Silent tidbitting in male fowl, Gallus gallus: a referential visual signal with multiple functions

Transcription

1 835 The Journl of Experimentl Biology 212, Published by The Compny of Biologists 2009 doi: /jeb Silent tidbitting in mle fowl, Gllus gllus: referentil visul signl with multiple functions Crolynn L. Smith* nd Christopher S. Evns Centre for the Integrtive Study of Animl Behviour, Deprtment of Brin, Behviour nd Evolution, Mcqurie University, Sydney, NSW, 2109, Austrli *Author for correspondence (e-mil: Accepted 6 Jnury 2009 SUMMARY With the notble exception of bee dnces, there re no estblished exmples of multimodl referentil signls. The food clls of mle fowl, Gllus gllus, re functionlly referentil nd the coustic component of multimodl disply. However, the specificity of the receiver s response to the visul component (tidbitting) hs never been tested. Here we provide the first detiled nlysis of tidbitting, nd test the hypothesis tht these chrcteristic movements re functionlly referentil. We conducted plybck experiment with five high-definition video stimuli: Silent tidbit, Mtched-frequency motion in the opposite direction, Silent crows, Inctive mle nd Empty cge. Femles serched for food more during Silent tidbitting thn under ny other condition, suggesting tht this visul disply specificlly predicts the presence of food nd hence hs similr functionl properties to food clls. Silent tidbitting ws lso singulrly effective t evoking pproch nd close inspection, which my enhnce signl memorbility. These socil responses suggest tht the visul component of the disply hs the unique function of triggering ssessment of signler identity nd qulity s potentil mte. The coustic nd visul components re hence redundnt s food signl, but synergistic when dditionl functions re considered. These findings emphsize the perceptul complexity of multimodl displys nd provide the first demonstrtion of multimodl referentil signling in vertebrte. Key words: multimodl communiction, visul displys, food clls, referentil signls. INTRODUCTION Fundmentl to ny nlysis of signl function is the issue of receiver response specificity. Signls produced solely in prticulr context cn llow receivers to identify the eliciting event, such s the discovery of food or the sudden ppernce of predtor. In mny cses, this predictive reltionship llows dptive decision mking [e.g. food serching or flight (Evns, 1997; Huser, 1997)]. Signls tht evoke such highly specific receiver responses in the bsence of contextul cues re considered functionlly referentil (Evns, 1997). Previous reserch hs demonstrted the existence of coustic (Evns et l., 1993; Evns nd Evns, 1999; Seyfrth nd Cheney, 1990) nd visul (Gould nd Gould, 1988) referentil signls nd these now pper to be txonomiclly widespred, occurring in species s diverse s non-humn primtes (e.g. Di Bitetti, 2003; Mcedoni, 1990; Seyfrth et l., 1980), birds (Bugnyr et l., 2001; Evns et l., 1993; Evns nd Evns, 2007), surictes (Mnser, 2001) nd socil insects (von Frisch, 1993). Some of the sme species tht produce referentil signls lso hve multimodl communiction. The honeybee, Apis mellifer, wggle dnce encodes the direction, distnce nd profitbility of food source (von Frisch, 1974; Gould nd Gould, 1988). During the wggle dnce, the direction of the wgging run indictes the direction of food. This visul disply is ccompnied by vibrtory bursts, the durtion of which is correlted with the distnce to food. The combined disply is hence n exmple of multimodl referentil communiction. Multimodl communiction hs two puttive benefits: ensuring tht the intended receiver perceives the signl, nd mximizing informtion content (Hebets nd Ppj, 2005; Prtn nd Mrler, 1999; Prtn nd Mrler, 2005; Rowe, 1999). Redundnt or bckup signls (Johnstone, 1996; Zuk et l., 1993) convey the sme informtion through ech modlity, thereby incresing the likelihood of signl detection in noisy environment. By contrst, nonredundnt or multiple messge displys (Johnstone, 1996; Møller nd Pominkowski, 1993; Zuk et l., 1993) trnsmit different informtion vi ech chnnel, potentilly incresing the rte of informtion trnsfer (Cndolin, 2003; Prtn nd Mrler, 1999; Prtn nd Mrler, 2005). If one component of redundnt multimodl disply is functionlly referentil, then the other component might be expected to elicit similrly specific receiver response when produced independently. However, this prediction hs never been experimentlly tested. We focus here on the visul component of food-relted udiovisul multimodl disply produced by fowl (tidbitting). The coustic component of this disply (food clling) consists of series of pulstile, cluck-like sounds (Mrler et l., 1986; Stokes nd Willims, 1971), which re known to be functionlly referentil. Audio plybcks re sufficient to evoke substrte-serch responses, in the bsence of other cues (Evns nd Evns, 1999), nd the behvior of the hens is medited specificlly by the expecttion of finding food (Evns nd Evns, 2007). Under nturl conditions, discovery of pltble item by mle in the presence of hen relibly elicits the multimodl tidbitting disply (Evns nd Mrler, 1994; Mrler et l., 1986; Mrler et l., 1986b; Stokes nd Willims, 1971). This performnce often entices one or more hens to pproch the tidbitting mle nd foodserch ner him, sometimes tking the food item directly from his mndibles (Gyger nd Mrler, 1988; Mrler et l., 1986; Mrler et l., 1986b). In this hierrchiclly structured system, hens prefer to mte with mles tht provide food (Pizzri, 2003) nd, in the

2 836 C. L. Smith nd C. S. Evns presence of hen, dominnt mles respond to subordinte s food clling nd tidbitting disply with overt ggression, often chsing him wy from the food nd then food clling themselves (Stokes nd Willims, 1972). On occsions, subordinte mles lter the signl by producing only the visul component, suppressing the cll. Experimentl plybcks demonstrte tht these unimodl displys still ttrct hens to the silently tidbitting mle (Smith nd Evns, 2008). In contrst to these detiled nlyses of food clling, the visul disply hs only been described in generl terms. Dvis nd Domm (Dvis nd Domm, 1943) chrcterized tidbitting s repeted, rhythmic motion of the hed nd neck, including repeted picking up nd dropping of the food item. Wood-Gush s (Wood-Gush, 1954) description ws similrly brief nd he treted the two disply components s single ction, wheres Dvis nd Domm (Dvis nd Domm, 1943) recognized tht these often occur seprtely nd treted them s distinct. Stokes nd Willims (Stokes nd Willims, 1971) provided the most detiled overview of the tidbitting disply, lthough they did not quntify the specific motions performed or determine if there ws ny consistency to the sequence order. Mny visul displys re stereotyped (Wiley, 1983) nd include n lerting component, which engges the ttention of the intended receiver (Fleishmn, 1988; Rowe, 1999) thereby improving signl detectbility in noisy environments. A more precise nlysis of the structure of the tidbitting disply is n essentil prerequisite for further study of signl design. We hve previously shown (Smith nd Evns, 2008) tht the visul disply elicits similr overll levels of food serching to the multimodl disply nd the coustic component lone. However, the multimodl nd visul disply evoke higher levels of inspection conspecific recognition behvior (Dwkins, 1995; Guhl nd Ortmn, 1953) thn the udio lone. Hence the coustic nd visul components of mle tidbitting cn be clssified s perceptully redundnt (sensu Prtn nd Mrler, 1999) with regrd to food serch durtion. However, the dditionl socil response suggests tht the visul disply hs n dditionl function. These findings re consistent with the ide tht the visul disply might be food-relted signl with response specificity comprble to tht of food clls, in which cse tidbitting would hve the unusul property of being multimodl signl in which the components in ech modlity were functionlly referentil. However, there re severl possible lterntive explntions for the hen s food serch response. These include relese from vigilnce in the presence of n lert compnion (Artiss nd Mrtin, 1995) or generl increse in forging in response to ny mle movement. In this study, we tested the specificity of hen responses to tidbitting motion. First, we clssified the most common movements nd described their typicl order over the course of the tidbitting disply. We gve mles food items in the presence of n unfmilir hen nd scored the types of motor pttern produced nd the probbility of trnsitions between them. This defined the gross structure of the visul disply nd tested whether the temporl sequence ws constrined. Results lso informed the design of the plybcks tht followed. We then conducted n experiment to test perceptul processing. Stimuli were designed to evlute rnge of motions, with vrying levels of sptiotemporl similrity to norml tidbitting. Hens were shown high-definition videos of mle performing four different movement ptterns nd control footge of n empty cge. We then nlyzed video recordings of the hens food serching nd socil behviorl responses. If the visul disply were functionlly referentil, we would predict n increse in the food serching behvior specific to norml tidbitting, such tht responses to this plybck would be greter thn those to ny other type of motion nd to the empty cge control. MATERIALS AND METHODS Subjects Mle nd femle golden Sebright bntm fowl, Gllus gllus (Linneus 1758), were the subjects for both experiments. The behvior of this strin closely resembles tht of the ncestrl form, the red junglefowl, Gllus gllus (Collis nd Joos, 1953; Collis, 1987; Andersson et l., 2001; Schütz nd Jensen, 2001) from which ll domesticted strins hve been derived (Fumihito et l., 1994; Fumihito et l., 1996). In prticulr, Sebrights hve not been subjected to rtificil selection for rpid growth or egg production. Birds were housed in 1.0m 1.0m 0.6m (length width height) cges in climte-controlled room mintined t 22 C on 12h:12h dy:night cycle nd given d libitum ccess to highprotein food (Gordon Specilty Feeds, Sydney, Austrli) nd wter in their home cges. Twenty-five mles prticipted in experiment I, which mesured the structure of tidbitting. Ech mle ws housed with single femle. Twenty-four different hens prticipted in experiment II, which tested specificity of response. These hens were house in sme-sex pirs. Recordings for cretion of plybck stimuli nd behviorl testing were conducted in sound-ttenuting chmber (Ampisilence S.p.., Roberssomero, Itly 2.38m 2.38m 2.15m), which ws lined with 10 cm Sonex fom bffles (Illbruck, Minnepolis, USA) on the side wlls nd 15cm bffles on the ceiling to prevent reverbertion. Experiment I: tidbitting performnce Test pprtus During ech tril, the mle ws confined to steel-frmed wire cge (0.60m 0.45m 0.86m), which hd n udience hen cge of similr size butting the left side wll nd food dispenser on the opposite wll (for detils, see Evns nd Evns, 1999). We monitored tests vi Sony 1450QM color monitor, connected to Pnsonic WV-CL320 video cmer in the sound chmber. A Pnsonic WJ-810 time-dte genertor provided stopwtch t the bottom of the video disply for timing test sessions. All tests were video recorded using Pnsonic AG-7750 VHS-formt deck. Design nd test procedure Mles were plced in the test cge for 15 min ech on three consecutive dys to hbitute them to the test environment. We plced ech mle s cge-mte in the djoining udience cge during these sessions to fcilitte hbitution. No food ws presented. On the fourth dy, the mle ws gin plced in the test cge, but with n unfmilir hen in the udience cge. Unfmilir hens were used s udience becuse mles re most likely to disply for n unfmilir femle, lthough the performnce does not differ bsed on fmilirity (C.S.E., unpublished). After 5 min dely, which llowed the mle to settle fter hndling, the food dispenser ws ctivted, delivering five corn kernels onto the floor of his cge. Anlysis of hed movements We recorded the type of movement nd position of the mle s hed in spce in every video frme (PAL stndrd; 40ms time intervl). Movements fell nturlly into three discrete clsses, which we coded s follows: twitch (Fig. 1A): rpid horizontl side-to-side motion of the hed with the neck held fully upright; short bob (Fig.1B): brupt verticl movement of the hed from fully upright position

3 Referentil multimodl signling 837 A B Initil/finl position Midpoint of movement on video screens (McQuoid nd Glef, 1993) nd tht video plybck evokes nturl nti-predtor nd socil responses (Evns nd Mrler, 1991; Evns et l., 1993; Evns et l., 1993b). The HDV formt used in this study provides n pproximtely four-fold increse in resolution over stndrd-definition digitl video nd hs successfully been used in multimodl plybcks with hens (Smith nd Evns, 2008). Decisions bout the overll lyout of the test setup were informed by the well-described properties of the fowl visul system. Hens recognize conspecifics using close binoculr inspection of the other bird s hed nd neck region (Guhl nd Ortmn, 1953), but they re myopic in the frontl field nd so unble to determine individul identity from distnces greter thn 30cm (Dwkins, 1995; Dwkins, 1996; Dwkins nd Woodington, 1997). We hence positioned the end of the test cge 30cm from the plsm disply, distnce t which hen should ttempt socil recognition by fixting on the screen. Note tht this sptil seprtion ws lso sufficient to prevent hens from resolving individul pixels, with concomitnt loss of verisimilitude. The long xis of the cge ws perpendiculr to the plsm disply nd the remote-controlled door ws t the end frthest from it. C Fig. 1. Clssifiction of tidbitting movements. (A) Twitch: side-to-side horizontl movement of the hed with the neck upright; (B) short bob: verticl movement of the hed from fully upright position, stopping bruptly when pproximtely horizontl; nd (C) long bob: verticl movement of the hed from upright through full rc towrds the floor. This often includes seizing the food item between the mndibles. The smll white object on the floor is chicken dropping. to point hlfwy bove the floor, returning to the upright position; nd long bob (Fig. 1C): verticl movement of the hed, plunging through the full rc towrd the floor, often picking up the food item with the mndibles, nd ending with the hed in the upright position. We recorded ll trnsitions between these motor ptterns nd clculted men probbilities, which were then summrized in kinemtic plot (Lehner, 1979). Results of this nlysis informed the design of the experiment II stimuli. Experiment II: femle response to visul tidbitting disply Test pprtus The test pprtus consisted of 106 cm Sony high-definition flt pnel plsm disply (resolution 1080 by 1960 pixels), mounted t floor level next to 1.2m 0.30m 0.5m cge with remotecontrolled wire door 0.4 m from one end. Previous reserch hs demonstrted the hens cn recognize the movements of conspecifics Stimulus design We used 3-CCD high-definition video cmer (Sony HDR-FX1) to mke high-definition ( lines) video recordings of 12 mle fowl, Gllus gllus, performing the tidbitting disply. During recording, the video signl ws monitored on the plsm disply lter employed for plybcks, llowing us to djust the cmer zoom so tht the imge of the mle ws precisely life-sized. Mles were confined in 0.60m 0.45m 0.86m wire cge, 0.8m from the cmer. The udience hen ws held in seprte cge, pproximtely 30cm from tht of the mle. After 10min cclimtion period, four melworms were delivered from remote-controlled food hopper mounted bove the mle s cge. This usully evoked tidbitting nd food clling from the mle. We selected four mles using the sme criteri s in our previous study (Smith nd Evns, 2008). Rw footge ws then edited using Finl Cut Pro 5.1 (Apple Computer) to crete five stimuli from ech mle, ech of 15min durtion. These consisted of n empty cge for the first 5min, to llow the hen to settle, followed by 5min of mle stnding lone, which provided bseline mesurement for food serching in the compny of vigilnt compnion, then 60s of one of the five test sequences, followed by 4min of empty cge. Ech successive phse of the test stimuli begn nd ended with 0.5 s fde trnsition to void evoking strtle response. The first 10min nd the lst 4min of every tril were hence identicl cross every condition. Only the 60 s test sequence differed cross conditions. Test sequences were designed systemticlly to ssess the specificity of femle responses to mle disply movements. The five stimuli were Silent tidbitting (norml signl), Mtchedfrequency motion in the opposite direction, Silent crows, Inctive mle nd Empty cge. Decisions bout design of the stimuli were informed by the results of experiment I, s detiled below. Silent tidbitting Silent tidbitting consisted of the mle performing the full nturl disply, including twitches, short bobs nd long bobs (Fig. 1). Mtched-frequency motion in the opposite direction Anlysis of the tibitting movements (experiment I) reveled tht the mjority (77%) of the movements during typicl tidbitting disply

4 838 C. L. Smith nd C. S. Evns include rpid downwrd sweep nd return upwrd of the hed nd neck. To test whether this chrcteristic, rpid, verticl movement ws crucil signl component, independent of temporl pttern, we creted stimuli with Mtched-frequency motion in the opposite direction. Ech of the Silent tidbitting exemplrs ws used s templte; we edited spontneous crowing of the sme mle so tht ech upwrd extension of the neck during the crowing motion corresponded to one of the long, downwrd sweeping motions in the Silent tidbitting sequence (Fig. 2A,B). Mtched pirs of completed stimuli contined movement sequences with similr timing nd mplitude of displcement in the verticl plne (rnge movements) but in opposite directions. Silent crows To ssess the role of movement frequency, we creted exemplrs of Silent crows, in which the mle performed four upwrd neck extensions with 12 s of resting behvior between them; thereby creting stimulus with fivefold reduction in the frequency of movements compred to the Silent tidbitting nd Mtched-frequency motion stimuli. Crowing ws chosen s the motor pttern for these two stimuli becuse it hs similr durtion nd distnce of displcement to the long bob motor pttern in tidbitting. It is lso nturl motion performed spontneously by mles. Unlike the coustic component of crowing, which hs well-estblished territoril function (Miller, 1978), there is no evidence tht the ccompnying movement hs specific socil slience in the bsence of vocliztions. Inctive mle To test the effect of n lert compnion, we creted the Inctive mle plybcks, in which the mle mintined sttionry lert posture, with occsionl spontneous side-to-side hed movements, ppering to scn horizontlly round the cge, but without ny verticl displcement of the hed. Empty cge The Empty cge video ws identicl to the other stimuli in every respect, except tht the mle ws bsent. We stndrdized the five test sequences for severl prmeters tht seemed likely to influence hen responses. To prevent potentil interctions between food clls nd the motor ptterns, which were the focus of the study, ll plybcks were silent. Melworms ppered nd remined on the video screen during ll five 60s test sequences to control for the presence of preferred food item. Finlly, to reduce the possibility of socil fcilittion of food serch cused by observtion of feeding compnion (Zjonc, 1965; McQuoid nd Glef, 1993), the mles were not shown consuming the melworms. Results from experiment I reveled no obligtory trnsitions between the three motor ptterns (Fig. 3), suggesting tht temporl sequence is unlikely to be n importnt spect of disply structure; hence we did not mnipulte trnsition ptterns in experiment II. Test procedure Hens were individully plced in the test cge for four 15min periods, t intervls of 48h, to cclimte them to the pprtus nd sound chmber. At the strt of ech session, the wire door ws closed nd the hen ws confined to the end section of the cge, preventing her from pproching the plsm disply. During ech session the wire door ws remotely relesed once nd the empty cge video, without melworms, plyed on the plsm screen. By the fourth cclimtion session, ll hens redily emerged nd wlked the length of the cge fter the door opened nd none exhibited signs of disturbnce such s wing-flpping or crouching. We used within-subjects design in which ech hen ws first ssigned one of the four mle exemplrs nd then unique rndom sequence of the five tretments (test sequences). To ensure tht the video mle ws unfmilir to the hen, we pplied the constrint tht she should hve hd no socil contct with the rel mle for t lest six months prior to the experiment. Hens were tested t the sme time of dy to minimize diel vrition in behvior. The inter-tril intervl ws 48h. Ech tril begn with the hen behind the closed wire door in the section of the cge frthest from the plsm screen. This stndrdized the hen s minimum distnce from the video mle t the beginning of the 60s stimulus. For the first 10min, the door remined closed while the empty cge nd then the mle video A Long bob B Crow Fig. 2. Imges of Silent tidbitting (A) nd Mtched motion (B) stimuli. Ech long bob in the Silent tidbitting exemplr ws pired with crow by the sme mle in the Mtched motion exemplr.

5 Referentil multimodl signling 839 Motor pttern key: Twitch: Short bob: Long bob: Fig. 3. Kinemtic plot of motor pttern frequency nd trnsition probbilities during tidbitting disply. Dimeter of circles is proportionl to the frequency of occurrence of ech motor pttern (twitch: 23%; short bob: 40%; long bob: 37%). Width of connecting brs is proportionl to frequency of trnsitions between the motor ptterns. Brs of sme shding dd to 100% of trnsitions. Specific trnsition frequencies were s follows: twitch to short bob (78.8%), to long bob (21.2%); short bob to twitch (63%), to long bob (37%); nd long bob to twitch (15%), to short bob (85%). plyed. The test sequence (i.e. one of the four movement types or Empty cge control) then begn nd the remote control door ws opened, llowing the hen to pproch the video mle. We monitored the tests using CCD cmer (Pnsonic WV- CL320) connected to Sony color monitor (Sony PVM-1450QM). The video output ws converted into MPEG-2 formt using Migli- EvolutionTV nd sved for lter nlysis. Behviors of interest were scored using JWtcher Video 1.0 (Blumstein et l., 2006), which reds the time-code of the video file to permit single-frme resolution. We mesured the durtion of food serching, which is chrcterized by distinctive close binoculr fixtion of the substrte (Evns nd Evns, 1999; Evns nd Evns, 2007) nd two socil responses: time spent in close proximity to video mle, indicted by pproch to within 0.1m of the end of the cge closest to the screen, nd inspection behvior, which ws chrcterized by the hen stretching her neck towrds the flt pnel monitor t the height of the mle s hed, exctly s hens scrutinize other flock members (Guhl nd Ortmn, 1953). There is considerble individul vrition mong hens in time llocted to forging. To increse the sensitivity of our response mesure, we used difference score, djusting responses to plybck with corresponding estimte of the spontneous level of food serching (Evns nd Evns, 1999; Evns nd Evns, 2007). This ws clculted by mesuring ech hen s totl durtion of food serching during the 60 s test sequence nd then subtrcting bseline estimte, which ws obtined by verging two smples: her time spent food serching during the first 60s fter the video mle ppered (min6 of 15min tril period) nd tht during the lst 60s before the test plybck begn (min9). This pproch lso further controlled for dy-to-dy vrition. Tests for overll tretment effects were conducted with repeted mesures ANOVAs (SPSS for Windows), using mle exemplr s blocking fctor. Exemplr ws never significnt, so ll dt were pooled for further nlysis. Significnt differences were further explored using Tukey s honestly significnt difference (HSD) to conduct multiple pir-wise comprisons; this mintined the overll lph level t the nominted vlue of RESULTS Experiment I We nlyzed the verge frequency of ech motor pttern in the tidbitting disply nd the trnsition probbilities between them (Fig. 3). Short bob nd long bob combined ccounted for the mjority of the movements (40% nd 37%, respectively). Although the individul motor ptterns pper stereotyped nd re redily distinguishble, the sequences re reltively unconstrined. Over the course of the disply, mles switched continully between the three motor ptterns. Trnsitions between twitch nd short bob were common nd short bob ws most frequently followed by long bob (Fig. 3). Trnsitions between long bob nd twitch were less frequent. Experiment II Food serch durtion Anlysis of food serching durtion, djusted for bseline rte, reveled tht Silent tidbitting plybcks evoked significntly higher response thn ny of the other test sequences (Fig.4). Responses to Inctive mle, Silent crows nd Mtched motion in the opposite Adjusted food serch rte (s) b,b,b c Fig. 4. Food serch durtion. Time spent food serching during 60 s plybcks (vlues re mens ± s.e.m.). Scores djusted for individul bseline level (see text for detils). Different letters indicte significnt differences (P<0.05) s determined by repetedmesures ANOVA, followed by post-hoc Tukey s HSD. Inset shows chrcteristic binoculr close inspection of substrte during food serch. 0 2 Empty cge Inctive Silent crows Mtched motion Silent TB Tretment

6 840 C. L. Smith nd C. S. Evns Time spent close (s) b Fig. 5. Close pproch to video mle. Time spent within 0.1 m of the video mle during 60 s test stimulus (vlues re mens ± s.e.m.). Different letters indicte significnt difference (P<0.05) s determined by repeted-mesures ANOVA, followed by post-hoc Tukey s HSD. Inset shows hen stnding in region closest to plsm screen (white line dded for illustrtion) Empty cge Inctive Silent crows Mtched motion Silent TB Tretment direction did not differ significntly from one nother, despite considerble vrition in the frequency of movement depicted in these three stimulus types. In ddition, Silent crows nd Mtched motion were not significntly different from Empty cge (F 4,88 =9.61, P<0.0001; Tukey s HSD, P<0.05). Proximity to video mle We mesured pproch to the video mle by scoring the time ech hen spent within 0.1m of the end of the cge closest to the plsm screen. Muchly s test of sphericity ws significnt (P<0.05), so we pplied Huynh-Feldt correction (ε=0.85). The overll tretment (stimulus type) effect ws highly significnt (F 3,75 =7.52, P<0.001). Post-hoc tests reveled tht hens spent significntly more time close to the simulted mle in the Silent tidbitting plybcks thn in ny other tretment. There were no significnt differences mong the other four stimuli (Tukey s HSD, P<0.05; Fig. 5). Inspection As complementry mesure of socil response, we mesured the mount of time hens spent in inspection behvior (Fig. 6). Muchly s test of sphericity ws significnt (P<0.05), so we pplied Huynh- Feldt correction (ε=0.78). The overll tretment effect ws highly significnt (F 3,68 =40.04, P<0.001). Post-hoc tests reveled tht femles spent significntly more time inspecting the mle on the plsm screen during the Silent tidbitting sequences thn in ny of the other plybcks. None of the other tretments were significntly different from one nother (Tukey s HSD, P<0.05). DISCUSSION Structure of the tidbitting disply The visul component of the tidbitting disply is mde up of three distinct motor ptterns: twitch, short bob nd long bob (Fig. 1). Displys re primrily composed of the short bob nd long bob motor ptterns, with reltively fewer twitches (Fig. 3). Over the course of the displys, the mles continuously trnsitioned between the three motor ptterns. These motor pttern sequences re highly vrible, with no indiction of temporl stereotypy (Fig. 3), suggesting tht none of them hs evolved specificlly to hve n lerting function (i.e. design for prticulr efficcy in engging visul grsp reflex), s in some other dynmic visul signls (Peters nd Evns, 2003) b Fig. 6. Inspection of video mle. Time hens engged in binoculr fixtion directed towrd the video mle during 60 s test stimulus (vlues re mens ± s.e.m.). Different letters indicte significnt difference (P<0.05) s determined by repeted-mesures ANOVA, followed by post-hoc Tukey s HSD. Inset shows hen inspecting video mle. Inspection (s) Empty cge Inctive Silent crows Mtched motion Silent TB Tretment

7 Referentil multimodl signling 841 One possible explntion for this my be tht the displys contin fr more motion in the verticl plne thn in the horizontl. Such signl design should generte lrge sweep re in the visul field of potentil receivers tht is robust to vrition in ngle of view nd therefore should be consistently conspicuous (Mrr, 1982; Peters et l., 2002; Peters nd Evns, 2007). A second possibility is tht the visul disply is typiclly ccompnied by food clls, which re esy to loclize nd udible over considerble distnce (Stokes nd Willims, 1971), thereby relxing selection for specilized initil movement. A third possibility is tht the wttles, which move considerbly, swinging from side to side, during ech type of motor pttern nd produce lrge proportion of the pprent motion during the disply (C.L.S., unpublished), my so enhnce detectbility tht no constrint on movement sequence is necessry. Plnned experiments will test these predictions. Referentil signling We tested whether the food serch response evoked by visul tidbitting is specificlly dependent on the sptiotemporl chrcteristics of the disply, including frequency nd direction. In its most restrictive form, this hypothesis requires tht hens respond significntly more to silently tidbitting mle thn to ny other motion type. Comprisons of food-serching durtion revel precisely the predicted pttern of responses (Fig. 4). It is prticulrly striking tht Mtched control movements t the sme frequency s tidbitting, but in the opposite direction, were much less effective (Fig.4). This implies tht the downwrd direction of motion is n importnt component of the signl. This experiment lso demonstrtes tht the hens incresed food serching is not simply consequence of decrese in vigilnce cused by the presence of n lert compnion (Artiss nd Mrtin, 1995). Although the difference between the Inctive mle nd Empty cge is consistent with such n effect, the fourfold increse in food serching to the Silent tidbitting disply over tht recorded with the Inctive mle revels n dditionl response specific to the tidbitting signl (Fig. 4). We conclude the food serching response to silently tidbitting mle is not cused by chnge in the forging/vigilnce trdeoff (Artiss nd Mrtin, 1995). In ddition, the increse in food serching response to Silent tidbitting cnnot be ttributed to socil fcilittion (Zjonc, 1965; McQuoid nd Glef, 1993) becuse the plybck mle ws shown signling, rther thn serching the substrte. We conclude tht the tidbitting disply is sufficient to evoke forging behvior nd tht responses to these movements hve specificity similr to tht previously demonstrted for food clls (Evns nd Evns, 1999). Tidbitting movements. therefore hve ll the chrcteristics of functionlly referentil visul signl. When combined with food clling, s in the mjority of displys, this constitutes the first experimentl demonstrtion of multimodl referentil signling in non-humn vertebrte. Socil responses For mles, chieving close proximity to hen is n importnt fctor in mting success (Grves et l., 1985). The plybck experiment revels tht tidbitting cn ply n importnt role: hens spent more time stnding close to the silent tidbitting mle thn they did during ny other tretment (Fig. 5). This visul disply ws uniquely effective, both reltive to other movements nd in bsolute terms; none of the other video mle stimulus evoked significntly more close pproch thn the Empty cge control sequence. Hens, therefore, responded very specificlly to the visul disply, nd not simply to rpid motion or to the presence of simulted compnion. It hs been suggested tht one of the functions of the food clling might be to lure the intended receiver close to the signler (Stokes nd Willims, 1971) nd mny descriptions of multimodl tidbitting hve noted tht the hen typiclly pproches the clling mle (Collis nd Collis, 1996; Stokes nd Willims, 1971; Wood-Gush, 1954). Our experimentl results confirm tht the visul disply lone is sufficient to evoke this socil response. Tidbitting my lso improve memorbility of both the signl nd signler, n importnt spect of visul signl design (Guilford nd Dwkins, 1991) tht hs primrily been explored in the context of posemtic colortion (Hlpin et l., 2008). Hens spent pproximtely four times longer inspecting the silent tidbitting mle thn ny of the other mle stimuli, none of which differed from Empty cge (Fig. 6). Close inspection, s indicted by the hen binoculrly fixting on the video mle s hed, is hence uniquely triggered by tidbitting. Previous reserch (Dwkins, 1995; Dwkins, 1996; Hodos, 1993) hs implicted binoculr fixtion with the frontl field s criticl process in conspecific recognition nd hs estblished tht hens use the re round the eye to identify flock mtes (Dwkins, 1996). Hens prefer mles tht tidbit more (Pizzri, 2003; Zuk et l., 1993), so it seems likely tht the inspection of the mle s hed evoked by tidbitting my fcilitte formtion of n ssocition between the ppernce of prticulr mle nd provisioning with food. Mting does not usully occur immeditely fter tidbitting (Stokes nd Willims, 1971), so femles must retin some memory of individul mle tidbitting performnce for preference subsequently to be expressed. Although the visul nd coustic components of this multimodl disply re redundnt (sensu Prtn nd Mrler, 1999), with regrd to predicting food vilbility (Smith nd Evns, 2008), the visul disply hs synergistic effect on socil responses tht the sounds do not (Evns nd Evns, 1999), incresing the time spent in close proximity nd inspecting the mle. This contrst presents chllenge for theoreticl models of multimodl signling, since ctegoriztion will be sensitive to the function(s) of interest. Unimodl production of the tidbitting signl: functionl implictions Short-term chnges in signl structure cn reduce potentil costs (Ryn et l., 1982). Conspicuous signls ttrct predtors (Byly nd Evns, 2003; Roberts et l., 2007), prsites (Bernl et l., 2006) nd competitors (Stokes nd Willims, 1971). In fowl, dominnt mles obtin the mjority of mtings (Pizzri, 2003) nd re ggressive towrds subordinte mles tht tidbit, often displcing them nd tking the food item (Stokes nd Willims, 1971). However, Johnsen et l. (Johnsen et l., 2001) observed tht subordinte mles were ble to tidbit s frequently s lphs when the subordinte mle could disply out of sight of the lph. In flocks living under nturlistic conditions, we hve observed tht subordinte mles tidbit without perceptible food clling. This behvior creted dditionl mting opportunities by ttrcting nerby hens, which pproched nd food serched ner him (C.L.S., unpublished). If the silent tidbitting signl is less conspicuous thn the multimodl disply, then this behvior my reflect trdeoff by subordintes between the benefit of ttrcting femles nd the socil cost of incresed conspicuousness to dominnt mle. Plnned studies will test the conspicuousness of unimodl nd multimodl signls nd the frequency of their occurrence s function of socil context. These experiments comply with the Principles of niml cre, publiction No , revised 1985, of the Ntionl Institute of Helth, nd the Austrlin Code of Prctice for the Cre nd Use of Animls for Scientific Purposes (NHMRC, 1997). We thnk R. Miller nd C. Jude for bird cre, nd R. Mrshll for veterinry

8 842 C. L. Smith nd C. S. Evns support. We lso thnk A. Tylor for ssistnce with the sttisticl nlysis. This reserch ws supported by grnt to C.S.E. from the Austrlin Reserch Council. REFERENCES Andersson, M., Nordin, E. nd Jensen, P. (2001). Domestiction effects on forging strtegies in fowl. Appl. Anim. Behv. Sci. 72, Artiss, T. nd Mrtin, K. (1995). Mle vigilnce in white-tiled ptrmign, Lgopus leucurus: mte gurding or predtor detection? Anim. Behv. 49, Byly, K. L. nd Evns, C. S. (2003). Dynmic chnges in lrm cll structure: strtegy for reducing conspicuousness to vin predtors? Behviour 140, Bernl, X., Rnd, A. S. nd Ryn, M. J. (2006). Acoustic preferences nd locliztion performnce of blood-sucking flies (Corethrell Coquillett) to túngr frog clls Behv. Ecol. 17, Blumstein, D., Evns, C. S. nd Dniel, J. C. (2006). JWtcher Bugnyr, T., Kijne, M. nd Kotrschl, K. (2001). Food clling in rvens: re yells referentil signls?. Anim. Behv. 61, Cndolin, U. (2003). The use of multiple cues in mte choice. Biol. Rev. 78, Collis, N. E. (1987). The vocl repertoire of the red junglefowl: spectrogrphic clssifiction nd the code of communiction. Condor 89, Collis, N. E. nd Collis, E. C. (1996). Socil orgniztion of red junglefowl, Gllus gllus, popultion relted to evolution theory. Anim. Behv. 51, Collis, N. E. nd Joos, M. (1953). The spectrogrphic nlysis of sound signls of the domestic fowl. Behviour 5, Dvis, D. E. nd Domm, L. V. (1943). The influence of hormones on the sexul behvior of domestic fowl. In Essys in Biology, pp Berkeley, CA: University of Cliforni Press. Dwkins, M. S. (1995). How do hens view other hens? The use of lterl nd binoculr visul fields in socil recognition. Behviour 132, Dwkins, M. S. (1996). Distnce nd socil recognition in hens: implictions for the use of photogrphs s socil stimuli. Behviour 133, Dwkins, M. S. nd Woodington, A. (1997). Distnce nd the presenttion of visul stimuli to birds. Anim. Behv. 54, Di Bitetti, M. S. (2003). Food-ssocited clls of tufted cpuchin monkeys (Cebus pell nigritus) re functionlly referentil signls. Behviour 140, Evns, C. S. (1997). Referentil signls. Perspect. Ethol. 12, Evns, C. S. nd Evns, L. (1999). Chicken food clls re functionlly referentil. Anim. Behv. 58, Evns, C. S. nd Evns, L. (2007). Representtionl signling in birds. Biol. Lett. 3, Evns, C. S. nd Mrler, P. (1991). On the use of video imges s socil stimuli in birds: udience effects on lrm clling. Anim. Behv. 41, Evns, C. S. nd Mrler, P. (1994). Food clling nd udience effect in mle chickens, Gllus gllus: their reltionships to food vilbility, courtship nd socil fcilittion. Anim. Behv. 47, Evns, C. S., Evns, L. nd Mrler, P. (1993). On the mening of lrm cells: functionl reference in n vin vocl system. Anim. Behv. 46, Evns, C. S., Mcedoni, J. M. nd Mrler, P. (1993b). Effects of pprent size nd speed on the response of chickens, Gllus gllus, to computer-generted simultions of eril predtors. Anim. Behv. 46, Fleishmn, L. J. (1988). Sensory influence on physicl design of visul disply. Anim. Behv. 36, Fumihito, A., Miyke, T., Sumi, S. I., Tkd, M., Ohno, S. nd Kondo, N. (1994). One subspecies of the red junglefowl (Gllus gllus gllus) suffices s the mtrirchic ncestor of ll domestic breeds. Proc. Ntl. Acd. Sci. USA 91, Fumihito. A., Miyke, T., Tkd, M., Shingu, R., Endo, T., Gojobori, T., Kondo, N. nd Ohno, S. (1996). Monophyletic origin nd unique dispersl ptterns of domestic fowls. Proc. Ntl. Acd. Sci. USA 93, Gould, J. L. nd Gould, C. G. (1988). Evolution of the dnce. In The Honey Bee, pp New York: Scientific Americn Librry. Grves, H. B., Hble, C. P. nd Jenkins, T. H. (1985). Sexul selection in gllus: effects of morphology nd dominnce on femle sptil behvior. Behv. Processes 11, Guhl, A. M. nd Ortmn, L. L. (1953). Visul ptterns in the recognition of individuls mong chickens. Condor 55, Guilford, T. nd Dwkins, M. S. (1991). Receiver psychology nd the evolution of niml signls. Anim. Behv. 42, Gyger, M. nd Mrler, P. (1988). Food clling in the domestic fowl, Gllus gllus: the role of externl referents nd deception. Anim. Behv. 36, Hlpin, C., Skelhorn, J. nd Rowe, C. (2008). Nïve predtors nd selection for rre conspicuous defended prey: the initil evolution of posemtism revisited. Anim. Behv. 75, Huser, M. D. (1997). The Evolution Of Communiction. Cmbridge, MA: The MIT Press. Hebets, E. I. nd Ppj, D. R. (2005). Complex signl function: developing frmework of testble hypotheses. Behv. Ecol. Sociobiol. 57, Hodos, W. (1993). The visul cpbilities of birds. In Vision, Brin nd Behvior in Birds (ed. by H. P. Zeigler nd H. J. Bischof), pp Cmbridge, MA: MIT Press. Johnsen, T. S., Zuk, M. nd Fessler, E. A. (2001). Socil dominnce, mle behviour nd mting in mixed-sex flocks of red jungle fowl. Behviour 138, Johnstone, R. A. (1996). Multiple displys in niml communiction: bckup signls nd multiple messges. Philos. Trns. R. Soc. Lond. B Biol. Sci. 351, Lehner, P. N. (1979). Hndbook of Ethologicl Methods, pp New York: Grlnd STPM Press. Mcedoni, J. M. (1990). Wht is communicted in the ntipredtor clls of lemurs: evidence from ntipredtor cll plybcks to ring tiled nd ruffled lemurs. Ethology 86, Mnser, M. B. (2001). The coustic structure of surictes lrm clls vries with predtor type nd level of response urgency. Proc. Biol. Sci. 268, Mrler, P., Dufty, A. nd Pickert, R. (1986). Vocl communiction in the domestic chicken. I. Does the sender communicte informtion bout the qulity of food referent to receiver? Anim. Behv. 34, Mrler, P., Dufty, A. nd Pickert, R. (1986b). Vocl communiction in the domestic chicken. II. Is sender sensitive to the presence nd nture of receiver? Anim. Behv, 34, Mrr, D. (1982). Vision: A Computtionl Investigtion into the Humn Representtion nd Processing of Visul Informtion. Sn Frncisco: W. H. Freemn. McQuoid, L. M. nd Glef, B. G., Jr (1993). Socil stimuli influencing feeding behviour of Burmese red junglefowl: video nlysis. Anim. Behv. 46, Miller, D. (1978). Species typicl nd individully distinctive coustic fetures of crow clls of red junglefowl. Z. Tierpsychol. 47, Møller, A. P. nd Pominkowski, A. (1993). Why hve birds got multiple sexul ornments? Behv. Ecol. Sociobiol. 32, Prtn, S. nd Mrler, P. (1999). Communiction goes multimodl. Science 283, Prtn, S. R. nd Mrler, P. (2005). Issues in the clssifiction of multimodl communiction signls. Am. Nt. 166, Peters, R. A. nd Evns, C. S. (2003). Introductory til-flick of the Jcky drgon visul disply: signl efficcy depends upon durtion. J. Exp. Biol. 206, Peters, R. A. nd Evns, C. S. (2007). Active spce of movement-bsed signl: response to Jcky drgon (Amphibolurus murictus) disply is sensitive to distnce, but independent of orienttion. J. Exp. Biol. 210, Peters, R. A., Clifford, C. W. G. nd Evns, C. S. (2002). Mesuring the structure of dynmic visul signls. Anim. Behv. 64, Pizzri, T. (2003). Food, vigilnce, nd sperm: the role of mle direct benefits in the evolution of femle preference in polygmous bird. Behv. Ecol. 14, Roberts, J. A., Tylor, P. W. nd Uetz, G. W. (2007). Consequences of complex signling: predtor detection of multimodl cues. Behv. Ecol. 18, Rowe, C. (1999). Receiver psychology nd the evolution of multicomponent signls. Anim. Behv. 58, Ryn, M. J., Tuttle, M. D. nd Rnd, A. S. (1982). Bt predtion nd sexul dvertisement in neotropicl nurn. Am. Nt. 119, Schütz, K. E. nd Jensen, P. (2001). Effects of resource lloction on behviourl strtegies: comprison of red junglefowl (Gllus gllus) nd two domesticted breeds of poultry. Ethology 107, Seyfrth, R. M. nd Cheney, D. L. (1990). The ssessment by vervet monkeys of their own nd nother species lrm clls. Anim. Behv. 40, Seyfrth, R. M., Cheney, D. L. nd Mrler, P. (1980). Monkey responses to three different lrm clls: evidence of predtor clssifiction nd semntic communiction. Science 210, Smith, C. L. nd Evns, C. S. (2008). Multimodl signling in fowl, Gllus gllus. J. Exp. Biol. 211, Stokes, A. W. nd Willims, H. W. (1971). Courtship feeding in Gllinceous birds. Auk 88, Stokes, A. W. nd Willims, H. W. (1972). Courtship feeding in Gllinceous birds. Auk 89, von Frisch, K. (1974). Decoding the lnguge of the bee. Science 185, von Frisch, K. (1993). The Dnce Lnguge nd Orienttion of Bees, pp New York: Hrvrd University Press. Wiley, R. H. (1983). The evolution of communiction: informtion nd mnipultion. In Communiction (ed. T. R. Hllidy nd P. J. B. Slter), pp Oxford: Blckwell. Wood-Gush, D. G. M. (1954). The courtship of Brown Leghorn Cock. Anim. Behv. 2, Zjonc, R. B. (1965). Socil fcilittion. Science 149, Zuk, M., Ligon, J. D. nd Thornhill, R. (1993). Effects of experimentl mnipultion of mle secondry sex chrcters on femle mte preference in red jungle fowl. Anim. Behv. 44,