You are on page 1of 10

Behavioral Ecology

doi:10.1093/beheco/arq221
Advance Access publication 8 February 2011

Original Article

Sound familiar? Acoustic similarity provokes


responses to unfamiliar heterospecific alarm
calls
Pamela M. Fallow, Janet L. Gardner, and Robert D. Magrath
Division of Evolution, Ecology and Genetics, Research School of Biology, The Australian National
University, Canberra, ACT 0200, Australia

Eavesdropping on the alarm calls of other species can provide vertebrates with valuable information about danger, and

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


often responses to heterospecific calls seem to be acquired through learning. However, animals might not require
learning to respond to heterospecific alarm calls that are acoustically similar to conspecific calls. Although previous work
suggests that learning is necessary for superb fairy-wrens to develop responses to heterospecific alarms, acoustic similarity
might also be important. We tested whether fairy-wrens responded to playback of unfamiliar alarm calls and if the
strength of responses was affected by acoustic similarity to conspecific calls. We then determined which acoustic proper-
ties were likely to be used in alarm call identification. Birds fled to cover after playback of calls that were acoustically
similar to their own but did not usually respond to less similar calls. Fairy-wren aerial alarm calls are high-frequency and
rapidly frequency modulated; their probability of fleeing to playback of unfamiliar calls increased in response to calls with
increasing peak frequencies, and they spent more time in cover following calls with a number of frequency cycles similar
to their own. Fairy-wrens also responded strongly to relatively dissimilar calls of 1 allopatric species, possibly because these
unfamiliar calls resembled those of a sympatric species to which they had learnt to respond. Our study shows that acoustic
similarity can prompt responses to heterospecific alarm calls regardless of experience and together with previous work
suggests that both acoustic similarity and learning are important for interspecific responses to alarm calls. Key words:
Acanthizidae, acoustic structure, alarm calls, interspecific eavesdropping, Maluridae, Malurus cyaneus. [Behav Ecol 22:
401–410 (2011)]

any vertebrates use alarm calls to alert others to the pres- with early warnings (Burger 1984), and information that com-
M ence of danger (Zuberbühler 2009), and heterospecifics
that share the same predators can benefit from responding to
plements that provided by conspecific calls (Nuechterlein
1981; Goodale and Kotagama 2005). The information gained
these calls. Alarm calls can convey knowledge about the type through interspecific eavesdropping can be so valuable that it
of threat (Seyfarth et al. 1980; Pereira and Macedonia 1991; influences the temporary or long-term formation of mixed-
Gill and Sealy 2004), the direction from which it is approach- species groups (Goodale et al. 2010).
ing (Evans et al. 1993), how urgent it is (Leavesley and Ma- The advantages of interspecific eavesdropping and its wide-
grath 2005; Templeton et al. 2005), or a combination of these spread occurrence raise the question of how animals recognize
types of information (Manser et al. 2002), which help re- heterospecific calls as alarm calls. So far, most studies suggest
ceivers choose appropriate responses. In addition to warning that responses to heterospecific alarm calls are acquired
conspecifics, other species that are vulnerable to the same through learning. For example, adult bonnet macaques,
threats can eavesdrop on these calls and gain information Macaca radiata, are better than juveniles at distinguishing het-
about danger (Sullivan 1984; Fichtel 2004; Randler 2006; Fal- erospecific mammalian alarm calls from nonalarms, and anti-
low and Magrath 2010). Responses to heterospecific alarm predator responses to playback of heterospecific calls occur
calls have been shown between species within and across ver- only in macaque populations that have frequent exposure to
tebrate taxa (Caro 2005): between birds (Hurd 1996; Magrath those heterospecific calls and therefore the opportunity for
et al. 2007; Templeton and Greene 2007), between mammals learning (Ramakrishnan and Coss 2000). Similarly, playback
(Shriner 1998; Ramakrishnan and Coss 2000; Zuberbühler experiments showed that infant vervet monkeys, Cercopithecus
2000), a bird responding to a mammal (Rainey et al. 2004), aethiops pygerythrus, exposed more often to alarm calls of su-
mammals responding to birds (Randler 2006; Müller and perb starlings, Spreo superbus, responded to them at a younger
Manser 2008), and reptiles responding to birds (Vitousek age potentially by learning from the behavior of other group
et al. 2007; Ito and Mori 2010). Heterospecific alarm calls members (Hauser 1988). Some species respond to heterospe-
not only allow eavesdroppers to devote less time to scanning cific alarm calls that are acoustically dissimilar to conspecific
for danger (Bshary and Noë 1997) but can also provide them calls (Shriner 1998; Magrath et al. 2009a, 2009b), or even
when they do not have calls of their own (Griffin et al.
Address correspondence to P.M. Fallow. E-mail: pamela.fallow@anu.
2005; Vitousek et al. 2007; Ito and Mori 2010), suggesting that
edu.au.
J.L. Gardner is now at Fenner School of Environment and Society, The acoustic similarity to conspecific calls is not a requirement of
Australian National University, Canberra, ACT 0200, Australia responding to heterospecific calls. Finally, the most direct
Received 18 August 2010; revised 30 November 2010; accepted 16 demonstration to date that animals can learn to respond to
December 2010. alarm sounds is shown in golden-mantled squirrels, Spermophi-
lus lateralis, which can learn to associate a novel sound with the
 The Author 2011. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com
402 Behavioral Ecology

appearance of a model predator (Shriner 1999). Thus, various This suggests that fairy-wrens are capable of high levels of
studies using different experimental methods provide both acoustic discrimination and that the acoustic similarity be-
direct and indirect evidence of the importance of learning tween these calls and conspecific calls does not prompt re-
for the development of adaptive responses to heterospecific sponses to scrubwren calls without experience. Playback
alarm calls. experiments also show that fairy-wrens flee in response to
Although learning appears to be important for interspecific acoustically very dissimilar alarm calls of sympatric New Hol-
responses to alarm calls, vertebrates might not require learning land honeyeaters, Phylidonyris novaehollandiae, (Magrath et al.
to respond to heterospecific calls if they can identify them 2009a), showing that call similarity is not essential for re-
based solely on their acoustic structure. Unlike other signals, sponse and again implying learning. Furthermore, fairy-wrens
such as those used in courtship, that are directed solely at usually did not respond to honeyeater calls played back in
conspecifics, alarm calls do not need to be species-specific reverse despite the fact that reversed calls had almost all of
and so the calls of sympatric species can share similarities with- the same acoustic properties as ‘‘normal’’ honeyeater calls
out reducing effectiveness (Marler 1957). Therefore, alarm (Magrath et al. 2009b). This implies that fairy-wrens learn to
calls of different species can be acoustically similar as a result respond to the very specific acoustic structure of normal hon-
of calls retaining features from a common ancestor (de Kort eyeater calls. However, although it is likely that scrubwren calls
and ten Cate 2001) or convergence in acoustic structure (dis- are the most acoustically similar heterospecific calls to those
cussed below), thus provoking interspecific responses. Birds of fairy-wrens in Canberra (Magrath et al. 2009b), fairy-wrens
can have an innate disposition to respond to conspecific alarm may not require experience to respond to heterospecific calls
calls (Davies et al. 2004) and might respond to unfamiliar that are even more similar to their own calls than those of
alarm calls because they are acoustically similar to their own scrubwrens, such as those of closely related species.

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


(McCracken and Sheldon 1997; de Kort and ten Cate 2001; We investigated whether experience was necessary for fairy-
Ghirlanda and Enquist 2003). For example, Australian apostle- wrens to respond to unfamiliar heterospecific alarm calls and
birds, Struthidea cinerea, responded to playback of a North specifically if a greater degree of acoustic similarity to conspe-
American Carolina wren, Thryothorus ludovicianus, mobbing call cific calls increased the strength of responses. We then ana-
despite having no opportunity to learn it (Johnson et al. 2003), lyzed whether particular acoustic properties of these
implying the importance of acoustic similarity for prompting unfamiliar calls were likely to have affected fairy-wren
responses to unfamiliar alarm calls (Johnson et al. 2003). Thus, responses.
animals might not need to experience alarm calls that are
acoustically similar to their own, and recognition of heterospe-
cific calls may be enabled by acoustic similarities between the METHODS
calls of different species. However, there have been no playback
Study area and species
experiments using a range of alarm calls varying in their degree
of acoustic similarity to calls of the receiving species. As a result, We studied superb fairy-wrens around Lake Burley Griffin and
it is unknown whether vertebrates respond similarly to all het- in the Australian National Botanic Gardens (lat 3516#S, long
erospecific calls that they identify as alarm calls or if the 1496#E) in Canberra, Australia. Fairy-wren groups living in
strength of responses is affected by acoustic similarity to con- the Botanic Gardens were identified by the inclusion of color-
specific calls. banded birds, whereas groups by the lake were distinguished
The design of alarm calls reflects their function as a signal by location. Territories within the Botanic Gardens have
and their efficacy in a particular context, and so the calls of a mean width of around 80 m (Mulder 1995) and, on a single
different species can be acoustically similar as a result of con- day, groups sampled within and outside the Gardens were
vergence on the most effective design for a particular environ- found at least 100 m apart. Fairy-wrens breed cooperatively,
ment (Marler 1955; Russ et al. 2004). Alarm calls have 2 with groups typically consisting of a dominant pair that may
common designs depending on their function: ‘‘mobbing’’ be assisted by up to several subordinate males (Cockburn et al.
alarm calls attract others to join the caller in harassing a pred- 2003). They forage for invertebrates primarily on the ground
ator that poses little immediate danger and so are typically and frequently venture out of cover and into open areas. Each
relatively low-frequency broadband calls that are repeated fairy-wren group defends its territory during the breeding
over a period of time, allowing calls to be locatable to both season until the cooler months when territories break down,
the predator and other prey, and attracting more receivers to and neighboring groups gather to form larger flocks that
the source of the call (Hurd 1996; Bradbury and Vehrencamp commonly include other species (Bell 1980, Fallow PM, per-
1998). Conversely, ‘‘flee’’ alarm calls, such as aerial alarm calls sonal observation).
produced in response to predators in flight, signal the need to Fairy-wrens signal the presence of predatory birds in flight by
immediately freeze or escape to cover, and are typically producing aerial alarm calls consisting of one or more high-fre-
high-frequency and short in duration, making the call difficult quency elements. Each element consists of a single band that is
to overhear and perhaps difficult to locate (Bradbury and rapidly frequency modulated about a constant carrier fre-
Vehrencamp 1998; Jurisevic and Sanderson 1998). Interspe- quency (mean peak frequency: 9.1 kHz, frequency range: 8–
cific responses to alarm calls might therefore occur as a result 11 kHz, Magrath et al. 2007). Calls contain more elements
of the calls of different species converging on broadly similar and have a higher maximum frequency when threats are
acoustic structures. closer, and fairy-wrens typically respond in a 2-stage process
Superb fairy-wrens, Malurus cyaneus, seem to require expe- by 1) immediately fleeing to cover; and 2) waiting in cover
rience to recognize at least some heterospecific alarm calls. before reemerging, with the amount of time spent in cover
When living sympatrically with white-browed scrubwrens, Ser- increasing as the number of elements in the call increases
icornis frontalis, they usually respond to playback of scrubwren (Fallow and Magrath 2010). In Canberra, fairy-wrens are vul-
aerial alarm calls by immediately fleeing to cover (Magrath nerable to predatory birds including pied currawongs, Strepera
et al. 2007). However, a population of fairy-wrens living in graculina; grey butcherbirds, Cracticus torquatus; laughing koo-
the absence of scrubwrens did not respond to playback of kaburras, Dacelo novaeguinea; and collared sparrowhawks,
scrubwren calls, despite the calls of these 2 species Accipiter cirrocephalus (Magrath et al. 2009b).
overlapping in frequency, being of similar duration and both To test the response of fairy-wrens to unfamiliar calls with
featuring rapid frequency modulation (Magrath et al. 2009b). varying degrees of acoustic similarity to conspecific calls, we
Fallow et al. • Acoustic similarity provokes responses to alarm calls 403

presented them with alarm calls from 7 allopatric species in brown gerygone, Batemans Bay, New South Wales; chestnut-
the same families as superb fairy-wrens and white-browed rumped thornbill and southern whiteface, Nallan Station,
scrubwrens. The allopatric species included 3 congeneric spe- Western Australia; and western thornbill, Flint State Forest,
cies in the family Maluridae: splendid fairy-wren, Malurus Western Australia. All calls from allopatric species were pro-
splendens; variegated fairy-wren, M. lamberti; and white-winged voked using a gliding life-sized model of the collared sparrow-
fairy-wren, M. leucopterus; and 4 species in the family Acanthi- hawk because, unlike currawongs, they occur throughout
zidae: brown gerygone, Gerygone mouki; chestnut-rumped Australia. Alarm calls prompted by the currawong and spar-
thornbill, Acanthiza uropygialis; western thornbill, A. inornata; rowhawk models are acoustically indistinguishable from each
and southern whiteface, Aphelocephala leucopsis. other and from calls prompted by real predators (Magrath
The superb fairy-wren is a sedentary species (Higgins et al. et al. 2007), and birds respond to playback of calls prompted
2001), and individuals in our study population were therefore by models in the same way as playback of calls prompted by
highly unlikely to have experienced calls of the allopatric spe- live predators (Leavesley and Magrath 2005; Magrath et al.
cies used in the playback experiment. However, in other areas 2007). The calls of all species were recorded on dry days with
of Australia, the geographic range of superb fairy-wrens over- low wind. While one person launched the model by hand,
laps with that of most of the playback species. Superb fairy- another recorded calls from a distance of 2–6 m from the
wrens are widespread in south-eastern Australia, so that some group or individual using a Sennheiser ME66 directional mi-
areas of its range overlap with splendid fairy-wrens and varie- crophone and a Marantz PMD670 digital recorder, sampling
gated fairy-wrens (South Australia, inland Victoria, New South wave files at 44.1 kHz and 16 bits. Prior to each model pre-
Wales, and southern Queensland), white-winged fairy-wrens sentation, we observed the behavior of the group for 5 min
(South Australia, northern Victoria, New South Wales, and and did not attempt to launch models if alarm calls were given

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


inland Queensland), brown gerygones (eastern coast of Vic- during observation or if any potentially threatening species
toria, New South Wales, and coastal Queensland), and chest- were seen nearby. This ensured that we recorded alarm calls
nut-rumped thornbills and southern whitefaces (South provoked by the model predator and not by other stimuli.
Australia, Victoria, New South Wales, and inland Queensland,
Barrett et al. 2003). The only playback species with
Acoustic analyses
a geographic range that has no overlap with that of superb
fairy-wrens is the western thornbill, which is restricted to the We cropped digital recordings in Cool Edit Pro v. 1.2a (Syntril-
southwestern corner of Western Australia. Superb fairy-wrens lium Software, Phoenix, AZ) to isolate the aerial alarm calls,
are the only malurid species commonly found in our study site leaving several seconds of background noise prior to the mod-
in Canberra (Fennell 2009), where the only other playback el’s release and following its landing. We used Raven Pro 1.3
species recorded, brown gerygones, are rare visitors and have (Charif et al. 2008) to generate spectrograms using a Black-
not been sighted for several years (Overs 2009). man window function, a temporal grid resolution of 0.295 ms
with 94.9% overlap and a frequency grid resolution of 86.1
Hz. As alarm calls were recorded from different bird groups,
Recording alarm calls
each call was likely to be produced by a different individual.
We recorded aerial alarm calls from superb fairy-wrens in the We randomly selected one element from calls featuring a high
Botanic Gardens over a period of 7 weeks during May and June signal-to-noise ratio, minimal reverberation, and no overlap-
2008, with calls provoked using a gliding, life-sized model of ping background sounds, leaving us with a sample of unique
the pied currawong (see Magrath et al. 2007 for model spec- high-quality elements for each species (Table 1; Figure 1). We
ification). Calls from the allopatric species were recorded in then defined a range selection around each element on the
various locations around Australia during 2006 and 2007: spectrogram and measured the following acoustic properties:
splendid fairy-wren, Perth, Western Australia; variegated and maximum frequency (kilohertz), minimum frequency
white-winged fairy-wrens, Guilderton, Western Australia; (kilohertz), peak frequency (the frequency at greatest

Table 1
Comparison of call elements of birds in this study

Frequency
Peak frequency Frequency range Element duration Number of cycles modulation rate
Species N (kHz) mean 6 SD (kHz) mean 6 SD (ms) mean 6 SD mean 6 SD (Hz) mean 6 SD

Family Maluridae
Superb fairy-wren 12 9.1 6 0.34 2.2 6 0.35 90.4 6 18.8 7.7 6 1.5 105.2 6 8.0
Splendid fairy-wren 12 8.9 6 0.25 1.8 6 0.29 80.7 6 16.1 7.5 6 1.3 107.7 6 10.0
Variegated fairy-wren 12 8.9 6 0.40 2.1 6 0.31 79.9 6 10.4 8.1 6 1.1 114.3 6 9.2
White-winged fairy-wren 12 9.2 6 0.31 1.8 6 0.55 73.9 6 27.4 — —
Family Acanthizidae
Brown gerygone 12 6.5 6 0.56 2.3 6 0.29 50.5 6 10.6 3.9 6 1.2 129.7 6 10.3
Western thornbill 12 8.3 6 0.50 2.7 6 0.57 64.3 6 15.0 — —
Southern whiteface 12 6.7 6 0.22 1.6 6 0.14 40.6 6 13.2 — —
Chestnut-rumped thornbill 12 6.8 6 0.35 1.5 6 0.48 91.4 6 50.1 — —
White-browed scrubwrena 7 7.8 6 1.00 3.0 6 0.56 83.0 6 16.8 6.8 6 1.9 105.5 6 7.9
Buff-rumped thornbilla 13 8.3 6 0.56 2.4 6 0.45 80.6 6 23.6 — —
Family Psittacidae
Crimson rosella (control)b 12 2.6 6 0.52 1.2 6 0.61 132.0 6 100.3 — —

a
Species living in sympatry with superb fairy-wrens in Canberra, examined in the ‘‘DISCUSSION.’’ b Measurements were taken from the loudest
element in the call.
404 Behavioral Ecology

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


Figure 1
Spectrograms of aerial alarm call elements of superb fairy-wrens, the allopatric species used in the playback experiment, and the sympatric
species examined in the ‘‘DISCUSSION.’’ Elements presented have measurements within 1 SD of the mean values for each species’ maximum
frequency (kilohertz), minimum frequency (kilohertz), element duration (milliseconds) and where applicable, frequency modulation rate
(hertz). Spectrograms were produced in Raven Pro 1.3 using settings as described under ‘‘METHODS’’ for Acoustic analyses.

amplitude, kilohertz), frequency range (kilohertz), element bills, and southern whitefaces. Superb fairy-wren alarm calls
duration (milliseconds), number of frequency cycles, and fre- were used as a positive control, and the piping or ‘‘bell’’ calls
quency modulation rate (number of frequency cycles per sec- (Higgins 1999; Magrath et al. 2007) of crimson rosellas,
ond, hertz). Measurements were taken from the entire Platycercus elegans, harmless parrots that live sympatrically with
element at the same scale on the screen and with standard fairy-wrens in Canberra, were used as a negative control
spectral settings. (Table 1).
We prepared 12 playback calls for each species, composing
each call from a unique element to avoid pseudoreplication of
Playback experiment
calls. Elements were selected from different recordings or from
The playback experiment aimed to test how superb fairy-wrens calls of 2 different birds in the same recording when there
responded to unfamiliar aerial alarm calls that varied in their were fewer than 12 recordings for a species. We filtered out all
degree of acoustic similarity to conspecific calls. If the response sound below 4 kHz in the passerine alarm calls and all sound
of fairy-wrens to playback of the heterospecific calls varied in below 2 kHz in the rosella control calls. Previous work has
strength according to the degree of acoustic similarity, it would shown that superb fairy-wren aerial alarm calls containing 4
indicate that similarity could provoke responses to heterospe- elements signal urgent danger, and fairy-wrens respond by
cific alarm calls. We composed playbacks from high-quality ele- fleeing to cover in 100% of cases (Fallow and Magrath
ments chosen from the natural calls of 9 species: splendid 2010). Therefore, we constructed 4-element passerine alarm
fairy-wrens, variegated fairy-wrens, white-winged fairy-wrens, calls so that it would be clear whether they responded to
brown gerygones, chestnut-rumped thornbills, western thorn- the playbacks. Each alarm call was constructed by pasting
Fallow et al. • Acoustic similarity provokes responses to alarm calls 405

the same element 4 times, with a 45 ms interval between the detect an effect on the amount of time fairy-wrens spent in
end of each element and the start of the next element. This cover after fleeing (using only playback calls that provoked
standard interval length was within the natural range of alarm birds to spend time in cover, n ¼ 53).
calls used to generate playbacks and ensured that differences We then tested whether specific acoustic properties influ-
in fairy-wren responses to playback were due only to variation enced antipredator responses. A GLM with binomial error
in the element structure of different species and not interval was used to analyze the binary responses (stay or flee), and
length. a GLM with a normal distribution to analyze variation in the
Playbacks were uncompressed wave files broadcast from amount of time fairy-wrens spent in cover. Both GLMs were
a Roland Edirol R-09 HR solid-state playback and recording fitted with the following acoustic variables: peak frequency
device (20–22 000 Hz) via a Kemo Electronics integrated am- (kilohertz), element duration (milliseconds), and number
plifier (20–25 000 Hz) and a Response Dome tweeter speaker of frequency cycles. Variables measured in kilohertz were log
(1500–20 000 Hz). Playback amplitude was measured in Raven transformed. We used the Wald statistic to test for fixed effects
Pro 1.3 and calibrated against a tone that had its amplitude then sequentially removed nonsignificant terms one by one
determined using a digital level meter. The playback equip- until only significant effects remained. Analyses were per-
ment was mounted around the observer’s waist and playbacks formed using GenStat Software v. 11 (VSN International
broadcast at an element amplitude of 61 dB at 5 m, which is Ltd., Hemel Hempstead, UK.) with significance levels set at
around 1 standard deviation (SD) above the mean amplitude P , 0.05.
for superb fairy-wrens (Magrath et al. 2007), to ensure that
playbacks were audible. We played back calls from the 9 spe-
RESULTS
cies to 108 fairy-wren groups in Canberra in June and July

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


2009. Playback calls were grouped into 12 sets of 9 calls, in- Acoustic similarity of calls from different species
cluding one call from each species. Playback sets were then
There was substantial variation in the acoustic similarity of
presented one at a time, with each completed within a single
alarm calls from the allopatric playback species to those of su-
day. Each playback call was presented to a different fairy-wren
perb fairy-wrens (Table 1; Figure 1). The canonical variates
group, and calls were presented in random order. Prior to
plot provided a 2D representation of acoustic similarity be-
each playback, the fairy-wren group was observed for 5 min
tween the calls of different species, with the 3 acoustic prop-
to ensure that birds were not agitated by conspecifics or the
erties explaining a total of 98.7% of variation between
observer and that no threatening species or other bird species
measurements taken from the calls of the 8 playback species
that could provide cues of danger were nearby (,10 m). If no
(Figure 2). The degree of acoustic similarity between calls was
alarm calls were detected during the observation period, the
represented by distance on the x axis (canonical variate 1),
playback was presented to a randomly chosen focal bird when
it was foraging on the ground at least 50 cm from cover. We
presented playbacks from a distance of around 8 m (mean 6
SD ¼ 8.21 6 0.91) from the focal bird and recorded one of
the following immediate responses: 0, no response; 1, scan; 2,
scan then flee; and 3, immediately flee to cover or away from
their original position. These graded responses were then
sorted into binomial responses for comparison with previous
studies: 0, nonflee (0, 1, and 2) and 1, flee (3). If the focal
bird fled to cover, the duration of time spent in cover (sec-
onds) before it emerged was recorded if possible but capped
at 60 s. To minimize stress caused to the birds, we provoked
alarm calls from different fairy-wren groups when recording
calls and during the playback experiment presented each
group with only 1 alarm call.

Statistical analyses
To compare the acoustic similarity of superb fairy-wren alarm
calls to those of the allopatric species used in the playback ex-
periment, we generated a canonical variates plot using acous-
tic measurements taken from aerial alarm calls of the 8
species. Calls were placed on a 2D plane according to the
amount of variation explained by: peak frequency (kilohertz),
element duration (milliseconds), and number of frequency
cycles. Peak frequency was log transformed. We chose these
variables to classify calls because they were not correlated with
each other. Minimum, maximum, and peak frequencies were
particularly highly correlated, and we used peak frequency Figure 2
because it was easiest to measure and the most repeatable. Canonical variates plot of aerial alarm calls of the species used in the
Using intergroup distances from the canonical variates plot playback experiment. In order from most to least similar to superb
as a proxy for acoustic similarity between calls, we then used fairy-wrens according to intergroup distances, these are superb fairy-
wren (:), splendid fairy-wren (1), variegated fairy-wren (h), brown
generalized linear models (GLMs) to test whether there was
gerygone (d), white-winged fairy-wren (n), western thornbill (s),
a relationship between acoustic similarity to superb fairy-wren southern whiteface ( ), and chestnut-rumped thornbill (e). Circles
calls and the strength of fairy-wren responses to playback. A represent confidence regions around mean values for each species’
GLM with binomial error was used to detect an effect of calls calculated using the following acoustic properties: peak
intergroup distance on the binary responses (stay or flee, frequency (kilohertz), element duration (milliseconds), and number
n ¼ 96)), and a GLM with a normal distribution was used to of frequency cycles.
406 Behavioral Ecology

which explained 75.3% of variation, and by distance on the y


axis (canonical variate 2), which explained 23.3% of variation.
Loadings in the plot of calls used in the playback experiment
were canonical variate 1 (peak frequency, 218.3; element du-
ration, 8.7; and number of cycles, 21.1) and canonical variate
2 (peak frequency, 242.0; element duration, 29.8; and num-
ber of cycles, 0.5).
Superb fairy-wren aerial alarm calls were more similar to
those of most of its congeners than to the calls of species in
the family Acanthizidae. Calls of most of the malurids were
clustered together, indicating high acoustic similarity within
the genus. The calls of white-winged fairy-wrens were separated
from those of the other malurids due to their lack of frequency
modulation (no frequency cycles). Calls of the acanthizids
were more variable than those of the malurids and so more
spread out. We used intergroup distances between the mean
values for calls of each playback species to estimate their de-
gree of acoustic similarity relative to superb fairy-wren calls.
Calls of the allopatric playback species in order from most to
least acoustically similar to superb fairy-wren calls with their

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


intergroup distances were 1) splendid fairy-wren, 0.64; 2) var-
iegated fairy-wren, 0.75; 3) brown gerygone, 7.93; 4) white-
winged fairy-wren 8.84; 5) western thornbill, 8.88; 6) southern
whiteface, 10.42; and 7) chestnut-rumped thornbill, 10.59.

Fairy-wren responses to unfamiliar calls


Fairy-wrens responded more strongly to playback of unfamiliar
calls with greater acoustic similarity to conspecific calls. Play-
back calls with lower intergroup distances to fairy-wren calls
provoked more fairy-wrens to flee to cover (GLM: v21 ¼ 16:79;
P , 0.001; Figure 3a) and to spend more time in cover after
fleeing (GLM: v21 ¼ 19:16; P , 0.001; Figure 3b). Fairy-wrens
always fled to cover after playback of conspecific calls (positive
control), to most calls of other malurids, and to 75% of calls
from western thornbills. However, they fled to only one brown
Figure 3
gerygone call and did not flee to playback of southern white- Relationships between the intergroup distance of each playback
face or chestnut-rumped thornbill calls (Figure 4a). Playback species’ calls to superb fairy-wren calls and fairy-wren responses to
of rosella calls (negative control) never elicited flight or even playback: (a) the proportion of focal birds immediately fleeing and
a vigilance response in fairy-wrens. (b) the mean duration of time spent in cover before reemerging
Although superb fairy-wrens had a high probability of flee- (seconds). Playback species were: Superb fairy-wren (:), splendid
ing to playback of calls of congeners and western thornbills, the fairy-wren (1), variegated fairy-wren (h), brown gerygone (d),
amount of time they spent in cover differed among species white-winged fairy-wren (n), western thornbill (s), southern
(analysis of variance excluding birds that did not flee or that whiteface ( ), and chestnut-rumped thornbill (e).
fled to a location outside of cover, F4,48 ¼ 6.79, P , 0.001;
Figure 4b). Fairy-wrens spent 16 6 8.6 s in cover (mean 6 SD) Acoustic properties influencing responses
after playback of conspecific calls, results practically identical Only 2 acoustic properties affected fairy-wren responses to play-
to those found in a previous study (mean 6 SD ¼ 16 6 5.2 s, back, with each property influencing different antipredator be-
Fallow and Magrath 2010). Their response to playback of haviors. The probability of fairy-wrens fleeing to cover was
splendid fairy-wren calls was no different to their response greater in response to playback of calls with higher peak frequen-
to conspecific calls (mean difference 6 least significant dif- cies (GLM: v21 ¼ 31:90; P , 0.001), and the amount of time
ference at 5% ¼ 0.8 6 5.1 s); they always fled to splendid fairy- spent in cover increased following playback of calls with a greater
wren calls and then spent the same amount of time in cover number of frequency cycles (GLM: v21 ¼ 11:94; P ¼ 0.001). The
(mean 6 SD ¼ 15.1 6 8.1 s). However, fairy-wrens spent less substitution of frequency modulation rate for the number of
time in cover after fleeing to playback of calls of other heter- frequency cycles made no difference to the outcome, whether
ospecifics (mean difference 6 least significant difference at it was treated as a continuous variable (GLM:v21 ¼ 11:48;
5%): variegated fairy-wren (7.9 6 5.2 s); white-winged fairy- P ¼ 0.001) or as a categorical variable (GLM with frequency
wren (9.0 6 5.5 s); and western thornbill (10.8 6 5.5 s). modulation as present or absent:v21 ¼ 12:46; P , 0.001).
In all cases where superb fairy-wrens did not flee in response
to playback of a malurid call, they responded with an increase
DISCUSSION
in vigilance, such as raising the head and scanning (Figure 4a).
Scanning was also seen in 2/11 cases when they did not flee to Our study supports the suggestion by Marler (1957) that ver-
playback of brown gerygone calls and in 1/3 cases when they tebrates will respond to heterospecific alarm calls if they are
did not flee to playback of western thornbill calls. Also, de- acoustically similar to conspecific calls. Superb fairy-wrens re-
spite not fleeing to any playbacks of calls of southern white- sponded more strongly to unfamiliar calls that were acousti-
faces and chestnut-rumped thornbills, fairy-wrens responded cally more similar to conspecific alarm calls. This shows how
with scanning in 1 and 2 cases, respectively. birds can respond to alarm calls of congeners despite having
Fallow et al. • Acoustic similarity provokes responses to alarm calls 407

can be attributed to specific acoustic properties or to combina-


tions of acoustic properties they contain (Brémond 1976; de
Kort and ten Cate 2001). Peak frequency and the number of
frequency cycles appeared to be the key acoustic properties
affecting fairy-wren responses to heterospecific alarm calls.
This offers proximate reasons for why fairy-wrens fled to the
unfamiliar calls of some species but not to others: the calls of
malurids and western thornbills had higher peak frequencies
than most of the acanthizids and apparently therefore re-
sulted in a higher probability of birds fleeing to these calls.
Additionally, although brown gerygone calls had about the
same amount of overall similarity to superb fairy-wren calls
as those of white-winged fairy-wrens (Figure 2), the low peak
frequency of gerygone calls appears to explain why fairy-wrens
usually did not flee to playback of these calls and yet fled to
most calls of white-winged fairy-wrens. Gerygone calls had the
lowest peak frequencies of all species in the playback experi-
ment, whereas the peak frequency of white-winged fairy-wren
calls was close to that of its congeners. Peak frequency might
also be used to identify alarm calls because it is relatively re-

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


sistant to degradation. Our conclusions are limited to the
amount of variation in the natural calls of the species we
examined and the acoustic properties that we measured. To
confirm that fairy-wrens base their responses on these partic-
ular acoustic properties, and how variation in these properties
can affect responses, we need additional experiments testing
the effect of varying specific acoustic properties while keeping
all other parameters constant.
Superb fairy-wrens responded cautiously to unfamiliar alarm
calls by varying their behavior in 2 separate stages. They ap-
peared to process calls in at least 2 steps to produce appro-
priate responses: 1) rapidly decide whether to flee or not,
Figure 4 depending on whether the peak frequency of a call was in
The observed responses of superb fairy-wrens to playback of the range of a conspecific call; and 2) once in the safety of
unfamiliar 4-element aerial alarm calls: (a) the proportion of focal
birds fleeing to cover or scanning; and (b) the mean (1 standard
cover, decide how likely the call was a false alarm based on the
error) duration of time (seconds) the focal bird spent in cover before number of frequency cycles and consequently how much time
reemerging. Playback calls are presented from left to right in order of to spend in cover before reemerging. For example, fairy-wrens
acoustic similarity to superb fairy-wren calls based on intergroup had a high probability of fleeing to playback of all malurid
distances calculated by the canonical variates plot of playback species calls because they all had high peak frequencies, but they
(Figure 2). Mean values for the acoustic properties affecting the spent less time in cover after variegated and white-winged
probability of fleeing, and the amount of time spent in cover are fairy-wren calls because, unlike splendid fairy-wrens calls,
presented for each species. Superb fairy-wren alarm calls served as these calls did not have a number of frequency cycles similar
a positive control. Playback of crimson rosella piping calls, the to that of conspecific calls. This cautious initial response fol-
negative control, did not elicit any response and are not represented
here.
lowed by a second stage of escape behavior is comparable with
the response of western grebes, Aechmophorus occidentalis, to
playback of alarm calls by Forster’s terns, Sterna forsteri: after
no experience of them. Responding to acoustic similarity is playback, most grebes either assumed an alarmed posture or
likely to benefit fairy-wrens in the large areas where their geo- left the nest, but on no occasion did playbacks provoke the
graphic range overlaps with that of other malurid species, full alarm response of diving into the water, presumably be-
which themselves probably also respond to calls of congeners, cause grebes rely on further information in the form of visual
including superb fairy-wrens. cues (Nuechterlein 1981). A 2-stage response process allows
Fairy-wrens responded to calls that were acoustically similar fairy-wrens to increase their level of safety in response to sig-
to their own, with the strength of response increasing with in- nals that might indicate urgent danger while reducing the
creasing similarity. This suggests that they generalize responses amount of time wasted in cover after false alarms. The re-
from conspecific calls to unfamiliar calls that are acoustically sponses of fairy-wrens to playback of unfamiliar heterospecific
similar (Ghirlanda and Enquist 2003). All playback calls had alarm calls lends support to a previous study, which suggested
the same amplitude, tempo, and number of elements, and so that they varied the amount of time spent in cover in response
variation in the responses they provoked were due to differ- to the degree of urgency in conspecific calls: although fairy-
ences in the acoustic structure of elements. Therefore, fairy- wrens fled to playback of all multielement conspecific calls,
wrens do not require learning to respond to heterospecific the amount of time spent in cover increased with an increas-
calls that are acoustically similar to their own, including those ing number of elements in the call, suggesting the assessment
of congeners. of danger (Fallow and Magrath 2010).
Fairy-wren responses to playback were affected by specific The stronger responses of superb fairy-wrens to playback
acoustic properties and how similar they were to conspecific of calls of congeners with increasing phylogenetic related-
calls. Fairy-wrens had a higher probability of fleeing to calls ness appear to be due to the calls of more closely related spe-
with high peak frequencies and then spent more time in cover cies being acoustically more similar. Superb fairy-wrens are
when calls had a number of frequency cycles similar to their the sister species to splendid fairy-wrens, intermediately re-
own calls. The species-specificity of calls and responses to them lated to variegated fairy-wrens, and most distantly related to
408 Behavioral Ecology

white-winged fairy-wrens (Gardner et al. 2010), and the de- innate response to stimuli, such as conspecific calls (Davies
gree of relatedness between superb fairy-wrens and its con- et al. 2004), as it allows animals to generalize responses to
geners corresponded roughly with both the degree of novel stimuli that resemble stimuli to which they have learnt
acoustic similarity between calls and the strength of re- to respond (Ghirlanda and Enquist 2003).
sponses to playback. An increase in response strength with In addition to provoking responses to unfamiliar alarm calls
increasing phylogenetic relatedness was also shown by the that are highly similar to conspecific alarm calls, acoustic sim-
responses of 2 turtle dove species, Streptopelia roseogrisea and ilarity might make heterospecific calls easier to learn to recog-
S. vinacea, to playback of heterospecific ‘‘perch-coos,’’ long- nize (Magrath et al. 2009b). Fairy-wrens that did not flee to
range territorial vocalizations; both species called more in playback of the dissimilar calls of acanthizids usually showed
response to sister species than to the calls of more distantly no change in behavior, but in all 3 cases when birds did not
related congeners (de Kort and ten Cate 2001). If the flee to playback of similar calls of congeners, they still re-
acoustic similarity of alarm calls is a result of phylogeny sponded by scanning. If unfamiliar heterospecific alarm calls
then the degree of relatedness may have affected fairy-wren have some degree of acoustic similarity to conspecific calls,
responses through changes in acoustic properties, such as they might provoke vigilance in individuals that have not
the number of frequency cycles. Otherwise, it is also possi- learnt to associate it with danger. An example of this can be
ble that phylogeny affected responses through variation in seen in the response of ringtailed lemurs, Lemur catta, to play-
acoustic features that we did not measure. Therefore, fu- back of alarm calls from Verreux’s sifakas, Propithecus verreuxi:
ture studies should consider controlling for effects of phy- wild lemurs live in sympatry with sifakas in Madagascar and
logenetic distance either by using synthetic calls or the calls respond appropriately to sifaka calls indicating aerial or ter-
of species with a similar degree of relatedness. restrial predators. Although a colony of lemurs born in Japan

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


The importance of acoustic similarity in allowing fairy-wrens had no experience of sifaka calls and did not respond appro-
to respond to unfamiliar calls might explain why fairy-wrens priately to sifaka calls indicating different predator types,
did not respond to the apparently similar calls of white-browed some individuals nevertheless responded to the unfamiliar
scrubwrens and yet responded strongly to the relatively dissim- heterospecific calls by looking up (Oda and Masataka 1996).
ilar calls of western thornbills. Although fairy-wrens in our In another example, chicks of the common cuckoo, Cuculus
study site respond strongly to scrubwren calls (Magrath canorus, which parasitizes the nests of reed warblers, Acrocepha-
et al. 2007; Fallow and Magrath 2010), a population of fairy- lus scirpaceus, have an innate response to alarm calls used by
wrens living in the absence of scrubwrens did not respond to parent warblers to warn nestlings of danger and suppress beg-
playback of scrubwren calls indicating urgent danger ging calls but require experience to develop appropriate re-
(Magrath et al. 2009b), implying that learning is a prerequisite sponses. Cuckoo chicks that were raised by other host species
for fairy-wrens to respond to these calls. However, fairy-wrens responded only to warbler alarm calls but by inappropriately
in our study responded strongly to the unfamiliar calls of increasing their begging rate (Davies et al. 2006). In cases
western thornbills, calls that are generally less similar to such as these, where there is a weak or incorrect innate re-
fairy-wren calls then those of scrubwrens. Western thornbills sponse to heterospecific alarm calls, experience is required
were the only acanthizid species with calls provoking strong for the development and refinement of appropriate responses
responses, possibly because their calls have a relatively high (Magrath et al. 2010).
peak frequency or because fairy-wrens in our study site were In conclusion, responding to sounds that are acoustically sim-
influenced by experience with the acoustically similar calls of ilar to conspecific alarm calls can allow birds to respond to het-
a sympatric species. Fairy-wrens in Canberra live sympatrically erospecific calls without having to learn them. In addition to
with buff-rumped thornbills, Acanthiza reguloides, a species that learning responses, this shows how vertebrates can distinguish
is closely related to western thornbills and so might have sim- relevant signals from the plethora of sounds in the acoustic en-
ilar calls. We generated a second canonical variates plot in- vironment. Furthermore, because the calls of different species
cluding calls of white-browed scrubwrens and buff-rumped can converge on similar acoustic structures, acoustic similarity
thornbills to see how similar the calls of these sympatric spe- may be as widespread as learning in allowing vertebrates to re-
cies were to those of fairy-wrens and the allopatric species spond to heterospecific calls. Further research on interspecific
used in the playback experiment. The plot suggested that eavesdropping could use synthetic calls to isolate the specific
scrubwren calls were generally quite similar to fairy-wren calls acoustic properties that prompt responses to unfamiliar alarm
(intergroup distance to fairy-wren calls: 2.9). However, the calls as well as investigating the learning mechanisms that allow
peak frequency of scrubwren calls (7.8 kHz) might be too interspecific responses to dissimilar alarm calls.
low to prompt fairy-wrens to flee to them without experience.
Although buff-rumped thornbill calls were generally quite dis-
similar to fairy-wren calls, they were indistinguishable from
western thornbill calls (intergroup distance to superb fairy- FUNDING
wren calls: western thornbill, 9.13; buff-rumped thornbill,
9.31). Therefore, we consider it likely that fairy-wrens in Can- Australian Research Council (Discovery Grant DP0665481 to
berra learnt to respond to calls of sympatric buff-rumped R.D.M.).
thornbills and then generalized responses to calls of allopatric
western thornbills because they are acoustically similar (Ghir- We thank Hwan-Jin Yoon for statistical advice and Tonya Haff,
landa and Enquist 2003). Fairy-wrens are known to respond to Branislav Igic, Naomi Langmore, Justin Welbergen, and 2 anonymous
the acoustically dissimilar calls of sympatric heterospecifics; reviewers for valuable feedback on the manuscript. Michael and
for example, they respond to aerial alarm calls of New Hol- Lesley Brooker, Alan Burbidge, Perry and Alma deRebeira and
land honeyeaters (Melphagidae), which are low frequency Eleanor Russell provided accommodation and advice on field sites,
with a monotonic decline in frequency over time (Magrath and Michael and Sandy Clinch at Nallan Station allowed us to record
on their property. Thanks to those who assisted us with recording calls
et al. 2009b). If fairy-wrens in our study site fled to playback of in the field: Demetrios Bertzeletos, Caroline Blackmore, Helen
western thornbill calls because they were similar to calls they Gardner, Peter Marsack, and Cathy O’Reilly. This study was carried
had learnt, this would be an example of a receiver bias gen- out with approval from the Australian National University Ethics Com-
erated through learning (ten Cate and Rowe 2007). Learning- mittee and under permits from the Australian Bird and Bat Banding
based bias differs from a neural template, where there is an Scheme, the Australian National Botanic Gardens, the Department of
Fallow et al. • Acoustic similarity provokes responses to alarm calls 409

Environment and Conservation (Western Australia; SFOO5652 and Hurd CR. 1996. Interspecific attraction to the mobbing calls of black-
CE001538), Environment ACT (Australian Capital Territory), and capped chickadees (Parus atricapillus). Behav Ecol Sociobiol. 38:
the New South Wales Department of Environment and Conservation 287–292.
(S11897). Ito R, Mori A. 2010. Vigilance against predators induced by eavesdrop-
ping on heterospecific alarm calls in a non-vocal lizard Oplurus
cuvieri cuvieri (Reptilia: Iguania). Proc R Soc B Biol Sci. 277:
REFERENCES 1275–1280.
Barrett G, Silcocks A, Barry S, Cunningham R, Poulter R. 2003. The Johnson FR, McNaughton EJ, Shelley CD, Blumstein DT. 2003. Mech-
new atlas of Australian birds. Melbourne (Australia): Royal Austral- anisms of heterospecific recognition in avian mobbing calls. Aust J
asian Ornithologists Union. Zool. 51:577–585.
Bell HL. 1980. Composition and seasonality of mixed-species feeding Jurisevic MA, Sanderson KJ. 1998. Acoustic discrimination of passer-
flocks of insectivorous birds in the Australian Capital Territory. ine anti-predator signals by Australian raptors. Aust J Zool. 46:
Emu. 80:227–232. 369–379.
Bradbury JW, Vehrencamp SL. 1998. Principles of animal communi- Leavesley AJ, Magrath RD. 2005. Communicating about danger: ur-
cation. Sunderland (MA): Sinaur Associates, Inc. gency alarm calling in a bird. Anim Behav. 70:365–373.
Brémond J-C. 1976. Specific recognition in the song of Bonelli’s war- Magrath RD, Haff TM, Horn AG, Leonard ML. 2010. Calling in the
bler (Phylloscopus bonelli). Behaviour. 58:99–116. face of danger: predation risk and acoustic communication by par-
Bshary R, Noë R. 1997. Red colobus and Diana monkeys provide ent birds and their offspring. In: Brockmann HJ, editor. Advances
mutual protection against predators. Anim Behav. 54:1461–1474. in the study of behavior. Burlington (MA): Academic Press. p.
Burger J. 1984. Grebes nesting in gull colonies: protective associations 187–253.
and early warning. Am Nat. 123:327–337. Magrath RD, Pitcher BJ, Gardner JL. 2007. A mutual understanding?
Caro TM. 2005. Antipredator defenses in birds and mammals. Interspecific responses by birds to each other’s aerial alarm calls.

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014


Chicago (IL): The University of Chicago Press. Behav Ecol. 18:944–951.
Charif RA, Waack AM, Strickman LM. 2008. Raven Pro 1.3 User’s Magrath RD, Pitcher BJ, Gardner JL. 2009a. An avian eavesdropping
Manual. Ithaca (NY): Cornell Laboratory of Ornithology. network: alarm signal reliability and heterospecific response. Behav
Cockburn A, Osmond HL, Mulder RA, Green DJ, Double MC. 2003. Ecol. 20:745–752.
Divorce, dispersal and incest avoidance in the cooperatively breed- Magrath RD, Pitcher BJ, Gardner JL. 2009b. Recognition of
ing superb fairy-wren Malurus cyaneus. J Anim Ecol. 72:189–202. other species’ aerial alarm calls: speaking the same language or
Davies NB, Madden JR, Butchart SHM. 2004. Learning fine-tunes learning another? Proc R Soc Lond B Biol Sci. 276:769–774.
a specific response of nestlings to the parental alarm calls of their Manser MB, Seyfarth RM, Cheney DL. 2002. Suricate alarm calls signal
own species. Proc R Soc Lond B Biol Sci. 271:2297–2304. predator class and urgency. Trends Cogn Sci. 6:55–57.
Davies NB, Madden JR, Butchart SHM, Rutila J. 2006. A host-race of Marler P. 1955. Characteristics of some animal calls. Nature. 176:6–8.
the cuckoo Cuculus canorus with nestlings attuned to the parental Marler P. 1957. Specific distinctiveness in the communication signals
alarm calls of the host species. Proc R Soc Lond B Biol Sci. 273: of birds. Behaviour. 11:13–38.
693–699. McCracken KG, Sheldon FH. 1997. Avian vocalizations and phyloge-
de Kort SR, ten Cate C. 2001. Response to interspecific vocalizations is netic signal. Proc Natl Acad Sci U S A. 94:3833–3836.
affected by the degree of phylogenetic relatedness in Streptopelia Mulder RA. 1995. Natal and breeding dispersal in a co-operative,
doves. Anim Behav. 61:239–247. extra-group-mating bird. J Avian Biol. 26:234–240.
Evans CS, Evans L, Marler P. 1993. On the meaning of alarm calls: Müller CA, Manser MB. 2008. The information banded mongooses
functional reference in an avian vocal system. Anim Behav. 46: extract from heterospecific alarms. Anim Behav. 75:897–904.
23–38. Nuechterlein GL. 1981. ‘Information parasitism’ in mixed colonies of
Fallow PM, Magrath RD. 2010. Eavesdropping on other species: mu- western grebes and Forster’s terns. Anim Behav. 29:985–989.
tual interspecific understanding of urgency information in avian Oda R, Masataka N. 1996. Interspecific responses of ringtailed lemurs
alarm calls. Anim Behav. 79:411–417. to playback of antipredator alarm calls given by Verreaux’s sifakas.
Fennell P, editor. 2009. Birds of Canberra gardens. 2nd ed. Canberra
Ethology. 102:441–453.
(Australia): Canberra Ornithologists Group.
Overs A. 2009. Annual bird report. Canberra Bird Notes 34(1). Can-
Fichtel C. 2004. Reciprocal recognition of sifaka (Propithecus verreauxi
berra (Australia): Canberra Ornithologists Group Inc.
verreauxi) and redfronted lemur (Eulemur fulvus rufus) alarm calls.
Pereira ME, Macedonia JM. 1991. Ringtailed lemur anti-predator calls
Anim Cogn. 7:45–52.
denote predator class, not response urgency. Anim Behav. 41:
Gardner JL, Trueman JWH, Ebert D, Joseph L, Magrath RD. 2010.
Phylogeny and evolution of the Meliphagoidea, the largest radia- 543–544.
Rainey HJ, Zuberbühler K, Slater PJB. 2004. Hornbills can distinguish
tion of Australasian songbirds. Mol Phylogenet Evol. 55:1087–1102.
Ghirlanda S, Enquist M. 2003. A century of generalization. Anim Be- between primate alarm calls. Proc R Soc Lond B Biol Sci. 271:
hav. 66:15–36. 755–759.
Gill SA, Sealy SG. 2004. Functional reference in an alarm signal given Ramakrishnan U, Coss RG. 2000. Recognition of heterospecific alarm
during nest defence: seet calls of yellow warblers denote brood- vocalizations by Bonnet Macaques (Macaca radiata). J Comp Psy-
parasitic brown-headed cowbirds. Behav Ecol Sociobiol. 56:71–80. chol. 114:3–12.
Goodale E, Beauchamp G, Magrath RD, Nieh JC, Ruxton GD. 2010. Randler C. 2006. Red squirrels (Sciurus vulgaris) respond to alarm
Interspecific information transfer influences animal community calls of Eurasian jays (Garrulus glandarius). Ethology. 112:411–416.
structure. Trends Ecol Evol. 25:354–361. Russ JM, Jones G, Mackie IJ, Racey PA. 2004. Interspecific
Goodale E, Kotagama SW. 2005. Alarm calling in Sri Lankan mixed- responses to distress calls in bats (Chiroptera: Vespertilionidae):
species bird flocks. Auk. 122:108–120. a function for convergence in call design? Anim Behav. 67:
Griffin AS, Savani RS, Hausmanis K, Lefebvre L. 2005. Mixed-species 1005–1014.
aggregations in birds: zenaida doves, Zenaida aurita, respond to the Seyfarth RM, Cheney DL, Marler P. 1980. Monkey responses to three
alarm calls or carib grackles, Quiscalus lugubris. Anim Behav. 70: different alarm calls: evidence of predator classification and seman-
507–515. tic communication. Science. 210:801–803.
Hauser MD. 1988. How infant vervet monkeys learn to recognize star- Shriner WM. 1998. Yellow-bellied marmot and golden-mantled
ling alarm calls: the role of experience. Behaviour. 105:187–201. ground squirrel responses to heterospecific alarm calls. Anim Be-
Higgins PJ, editor. 1999. Handbook of Australian, New Zealand and hav. 55:529–536.
Antarctic Birds. Volume 4: Parrots to Dollarbird. Melbourne Shriner WM. 1999. Antipredator responses to a previously neutral
(Australia): Oxford University Press. sound by free-living adult golden-mantled ground squirrels, Spermo-
Higgins PJ, Peter JM, Steele WK, editors. 2001. Handbook of Austra- philus lateralis (Sciuridae). Ethology. 105:747–757.
lian, New Zealand and Antarctic Birds. Volume 5: Tyrant-flycatchers Sullivan KA. 1984. Information exploitation by downy woodpeckers in
to Chats. Melbourne (Australia): Oxford University Press. mixed-species flocks. Behaviour. 91:294–311.
410 Behavioral Ecology

Templeton CN, Greene E. 2007. Nuthatches eavesdrop on variations Vitousek MN, Adelman JS, Gregory NC, St Clair JJH. 2007. Hetero-
in heterospecific chickadee mobbing calls. Proc Natl Acad Sci U S specific alarm call recognition in a non-vocal reptile. Biol Lett. 3:
A. 104:5479–5482. 632–634.
Templeton CN, Greene E, Davis K. 2005. Allometry of alarm calls: Zuberbühler K. 2000. Interspecies semantic communication in two
black-capped chickadees encode information about predator size. forest primates. Proc R Soc Lond B Biol Sci. 267:713–718.
Science. 308:1934–1937. Zuberbühler K. 2009. Survivor signals: the biology and psychology of
ten Cate C, Rowe C. 2007. Biases in signal evolution: learning makes animal alarm calling. In: Naguib M, Janik VM, editors. Advances in
a difference. Trends Ecol Evol. 22:380–387. the study of behavior. Burlington (MA): Academic Press. p. 277–322.

Downloaded from http://beheco.oxfordjournals.org/ by guest on July 8, 2014

You might also like