Barking and mobbing

Abstract

Barking is most often associated with the domestic dog Canis familiaris, but it is a common mammalian and avian vocalization. Like any vocalization, the acoustic character of the bark is likely to be a product of adaptation as well as an expression of the signaler's internal motivational state. While most authors recognize that the bark is a distinct signal type, no consistent description of its acoustic definition or function is apparent. The bark exhibits considerable variability in its acoustic form and occurs in a wide range of behavioral contexts, particularly in dogs. This has led some authors to suggest that dog barking might be a form of referential signaling, or an adaptation for heightened capability to communicate with humans. In this paper we propose a general 'canonical' acoustic description of the bark. Surveying relevant literature on dogs, wild canids, other mammals and birds, we explore an alternative functional hypothesis, first suggested by [Morton, E.S., 1977. On the occurrence and significance of motivation-structural rules in some bird and mammal sounds. Am. Nat. 111, 855-869] and consistent with his motivational-structural rules theory: that barking in many animals, including the domestic dog, is associated with mobbing behavior and the motivational states that accompany mobbing.

Full text

Behavioural Processes 81 (2009) 358–368
Contents lists available at ScienceDirect
Behavioural Processes
journal homepage: www.elsevier.com/locate/behavproc
Barking and mobbing
Kathryn Lorda,∗, Mark Feinsteinb, Raymond Coppingerb
aUniversity of Massachusetts, Amherst, United States
bHampshire College, Amherst MA, United States
article info
Article history:
Received 2 December 2008
Received in revised form 6 April 2009
Accepted 8 April 2009
Keywords:
Dog
Bark
Mobbing
Morton’s rules
Motivational-structural rules
abstract
Barking is most often associated with the domestic dog Canis familiaris, but it is a common mammalian
and avian vocalization. Like any vocalization, the acoustic character of the bark is likely to be a product
of adaptation as well as an expression of the signaler’s internal motivational state. While most authors
recognize that the bark is a distinct signal type, no consistent description of its acoustic definition or
function is apparent. The bark exhibits considerable variability in its acoustic form and occurs in a wide
range of behavioral contexts, particularly in dogs. This has led some authors to suggest that dog barking
might be a form of referential signaling, or an adaptation for heightened capability to communicate with
humans. In this paper we propose a general ‘canonical’ acousticdescription of the bark. Surveying relevant
literature on dogs, wild canids, other mammals and birds, weexplore an alternative functional hypothesis,
first suggested by [Morton, E.S., 1977. On the occurrence and significance of motivation-structural rules in
some bird and mammal sounds. Am. Nat. 111, 855–869] and consistent with his motivational-structural
rules theory: that barking in many animals, including the domestic dog, is associated with mobbing
behavior and the motivational states that accompany mobbing.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
Barking is a universally recognizedhallmark of the domestic dog,
Canis lupus familiaris. From the casual human listener’s standpoint,
barking seems readily distinguishable from other vocalizations. But
the terms “bark” and “barking” are often used in the scientific
literature without a precise definition of a bark’s structure. The
function of this vocalization is variously analyzed as an alarm call
(Cohen and Fox, 1976; Tembrock, 1976; Lehner, 1978; Schassburger,
1987, 1993; Harrington and Asa, 2003); a territory-marking sig-
nal (Lehner, 1978; Cohen and Fox, 1976); a rally call (Schassburger,
1987; Cohen and Fox, 1976); or an indicator of motivational state
(Morton, 1977; Bleicher, 1963; Tembrock, 1976). Coppinger and
Feinstein (1991) argue that dog barking is a developmental artifact
with no intrinsic function; Yin and McCowan (2004),Yin (2002)
and Feddersen-Petersen (2000) speculate that barks may have ref-
erential content; and Pongrácz et al. (2005) suggest that barks are
co-adapted signals between dogs and humans.
Several descriptions of the acoustic properties of barking and
other canid vocalizations have appeared in the recent literature:
Reide and Fitch (1999) looked at correlations between vocal tract
anatomy and vocalization in the dog; Riede et al. (2001) and Riede et
al. (2005) examined harmonic-to-noise ratios in the bark; Yin and
McCowan (2004) carried out an extensive analysis of the acous-
∗Corresponding author.
E-mail address: klord3@mac.com (K. Lord).
tic character of barking in 10 adult dogs of 6 breeds, looking at
the interaction of multiple parameters. In our laboratory we made
sonograms of barking and other vocalizations of several breeds at
various ages, examining tonal and noisy qualities.
Despite this recent attention there does not yet appear to be a
common and useful definition of the acoustic structure of the bark.
We offer one here, acknowledging that barking is a highly variable
phenomenon: Yin and McCowan (2004) and many others provide
ample evidence of individual differences in dogs; Mitchell et al.
(2006) do the same for the coyote. Casual listeners readily conclude
that individual dogs bark differently in different circumstances,and
that different breeds have distinctive barks. Dog barks resemble
other calls, described as “woofs,” “yelps” or “cries.” Some investi-
gators treat them as functionally related graded variants of a single
call-type. But we show that the bark nonetheless has basic, largely
invariant structural properties which distinguish it systematically
from other signals. Moreover, by taking into account this distinctive
acoustic character–ageneralized form that can be termed “canon-
ical” because it occurs not only in the domestic dog and related
canids but also more widely in mammals and birds – we are able to
propose a functional explanation for barking as well as an account
of its contextual variability.
2. Acoustic features of barking
Sounds are generally characterized along three acoustic dimen-
sions—frequency, amplitude and duration. Within each dimension,
certain parameters capture the acoustic natureof particular signals.
0376-6357/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.beproc.2009.04.008

K. Lord et al. / Behavioural Processes 81 (2009) 358–368 359
These parameters include the following (with abbreviations wewill
adopt in later discussion).
Frequency
Tonality T
Noise N
Mean Pitch MP
Frequency Modulation FM
Amplitude
Relative Amplitude RA
Abrupt Onset AO
Duration
Pulse Duration PD
Pulse Repetition PR
3. Frequency
The frequency or pitch of a sound is determined by the rate
of periodic oscillation (vibration) of some structure (e.g. laryngeal
vocal folds in mammals) disturbing a medium such as air; it is mea-
sured in cycles per second or Hertz (Hz). Slow vibrations sound low
in pitch, fast vibrations sound high. This rate depends on anatomi-
cal and physiological factors (e.g., vocal fold length and thickness),
which vary in individuals. Vibration rate in many animals is also
under voluntary control, thus it can vary from production to pro-
duction in a single call-type, as well as over time within a single
signal.
Tonality (T). Tonal vocalizations are produced by structures that
vibrate periodically (regularly). A tonal sound may be a simple
wave, exhibiting a single frequency. This basic rate of vibration
is the fundamental frequency (F0), which gives rise to the percep-
tion of the sound’s relative pitch. But in the majority of biological
systems, the oscillating structure generates a complex wave with
a series of harmonic overtones in addition to F0. The overtones
are integral multiples of F0. Harmonic overtones in particular fre-
quency ranges may have differing levels of energy; such bands are
referred to as formants. Harmonic complexity and formant struc-
ture contribute to the overall auditory impression of a sound—its
timbre. The timbre of a simple-wave (or narrowband) tonal sound
is often described as “pure”; the more harmonically complex (or
broadband) the tonal sound, the richer its timbre as perceived by a
(human) receiver.
Barking typically contains a tonal component in which the
fundamental frequency and a set of harmonic overtones, usually
broadly distributed over the spectrum, are discernible in the signal
(Bleicher, 1963; Lehner, 1978; Morton, 1977; Pongrácz et al., 2005).
This tonality may persist throughout the call, occur only in a por-
tion of the signal, or may be masked by noise (Fig. 1); in some cases,
a tonal component is absent.
Noise (N). Noisy sounds are generated by structures that vibrate
irregularly (aperiodically), producing complex waveforms with
components at many non-harmonically related frequencies. By def-
inition noisy signals do not have a single fundamental frequency,
but dominant frequencies – regions of the acoustic spectrum with
the greatest energy – may produce an impression of relative pitch
in noisy sounds. Noise is often broadband, with frequency com-
ponents in a wide range of the acoustic spectrum, or narrowband,
concentrated in a particular frequency range. Noise may co-occur
in a signal with harmonically related tonality (Fig. 1).
Barking is described in the literature as a noisy signal containing
multiple non-harmonically related frequencies over a broad range
of the spectrum (Bleicher, 1963; Lehner, 1978; Pongrácz et al., 2005;
Schassburger, 1987). Some writers do not distinguish the bark from
the yelp (Feddersen-Petersen, 2000), but the yelp is purely tonal
with no noisy components. Our view is that the presence of noise
(typically though not always co-existing with a tonal component)
is a diagnostic feature of the canonical definition of a bark (Fig. 2).
Fig. 1. Bark of an adult male malamute dog. The abrupt onset is characterized by
an initial noisy burst (a); it is followed by a tonal component (b) with frequency
modulation evident in multiple harmonics; and some broadband noise (c) can be
seen throughout much of the signal, particularly in the upper frequencies. The barks
shown in this and the other two figures are illustrative of the phenomenon under
discussion and were chosen for those characteristics. They are not directly compa-
rable.
Fig. 2. Two barks from an adult female Anatolian guarding dog. Left: broadband
noise (a) occurs along with two distinct tonal harmonic components (b). Right: noise
dominates throughout the frequency range, with little if any tonality evident in the
signal.
Noise is characteristically present throughout the signal, or a noisy
period can precede a tonal period (Fig. 3).
Mean Pitch (MP). Pitch is the perception of frequency. Both
tonal and noisy signals give an impression of pitch. In tonal sig-
nals pitch is often a direct function of the fundamental frequency
(F0), the base frequency at which the vocal folds are vibrating.
Pitch may also be determined by harmonics in the upper frequency
range that have greater acoustic energy than the fundamental (for-
mant regions), hence pitch may be relative with respect to F0. In
Fig. 3. Pulse repetition in an adult female Maremma guarding dog. This series of
barks is characterized by regularlytimed inter-pulse intervals, and decreasing ampli-
tude. Note also the abrupt onset and initial noisy component (a) in the first pulse,
followed by a predominantly tonal frequency-modulated period (b) with multiple
harmonics.

360 K. Lord et al. / Behavioural Processes 81 (2009) 358–368
noisy signals the dominant frequency, or the frequency range at
which the most energy is present, also creates an impression of
pitch. For our purposes we define mean pitch as the mean fun-
damental frequency in signals containing tonal components and
the mean dominant frequency for signals that lack tonal compo-
nents.
Barking is highly variable in pitch. Feddersen-Petersen (2000)
reports F0 values varying from 120 to 1640 Hz, and dominant fre-
quency values ranging from 200 to 2360Hz in 6 breeds; Bleicher
(1963) reports a dominant frequency range between 200 and
3000 Hz. Sales et al. (1997) report a dominant frequency range
of 250–4000 Hz in the barking of kenneled dogs. This variation
is likely a function of the size, age and breed of an individual; it
may also relate to the animal’s motivational state (Morton, 1977).
Moreover, individual barks themselves typically vary in frequency
(FM). In general, barking is described in the literature as relatively
high-pitched. Bleicher (1963) reports a mean fundamental pitch of
650 Hz. In our own lab (Clemence, unpub. 1992) the fundamental
frequencies of 25 randomly sampled barks from each of 86 indi-
viduals (recorded on a Sony Professional Walkman recorder with
a Sony 0–15 kHz omnidirectional microphone) were extracted in
MacSpeech Lab II (GW Instruments, Camridge, MA) and analyzed
in Statview (SAS Institute, Cary, NC). We found an average funda-
mental frequency of 715Hz (SD115).
Frequency Modulation (FM). The frequency (hence relative
pitch) of a given tonal vocalization can vary over time (Fig. 3). These
modulations can be considerable, with rapid shifts in F0 over many
hundreds of Hz, or more gradual. Because harmonic overtone struc-
ture is integrally related to F0, modulation will affect these upper
frequencies as well. Although a very small degree of FM, or jitter,
is present in most biologically generated tonal signals, some sig-
nals are relatively “steady-state” or monotone, and exhibit minimal
modulation.
The bark often has a characteristic “chevron-shaped” modu-
lated frequency contour, with a relatively rapid rise and fall. Yin
and McCowan (2004) report frequency ranges of 766 Hz and higher
within a single bark; Schassburger (1987) also reports a wide intra-
signal frequency range. Monotone barks occur in some individuals,
but some degree of FM is the general rule.
4. Amplitude
Amplitude is a function of the amount of energy in the sig-
nal as emitted by the sender. It is not a constant, and may vary
over time within a signal, or when a call is repeated at brief
intervals. Moreover, amplitude decreases with distance by the
inverse-square law, therefore a signal heard 10m from its source
would be significantly louder than if it were detected at 100 m.
Amplitude can also be strongly affected by external factors such
as environmental conditions and habitat obstacles as well as
internal factors such as available respiratory volume and muscle
effort.
Relative Amplitude or Loudness (RA). Given that loudness, like
pitch, is a perceptual phenomenon and that this perception is
affected by the extrinsic physical factors noted above, it is appro-
priate to think in relative terms such as “softer” or “louder” when
characterizing amplitude. There is general agreement in the litera-
ture that barking is a relatively loud signal (Lehner, 1978; Sales et
al., 1997; Tod et al., 2005).
Abrupt Onset (AO). A signal can reach its maximum ampli-
tude gradually or suddenly. Sounds that are described as having
an abrupt onset are characterized by a steep “rise-time” for ampli-
tude. Some AO signals are produced by a build-up and rapid release
of pressure resulting in a very brief broadband burst of noise at
the beginning of the signal, often perceived as click-like. Non-AO
signals have a gradual rise-time.
Barks are reported to have abrupt or explosive onsets (Scott
and Fuller, 1965; Lehner, 1978; Schassburger, 1987; Robbins and
McCreery, 2003).
5. Duration
Duration is the time aspect of the signal, i.e. the period during
which the signal is emitted. The duration of biological signals may
be as brief as a few milliseconds, as long as several seconds or even
several minutes.
Pulse Duration (PD). The energy expended in a call may occur
in a single “pulse” of some length. For comparative purposes we
define the duration of a call in terms of the time in which a single
pulse is active.
A single bark is a short event, lasting from 100 to 500 ms (Scott
and Fuller,1965; Tembrock,1976;Lehner, 1978;Schassburger, 1987;
Robbins and McCreery, 2003; Yin and McCowan, 2004).
Pulse Repetition (PR). Calls with pulse repetition may repeat
with short silent inter-pulse intervals. The absolute duration of
pulse-repeating calls is longer than the basic value of PD for that
call. The rate of PR is a function of the number of individual pulses
per unit of time.
Dogs typically bark serially, with very short silent inter-pulse
intervals, which may vary in length both within and between ani-
mals. The rate of repetition can be very high: Scott and Fuller (1965)
for example report a cocker spaniel barking at 900 pulses per hour.
Increased barking rate, i.e., shorter pulse-interval,appears to be cor-
related with increased arousal level (Gould, 1983). But PR is not an
invariant property of barking: it is a “canonical” and typical fea-
ture of domestic dog vocalization, although single barks (pulses)
do occur in dogs. It is also important to note that PR increases infor-
mational reliability (redundancy) as well as the perceptual salience
of a signal.
6. The canonical form of the bark
What, then, are the acoustic features that distinguish the bark
from other types of vocalizations? We maintain the bark is a signal
that is always abrupt in onset and short in duration. It may exhibit
both tonality and noise in some degree; is relatively high-pitched;
is subject to frequency modulation; is relatively loud; and is sub-
ject to rapid repetition. Taken together, these properties uniquely
define the bark. We are aware of no other vocalization described in
the literature that generally exhibits all of these features.For simplicity
of exposition, we treat each of the acoustic parameters described
above as binary—marked either [+] or [−] depending on whether it
is present or meets a particular threshold value for that parame-
ter (granting that certain features are subject to variation). In this
framework the “ideal” bark may be described schematically as in
Table 1.
We refer tothis as the canonical bark—a specific range of intrinsic
acoustic properties and possibilities that define a particular signal
type occurring in the dog, related canids, and many other mammals
and birds. Of course not every observed instance of the bark will
exhibit all of these characteristics. We do not intend the notion of
“canonical bark form” to suggest that barking is a stereotyped phe-
nomenon, in the way that manyalarm calls or care-soliciting signals
may be stereotyped. The domestic dog bark is highly variable with
respect to some of the features noted above, as Yin and McCowan
(2004) and many others have observed. Some of these variations
arise from intrinsic anatomical differences between individuals.
Animals can also alter components of the signal under particu-
lar motivational or behavioral conditions. For instance, the binary
feature [+PR] in Table 1 indicates that pulse repetition is a typical
realization of the canonical bark. Indeed it may be a common behav-

K. Lord et al. / Behavioural Processes 81 (2009) 358–368 361
Table 1
Comparison of canonical barking with other Canis vocalizations.
Vocalization T N MP FM RA AO PD PR
Bark +++(≥600 Hz) ++(≥70dB) +–(<500 ms) +
Whine +a,f–a,f+a,d,g+a–b+a,b+d,f+a
Yelp +f–f+b,e+f+a,b+e–c,f+b
Growl –e,f+a,f–b,f–a,b,c–b–c,e+c,d–a
Howl +a,f–a,b,f–d,f,g+h+b–b+a,d,f+i
Woof (huff)2–b,e,f+b,e,f–f–b–b–b–b,f–b,i
aCohen and Fox (1976).
bLehner (1978).
cTembrock (1976).
dFeddersen-Petersen (2000).
eBleicher (1963).
fSchassburger (1987).
gTheberge and Falls (1967).
hMitchell et al. (2006).
iLord (pers. obs.).
ioral outcome. But whether a given animal emits a repetitive signal,
or the actual rate of repetition, will depend on specific conditions
of its production.
It is also important to separate the terms “bark” and “barking.”
To us, “a bark” is an ideal signal type that generally fits the canon-
ical form in Table 1, and is distinct from other calls in an animal’s
repertoire. But by “barking” we mean the actual engagement of this
signal type in behavior—and it is here that significant variation can
arise.
As noted above, other calls described in the literature differ
from the bark in just one or two canonical features. It is not sur-
prising that vocalizations like the bark, yelp, huff and cry share a
degree of perceptual similarity to human observers and all may be
referred to as “barks” (even reports in the literature may fail to dis-
tinguish amongst them, e.g., Feddersen-Petersen, 2000). Moreover,
because of the inherent variability of signals in actual behavior they
may appear to grade into one another. A particular bark with little
or no discernible tonality and relatively low amplitude may not
be readily distinguishable from a woof. A beagle cry or bay that
is made with some small degree of noise and a relatively short
pulse duration may not be readily distinguishable from a bark; a
bark made with little noise may be difficult to distinguish from
a yelp. Graded signals are not uncommon in mammalian vocal-
behavioral systems (Marler, 1967). We discuss this question further
along with motivational conflict in the acoustic structure of mob-
bing calls.
7. Who barks?
Barking, as defined above, is not unique to the domestic dog.
It is reasonably common in a variety of taxa. Signals consis-
tent with canonical barking are found in numerous non-canid
mammalian species including distantly related carnivores, cervids,
primates and rodents, and are common in avian species. The
presence of barking in the vocal repertoire of the wild canids is
not surprising, on the uncontroversial assumption that domestic
dogs and their wild counterparts share a relatively recent com-
mon ancestor. But phylogenetic homology (rather than analogy)
is not likely to be the explanation for the occurrence of barking
calls in distantly related mammals or birds. A broader compar-
ative approach helps to understand the nature of an apparent
convergence, and also helps to explain the function and biolog-
ical history of the barking signal. The examples in Table 2,as
well as our general conclusions about the occurrence of bark-
ing calls, are based on inspection of calibrated spectrograms
(sonograms) and acoustic descriptions that have appeared in the
literature.
8. Wild canids
Barking is reported in the vocal repertoire of almost all the wild
Canidae (Cohen and Fox, 1976), including wolves (Cohen and Fox,
1976; Schassburger, 1993; Feddersen-Petersen, 2000; McCarley,
1978), coyotes (Lehner, 1978; Mitchell et al., 2006; McCarley, 1975),
foxes (Cohen and Fox, 1976; Frommolt et al., 2003; Murdoch
et al., 2008; Brady, 1981; Darden and Dabelsteen, 2006), jack-
als (Tembrock, 1976), African wild dogs, Lycaon pictus (Robbins,
2000), bush dogs, Spethos venaticus (Brady, 1981), dingos (Corbett,
2004), and New Guinea singing dogs (Koler-Matznick et al., 2003;
Brisbin et al., 1994). Virtually identical to domestic dog barking
in its acoustic character, each species appears to exhibit a some-
what narrower range of variability in tonality/noise and mean
pitch. For example, wolf barking is often (but not always) relatively
noisy and low-pitched (Cohen and Fox, 1976; Schassburger, 1993;
Feddersen-Petersen, 2000). This reduction in acoustic variability
can be expected given that, unlike the domestic dog, adults within
wild canid species tend to be similar in size and shape.
9. Other mammals
The literature on mammalian vocal behavior is replete with
examples of vocalizations that display the acoustic features we
ascribe to canonical domestic dog barking. Representative cases
among primates are reported for squirrel monkeys, Saimiri sciureus
(Winter et al., 1966), the black-and-white ruffed lemur, Varecia var-
iegata (Pereira et al., 1988), female chacma baboons, Papio ursinus
(Fischer et al., 2001), and Garnett’s greater bush baby, Otolemur
garnettii (Becker et al., 2003). Digweed et al. (2005) describe an
alerting signal in the white-faced capuchin monkey, Cebus capuci-
nus, as “shorter and more plosive (abrupt-onset)” than a related
alarm call, “giving it a conspicuously bark-like quality”. It is typi-
cally given repeatedly in protracted bouts of calling. The capuchin
signal is both noisy and tonal.
Reby et al. (1999a) note that “most Cervinae species bark”, how-
ever, only a few have been described with enough detail to be
categorized as canonical barking. The barking vocalization of both
the Indian muntjac, Muntiacus muntjac (Wiles and Weeks, 1981;
Oli and Jacobson, 1995), and Chinese muntjac, Muntiacus reevesi
(Yahner, 1980), match our definition. Roe deer barking also corre-
sponds with the acoustic characterizations of barking (Reby et al.,
1999b).
Additional mammalian examples in diverse taxa include red
squirrels, Sciurus vulgaris (Greene and Maegher, 1998) and tree
shrews, Tupaia belangeri (Schehka and Esser, 2007). Mammals
such as the meerkat, Suricata suricatta (Graw and Manser, 2007;

362 K. Lord et al. / Behavioural Processes 81 (2009) 358–368
Table 2
Characteristics of the bark in non-canid mammals and bird species.
Species T N MP FM RA AO PD PR Context
Canonical bark + + ≥600 Hz + ≥70 dB + <50 0 ms + Conflict/mobbing
Squirrel monkey (yap)1,2 ++ ≥600 Hz*+ NA + 100–250 ms1+ Mobbing1
Black-and-white ruffed lemur (pulsed squawks)3++≥600 Hz*+ “High amplitude” + <400ms + Squawk at potential predators; approach and squawk in
response.3
Chacma baboon (female loud bark)4++ ≥600 Hz*+ Mean 69 dB + ∼200 ms + Do use in mobbing (response varies depending on context,
look for eliciting stimuli)4
Garnett’s greater bush baby (bark)5+ + 589 Hz + “Loud” + ∼400 ms + NA
White-faced capuchin (alerting call)6++ ≥600 Hz*– “Loud” + <150ms + Non referential predator call and mobbing call6
Indian muntjac (bark)7,8 + 588 ±43 Hz8+ “Loud” + 620ms ±64 ms8+ Potential predator (usually when it can’t tell what is
coming)7
Chinese muntjac (bark)9+ + 670–1120Hz + NA 270–510ms + Novelty (hypothesized to interrupt predator)9
Roe deer (bark)10 + + 1600–2500 Hz – “Loud” + <500 ms*+ Unknown disturbance (hypothesized Predator
deterrence)11
Red squirrel (bark)12 + + 30 00 Hz + “Loud” + 20–400 ms + Potential terrestrial predators
Tree shrew (high arousal threat squeak)13 + + 2313–2413Hz + “Loud” + <70 ms + Conflict13
American crow (inflected caw)14 + + “High pitched” + “High amplitude” + <200ms + Owls, hawks and humans14
Southern lapwing (mobbing call)15 + + >1000 Hz*+ NA + <500 ms + Mobbing15
Black-capped chickadee (mobbing call)16 + + >200 0 Hz + NA + <400 ms + Mobbing16
Mexican chickadee (mobbing call)18 – + 3000 Hz*– Loud*+200ms
*+ Mobbing18
Red vented bulbul (low pressure alarm call)17 + + >940 Hz + NA + ∼100ms + Potential threat17
Blue jay (mobbing call)18 + + 2000 Hz*+ Loud*+400ms
*+ Mobbing18
Titmouse (mobbing call)18 + + 1000Hz*+ Loud*+ 220 ms*+ Mobbing18
Western tanager (Mobbing call)18 + + 2200 Hz*– Loud*+100ms
*+ Mobbing18
Red-breasted nuthatch (Mobbing call)18 ++ 1500Hz
*+ Loud*+ 400 ms + Mobbing18
1Winter et al. (1966);2Fichtel et al. (2005);3Pereira et al. (1988);4Fischer et al. (2001);5Becker et al. (2003);6Digweed et al. (2005);7Wiles and Weeks (1981);8Oli and Jacobson (1995);9Yahner (1980);10Reby et al.
(1999a);11Reby et al. (1999b);12Greene and Maegher (1998);13Schehka and Esser (2007);14Yorzinski et al. (2006);15Walters (1990);16Hurd (1996);17Walters (1990);18Kroodsma (2004).
*Estimates based on spectrogram.

K. Lord et al. / Behavioural Processes 81 (2009) 358–368 363
Manser, 2001), raccoons, Procyon lotor (Sieber, 1984) and Gunni-
son’s prairie dogs, Cynomys gunnisoni (Kiriazis and Slobodchikoff,
2006), are reported as displaying vocalizations that seem to match
barking, but are not described in enough detail to categorize
them definitively. We suspect that a similar degree of variability
may be characteristic of barking behavior in general—for rea-
sons that will become clear in our discussion of the function of
barking.
10. Birds
While the descriptive term “barking” is usually associated with
mammals, signals with the canonical features of barking occur
widely in birds; the onomatopoetic label “barking” is often used
to describe them. Avian barking shares the features [+AO, +RA, +PR,
+FM, −PD], but as is the case with dogs and other mammals the val-
ues of [N] and [T] (harmonic-to-noise ratio) and mean pitch [MP]
are variable. Many authors (e.g. Marler, 1955; Leger and Carroll,
1981) have noted this convergence on canonical barking features
across a wide range of avian species, especially in the context
of predator-avoidance behavior. Morton (1977) notes the acous-
tic (and motivational) similarity between mammalian barking and
certain avian calls.
Among the numerous birds in many taxa that exhibitbarking are
the American crow, Corvus brachyrhynchos (Yorzinski et al., 2006);
the lapwing, Vanellus vanellus (Walters, 1990); the black-capped
chickadee, Parus atricapillus (Hurd, 1996; Ficken et al., 1976); and
the red-vented bulbul, Pycnonotus cafer (Kumar, 2004). Additional
examples, from Kroodsma (2004), include the blue jay (Cyanocitta
cristata); the tufted titmouse, Baeolophus bicolor; western tanager,
Piranga ludoviciana; the Mexican chickadee, Poecile sclateri, and the
red-breasted nuthatch, Sitta canadensis. A number of other birds
appear to bark, but lack detailed acoustic descriptions—such as the
Carib grackle, Quiscalus lugubris (Griffin et al., 2005); the greater
racket-tailed drongo, Dicrurus paradiseus (Goodale and Kotagama,
2006); the recently discovered Ecuadorian antpitta, Grallaria ridge-
lyi (Krabbe et al., 1999); and the gray catbird, Dumetella carolinensis
(Hurd, 1996).
11. Why barking?
Animals produce sounds for many reasons. Some are com-
municative signals involving a transfer of information between a
sender and a receiver that confer a selective advantage on both.
These may be heritable products of natural selection, e.g., innate
alarm or care-soliciting calls. But animals also produce vocaliza-
tions that have no primary adaptive communicative value for a
sender or receiver. Biological sounds may arise, for instance, as
a by-product of a normal physiological function such as relaxed
breathing. Canid moans may be sounds of this sort (Koler-Matznick
et al., 2005). Likewise, sounds such as yelps and moans may be
emitted by animals in the context of acute pain because a common
mid-brain region, the periaqueductal gray (PAG), aids aspects of
vocalization as well as pain reception and modulation (Behbehani,
1995). In such cases there is often no receiver, and no significant
functional advantage is gained by vocalizing. But even without an
intrinsic communicative function, these vocalizations may some-
times adventitiously produce a beneficial response, as when a loud
yelp of pain from a prey animal incidentally provokes a startle
response in a predator, allowing the prey animal to escape. Thus
they may be become exaptations in the sense of Gould and Vrba
(1982) and Gould (2002). Exaptations arise when an existing adap-
tive trait or developmental outcome is recruited (or “co-opted”)
to a new and perhaps quite distinct functional end with no inter-
vening process of selection. Finally, wholly novel signals may be
learned as a consequence of experience (and perhaps “culturally”
transmitted).
The wide-spread occurrence of the barking signal across many
(unrelated) taxa may thus be plesiomorphic, i.e., an evolutionary
trait that is homologous within a particular group of organisms but
not unique to members of that group. Alternatively, it may be a con-
sequence of convergent selection pressures or exaptive solutions.
Or, multiple explanations may be at work in different taxa.
12. Function of barking
Barking occurs in a remarkable range of contexts, and barking
has been described as a “hypertrophied” behavior in dogs compared
with the wild canids (Cohen and Fox, 1976). They bark near feeding
time and they bark, sometimes incessantly, when left alone. They
bark when it is time for a walk, they bark at approaching human
strangers, on recognition of familiar humans, at cars (both familiar
and unfamiliar) coming up a road, at sudden changes in the environ-
ment (wind noise, a bright moon, detection of an odor of interest),
in the presence of conspecifics and non-conspecifics alike, and in
many situations where there does not appear tobe an eliciting envi-
ronmental signal. Some dogs will bark in a particular location, daily,
over years, with no evident external stimulus. And barking itself –
or sounds like the hooting of an owl – may be the stimulus, as one
animal triggers a chain response of barking in others.
The apparent multiplicity of contexts associated with dog bark-
ing might seem to mitigate against assigning any single function
to barking. This (among other observations) led Coppinger and
Feinstein (1991) to doubt that dog barking represents a single adap-
tive consequence of natural selection. Certainly there are cases of
single animal signals serving multiple functions in multiple con-
texts. Bird song is a prime example, playing both territorial and
reproductive roles – excluding males while attracting females –
in many species (Searcy and Nowicki, 2008). Nevertheless it is
unusual for a single animal signal to have as wide an array of dis-
tinct contextual triggers as dog barking. The multiplicity of contexts
has suggested to some researchers (including Feddersen-Petersen,
2000; Pongrácz et al., 2005, 2006; Yin and McCowan, 2004; Yin,
2002) that dog barking might possibly function as a form of ref-
erential communication, with each context having an acoustically
different bark, both intra- and inter-specific (between dogs and
humans) (see Section 20 for further discussion).
13. Mobbing behaviors
We propose here a simpler functional explanation—barking is
a signal associated with mobbing behavior. Its canonical acoustic
shape and contextual variability are consequences of the functional
requirements and motivational states that underlie mobbing.
Widely reported in birds and in many mammals, mobbing is
defined as a form of cooperative anti-predator behavior (Caro,
2005). It is elicited by the approach of a predator or an unknown
stimulus, including unfamiliar members of the same species or
other species: more generally, an “intruder”. Mobbing involves
multiple group behavioral responses and is characterized by con-
spicuous displays: rapid and abrupt movement, and on occasion
joint physical attack (Table 2). It is typically initiated by a sin-
gle individual who has first detected an intruder, and is signaled
by means of vociferous and conspicuous vocalization. Mobbing
calls are received by both the intruder and the sender’s con-
specifics (Klump and Shalter, 1984). Conspecifics may respond by
approaching the sender, joining in the production of the mobbing
vocalization, and repeatedly approaching and withdrawing from
the intruder. The intruder’s approach (often predatory) behavior is
interrupted.

364 K. Lord et al. / Behavioural Processes 81 (2009) 358–368
Mobbing tends to occur in situations where the animal has con-
flicting motivations: when an individual is motivated to escape but
also to stand its ground (e.g., when a parent, at a den site with
offspring, confronts a predator). Conflict can also be elicited when
the animal is not fearful enough to run (e.g., when an intruder is
detected but has not yet come closer than the animal’s flight dis-
tance, or there is not enough information to determine the threat of
the intruder), or when it is physically constrained in some way and
cannot engage in normal approach/withdrawal responses. There-
fore, mobbing calls may regularly occur in situations of conflict
that are not in response to actual or even perceived predators (See
Section 19 for further examples).
14. The acoustic structure of mobbing calls
Why then have so many different species converged on barking-
like signals in mobbing contexts? The literature suggests two
possibilities. First, from a functional acoustic standpoint, mobbing
vocalizations need to be highly salient and easily localized. Marler
(1955) observed that mobbing calls tend to be characterized by
wide frequency-spectra ([+N] and/or [+FM] in our terms), sharp
onset [+AO], brief duration [−PD], high amplitude [+RA], and rapid
and persistent repetition[+PR]. All these characteristics are well-
suited for detection and localization by conspecifics and also for
attention by the intruder. Ficken and Popp (1996), in an acoustic
analysis of 52 mobbing calls in passerine birds, report considerable
variation in the details of acoustic structure of calls across species
but virtually all are readily detectable and localizable. The majority
of these calls share all or most of the acoustic properties of canonical
barking.
Second, there is a motivational basis for the acoustic structure,
illuminated by the motivation-structural (MS) theory of Mor-
ton (Morton, 1977; Owings and Morton, 1998). MS theory holds
that factors shaping internally motivated vocalizations have phy-
logenetically deep adaptive roots, and that close-contact signals
sub-divide along two major acoustic axes: harshness (harmonic-
to-noise ratio), and frequency. The end-points of these axes reflect
specific groups of motivational states. High frequency and tonal-
ity (without noise) correlate with affiliative behaviors, including
appeasement, submission and care-solicitation, all of which are
associated with approach on the part of the receiver. Low frequency
and noise (with no tonal component) correlate with aggressive
behaviors, including dominance and threat, and are associated with
withdrawal by the receiver. Morton supports his hypothesis with
a survey of vocalizations from a large variety of birds and mam-
mals. Although factors such as sensory and ecological constraints
on perceptibility and transmissibility may sometimes take prece-
dence, Morton’s motivational rules have been largely supported
by further research on both birds and mammals, including the
canids: Cleveland and Snowdon (1982),Sieber (1984),August and
Anderson (1987) and Robbins and McCreery (2003), among oth-
ers.
MS rules potentially impose significant constraints on signal
form: high-frequency vocalizations, for example, are functionally
inconsistent with aggression, since this form tends to attract rather
than deter a threat to the signaler by encouraging approach. By
the same logic, low-frequency noisy vocalizations are inconsistent
with care-solicitation behavior. But, crucially for our purposes, the
MS model predicts that there can also be signals that tend toward
the midpoints of the two motivational axes—e.g., which rise and
fall in frequency, and/or which are not exclusively tonal or noisy.
The bark is a signal of precisely this sort. Morton (1977) suggests
that such “composite” or “midpoint” forms indicate motivational
conflict:
“Presumably, a sound indicating ambivalence, such as occurs in
mobbing behavior (e.g., Andrew, 1961), may acquire a steep slope
so as to become nearly a pulse if selection pressure derived from
the sound’s function favors qualities that enhance the sender’s loca-
tion by the receiver (Marler, 1956)...In mammals, the intermediate
structure tends to be frequency constant but still short or abrupt,
and the sounds are termed barks or grunts. For both birds and mam-
mals, this sound type indicates the sender is indecisive (i.e., it may
either go toward or away from or become more or less aggressive
or appeasing toward the stimulus), usually because the stimulus
is too far from the sender for it to make an adaptive response” (p.
861–862).
A motivationally ambivalent composite signal like barking is
thus ideally suited to mobbing (Owings and Morton, 1998). It
may be viewed as directed at two different receivers: conspecifics
and intruders. A tonal component [+T], and higher pitch [+MP]
encourage the approach of conspecifics (and also non-conspecifics
who convergently recognize the signal) to join in the collec-
tive deterrence behavior. A noise component [+N] and lower
pitch [−MP] encourages withdrawal in an intruder. Along with
perceptually salient features such as abrupt onset [+AO], high
amplitude [+RA] and pulse repetition [+PR], these characteris-
tics are also likely to attract and engage an intruder’s attention,
a condition under which predatory sequences are often dis-
rupted (Clark, 2005; Woodland et al., 1980; Zuberbühler et al.,
1999).
This motivational perspective also explains variation in the bark
itself (Owings and Morton, 1998). An animal always has some moti-
vation for signaling and the character of its vocalization is often a
reflection of motivational state. But a bark is by its nature a reflec-
tion of conflict—a struggle between differing motivational states.
The conflict can be significant, e.g. between hostile (lower pitch and
more noise) and soliciting (higher pitch and less noise). Moreover,
it can be expressed while in a state of higher arousal (with high
pulse repetition, high amplitude, high pitch), or of lower arousal.
The degree of conflict itself can even vary. We suspect that some
features of the bark, such as its invariantly abrupt onset and short
duration of individual pulses, may not be directly related to moti-
vational state. But those acoustic characteristics that are indeed
correlated with motivation can vary together (or separately) and
can fall anywhere along their individual scales, generating a high
degree of potential variability. By contrast, the bark’s canonical form
and potential for variation are distinct fromsignals such as the care-
soliciting, high-arousal yelp (pure tonal, high pitch,high amplitude)
and the low-arousal but aggressive woof (entirely noisy, low ampli-
tude, low pitch). These relatively invariant vocalizations have fixed
functions and do not signal conflicting degrees of hostility, solicita-
tion or arousal.
15. Mobbing in wild Canis
If the dog bark is homologous with mobbing calls we would
expect that dogs’ closest relatives would bark in mobbing contexts.
Therefore the genus Canis is of particular interest. All members of
the genus Canis (including dogs) are karyotypically identical and
can reproduce fertile offspring with one another (Chiarelli, 1975).
Although the origin of the dog is continuously debated, it is obvious
that the form evolved recently from some generalized species of
Canis (Coppinger et al., 2009).
Though he does not explicitly call it a mobbing vocalization,
Schassburger (1987) suggests that wolf barking might function to
signal a “call to arms of distant pack members” and to elicit the
“withdrawal of intruders”. McNay (2002) conducted a review of
wolf-human interactions in Alaska and Canada from 1900 to 2001,
and found that in all cases where wolves were documented as
defending either a den or rendezvous site, they barked loudly.
He also notes that wolves often ran towards and then away from
humans near dens. These descriptions closely fit the classical

K. Lord et al. / Behavioural Processes 81 (2009) 358–368 365
picture of mobbing behavior, and are strongly suggestive of moti-
vational conflict.
Lehner (1978)reports that when coyotes are approached at their
den sites they typically use the “woof”, a low amplitude noisy signal
that induces pups to hide, and then run a considerable distance
from the den site and bark repeatedly. Lehner also suggests that
coyote barking occurs during hunting, agonistic interaction and in
territorial displays, but his descriptions are not specific enough to
determine if the vocalizations fit the canonical form of the bark
as we have defined it, or if they are associated with mobbing-like
behavior.
While descriptions of the context of barking in jackals are not
detailed, barking is regularly reported at the den site. Loveridge
and Nel (2004) report that black-backed jackals (Canis mesome-
las) bark when threatened at the den site. Estes (1991) reports that
golden Jackals (Canis aureus) will growl and bark when they sense
danger near the den. Moehlman (1983) states that a single adult
jackal (either black-backed or golden) at a den can protect pups
from intruders by growling and barking.
16. Mobbing in domestic dogs
A fundamental question about barking in the domestic dog is
why it is so much more frequent than in other members of Canis.
In light of the mobbing hypothesis advanced here, we rephrase
the question: Why do dogs seem to be (or to perceive them-
selves to be) so frequently beset by intruders? In one respect, the
answer seems transparent: mobbing is induced when intruders
approach places like den sites (a territory) where escape behav-
ior is inhibited or impossible. It could be argued that the domestic
environment dramatically increases the number of situations that
elicit mobbing and its associated vocal behavior (Convergent with
Yin, 2002).
As noted earlier, mobbing vocalizations are produced when an
animal is in conflict, unable to escape an intruder (real or perceived).
For wolves this tends to occur when they are in the presence of
their offspring and threatened by an intruder. Dogs are routinely
confined or constrained with no opportunity for escape. Kept in
a kennel, a crate, a house, a fenced yard or tied up, they cannot
run from approaching unfamiliar “intruders”, who are virtually
omnipresent in human environments. The relatively close living
quarters of captive dogs thus facilitates group vocal response to a
mobbing signal, accounting for the cacophony that often followsthe
initial barking of a single animal. It should be noted that free-living
“village dogs” in non-western societies, described by Coppinger
and Coppinger (2001), are not constrained or confined and exhibit
much lower levels of barking. Boitani et al. (1995) noted that free-
ranging village dogs rarelybarked or approached humans or strange
dogs except in the core of their territories. Ortolani et al. (in press)
found village dogs in Ethiopia significantly more likely to vocalize if
approached while in a house or constrained than when approached
on a street.
The less fearful an animal, the more likely it will hold its ground
(Stankowich and Coss, 2007) and produce a mobbing vocalization
rather than a simple alarm call and flight (Knight et al., 1987). Calling
rates increase with decreasing distance to a predator (Curio and
Regelmann, 1985). The shorter an animal’s flight distance – that is,
the closer an animal allows an intruder to get – the more likely it is
to mob and vocalize.
The relationship between decreased flight distance and
increased barking is also supported by the work of Belyaev (Trut,
1999) who bred Siberian silver foxes (Vulpes vulpes) specifically
for decreased flight distance in order to improve their tractabil-
ity during handling. After 30 generations the offspring from
this line of foxes displayed many unexpected morphological and
behavioral phenotypes associated with domestic dogs, includ-
ing reduced flight distance. They also exhibited a significantly
increased tendency to bark, compared with non-selected ani-
mals.
Graw and Manser (2007) suggest that along with deterring
predators, mobbing may also allow animals to approachand inspect
novel stimuli to determine if they are threatening. By mobbing,
they not only avoid being surprised by an actual predator, but they
can also inspect the intruder. Curio et al. (1978) has shown that
the target of mobbing in captive European blackbirds is cultur-
ally transmitted. A “teacher bird”, which appeared to be mobbing
a non-threatening novel object, taught the native birds to mob
non-threatening objects. Other species such as capuchin monkeys
(Digweed et al., 2005) have been regularly observed to mob animals
that do not resemble known predators.
The hypothesis is that barking (as a component of mobbing
behavior) enhanced the fitness of the canid ancestor and is retained
in descendent populations of dogs. Hence barking, or at least the
disposition to bark, must be genetically transmitted. In our own
observations of congenitally and profoundly deaf dogs, we see evi-
dence that the form of the bark itself is inherited. These animals
first exhibit barking behavior at precisely the same developmental
stage as normally hearing animals, and they participate in barking
bouts with other dogs which are essentially indistinguishable from
that of normal dogs of the same breed and size.
However, we do not believe that selection pressure has led to
hypertrophied barking in the dog. Rather, the increased frequency
of dog barking is a consequence of a domestic environment (in
which conflict and novel stimuli are commonplace) and the process
of domestication. Not only aredogs more likely to be placed in a con-
flicting situation as a result of being artificially restrained, but dogs
are also more likely to place themselves in a conflicting situation.
As noted earlier, decreased fear of novelty – which occurred during
the domestication and/or development process – encourages mob-
bing behavior. Because dogs are less likely to run from novelty than
wolves, dogs are more likely to put themselves in a conflicting sit-
uation even when they have the option of escape. These factors not
only have an immediate effect on the display of barking behavior
in the adult, but also on its development.
Adult behavioral complexes including social behaviors such as
mobbing first begin to appear during the juvenile period between
approximately 4 and 8 weeks. At onset, these patterns occur out
of context, and may be performed along with components of
behavioral sequences persisting from earlier periods in ontogeny
(Burghardt, 2005). Juvenile “play” has been interpreted as the mix-
ing and repetition of motor patterns during this developmental
period (Coppinger and Smith, 1990; Burghardt, 2005). By experi-
menting with onsetting behavioral motor patterns, juveniles learn
combinations that are rewarding when performed in specific con-
texts.
Earlier display can lead to greater frequency of a display (hyper-
trophy). While the onset of barking in dogs and wolves is the same
(dogs, 18–24 days: Ohl, 1996; Bleicher, 1963; wolves, 11–28 days:
Frommolt et al., 1988; Harrington and Asa, 2003; Lord, pers. obs.)
the frequency of display is also dependent on environmental fac-
tors. Decreased fear and increased exposure to novelty would give
dogs more opportunity to incorporate barking into other behav-
ior sequences increasing the likelihood that barking will be more
broadly represented in the adult behavioral repertoire. In recent
observations in our lab, both wolf and dog pups participate in
raucous barking during play, although it occurs with much more
regularity in the former. In addition, dogs, like any other canid, can
learn to use any behavior in a new situation provided that it is rein-
forced. It is likely there are numerous circumstances that induce
conflict in the dog and elicit barking and subsequent attention from
human beings.

366 K. Lord et al. / Behavioural Processes 81 (2009) 358–368
17. Other perspectives
The mobbing hypothesis offers a simple structural and func-
tional account of barking behavior in dogs, related canids, and other
mammals and birds. This is important, given the recent attention
in the literature to hypotheses that attribute greater complex (cog-
nitive) functions to barking in the domestic dog. We have noted
the considerable evidence that the acoustic structure of dog bark-
ing varies with context; some authors have further suggested that
such variation may have a referential function. Feddersen-Petersen
(2000),Yin (2002) and Yin and McCowan (2004) all address this
question directly or indirectly. Several authors (Pongrácz et al.,
2005, 2006) argue that barking has context-specific effects on
both canine and human receivers, and speculate that barking was
an adaptation to canid-human communicative requirements dur-
ing the course of canid domestication. Because these approaches
include some of the most extensive and acoustically detailed inves-
tigations of barking to appear in the literature in recent years –
and because they offer accounts that compete with the mobbing
hypothesis as general explanations for barking – we discuss them
in some detail.
18. Referential communication
Biological signals may be components of functional behavior,
expressions of an organism’s internal state, or they may in some
cases be referential signs that relate to an event or property of the
external world. Referential signals are emitted under the stimulus
of particular objects or events, and therefore will vary with context
(Evans, 1997). Such signals are adaptive for both the sender and
receiver (Marler, 1967), providing information about the world that
can facilitate an adaptive response (Seyfarth and Cheney, 2003). By
contrast, non-referential signals are adaptive for the sender and
may initiate an adaptive change in the receiver’s behavior, but nei-
ther the signal nor the response is specific to a particular eliciting
stimulus or context (although the signal may vary with the nature
and intensity of the sender’s internal state).
19. Is (any) mammalian barking referential?
Referential alarm calls have been reported in the literature.
Seyfarth and Cheney (2003) observed that vervet monkeys have
acoustically distinct alarms for predatory mammals, birds and
snakes. Similar claims have been made for a number of other mam-
mal and bird species (Marler et al., 1992; Evans, 1997; Hauser, 1997).
The white-faced capuchin produces two acoustically distinct anti-
predator calls: one is elicited by avian predators and the other by
potential terrestrial predators. The avian call – acoustically dissim-
ilar to the bark – appears to be referential, while the terrestrial call,
a bark-like signal, does not (Digweed et al., 2005). The avian call
is produced only when the monkey sees low flying or diving birds
that resemble predatory raptors. The barking call is produced in
a much broader array of contexts including to potential terrestrial
predators and non-threatening mammals such as coatis and pecca-
ries that do not resemble predators or conspecifics. It may be noted
that these cases do not fit the standard picture of mobbing con-
texts. Rather, they are all potentially conflicting situations in which
the animal may not know what reaction is called for (on approach
of an unknown conspecific group) or may not be able to retreat
(when constrained by humans or a large number of other animals).
Thus the internal state of the caller is much the same as in a mob-
bing situation. The response to the bark call is also variable and
depends upon the eliciting stimuli. On some occasions monkeys
respond by running away; on others they respond by mobbing. In
both cases monkeys first confirmed the threat visually before pro-
ducing a response (Digweed et al., 2005; Fichtel et al, 2005). Thus,
barking does not signal conspecifics to mob, but attracts the atten-
tion of conspecifics who need to see the eliciting stimuli before
respond to it. Except for dogs (see below), there is no claim in the
literature for barking being referential.
20. Is dog barking referential?
Several authors (Feddersen-Petersen, 2000; Yin, 2002; Yin and
McCowan, 2004) noted that the acoustic structure of dog barking
varied with context. Feddersen-Petersen noted different subsets of
subcategories in the repertoires of different breeds, but most sig-
nificantly found that warning, threat, and play-fighting barking was
relatively noisy, while play-solicitation barking was more tonal. Yin
(2002) found that the acoustic structure of barking varied with con-
text: play and isolation barking were higher in frequency and more
tonal than disturbance barking. Yin and McCowan(20 04) found that
disturbance barking was low-pitched and noisy, with little mod-
ulation; isolation and play barking were high-pitched, tonal, and
more highly modulated. They also found that inter-bark intervals
of disturbance barking were shortened. (In some cases the inter-
bark interval was so short it was unperceivable to the human ear,
producing what the authors referred to as ‘superbarks’).
While these works contribute to the data on signal variabil-
ity, they do not provide evidence that barking has a referential
function. The contextual cases are entirely accounted for by Mor-
ton’s rules and arousal levels. Moreover, the behavioral contexts in
which barking occurs in these studies are consistent with the mob-
bing hypothesis and auxiliary assumptions about development and
exaptation.
Pongrácz et al. (2005) examined whether humans could deter-
mine the context of a given barking sequence in a playback
experiment. The authors played recordings of barks from 19 dogs
all from a single breed, the Mudi. The barks were recorded in six
distinct situations: presence of a stranger; “schutzhund”, where a
trainer acted menacingly and encouraged the dog to bite a padded
arm band; going for a walk; isolation; presence of a ball; and “play”.
Human subjects, split into three groups (mudi owners, other dog
owners, and non-owners) listened to 3 different barking sequences
from each context (for a total of 18 barks). They were asked to rate
the level of aggressiveness, fearfulness, despair, playfulness, and
happiness of the barking on a scale from 1 to 5, and to categorize
each barking into one of the six situations in which the barking was
originally recorded.
The authors found that each group was able to assign the barks
to their appropriate categories approximately 40% of the time, this
rate is better than chance (or 16.67% or 3 out of 18 cases). By
combining the data of all three groups Pongrácz et al. found that
58.33% of stranger barks, 48.15% of the schutzhund barks, 23.15%
of the walk barks, 47.22% of the alone barks, 25% of the ball barks,
and 37.04% of play barks were categorized correctly. The authors
conclude that dog barking has been selected to communicate infor-
mation to humans, and suggest a process of co-adaptation between
dogs and humans in the course of domestication.
The ability to guess the context of a dog’s barking with 23.15%
– or even 58.33% – success may be statistically significant, but cer-
tainly does not suggest that barking is a co-adaptation between
dogs and humans, let alone that it is systematically referential.
Rather we may conclude from the authors’ data that Morton’s rules
– plesiomorphic in the mammals – were at work both for dogs and
humans. In other words these results showthat humans can discern
that barks are generally more fearful or more aggressive, but they
cannot tell what stimuli is eliciting the bark, therefore they cannot
respond adaptively without further visual information.
Pongrácz et al. (2006) went on to investigate whether humans
were in fact using Morton’s rules to identify the emotional

K. Lord et al. / Behavioural Processes 81 (2009) 358–368 367
content of barking. The same three groups of listeners were used
as in the previous paper (mudi owners, other dog owners, and
non-owners). The investigators recorded barks from the same six
situations (stranger, dog attacks human, walk, alone, ball, and play).
However, these barks were then rated as having high, medium, or
low pitch and high, medium or low levels of noisiness. The barks
were placed into one of nine subsets based on these characteris-
tics (e.g. high pitch, low noisiness; high pitch, medium noisiness;
high pitch, high noisiness). Ten barks from each subset were played
back with either short, medium, or long interbark intervals. This
arrangement resulted in 27 different structural possibilities, which
were played to the human subjects for rating. Human subjects were
once again asked to rate barking on a scale of 1–5 for the same five
emotional components (aggressiveness, fearfulness, despair, play-
fulness and happiness). The results showed that the importance
of structural components to human listeners varied with context
(e.g. “aggression” scores relied on pitch and inter-bark interval,
while “despair” relied on pitch, inter-bark interval, and tonality).
In every case, emotional ratings were consistent with Morton’s
rules. It is unclear how anthropocentric emotional categories such
as despair and happiness fit into Morton’s continuum of aggres-
sion/affiliation.
Be that as it may, these results provide little basis for the claim
that there was selection for diversification of barking in dogs or
for referential content in those barks, either through selection by
humans for more understandable dogs or as an adaptation of dogs
to increase their ability to cooperate with humans.
In summary, Yin (2002),Yin and McCowan (2004), and
Feddersen-Petersen (2000) offer evidence that dog barking can
vary contextually. They show that robustly distinct barks (differing
from one another in noise/tonality ratio and inter-pulse interval, or
pulse-repetition rate) occur in distinct interactional and social set-
tings, including interactions with humans. Pongrácz et al. (2005,
2006) show that humans can pick up on some of these contextual
differences in playback studies. While the authors are all careful
to allow that these differences may arise from differing “affec-
tive” or motivational states (just as Morton would predict), they
also nevertheless suggest that contextual variations in the bark
may be “intentional”, i.e., representing information about the con-
text, rather than simply signaling the internal state of the sender
in a particular context. Yin and McCowan (2004:353) write that
“Co-variation between context and bark structure suggests that
dogs may perceive meaningful [our emphasis] differences between
contexts and adjust their barks accordingly” and Pongracz et al.
(2006:238) write that their “results do not exclude referential com-
munication”.
We would argue that these studies do not provide evidence that
barking in the dog supports a referential hypothesis, or that dog
vocalization has evolved to be a special case.
21. Conclusion
We have hypothesized that barking is associated with the func-
tional requirements and motivational states that underlie mobbing
(i.e. conflict, as suggested by Morton, 1977; Owings and Morton,
1998). In wild Canis barking occurs in this mobbing context. In dogs,
the underlying motivationalstate associated with barking occurs on
a daily basis. This increase in conflicting motivations leads directly
to increased barking as well as a greater developmental propen-
sity for the behavior. This mobbing hypothesis offers a conceptually
and empirically simple account of the structure and the function of
barking in the domestic dog and in a wide variety of other species.
Experiments testing for differences in development, eliciting stim-
uli, and environment on the quality, frequency and sequencing of
dog barking would be valuable.
References
Andrew, R.J., 1961. The motivational organization controlling the mobbing calls of
the blackbird (Turdus merula). I, II, III, and IV. Behavior17, 224–246;17, 228–321;
18, 25–43; 18, 161–176.
August, P.V., Anderson, J.G., 1987. Mammal sounds and motivation-structural rules:
a test of the hypothesis. J. Mammal. 68, 1–9.
Becker, M., Buder, E., Ward, J., 2003. Spectrographic description of vocalizations in
captive Otolemur garnettii. Int. J. Primatol. 24, 415–446.
Behbehani, M., 1995. Functional characteristics of the midbrain periaqueductal grey.
Prog. NueroBiol. 46, 575–605.
Bleicher, N., 1963. Physical and behavioral analysis of dog vocalizations. Am. J. Vet.
Res. 24, 415–427.
Boitani, L., Francisci, F., Ciucci, P., Andreoli, G., 1995. Population biology and ecol-
ogy of feral dogs in Italy. In: Serpell, J.A. (Ed.), The Domestic Dog: Its Evolution,
Behavior, and Interactionswith People. Cambridge University Press, Cambridge,
pp. 218–244.
Brady, C.A., 1981. The vocal repertoires of the bush dog (Spethos venticus), crab-
eating fox (Cerdocyon thous), and maned wolf (Chrysocyon brachyurus). Anim.
Behav. 29, 649–669.
Brisbin, I., Coppinger, R., Feinstein,M., Austad, S., 1994. The New Guinea singing dog:
taxonomy, captive studies and conservation priorities. Sci. N.G. 20, 27–38.
Burghardt, G.M., 2005. The Genesis of Animal Play. Bradford Books (MIT Press),
Cambridge, 501 pp.
Caro, T., 2005. Anti-predator Defenses in Birds and Mammals. University of Chicago
Press, Chicago, 591 pp.
Chiarelli, A.B., 1975. The chromosomes of the canidae. In: Fox, M.W. (Ed.), The Wild
Canids. Rober E. Krieger Publishing Company, Florida, pp. 40–53.
Clark, R., 2005. Pursuit-deterrent communication between prey animals and timber
rattlesnakes (Crotalus horridus): the response of snakes to harassment displays.
Behav. Ecol. Sociobiol. 59, 258–261.
Clemence, E., 1992.Vocalizations of domestic dogs (Canis familiaris): an investigation
of intraspecific variation. Unpublished Thesis, Hampshire College, Amherst MA.
Cleveland, J., Snowdon, C., 1982.The complex vocal repertoire of the adult cotton-top
tamarin (Saguinas oedipu oedipus). Z. Tierpsychol. 58, 231–270.
Cohen, J., Fox, M., 1976. Vocalizations in wild canids and possible effects of domes-
tication. Behav. Process. 1, 77–92.
Coppinger, R.P., Smith, C.K., 1990. A model for understanding the evolution of mam-
malian behavior. In: Genoways, H. (Ed.), Current Mammalogy,vol. 2. Plenum, NY,
pp. 335–374.
Coppinger, R., Feinstein, M., 1991. Hark! hark! the dogs do bark. and bark and bark.
Smithsonian 21, 119–129.
Coppinger, R., Coppinger, L., 2001. Dogs: A Startling New Understanding of Canine
Origin, Behavior and Evolution. Scribner, NY, 352 pp.
Coppinger, R., Spector, L., Miller, L., 2009. What if anything is a wolf? In The Worldof
Wolves: New Perspectives on Ecology, Behaviour and Management. University
of Calgary Press, Calgary, Canada, 412 pp.
Corbett, L., 2004. Dingo. In: Sillero-Zubiri, C., Hoffmann, M., Macdonald, D.W. (Eds.),
Status Survey and Conservation Action Plan—Canids: Foxes, Wolves, Jackals and
Dogs. Canid Specialist Group (CSG) of the International Union for the Conserva-
tion of Nature (IUCN), Gland, Switzerland, pp. 223–230.
Curio, E., Ernst, U., Vieth, W., 1978. Cultural transmission of enemy recognition: one
function of mobbing. Science 202, 899–901.
Curio, E., Regelmann, K., 1985. The behavioural dynamics of great tits (Parus major)
approaching a predator. Z. Tierpsychol. 69, 3–18.
Darden, S., Dabelsteen, T., 2006. Ontogeny of swift fox Vulpes velox vocalizations:
production, usage and response. Behaviour 143, 659–681.
Digweed, S.M., Fedigan, L.M., Rendall, D., 2005. Variable specificity in the anti-
predator vocalizations and behaviour of the white-faced capuchin, Cebus
capucinus. Behaviour 142, 997–1021.
Estes, R.D., 1991. The Behavior Guide to African Mammals Including Hoofed
Mammals, Carnivores, Primates. The University of California Press, Berkley,
611 pp.
Evans, C.S., 1997. Referential signals. Perspect. Ethol. 12, 99–143.
Feddersen-Petersen, D.U., 2000. Vocalization of European wolves (Canis lupus lupus
L.) and various dog breeds (Canis lupus f. fam.). Arch. Tierz., Dummerstorf 43,
387–397.
Fichtel, C., Perry, S., Gros-Louis, J., 2005. Alarm calls of white-faced capuchin mon-
keys: an acoustic analysis. Anim. Behav. 70, 165–176.
Ficken, M.S., Popp, J., 1996. A comparative analysis of passerine mobbing calls. Auk
113, 370–380.
Ficken, M.S., Ficken, R.W., Witkin, S.R., 1976. Vocal repertoire of the black-capped
chickadee. Auk 95, 34–48.
Fischer, J., Hammerschmidt, K., Cheney, D.L., Seyfarth, R.M., 2001. Acoustic features
of female chacma baboon barks. Ethology 107, 33–54.
Frommolt, K.-H., Goltsman, M.E., Macdonald, D.W.,20 03. Barkingfoxes, Alopex lago-
pus: field experiments in individual recognition in a territorial mammal. Anim.
Behav. 65, 509–518.
Frommolt, K.H., Kaal, M.I., Paschina, N.M., Nikolskii, A.A., 1988. Sound development
of the wolf (Canis lupus L. Canidae L.) during the postnatal ontogeny. Zool. Jb.
Physiol. 92, 105–115.
Goodale, E., Kotagama, S.W., 2006. Context-dependent vocal mimicry in a passerine
bird. Proc. Biol. Sci. 273, 875–880.
Gould, E., 1983. Mechanisms of mammalian auditory communication. In: Eisenberg,
J.F., Kleiman, D.G. (Eds.), Advances in The Study of Mammalian Behavior, special
publication no.7, Am. Soc. Mammal., pp. 265–342.

368 K. Lord et al. / Behavioural Processes 81 (2009) 358–368
Gould, S., Vrba, E., 1982. Exaptation—a missing term in the science of form. Paleobi-
ology 8, 4–15.
Gould, S., 2002. The Structure of Evolutionary Theory. Belknap Press of Harvard
University Press, Cambridge, 1464 pp.
Graw,B., Manser, M.B., 2007.The function of mobbing in cooperative meerkats. Anim.
Behav. 74, 507–517.
Greene, E., Maegher, T., 1998. Redsquirrels, Tamiasciurus hudsonicus, produce preda-
tor class specific alarm calls. Anim. Behav. 55, 511–518.
Griffin, A.S., Savani, R.S., Hausmanis, K., Lefebvre, L., 2005. Mixed-species aggrega-
tions in birds: zenaida doves (Zenaida aurita) respond to the alarm calls of carib
grackles (Quiscalus lugubris). Anim. Behav. 70, 507–515.
Harrington, F.H., Asa, C.S., 2003. Wolf communication. In: Mech, D., Boitani, L.
(Eds.), Wolves: Behavior, Ecology, and Conservation.University of Chicago Press,
Chicago, pp. 66–103.
Hauser, M.D., 1997. The Evolution of Communication. MIT Press, Cambridge, 771pp.
Hurd, C., 1996. Interspecific attraction to mobbing calls of black-capped chickadees
(Parus atricapillus). Behav. Ecol. Sociobiol. 38, 287–292.
Klump, G., Shalter, M., 1984. Acoustic behaviour of birds and mammals in the preda-
tor context. Z. Tierpsychol. 66, 189–226.
Kiriazis, J., Slobodchikoff, C., 2006. Perceptual specificity in the alarm calls of Gun-
nison’s prairie dogs. Behav. Process. 73, 29–35.
Knight, R.L., Grout, D.J., Temple, S.A., 1987. Nest-defense behavior of the American
crow in urban and rural areas. Condor 89, 175–177.
Koler-Matznick,J., Brisbin, I., Feinstein, M., Bulmer, S., 2003. An up-dated description
of the New Guinea singing dog (Canis hallstromi). J. Zool. (London) 261, 109–118.
Koler-Matznick, J., Brisbin, I., Feinstein, M., 2005. An Ethogram for the New Guinea
Singing (Wild) Dog (Canis hallstromi). The New Guinea Singing Dog Conservation
Society, U.S.A., 49 pp.
Krabbe, N., Agro, D.J., Rice, N.H., Jacome, M., Navarrete, L., Sornoza, F., 1999. A
new species of Antpitta (Formicariidae: Grallaria) from the Southern Ecuadorian
Andes. Auk 116, 882–890.
Kroodsma, D., 2004. Vocal behavior. In: Padulka, S., Rohrbaugh Jr., R.W., Bonney, R.
(Eds.), Handbook of Bird Biology. Cornell Laboratory of Ornithology, Ithaca, p.
22.
Kumar,A ., 2004. Acoustic communication in the red-vented Bulbul Pycnonotuscafer.
An. Acad. Bras. Cienc. 76, 350–358.
Leger, D.W., Carroll, L.F.,1981. Mobbing calls of the Phainopepla. Condor 83, 377–380.
Lehner, P.N., 1978. Coyote vocalizations: a lexicon and comparisons with other
canids. Anim. Behav. 26, 712–722.
Loveridge, A.J., Nel, J.A.J., 2004. Black-backed jackal (Canis mesomelas). In: Sillero-
Zubiri, C., Hoffmann M, Macdonald, D.W. (Eds.), Canids: Foxes, Wolves,
Jackals and Dogs Status Survey and Conservation Action Plan, Gland, Switzer-
land, and Cambridge, UK, IUCN, Gland, Switzerland, and Cambridge, UK, pp.
161–166.
Manser, M.B., 2001. The acoustic structure of suricates’alarm calls varies with preda-
tor type and the level of repsonse urgency. Proc. R. Soc. Lond. B 268, 2315–2324.
Marler, P., 1955. Characteristics of some animal calls. Nature 176, 6–8.
Marler, P., 1956. The voice of the chaffinch and its function as a language. Ibis 98,
231–261.
Marler, P., 1967. Animal communication signals. Science 157, 769–774.
Marler, P., Evans, C.S., Hauser, M.D., 1992.Animal signals: motivational, referential or
both? In: Papousek, H., Jurgens, U.,Papousek, M. (Eds.), Nonverbal Vocal Commu-
nication: Comparative and Developmental Approaches. Cambridge University
Press, Cambridge, pp. 66–86.
McCarley, H., 1975. Long-distance vocalizations of coyotes (Canis latrans). Am. Soc.
Mammal. 56, 847–856.
McCarley, H., 1978. Vocalizations of red wolves (Canis rufus). J. Mammal. 59, 27–35.
McNay, M.E., 2002. Wolf-human interactions in Alaska and Canada: a review of the
case history. Wildlife Soc. B 30, 831–843.
Mitchell, B., Makagon, M., Jaeger, M., Barrett, R.H., 2006. Information content of
coyote barks and howls. Bioacoustics 15, 289–314.
Moehlman, P.D., 1983. Socioecology of silverbacked and golden jackals (Canis
mesomelas and Canis aureus). In: Eisenberg, J.F., Kleiman, D.G. (Eds.), Recent
Advances in The Study of Mammalian Behavior. American Society of Mammal-
ogists, Lawrence, KS, pp. 423–453.
Morton, E.S., 1977.On the occurrence and significance of motivation-structural rules
in some bird and mammal sounds. Am. Nat. 111, 855–869.
Murdoch, J., Ralls, K., Cypher, B., Reading, R., 2008. Barking vocalizations in San
Joaquin kit foxes (Vulpes marcrotis mutica). Southwest. Nat. 53, 118–124.
Ohl, F., 1996. Ontogeny of vocalizations in domestic dogs, breed standard-poodle
(Canis lupus f. familiaris). Zool. Beit. 37, 199–215.
Oli, M.K., Jacobson, H.A., 1995. Vocalization of barking deer (Muntiacus muntjak).
Mammalia 59, 179–186.
Ortolani, A., Vernoij, H., Coppinger, R., in press. Ethiopian village dogs: behavioural
responses to a stranger’s approach. Appl. Anim. Behav. Sci.
Owings, D.H., Morton, E.S., 1998. Animal Vocal Communication: A New Approach.
Cambridge University Press, Cambridge, UK, p. 284.
Pereira, M.E., Seeligson, M.L., Macedonia, J.M., 1988. The behavioral repertoire
of the black-and-white ruffed lemur, Varecia variegata variegata (primates:
Lemuridea). Folia Primatol. 51, 1–32.
Pongrácz, P., Molnár, C., Miklósi, Á., 2006. Acoustic parameters of dog barks carry
emotional information for humans. Appl. Anim. Behav. Sci. 100, 228–240.
Pongrácz, P.,Molnár, C., Miklósi, Á., Csányi, V., 2005. Human listeners are able to clas-
sify dog (Canis familiaris) barks recorded in different situations. J. Comp. Psychol.
119, 136–144.
Reby, D., Cargnelutti, B., Joachim, J., Aulagnier, S., 1999a. Spectral acoustic struc-
ture of barking in roe deer (Capreolu capreolus). sex-,age- and individual-related
variations. C. R. Acad. Sci. Paris, Sciences de la vie 322, 271–279.
Reby, D., Cargnelutti, B., Hewison, A., 1999b. Contexts and possible functions of
barking in roe deer. Behaviour 57, 1121–1128.
Reide, T.,Fitch, T., 1999. Vocal tractlength and acoustics of vocalization in the domes-
tic dog (Canis familiaris). J. Exp. Biol. 202, 2859–2867.
Riede, T., Herzel, H., Hammerschmidt, K., Brunnberg, L., Tembrock, G., 2001. The
harmonic to-noise ratio applied to dog barks. J. Acoust. Soc. Am. 110,2191–2197.
Riede, T., Mitchell, B., Tokuda, I., Owren, M., 2005. Characterizing noise in nonhuman
vocalizations: acoustic analysis and human perception of barks by coyotes and
dogs. J. Acoust. Soc. Am. 118, 514–522.
Robbins, R.L., 2000. Vocal communication in free-ranging African wild dogs (Lycaon
pictus). Behaviour 137, 1271–1298.
Robbins, R.L., McCreery, E.K., 2003. African wild dog pup vocalizations with special
reference to Morton’s model. Behaviour 140, 333–351.
Sales, G., Hubrecht, R., Peyvandi, A., Milligan, S., Shield, B., 1997. Noise in dog ken-
nelling: is barking a welfare problem for dogs? Appl. Anim. Behav. Sci. 52,
321–329.
Schassburger, R., 1987. In: Frank, H. (Ed.), Wolf Vocalization: An Integrated Model
of Structure, Motivation and Ontogeny. Man and Wolf, Dr. W. Junk Publishers,
Dordrecht, Netherlands, pp. 313–347.
Schassburger, R., 1993. Vocal communication in the timber wolf, Canis lupus (Lin-
naeus). In: Advances in Ethology, Ser. 30. Structure, Motivation, and Ontogeny.
Paul Parey, Berlin, 84 pp.
Schehka, S., Esser, K.-H., 2007. Acoustical expression of arousal in conflict situations
in tree shrews (Tupaia belangeri). J. Comp. Physiol. A 193, 845–852.
Scott, J.P., Fuller, J.L., 1965. Genetics and the Social Behavior of the Dog. University of
Chicago Press, Chicago, 468 pp.
Searcy, W.A., Nowicki, S., 2008. Bird song and the problem of honest communication.
Am. Sci. 96, 114–121.
Seyfarth, R.M., Cheney, D.L., 2003. Signalers and receiversin animal communication.
Annu. Rev. Psychol. 54, 145–173.
Sieber, O.J., 1984. Vocal communication in raccoons (Procyon lotor). Behaviour 90,
80–113.
Stankowich, T., Coss, R.G., 2007. Effect of risk assessment, predator behavior, and
habitat on escape behavior in Columbian black-tailed deer. Behav. Ecol. 18,
358–367.
Tembrock, G., 1976. Canid vocalizations. Behav. Process. 1, 57–75.
Theberge, J.B., Falls, J.B., 1967. Howling as a means of communication in timber
wolves. Am. Zool. 7, 331–338.
Tod, E., Brander, D., Waran, N., 2005. Efficacy of dog appeasing pheromone in reduc-
ing stress and fear related behaviour in shelter dogs. Appl. Anim. Behav. Sci. 93,
295–308.
Trut, L., 1999. Early canid domestication: the farm fox experiment. Am. Sci. 87,
160–169.
Walters, J.R., 1990. Anti-predatory behavior of lapwings: field evidence of discrimi-
native abilities. Wilson Bull. 102, 49–70.
Wiles, G., Weeks, H., 1981. Barkingb ehaviorin Indian muntjacs (Muntiacus muntjak).
Am. Soc. Mammal. 62, 208–211.
Winter, P., Ploog, D., Latta, J., 1966. Vocal repertoire of the squirrel monkey (Saimiri
sciureus) its analysis and significance. Exp. Brain Res. 1, 359–384.
Woodland, D., Jaafar,Z., Knight, M.-L., 1980. The “pursuit deterrent” function of alarm
signals. Am. Nat. 115, 748–753.
Yahner, R., 1980. Barking in a primitive ungulate, Muntiacus reevesi: function and
adaptiveness. Am. Nat. 116, 157–177.
Yin, S., 2002. A new perspective on barking in dogs (Canis familiaris). J. Comp. Psychol.
116, 189–193.
Yin, S., McCowan, B., 2004. Barking in domestic dogs: context specificity and indi-
vidual identification. Anim. Behav. 68, 343–355.
Yorzinski, J.L., Vehrencamp, S.L., Clarke, A.B., 2006. The inflected alarm caw of the
American crow: differences in acoustic structure among individuals and sexes.
Condor 108, 518–529.
Zuberbühler, K., Jenny, D., Bshary, R., 1999. The predator deterrence function of
primate alarm calls. Ethology 105, 477–490.