmice. Tori (1980) 29:87–90.
Møller AP. False alarm calls as a means of resource usurpation in the great tit Parus major. Ethology (1988) 79:25–30.[Web of Science]
Munn CA. Birds that ‘cry wolf’. Nature (1986) 319:143–145.[CrossRef]
Nuechterlein GL. ‘Information parasitism’ in mixed colonies of western grebes and Forster's terns. Anim Behav (1981) 29:985–989.[CrossRef][Web of Science]
R Development Core Team. R: a language and environment for statistical computing. (2006) Vienna (Austria): R Foundation for Statistical Computing.
Rainey HJ, Zuberbühler K, Slater PJB. Hornbills can distinguish between primate alarm calls. Proc R Soc Lond B (2004) 271:755–759.[Medline]
Rasa OAE. Dwarf mongoose and hornbill mutualism in the Taru desert, Kenya. Behav Ecol Sociobiol (1983) 12:181–190.[CrossRef][Web of Science]
Rowley I. The life history of the superb blue wren Malurus cyaneus. Emu (1965) 64:251–297.
Rydén O. Differential responsiveness of great tit nestlings, Parus major, to natural auditory stimuli. Z Tierpsychol (1978) 47:236–253.[Web of Science]
Shriner WM. Yellow-bellied marmot and golden-mantled ground squirrel responses to heterospecific alarm calls. Anim Behav (1998) 55:529–536.[CrossRef][Web of Science][Medline]
Sullivan K. Information exploitation by downy woodpeckers in mixed-species flocks. Behaviour (1984) 91:294–311.
Sullivan K. Selective alarm calling by downy woodpeckers in mixed-species flocks. Auk (1985) 102:184–187.[Web of Science]
Taylor and Canberra Ornithologists Group 1992. Birds of the Australian Capital Territory: an atlas. Canberra (Australia): Canberra Ornithologists Group Inc.
Templeton CN, Greene E, Davis K. Allometry of alarm calls: black-capped chickadees encode information about predator size. Science (2005) 308:1934–1937.
Zuberbühler K. Interspecies semantic communication in two forest primates. Proc R Soc Lond B (2000) 267:713–718.[Medline]
Zuberbühler K. Predator-specific alarm calls in Campbell's monkeys, Cercopithecus campbelli. Behav Ecol Sociobiol (2001) 50:414–422.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Hingee and R. D. Magrath Flights of fear: a mechanical wing whistle sounds the alarm in a flocking bird Proc R Soc B, December 7, 2009; 276(1676): 4173 - 4179. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. B. V. Bell, A. N. Radford, R. Rose, H. M. Wade, and A. R. Ridley The value of constant surveillance in a risky environment Proc R Soc B, August 22, 2009; 276(1669): 2997 - 3005. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. D. Magrath, B. J. Pitcher, and J. L. Gardner An avian eavesdropping network: alarm signal reliability and heterospecific response Behav. Ecol., July 1, 2009; 20(4): 745 - 752. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. D Magrath, B. J Pitcher, and J. L Gardner Recognition of other species' aerial alarm calls: speaking the same language or learning another? Proc R Soc B, February 22, 2009; 276(1657): 769 - 774. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. J. Lea, J. P. Barrera, L. M. Tom, and D. T. Blumstein Heterospecific eavesdropping in a nonsocial species Behav. Ecol., September 1, 2008; 19(5): 1041 - 1046. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Goodale and S. W. Kotagama Response to conspecific and heterospecific alarm calls in mixed-species bird flocks of a Sri Lankan rainforest Behav. Ecol., July 1, 2008; 19(4): 887 - 894. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||