‘dc

Aa

b

AAA

KFVPGSH

1

lotisb

sf

mm(

0h

Applied Animal Behaviour Science 148 (2013) 128– 137

Contents lists available at ScienceDirect

Applied Animal Behaviour Science

journa l h omepa g e: www.elsev ier .com/ locate /applan im

Beware, I am big and non-dangerous!’ – Playfully growlingogs are perceived larger than their actual size by theiranine audience

nna Bálinta, Tamás Faragób, Antal Dókaa, Ádám Miklósia,b, Péter Pongrácza,∗

Department of Ethology, Eötvös Loránd University, Budapest, HungaryMTA-ELTE Comparative Ethological Research Group, Hungary

a r t i c l e i n f o

rticle history:ccepted 27 July 2013vailable online 8 August 2013

eywords:amily dogocal communicationlay signalrowlize informationonesty

a b s t r a c t

Nonhuman animals often use specific signals to initiate playful interactions. There isevidence also for different forms of play-maintenance. Playful encounters include out-of-context and exaggerated behavioural sequences. Scientists have already collectedknowledge about virtual size modification via acoustic signalling in particular animalspecies during competitive/agonistic interactions, but the same was unknown in playfulencounters. Using the cross-modal matching paradigm, we tested whether dogs prefer tolook at the picture of a matching size dog when they are offered two differently sized pro-jected pictures simultaneously with a playback of a playful or a food-guarding growl. Wefound that dogs looked at the matching picture when they heard the food-guarding growl,but they looked at rather the larger than the matching size dog when play growls were

played back. These are the first results to show that dogs may communicate an exagger-ated body size by the means of their growls during play, which may help in maintainingor enhancing the playful interaction. As agonistic dog growls were proven to be honestregarding their referential and size-related information content, our results gave evidencethat exaggeration may work as play signal in the case of animal vocalizations.

. Introduction

The source-filter theory of vocal production (Fant, 1960)inks acoustic features of mammal calls to the anatomyf the vocal apparatus that creates them. Thus, vocaliza-ions have the potential to provide receivers with direct

nformation on the callers phenotype and/or motivationaltate (Charlton et al., 2010; Fitch and Hauser, 2003). It haseen shown, that the average spacing between succeeding

∗ Corresponding author at: Department of Ethology, Pázmány Péterétány 1/C, H-1117 Budapest, Hungary. Tel.: +36 1 381 2179;ax: +36 1 381 2180.

E-mail addresses: nani.balint.@gmail.com (A. Bálint),ustela.nivalis.@gmail.com (T. Faragó), dokaantal.@gmail.com (A. Dóka),iklosi62.@gmail.com (Á. Miklósi), peter.celeste.pongracz@gmail.com

P. Pongrácz).

168-1591/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.applanim.2013.07.013

© 2013 Elsevier B.V. All rights reserved.

formants (spectral peeks in the vocal signal), termed ‘for-mant dispersion’ is directly linked to the vocal tract length,which in turn is in close association with the overall bodysize and acts as an indexical cue in several mammalianspecies (e.g., dog (Canis familiaris): Riede and Fitch, 1999;american bison (Bison bison): Wyman et al., 2012; koala(Phascolarctos cinereus): Charlton et al., 2011). More closelyspaced formants reflect to a longer vocal tract, which isrelated to a larger body size (Taylor and Reby, 2010).

Despite the association between an acoustical signaland the caller’s physical parameters, the strength of thisrelationship might depend on certain anatomical adapta-tions, and the caller may be able to modulate the vocal

signal depending on the context. As a result, the vocal signalmay not reflect the callers’ real body size in certain situa-tions (Taylor and Reby, 2010). For example, in male red(Cervus elaphus) and fallow deer (Dama dama), the larynx

Behavio

A. Bálint et al. / Applied Animal

rests in an unusually low position in the neck, and it canbe further retracted in the throat during the productionof mating calls (Reby et al., 2005). Other mechanisms mayalso lead to this acoustical ‘exaggeration’; e.g., male saigaantelopes (Saiga tatarica) use a specific vocal posture whileproducing mating calls (Volodin et al., 2009), while ele-phant seals (Mirounga leonina) have an elongated nasalregion that is able to potentially influence the spacing offormants (Sanvito et al., 2007; Taylor and Reby, 2010).

Domestic dogs have a rich vocal repertoire (Cohen andFox, 1976). Some of their vocalizations, like barking, weresubject to considerable alterations due to the domestica-tion (see for review Pongrácz et al., 2010). Other typesof dog vocalizations, like growls are thought to be moreconservative, being used in similar contexts in dogs asin their wild relatives (Cohen and Fox, 1976). Accordingto the so-called structural–motivational rules of Morton(1977), specific acoustic parameters (like the fundamentalfrequency and the tonality) have well-defined relation-ship with particular inner states among a multitude ofmammalian species. For example, deep and atonal (hoarse)sounds refer to aggression/agonistic intents, while high-pitched, tonal vocalisations signal the lack of aggression,subordinance, and amicable intentions. Some of the dogvocalisations fall clearly into one or the other category,like whining, which is typically a high pitched and tonalsound – and it is emitted in clearly non-agonistic contexts(Cohen and Fox, 1976). Based on the acoustic setup, barksoriginally were clearly agonistic calls (being dominantlylow-pitched and noisy sounds, like in the wolves). How-ever, in dogs the acoustic spectrum of barks reached outtowards higher frequencies and more tonal calls, wherethe information content may be deciphered only with thehelp of additional acoustic parameters (Molnár et al., 2008),or perhaps by getting visual information also from thesignalling dog. Acoustically, dog growls are low-frequencybroadband, noisy signals, built up of sequences of vari-able duration, divided by pauses (Riede and Fitch, 1999).Dogs might growl under a variety of different (but mostlyagonistic) circumstances: during social conflict, as a threatsignal, while guarding food or during social play (reviewedin Yeon, 2007). As the acoustic characteristics of dog growlsfit mostly to the aggressive inner state, one may hypothe-size that growls in a playful context have to be supported byother, most likely visual signals signalling the playful inten-tion of the caller. Alternatively, playful growls should havemarkedly characteristic, non-aggressive features. Experi-mental evidences show that dogs are able to differentiatebetween contextually different growl types (Faragó et al.,2010b). Moreover, it has been revealed that dog growlsconvey size information of the caller (Riede and Fitch,1999), which is available either for humans (Taylor et al.,2008), and conspecifics (Faragó et al., 2010a). It was foundthat dogs recognize the real size of another dog that wasemitting food guarding growls (Faragó et al., 2010a). Intheir cross-modal matching experiment dogs preferred tolook at the dog picture, which showed a similar size dog

to the one whose agonistic growls were played back tothe subject. An acoustical analysis of contextually differ-ent growls showed that ‘Play’ growls (PL) were shorter andhad a narrower formant dispersion than ‘Food guarding’

ur Science 148 (2013) 128– 137 129

(FG) growls (Faragó et al., 2010b). The difference in theacoustical parameters of the two growl types may indi-cate a larger body size when producing PL than in thecase of FG growls. This result was interesting from twoaspects: first, it indicated that dogs may modify the indexi-cal (i.e., size-related) information content of their growlsdepending on the context; and second, the exaggeratedsize information appears usually not in an agonistic context(Clutton-Brock et al., 1979; Enquist and Leimar, 1983), butduring play. Whether such an acoustical size-manipulationis congruent with playful behaviour, at first we shouldtake a closer look at the characteristics of animal playitself.

Play is known to be a behaviour that appears in diverseforms within and across species (Burghardt, 2005). It hasvarious types and manifestations, and despite its con-spicuous, prevalent features, it is not easy to deliver anunambiguous and simple definition. Regarding the func-tional aspects of play, a number of different theories havebeen suggested. Animals for example, can get physicalexercise while playing (Brownlee, 1954), or acquire spe-cial skills, that facilitate the survival and activity in theirenvironment (Ripley, 1967). It appears that play experienceincreases the efficiency of executing certain behaviouralsequences (e.g., the killing of prey) (Beach, 1968), andalso enhances the flexibility of an individual’s behaviouralrepertoire (Miller, 1973). Moreover, play with conspecificsproves to be valuable in learning different social skills(Biben, 1998), to strengthen (Bekoff, 1984) or test socialbonds (Pozis-Francois et al., 2004; Zahavi, 1977), or as a wayto assess one’s own capabilities relative to others’ (Palagiet al., 2004; Smith et al., 1999; Thompson, 1998).

Two very common formal appearances are object andsocial play. In the first case, animals manipulate differ-ent objects, that otherwise do not seem to fulfil any otherimmediate need or provide any benefit to the animal(Bekoff, 1974). Social play is directed towards other livingbeings, and is frequently seen in many mammalian species.During social play, animals frequently use action patternsthat are borrowed from other contexts, like courtship,fight or elements of the predatory behavioural sequence(Sutton-Smith, 2001). The various incorporated elementsare typically mixed into unpredictable temporal sequences(Bekoff, 1974). Often unusual, elaborate, repetitive motorpatterns and movements are performed, and observa-tions have shown that exaggerations and role reversalsmight also take place in this behaviour (Bekoff, 2004). Self-handicapping means that participants make themselvesmore vulnerable to attacks by their opponents, while rolereversals refer to participants alternating dominant andsubordinate positions during play, often suspending orbending their real-life social status (Biben, 1998; Fagen,1981; Symons, 1978). Some researchers have even sug-gested that individuals follow a so-called 50:50 rule, sothat the winning encounters are balanced between the par-ticipants (Altmann, 1962; Pellis and Pellis, 1991). Socialplay has been hypothesized to affect the players’ relation-ships in a variety of ways (Fagen, 1981). One hypothesis is

that dominance relationships are established and alteredduring play (Smith, 1982). However, recent evidence sug-gests, that play behaviour does not affect dominance,

1 Behavio

bha1

cSditsSaPbfArttp2(pcp

piib‘mCpi1ffamsv(Saaboi

smbWrbFp2o

30 A. Bálint et al. / Applied Animal

ut rather reflects existing dominance relationships (e.g.,umans; Clark et al., 1969: rats, Rattus norvegicus; Pellisnd Pellis, 1991: squirrel monkeys, Saimiri sciureus; Biben,998).

Dogs show high levels of both intra- and interspe-ific play, both in their young-, and adulthood (Bauer andmuts, 2007; Horowitz, 2009). Object (directed towardsifferent objects) and social (directed towards other liv-

ng beings) play is also commonly seen, and the differentypes often mix in the play bouts. Besides having con-pecifics as playmates (as generally in other mammals;uomi, 1982), dogs also engage in play between species,s for example, between human caregivers and their pets.lay occupies a large proportion of the interaction timeetween dog and owner (Hart, 1995), for example, in theorm of tug-of-war, pulling or chase games (Bekoff andllen, 1998). It was found that play interactions tend toeflect the general patterns of social hierarchy, contraryo the suggested ‘50:50 rule’. That is, dominant dogs tendo maintain their advantageous position over subordinateartners, even during the play bouts (Bauer and Smuts,007). Rooney et al. (2000) and Rooney and Bradshaw2002) showed that dog–human play are structurally (andresumably motivationally) different, and seem to be lessompetitive regarding possession demands, than dog–doglay.

Object play in dogs appears in the form of mouthing,awing, pushing, pulling, grasping, lifting, hitting, carry-

ng, etc. different items, and as well as in other carnivores,t often incorporates predatory movements (stalking, grab-ing, shaking). Therefore, it is also commonly called

predatory play’. Generally, foraging movements play aajor role in the object play repertoire (Burghardt, 2005).

anine social play usually includes contact-chase andlay-fighting games, mounting behaviour (i.e., mimick-

ng copulatory movements) or inhibited biting (Bekoff,974; Burghardt, 2005). In dyadic play bouts the play-ul interaction can appear in the form of offensive (e.g.,orced downs, mounts, chin-overs, full body-overs, chases)nd affiliative/self-handicapping behaviours, for example,uzzle licks or voluntary downs (Bauer et al., 2009). In

ocial play-fighting, the behaviours adopted by the indi-iduals often resemble real-life aggressive confrontationsAldis, 1975; Bekoff, 1995; Fagen, 1981; Power, 2000;ymons, 1978), yet such play rarely escalates to realggression (Power, 2000). Although the various playfulctions mix unpredictably, it has been shown, that someehaviours are more flexibly modified and changed, thanthers. Possibly the more rigid actions serve as invariantndicators of formal dominance also in the playful context.

Studies revealed that the playful behavioural repertoirehows great individual variation. Some of this variationight be attributable to breed differences, since it has

een shown, that breeds might differ in their playfulness.hile selection for purposes dog showing is negatively cor-

elated with playfulness, working dog trials and popularreeds have higher playfulness scores (Svartberg, 2006).

urthermore, since dogs have been found having distinctersonality traits (Gosling et al., 2003; Jones and Gosling,005), this can also lead to differences in the play behaviourf dogs.

ur Science 148 (2013) 128– 137

Some of the findings suggest that the style and possi-bly the function of play may change over time. It has beenshown for example, that some behaviours were more com-mon in earlier developmental stages (e.g., forced downsand overs) and became less common as puppies matured.On the other hand, other behaviours became more com-mon with time (e.g., mounts; Ward et al., 2008). Bauer et al.(2009) observed that those behaviours that are commonlylinked to dominance related interactions in dogs were mostfrequent at a developmental stage where dogs were form-ing their dominance relationships. A decrease could befound in the rate of these behaviours in later development.They concluded that the frequent usage of such behaviourscould be relevant at a stage, when dogs are solidifying theirdominance relationships. The decrease later on might indi-cate a tendency for milder play, at a time when dominancechallenges could have severe consequences. Other studiessuggested, that as puppies mature, winning or being in thedominant position during play becomes more important,and play becomes more competitive. The authors hypoth-esized, that this might be due to dogs’ increasing awarenessof their intragroup social status (Fagen, 1981; Rothstein andGriswold, 1991).

Since among canids, males and females show little sizedimorphism and engage in similar roles (Derix et al., 1993;McLeod and Fentress, 1997), some research has suggested,that there are no sex differences in the play style of maleand female dogs (Bauer and Smuts, 2007; Bekoff, 1974;Biben, 1983). However, some studies have shown thatinfant domestic dogs do show sex differences in their playbehaviour (Lund and Vestergaard, 1998; Pal, 2008).

In another study, they examined inter- and intrasex-ual dyadic play bouts in dog littermates. They found thatdogs showed a preference for same-sex play initiations.Since dogs form intrasexual dominance relationships (Palet al., 1998), this might suggest, that intrasexual playbehaviour may function in the formation of these domi-nance relationships (Bekoff, 1972). It has also been shown,that in intersexual play bouts, males initiate play at higherrates than females, and males display self-handicappingmore often than females. They hypothesized that duringplay, males have an opportunity to learn characteristics offemale behaviour and this could improve their reproduc-tive success later in life, through female mate preferences(Ghosh et al., 1984; Pal et al., 1999).

In order to maintain the playful ‘mood’ and uphold thegamesome attitude of all joining members, it is importantto utilize different signals all over the course of play (Bekoffand Allen, 1998). This is especially important in social play,which often includes motor patterns used in predatory,sexual and agonistic contexts, which could be otherwiseeasily misinterpreted (Bekoff and Allen, 1998; West, 1974).These play markers may appear in various forms, such aspostures, ritualized movements (e.g., play bow in dogs:Bekoff, 1977), olfactory cues (Wilson and Kleiman, 1974),or vocal signals. Play bow, a common play signal in canines,has been found to communicate playful intent in asso-

ciation with behaviours that otherwise would be likelyto be misinterpreted as real aggression (Bekoff, 1977,1995). Besides the visual play-signals, other senses can beinvolved in detecting playful intentions. In rodents, it has

A. Bálint et al. / Applied Animal Behaviour Science 148 (2013) 128– 137 131

Table 1The basic data of the test subject dogs in the two experimental groups.

Food guarding growl Playful growl

Name Dog breed Gender Age (year) Name Dog breed Gender Age (year)

Borisz Whippet Male 5 Doki Mixed Male 14Whiskey Whippet Male 14 Dézike Mixed Female 11Daeron Belgian malinois Male 4 Bosko Great dane Male 4Rege Transylvanian hound Male 6 Gomez French bulldog Male 10Csitri Transylvanian hound Female 1 Ékes Hungarian greyhound Female 5Kósza Transylvanian hound Male 3 Tódor Border collie Male 6Rege Mixed breed Male 5 Álom Hungarian greyhound Female 1Betyár Mixed breed Male 7 Arwen Collie Female 2Szuzi Mixed breed Female 3 Gazsi Hungarian vizsla Male 6Csipke Mixed breed Female 3 Kópé Cairn terrier Male 4Lexi Great dane Female 5 Artha German shepard Female 3Manna Pug Female 2 Moji Beagle Female 2Monty4 Border collie Male 2 Samu Mixed Male 11Ribizli Mixed breed Female 1 Misi Mixed Male 1Málna Hungarian vizsla Female 1 Zadar Weimeraner Male 4Brúnó Labrador retriever Male 6 Odett Golden retriever Female 6Daniel Labrador retriever Male 2 Fecske Mudi Female 10Alma Labrador retriever Female 4 Plüsi Golden retriever Male 10Boni Foxterrier Female 2 Benji Mixed Male 12Mabu Foxterrier Female 5 Rupert Border collie Male 1Joker Golden retriever Male 2 Teo German shepard Male 5

Bordy Border collie Male 1

Peszka Labrador retriever Male 1

Hektor Mixed breed Male 3

been suggested, that different body odours might serve asplay soliciting signals (Wilson and Kleiman, 1974). Acous-tical play signals have been identified in several species(for a review see Burghardt, 2005). For example, mon-gooses emit a whistle during social play (Rasa, 1984), whilesquirrel monkeys also have a specific vocalizations onlyused during play (Biben and Symmes, 1986). It has beenreported that most dog breeds develop bark-games (Zimen,1981), i.e., they use ‘play-sounds’ (harmonic barks, growl-ing, vibrato-sounds), which might announce longer lastingplay sequences (see in: Jensen, 2007). A main characteris-tic of play is that it is always ‘non-serious’ (‘non-literal’),both in actions and in consequences (Bekoff and Allen,1998; Burghardt, 2005; Saracho and Spodek, 1998). Playsignals have great importance in initiating and maintain-ing this special atmosphere (Pellis and Pellis, 1996), oftenby means of exaggerated or modified movements, vocali-zations (Bekoff and Allen, 1998; Burghardt, 2005; Sarachoand Spodek, 1998). Juvenile wolves have been found toplay communication or facial expression games for exam-ple, in which they mutually display playful and exaggeratedsignals exclusively from the facial area (Jensen, 2007). Forexample, in squirrel monkeys, play vocalizations are muchlouder than needed for the play partners to hear them(Biben and Symmes, 1986). Besides the ‘play-barks’ dogsuse during playful encounters, in some cases play bouts areaccompanied by growling (Taylor et al., 2010; Yeon, 2007),as for example, in tug-of-war games (Faragó et al., 2010b).

As our earlier acoustical analysis showed that accord-ing to vocal parameters, these playful growls seem todepict a larger body size than an agonistic (food guard-

ing) growl type (Faragó et al., 2010a), we hypothesizedthat this size-manipulation can be the part of the complexplay signalling system during canine play behaviour. In thisstudy, we investigated whether other dogs are sensitive to

Berry French bulldog Male 1Sherry Mixed Female 1Nati Dwarf poodle Female 2

the indexical cues encoded in play growls. By the means ofa cross-modal matching experimental design (Faragó et al.,2010a), we tested dogs with simultaneously projected dogpictures and playbacks of growls, expecting that dogs showa preference for the matching size picture when hearing FGgrowls, but they prefer to look at the larger picture whenPL growls are played back.

2. Methods

2.1. Ethical note

This research consisted of non-invasive experiments.During the experimental work the owners of the dogs werepresent and they were informed about the aims and cir-cumstances of the investigation. The authors adhered to theASAB/ABS Guidelines for the Use of Animals in Research,the legal requirements of Hungary about protecting animalwelfare, and the ethical guidelines of the University.

2.2. Subjects

Our subjects were adult (>1 year old) family dogs fromvarious breeds. In total, 48 dogs were included in theexperiment (24 dogs were exposed to PL growls and 24 toFG growls). Each dog was tested once and participated inonly one group (see the detailed list of subjects in Table 1).We purposely avoided the repeated testing of our subjects.Practical experience suggests that it involves the loss ofinformation due to the appearance of order effect (as aconsequence of habituation of the participants). The gain

we could expect from the repeated design would be mostprobably lost as a consequence of the order effect. Insteadof testing each dogs in both conditions (FG and PL), weused larger sample sizes with a representative selection of

132 A. Bálint et al. / Applied Animal Behavio

Fig. 1. An example of the projections of dog pictures used during thecross-modal matching experiments. Differently sized copies of the samepdd

vdwSht

2

ebecsSlptaFPltdIta

((tdge

wgasT1sate

3. Results

ictures were presented to the subjects with a video projector. The sizeifference between the pictures was 30% measured at the withers of theog. Both pictures remained in the natural size range.

arious breeds and sizes of dogs. The size of the subjectsid not differ from each other in the two groups. (FG group,eight: mean: 24.06, SD = 11.43; height: mean: 53.31,

D = 11.51. PL group, weight: mean: 22.79, SD = 12.96;eight: mean: 53.29, SD = 13.93. Unpaired t-tests, weight(46) = 0.38; P = 0.71; and the height t(46) = 0.02; P = 0.99).

.3. Experimental procedure

We applied a modified version of the Looking Prefer-nce method used by (Faragó et al., 2010a). The set-up wasasically the same, the experimental room was 6 m × 3 m,quipped with a projector (Optima ES22), quad closedircuit camera system (Panasonic VJ420) and a speakerystem powered with Technics amplifiers (SU-C909U andE-A909S). A chair for the owner was placed in the mid-ine of the room, in 4 m distance from the canvas. Theair of speakers (Technics SB-M300M2) was hidden behindhe canvas, in the midline, emitting the sound playbackpproximately from the centre point of the projection.rom the four cameras available two was used, one (ELMOTC150S) recording the projections, and one that has zeroux capability (Panasonic NVDS29) was placed in front ofhe canvas on the floor on a short tripod, pointing at theog. This cameras view was assisted with a custom made

R lamp. The experimenter controlled the playback fromhe adjacent room and observed the events in the room on

PC computer used for video recording (VirtualDub 1.8.8).The subjects were presented with Power Point slides

Microsoft Office 2010) consisting of two dog picturesFig. 1) and a pre-recorded growl playback. The pic-ures were projected at floor level on the canvas. Eachog received a different picture–sound combination. Therowls used during playback were collected in two differ-nt contexts.

In the playful situation dogs were playing a tug-of-ar game with their owners PL. In the other situation the

rowling dog defended a bone from a conspecific that waspproaching it (FG growls, see Faragó et al., 2010b). Fig. 2hows sonograms of a typical exemplar of both contexts.he acoustical stimuli were collected in both cases from2 adult family dogs, indoors, with a Zoom H4n handheldound recorder (see the detailed list of dogs that served

s sources of growls in Table 2). The size of the dogs inhe two groups did not differ from each other significantly,ven after we removed those dogs from the analysis, which

ur Science 148 (2013) 128– 137

provided both types of growls (unpaired t-tests, height:t(12) = 0.43; P = 0.68; weight: t(12) = 0.15; P = 0.89).

The pictures were of two differently sized (within real-istic range) versions of the same photograph depicting 12dogs of various breeds in a sitting or standing position.The size (height at the withers) of one of the two pic-tures matched the size of the dog whose growl was coupledto the particular projection (“sound-matching”). The otherpicture was either 30% larger or smaller. Thus, the size ofthe pictures and their “sound-matching” were not mutuallyexclusive, in half of the cases the larger, in the other half thesmaller picture was the “sound-matching”. The side (leftor right) of the matching picture was equally distributedamong the presentations.

All projections started with an attractor, which wasa bouncing yellow circle, intended to focus the dogs’attention to the canvas. After that, the dog pictures wereprojected for 10 s (‘Only Projection’ phase, OP), before thegrowl playback started. The pictures were visible for 20 smore after the growl (‘After Sound’ phase, AS) then theprojection (and the test) ended. Each growl lasted approx-imately 2 s.

2.4. Data collection and statistical analysis

Behavioural coding was performed by SolomonCoder 11.07.04 software, developed by András Péter(http://solomoncoder.com/). We coded the subjects’ look-ing directions throughout the OP and AS phases. Fourdifferent directions were distinguished: ‘Left’ (dog looksat the left picture), ‘Right’ (dog looks at the right picture),‘Middle’ (dog looks at the middle of the canvas), and‘Other’, when the dog was not looking at the canvas, orits’ head or eyes were not apparent (16.93% of the totaltime recorded). The video footages had a 0.2 s scaling. Thelooking preference (or proportion of looking time) of thedogs was calculated by dividing the looking duration (s) atthe picture of interest by the sum of the looking durationsat the two pictures. This proportion of time was set against50% of the total time spent looking at the two pictures(the expected proportion of time if the dogs did not showa preference for a particular picture). The deviation of themeasured ratio from the estimated fraction of time (50% oftotal time) expressed the looking preference of the dogs.The two phases of the test were analyzed separately. Wealso analyzed whether dogs changed their gaze when thegrowl playback started, but only in those cases when thesubject was looking at one of the pictures at the onset of thegrowl playback. All statistical analyses were performed bySPSS 17.0. Two-tailed, one sample Wilcoxon Signed Ranktest was used to analyze the dogs’ looking preferences, dueto the data limited and not normal nature. The number ofdogs that altered their gazing direction after they heard thegrowl playback, and those that did not change directionwas compared to the chance level (0.5) with Binomial test.Alpha level was 0.05 in all comparisons.

First we measured whether dogs were lookingat the pictures during most of the time in both

A. Bálint et al. / Applied Animal Behaviour Science 148 (2013) 128– 137 133

arding ( betwee

Fig. 2. Growl sonograms of the same dog in playful (above) and food guare indicated with arrows. The formant dispersion is markedly differentindicates larger body size.

parts of the test. In the PL group, during the OPphase: mean = 7.6417, SD = 3.1818, during the AS phase:mean = 15.8083, SD = 4.4130. In the FG group, during theOP phase: mean = 7.4883, SD = 2.9133, and during the ASphase: mean = 17.2283, SD = 3.6250.

In the PL playback group during the OP phase, dogs didnot show preference for the ‘larger picture’ (One sampleWilcoxon signed rank test; Z = −0.234, N = 24, P = 0.823),but in the AS phase they looked significantly longer atthe ‘larger picture’ (Z = −2.487, N = 24, P = 0.011). Dogs didnot show a preference for the ‘sound-matching picture’ ineither of the phases (OP phase: Z = −0.834, N = 24, P = 0.414;AS phase: Z = −1.043, N = 24, P = 0.306) (Fig. 3).

In the FG growl playback group dogs did not showpreference for the ‘larger picture’ (OP phase: One sampleWilcoxon signed rank test: Z = −0.755, N = 24, P = 0.462; ASphase: Z = −1.219, N = 24, P = 0.231), or the ‘sound-matchingpicture’ in either of the phases (OP phase: Z = −0.115, N = 24,

P = 0.916; AS phase: Z = −1.577, N = 24, P = 0.118).

Regarding gaze alternations after the onset of the growlplaybacks, we found significant differences in both groups(Fig. 4). We sorted the dogs into two categories in both

below) context are shown. Formants (darker, denser horizontal stripes)n the two growls: narrower formant dispersion (like in the play growl)

groups (‘match first’: dogs looking at ‘matching picture’at the start of the growl playback; ‘non-match first’: dogslooking at ‘non-matching picture’ at the start of the growlplayback) and analyzed whether the proportion of dogsthat switched their gaze towards the other picture differsfrom the chance level (0.5). In the FG growl group, dogschanged their gaze direction in the ‘non-match first’ cate-gory above the chance level (ten out of 11, Binomial testP < 0.05), while in the ‘match first’ category the proportionof the switchers and non-switchers did not differ from eachother (six out of ten dogs switched, Binomial test P = 0.75).In the PL growl group the proportion of dogs that switchedtheir gaze did not differ from the chance level in the ‘non-match first’ category (five out of six, Binomial test P = 0.22),but in the ‘match first’ category we found a significant pro-portion of dogs that switched their gaze to the other picture(ten out of 12, Binomial test P < 0.05).

4. Discussion

In a cross modal matching experiment we found thatdogs prefer to look at the larger dog picture than at a

134 A. Bálint et al. / Applied Animal Behaviour Science 148 (2013) 128– 137

Table 2The list and basic size data of those dogs from which the growls were collected. PL = growls from a playful (tug-of-war) context; FG = growls from anagonistic food-guarding context.

Name Breed Age (year) Gender Height (cm) Weight (kg)

‘PL’Bodza Pumi 2 Female 48 18Fecske Mudi 7 Female 50 17Gréti Schnauzer 3 Female 43 17Totó West Highland white terrier 4 Male 35 10Mopi Mixed breed ∼1 Female 30 6Stokes English bulldog 5 Male 41.5 24Bosko Hungarian vizsla 10 Male 63 27Jamile Mixed breed 4 Male 60 35Kevin Mixed breed 5.5 Male 61 32Mio Mixed breed 2.5 Male 60 30Odin Hungarian vizsla 4 Male 62.5 25.5Pedro Mixed breed 1.5 Male 61.5 25

‘FG’Angel Mudi 7 Female 44 13Kira Mixed breed 4.5 Female 45 12Xena Jack Russell terrier 6.5 Female 32 6.5Fecske Mudi 7 Female 50 17Bosko Hungarian vizsla 10 Male 63 27Linka Mixed breed 5 Female 64 34Guru Belgian shepherd 2.5 Male 64 29Edgar Fox terrier 4 Male 40 8Lucy Border collie 4 Female 52 18

sI‘f

FbgtDapssdg

matching size pictures.

Mio Mixed breed 2.5

Kevin Mixed breed 5.5Jamile Mixed breed 4

maller one if they hear dog growls from a playful context.n a similar experimental setup dogs looked rather at the

sound-matching size’ dog picture when they heard a growlrom an agonistic context. Before the sound playbacks dogs

ig. 3. Looking preference for the ‘larger’ and the ‘matching size’ dogefore and after the sound presentation in the food guarding and play-rowl group. The horizontal line shows the expected ratio of lookingime if the dogs did not show preference for a particular picture (0.50).ogs spent significantly longer time looking at the photograph depicting

larger dog, than the real size of the growling dog was in the originallayful situation. The same dogs did not show a preference for the picturehowing a matched size dog to the original growling dog’s size. (Wilcoxonigned rank test, *P < 0.05). When dogs heard Food guarding growls theyid not show preference for any of the pictures. FG = food-guarding growlroup; PL = play-growl group.

Male 60 30Male 61 32Male 60 35

did not show preference towards either the larger, or the

Our results confirmed that while the size informationencoded in one of the growl types (‘play-growl’) seemsto advert a larger animal (at least in a situation when the

Fig. 4. Number of dogs that altered their gaze from the matching or thenon-matching picture to the other when the growl playback started incontrast with dogs showing no alteration. In the food-guarding growlgroup, a significantly higher proportion of dogs shifted their gaze fromthe ‘non-matching size’ picture to the ‘matching size’ picture. In the play-growl group most of the dogs looked away from the matching picture.(Binomial test, *P < 0.05). NS = non-significant; FG = food-guarding growlgroup; PL = play-growl group; matching = dogs looking at the ‘match-ing size’ picture when the growl playback started; non-matching = dogslooking at the ‘non-matching size’ picture when the growl playbackstarted; switch = dog changes its looking direction between the two pic-tures; no switch = dog does not switch looking between the two pictures,# = expected ‘correct’ reaction if we assume the ability of cross-modalmatching. The black horizontal lines represent the random level in eachgroup.

Behavio

A. Bálint et al. / Applied Animal

subject was faced with two choices), the size informationconveyed by another growl type (‘food-guarding growl’)depicts the adequate size of the caller. At least in the caseof the food guarding growls one could hypothesize thatdogs may avoid looking at the larger picture. However,our experimental design used a balanced amount of largerpicture-matching size and smaller picture-matching sizepresentations, therefore the results showing that dogs pre-ferred the matching size picture when they heard FG growlsproved that they did not simply averted their gaze fromthe larger picture. These results contribute to the currentknowledge about the playful signals of dogs, and they haveinteresting implications regarding the role and function-ing of contextually different growl types. We suggest thatthe earlier found acoustical alteration in ‘play growls’ isa special feature of the playful vocalization. It is possiblethat just like certain movements and gestures can be exag-gerated during play (Fagen, 1981), the size, encoded by thevocal signal might also be augmented, and thus might serveas a playful sign. Since growls are most suitable in short-range communication (Taylor et al., 2009), the sender ofthe vocalization is well visible for the other participantsof the social interaction. The obvious dichotomy betweenthe acoustical and visual information may be an impor-tant factor in maintaining the ‘non-serious’ manner of play,even for a longer duration. Also the higher fundamental fre-quency and more pulsing rhythm of the playful growls canensure the avoidance of misinterpretation (Faragó et al.,2010b). These play ‘markers’ might be especially impor-tant during play sequences that strongly resemble agonisticinteractions (like the tug-of-war game), thus the play-ing parties benefit from repeatedly re-assuring each otherabout their on-going playful intentions.

In contrast to the playful context, and in concurrencewith the results of (Faragó et al., 2010a), we found thatgrowls emitted in the food guarding situation convey ade-quate size information of the caller. One might hypothesizethat it would be beneficial for the emitter to send an‘exaggerated’ signal in an agonistic situation such as this.However, other aspects also have to be taken into consid-eration. In dogs, situations that involve agonistic growls(like food guarding) are typically very short range inter-actions; therefore there is direct visual information aboutthe caller’s body size besides the acoustical signal, in con-trast with the long-range calls of the deer for example. Inthe latter species the calls conveying size information areused without visual contact, thus the size exaggeration canbe effective (Reby et al., 2005). It is possible that in thecase of dogs during agonistic encounter the potential costof ‘cheating’ (like advertising a larger body size) would betoo high to risk, since it could easily escalate into a seriousfight. Several theoretical models suggest, that this effectmight preclude ‘dishonest’ signalling in a number of dif-ferent situations (for a review, see: Számadó, 2000, 2011).A modification that might add to the playful and off-handmood in one context (during play), might have severe con-sequences in another (guarding of food).

Scientists have also collected knowledge about howanimal species can be adapted to modify their formantdispersion with passive and active elongation of the vocaltract (Fitch and Hauser, 2003; Fitch and Reby, 2001). For

ur Science 148 (2013) 128– 137 135

example, we can find elongated trachea in several birdspecies (Fitch, 1999), and lowered larynx in mammals (e.g.,Charlton et al., 2011). It was also shown in multiple speciesthat animals can actively modify the position of their larynx(Fitch, 2000), raising the possibility to virtual size alteration(Sanvito et al., 2007; Taylor and Reby, 2010; Volodin et al.,2009). In the case of dogs, the exploration and unfoldingof the possible underlying mechanisms of such acousti-cal alterations is still in its infancy, therefore a numberof potential explanations might be addressed about theexact mechanism how the modified formant dispersionsare being produced during play. It is known, that dogs areable to lower their larynx during vocalization, at least incertain vocalization types (e.g., barks, whines). This low-ering is more typical and pronounced in loud calls (Fitch,2000), thus it is possible that during the vigorous playgrowls, dogs lower their larynges to a notable extent. Itis also feasible, that the special posture dogs display dur-ing the tug-of-war game, and/or the tense neck and bodycontribute to the formant disposition. The toy held in thedogs’ mouth could also cause deviations in the formantpattern, but since all cases were somewhat different (toy’sdifferent position in mouth, dogs hold the toy with differ-ent strength, etc.), it is hard to conclude how it affects theformant pattern.

In this study, we were able to demonstrate that dogs’perception of contextually different growl types is in accor-dance with the measured acoustical difference betweenthem. Our results further strengthen the idea that dogs’vocal communication system is very flexible, and pointout the importance of the link between the physical signal(the vocalization) and its biological effect. Even seeminglysubtle differences might have a substantial communica-tive role. The cross-modal matching experimental designwe used here appears to be a highly suitable methodin ethological experiments (e.g., in the study of commu-nication systems). In the future, research of the role ofacoustic signals in intra- and interspecific communicationand recognition capacity of dogs will follow based on thedescribed research paradigm.

Acknowledgements

This paper was supported by the János Bolyai ResearchScholarship from the Hungarian Academy of Sciences.The study was also funded by grants from the EuropeanUnion FP7-ICT-2007 LIREC 215554, ESF Research Net-working Programme “CompCog”: The Evolution of SocialCognition (www.compcog.org) (06-RNP-020) and the Hun-garian Ministry of Education OTKA K82020. Support for theHungarian Academy of Sciences MTA-ELTE ComparativeEthological Research Group (01 031) is also acknowledged.

References

Aldis, O., 1975. Playfighting. Academic Press, New York.

Altmann, S.A., 1962. Social behavior of anthropoid primates: analysis of

recent concepts. In: Bliss, E.L. (Ed.), Roots of Behavior. Harper, NewYork, pp. 277–285.

Bauer, E.B., Smuts, B.B., 2007. Cooperation and competition during dyadicplay in domestic dogs, Canis familiaris. Anim. Behav. 73, 489–499.

1 Behavio

B

B

B

B

B

BB

B

B

B

B

B

B

B

C

C

C

C

C

D

E

F

F

F

F

F

F

F

F

G

G

36 A. Bálint et al. / Applied Animal

auer, E., Ward, C., Smuts, B., 2009. Play like a puppy, play like a dog. J.Vet. Behav. 4, 61–62.

each, F.A., 1968. Coital behavior in dogs: III. Effects of early isolation onmating in males. Behaviour 30, 217–238.

ekoff, M., 1972. Development of social interaction, play, and metacom-munication in mammals: an ethological perspective. Q. Rev. Biol. 47,412–434.

ekoff, M., 1974. Social play and play-soliciting by infant canids. Am. Zool.14, 323–340.

ekoff, M., 1977. Social communication in canids: evidence for the evolu-tion of a stereotyped Mammalian display. Science 197, 1097–1099.

ekoff, M., 1984. Social play behavior. Bioscience 34, 228–233.ekoff, M., 1995. Play signals as punctuation: the structure of social play

in canids. Behaviour 132, 419–429.ekoff, M., 2004. Wild justice and fair play: cooperation, forgiveness, and

morality in animals. Biol. Philos. 19, 489–520.ekoff, M., Allen, C., 1998. Intentional communication and social play: how

and why animals negotiate and agree to play. In: Bekoff, M., Byers, R.D.(Eds.), Animal Play. Cambridge University Press, Cambridge, UK, pp.97–114.

iben, M., 1983. Comparative ontogeny of social behaviour in three South-American canids, the maned wolf, crab-eating fox and bush dog:implications for sociality. Anim. Behav. 31, 814–826.

iben, M., 1998. Squirrel monkey play-fighting: making the case for acognitive training function for play. In: Bekoff, M., Byers, J.A. (Eds.),Animal Play: Evolutionary, Comparative, and Ecological Perspectives.Cambridge University Press, Cambridge, U.K, pp. 161–182.

iben, M., Symmes, D., 1986. Play vocalizations of squirrel monkeys(Saimiri sciureus). Folia Primatol. 46, 173–182.

rownlee, A., 1954. Play in domestic cattle in Britain: an analysis of itsnature. Br. Vet. J. 110, 48–68.

urghardt, G.M., 2005. The Genesis of Animal Play: Testing the Limits. TheMIT Press, Cambridge, MA.

harlton, B.D., Ellis, W.A.H., McKinnon, A.J., Cowin, G.J., Brumm, J., Nilsson,K., Fitch, W.T., 2011. Cues to body size in the formant spacing of malekoala (Phascolarctos cinereus) bellows: honesty in an exaggerated trait.J. Exp. Biol. 214, 3414–3422.

harlton, B.D., Keating, J.L., Rengui, L., Huang, Y., Swaisgood, R.R., 2010.Female giant panda (Ailuropoda melanoleuca) chirps advertise thecaller’s fertile phase. Proc. R. Soc. B 277, 1101–1106.

lark, A.H., Wyon, S.M., Richards, M.P.M., 1969. Free play in nursery schoolchildren. J. Child Psychol. Psychiatry 10, 205–216.

lutton-Brock, T.H., Albon, S., Gibson, R., Guinness, F., 1979. The logicalstag: adaptive aspects of fighting in red deer (Cervus elaphus L.). Anim.Behav. 27, 211–225.

ohen, J.A., Fox, M.W., 1976. Vocalizations in wild canids and possibleeffects of domestication. Behav. Process. 1, 77–92.

erix, R., van Hooff, J., de Vries, H., Wensing, J., 1993. Male and female mat-ing competition in wolves: female suppression vs. male intervention.Behaviour 127, 141–174.

nquist, M., Leimar, O., 1983. Evolution of fighting behaviour: decisionrules and assessment of relative strength. J. Theor. Biol. 102, 387–410.

agen, R., 1981. Animal Play Behaviour. Oxford University Press, New York,NY.

ant, G., 1960. Acoustic Theory of Speech Production. De Gruyter Mouton,The Hagen, The Netherlands.

aragó, T., Pongrácz, P., Miklósi, Á., Huber, L., Virányi, Z., Range, F., 2010a.Dogs’ expectation about signalers’ body size by virtue of their growls.PLoS ONE 5, e15175.

aragó, T., Pongrácz, P., Range, F., Virányi, Z., Miklósi, Á., 2010b. The boneis mine: affective and referential aspects of dog growls. Anim. Behav.79, 917–925.

itch, W.T., 1999. Acoustic exaggeration of size in birds via tracheal elon-gation: comparative and theoretical analyses. J. Zool. 248, 31–48.

itch, W.T., 2000. The phonetic potential of nonhuman vocal tracts:comparative cineradiographic observations of vocalizing animals.Phonetica 57, 205–218.

itch, W.T., Hauser, M.D., 2003. Unpacking honesty: vertebrate vocal pro-duction and the evolution of acoustic signals. In: Simmons, A.M.,Popper, A.N., Fay, R.R. (Eds.), Acoustic Communication. Springer, NewYork, pp. 65–137.

itch, W.T., Reby, D., 2001. The descended larynx is not uniquely human.Proc. R. Soc. B 268, 1669–1675.

hosh, B., Choudhuri, D.K., Pal, B., 1984. Some aspects of the sexual

behaviour of stray dogs, Canis familiaris. Appl. Anim. Behav. Sci. 13,113–127.

osling, S.D., Kwan, V.S.Y., John, O.P., 2003. A dog’s got personality: a cross-species comparative approach to personality judgments in dogs andhumans. J. Pers. Soc. Psychol. 85, 1161–1169.

ur Science 148 (2013) 128– 137

Hart, L.A., 1995. Dogs as companions: a review of the relationship. In:Serpell, J. (Ed.), The Domestic Dog: Its Evolution, Behaviour and Inter-actions with People. Cambridge University Press, Cambridge, pp.161–179.

Horowitz, C.A., 2009. Attention to attention in domestic dog (Canis famil-iaris) dyadic play. Anim. Cogn. 12, 107–118.

Jensen, P., 2007. The Behavioural Biology of Dogs. CABI, Wallingford, UK.Jones, A.C., Gosling, S.D., 2005. Temperament and personality in dogs

(Canis familiaris): a review and evaluation of past research. Appl. Anim.Behav. Sci. 95, 1–53.

Lund, J.D., Vestergaard, K.S., 1998. Development of social behaviour in fourlitters of dogs (Canis familiaris). Acta Vet. Scand. 39, 183–193.

McLeod, P.J., Fentress, J.C., 1997. Developmental changes in the sequentialbehavior of interacting timber wolf pups. Behav. Process. 39, 127–136.

Miller, S., 1973. Ends, means, and galumphing: some leitmotifs of play.Am. Anthropol. 75, 87–98.

Molnár, C., Kaplan, F., Roy, P., Pachet, F., Pongrácz, P., Dóka, A., Miklósi, Á.,2008. Classification of dog barks: a machine learning approach. Anim.Cogn. 11, 389–400.

Morton, E.S., 1977. On the occurrence and significance ofmotivation—structural rules in some bird and mammal sounds.Am. Nat. 111, 855–869.

Pal, S.K., 2008. Maturation and development of social behaviour duringearly ontogeny in free ranging dog puppies in West Bengal, India. Appl.Anim. Behav. Sci. 111, 95–107.

Pal, S.K., Ghosh, B., Roy, S., 1998. Agonistic behaviour of free-ranging dogs(Canis familiaris) in relation to season, sex and age. Appl. Anim. Behav.Sci. 59, 331–348.

Pal, S.K., Ghosh, B., Roy, S., 1999. Inter- and intra-sexual behaviour of free-ranging dogs (Canis familiaris). Appl. Anim. Behav. Sci. 62, 267–278.

Palagi, E., Cordoni, G., Borgognini, T., 2004. Immediate and delayedbenefits of play behaviour: new evidence from chimpanzees (Pantroglodytes). Ethology 110, 949–962.

Pellis, S.M., Pellis, V.C., 1991. Role reversal changes during the ontogeny ofplay fighting in male rats: attack versus defence. Aggress. Behav. 17,179–189.

Pellis, S.M., Pellis, V.C., 1996. On knowing it’s only play: the role of playsignals in play fighting. Aggress. Violent Behav. 1, 249–268.

Pongrácz, P., Molnár, C., Miklósi, Á., 2010. Barking in family dogs: an etho-logical approach. Vet. J. 183, 141–147.

Power, T.G., 2000. Play and Exploration in Children and Animals. L. Erl-baum, New Jersey.

Pozis-Francois, O., Zahavi, A., Zahavi, A., 2004. Social play in Arabian bab-blers. Behaviour 141, 425–450.

Rasa, O.A.E., 1984. A motivational analysis of object play in juvenile dwarfmongooses (Helogale undulata rufula). Anim. Behav. 32, 579–589.

Reby, D., McComb, K., Cargnelutti, B., Darwin, C.J., Fitch, W.T., Clutton-Brock, T.H., 2005. Red deer stags use formants as assessment cuesduring intrasexual agonistic interactions. Proc. R. Soc. B 272, 941–947.

Riede, T., Fitch, W.T., 1999. Vocal tract length and acoustics of vocalizationin the domestic dog (Canis familiaris). J. Exp. Biol. 202, 2859–2867.

Ripley, S., 1967. The leaping of langurs: a problem in the study of locomotoradaptation. Am. J. Phys. Anthropol. 26, 149–170.

Rooney, N.J., Bradshaw, J.W.S., 2002. An experimental study of the effectsof play upon the dog–human relationship. Appl. Anim. Behav. Sci. 75,161–176.

Rooney, N.J., Bradshaw, J.W.S., Robinson, I.H., 2000. A comparison ofdog–dog and dog–human play behaviour. Appl. Anim. Behav. Sci. 66,235–248.

Rothstein, A., Griswold, J.G., 1991. Age and sex preferences for social part-ners by juvenile bison bulls, Bison bison. Anim. Behav. 41, 227–237.

Sanvito, S., Galimberti, F., Miller, E.H., 2007. Vocal signalling of male south-ern elephant seals is honest but imprecise. Anim. Behav. 73, 287–299.

Saracho, O.N., Spodek, B., 1998. Multiple Perspectives on Play in EarlyChildhood Education. State University of New York Press, Albany, NY.

Smith, P.K., 1982. Does play matter? Functional and evolutionary aspectsof animal and human play. Behav. Brain Sci. 5, 139–184.

Smith, L.K., Fantella, S.L.N., Pellis, S.M., 1999. Playful defensive responsesin adult male rats depend on the status of the unfamiliar opponent.Aggress. Behav. 25, 141–152.

Svartberg, K., 2006. Breed-typical behaviour in dogs—historical remnantsor recent constructs? Appl. Anim. Behav. Sci. 96, 293–313.

Suomi, S.J., 1982. Why does play matter? Behav. Brain Sci. 5, 169–170.Sutton-Smith, B., 2001. The Ambiguity of Play, 2nd ed. Harvard University

Press, Cambridge, MA.Számadó, S., 2000. Cheating as a mixed strategy in a simple model of

aggressive communication. Anim. Behav. 59, 221–230.Számadó, S., 2011. The cost of honesty and the fallacy of the handicap

principle. Anim. Behav. 81, 3–10.

Behavio

A. Bálint et al. / Applied Animal

Symons, D., 1978. Play and Aggression: a Study of Rhesus Monkeys.Columbia University Press, New York.

Taylor, A.M., Reby, D., 2010. The contribution of source-filter the-ory to mammal vocal communication research. J. Zool. 280,221–236.

Taylor, A.M., Reby, D., McComb, K., 2009. Context-related variation inthe vocal growling behaviour of the domestic dog (Canis familiaris).Ethology 115, 905–915.

Taylor, A.M., Reby, D., McComb, K., 2008. Human listeners attend tosize information in domestic dog growls. J. Acoust. Soc. Am. 123,2903–2909.

Taylor, A.M., Reby, D., McComb, K., 2010. Size communicationin domestic dog (Canis familiaris) growls. Anim. Behav. 79,

205–210.

Thompson, K.V., 1998. Self-assessment in juvenile play. In: Bekoff, M.,Byers, J.A. (Eds.), Animal Play: Evolutionary, Comparative, and Eco-logical Perspectives. Cambridge University Press, Cambridge, UK, pp.183–204.

ur Science 148 (2013) 128– 137 137

Volodin, I.A., Volodina, E.V., Efremova, K.O., 2009. Antelope, callingthrough the nose: structure of sounds and effect of sexual selectionon the vocal behavior of the saiga (Saiga tatarica). Zoologiceskij zurnal88, 113–124.

Ward, C., Bauer, E.B., Smuts, B.B., 2008. Partner preferences and asym-metries in social play among domestic dog (Canis lupus familiaris)littermates. Anim. Behav. 76, 1187–1199.

West, M., 1974. Social play in the domestic cat. Am. Zool. 14, 427–436.Wilson, S.C., Kleiman, D.G., 1974. Eliciting play: a comparative study.

Integr. Comp. Biol. 14, 341–370.Wyman, M.T., Mooring, M.S., McCowan, B., Penedo, M.C.T., Reby, D., Hart,

L.A., 2012. Acoustic cues to size and quality in the vocalizations of maleNorth American bison, Bison bison. Anim. Behav., 1–11.

Yeon, S.-C., 2007. The vocal communication of canines. J. Vet. Behav. 2,141–144.

Zahavi, A., 1977. Testing of a bond. Anim. Behav. 25, 246–247.Zimen, E., 1981. The Wolf: A Species in Danger. Delacorte Press, New York,

NY.