ANIMAL BEHAVIOUR, 2001, 61, 949–958 doi:10.1006/anbe.2000.1668, available online at http://www.idealibrary.com on
Food calling in ravens: are yells referential signals? THOMAS BUGNYAR, MAARTJE KIJNE & KURT KOTRSCHAL
Konrad Lorenz Research Station, Gru ¨ nau and Institute of Zoology, University of Vienna (Received 27 March 2000; initial acceptance 22 May 2000; final acceptance 22 November 2000; MS. number: 6536R)
Ravens, Corvus corax, yell when they approach rich but defended food sources. As in other species, such food-associated calls attract conspecifics. These calls may provide information about the sender, such as its behaviour or motivation, and about the type of stimuli to which the caller is responding, such as the location, quality or quantity of a food source. We investigated whether yells convey information about different types and amounts of food. We experimentally exposed a group of free-ranging ravens foraging in a game park near Gru ¨ nau, Austria to six feeding situations. Food of one of three types (meat, kitchen leftovers, wild boar chow) and either of two quantities (one or three buckets) was shown to the ravens 10 min before they could gain access to it during the feeding of wild boars, Sus scrofa. Ravens responded to the sight of food with one type of yell, the long haa call. Haa calling rates varied with the type but not with the amount of food, and decreased during feeding. Although juveniles produced long yells (chii calls) in response to food they changed the context of calling with increasing independence from their parents. Ravens gave short ‘who’ yells when approaching the food. Who calling was thus not affected by the sight of food but by the feeding situation in general. This is comparable to other calls given in a foraging context, such as appeasement and intimidation calls. The different use of long and short yells relative to food availability suggests that who calls provide information about the caller, such as its behaviour at food, whereas haa calls may also provide information about the food itself. Our data are consistent with the idea that haa calls are functionally referential.
source is restricted (e.g. ravens, Corvus corax: Heinrich 1988). This variation in the context of calling raises the question of what information is encoded in food calls. Signals may encode information about attributes of the sender, such as its motivational state, behaviour or identity, and about stimuli or events in the environment, such as the location, quality or quantity of food (e.g. Marler 1967; Smith 1981; Maynard Smith & Harper 1995; Evans 1997). If such signals provide receivers with sufficient information to determine the context underlying signal production which, in turn, allows them to predict environmental events, the signals are regarded as functionally referential (Hauser 1996). Functionally referential signals have been documented mainly in the context of encounters with avian and terrestrial predators (e.g. Seyfarth et al. 1980; Macedonia 1990; Evans et al. 1993). In the context of food, referential signals have been proposed for only a few species that seem to meet the criteria of stimulus specificity for call production (review in Hauser 1996) and also the perceptual salience and response specificity to these calls (e.g. Hauser 1998). However, in most cases, the data do not provide a clear picture of what food calls mean (e.g.
Calling near food sources may attract competitors (e.g. Hauser 1996; but see Boinsky & Campbell 1996) which may benefit the signallers, as the formation of groups may either directly reduce predation risk (e.g. Elgar 1986) or facilitate access to defended food (Marzluff & Heinrich 1991). In addition, calling at food sources may attract potential mates (Stokes 1971; Marler et al. 1986b; Van Kampen 1994) or display status (Heinrich & Marzluff 1991; Clark & Wrangham 1994). Food-associated calls have been reported mainly in primates and birds (reviews in Marler et al. 1992; Hauser 1996). In almost all species observed, these calls are given at abundant food (e.g. toque macaques, Macaca sinica: Dittus 1984), divisible food (e.g. house sparrows, Passer domesticus: Elgar 1986; chimpanzees, Pan troglodytes: Hauser et al. 1993), preferred food (e.g. chickens, Gallus g. domesticus: Marler et al. 1986a; cottontop tamarins, Saguinus oedipus: Elowson et al. 1991; golden lion tamarins, Leontopithecus rosalia: Benz 1993; rhesus macaques, Macaca mulatta: Hauser & Marler 1993), or if access to the Correspondence: T. Bugnyar, Konrad Lorenz Research Station, A-4645 Gru¨nau 11, Austria (email:
[email protected]). 0003–3472/01/050949+10 $35.00/0
2001 The Association for the Study of Animal Behaviour
949
2001 The Association for the Study of Animal Behaviour
950
ANIMAL BEHAVIOUR, 61, 5
pant hoots of chimpanzees; Clark & Wrangham 1993, 1994; Hauser et al. 1993). Conclusive evidence that food calls provide sufficient information to evoke anticipatory feeding behaviour from conspecifics has been obtained from only one species, the domestic chicken (Evans & Evans 1999). Common ravens utter loud yells when hungry and when they see food to which they cannot gain access (Heinrich 1988; Heinrich & Marzluff 1991). Juvenile ravens yell when their parents return from foraging (Gwinner 1964); subadult and adult nonbreeders yell when they are confronted with food defended by dominant conspecifcs (Heinrich & Marzluff 1991); and adult pair-bonded females yell when begging food from their mates (Goodwin 1986). Although yells are structurally discrete, and therefore easily discernible to the human ear, variations in tone and length are apparent. Differences in tone may be due to individual characters (Heinrich & Marzluff 1991) and developmental state, since yells of juveniles are more raspy and high pitched than those of subadults and adults (Heinrich & Marzluff 1995). Differences in length may be due to social factors and motivational states: short yells (sounding like an emphatic who) are given by aggressive dominant ravens, whereas yells by hungry but nonaggressive birds are longer (haa; Heinrich & Marzluff 1991). Experiments have shown that yelling attracts nearby foraging ravens to a particular feeding site or, in playbacks, directly to a loudspeaker (Heinrich 1988). We examined whether the yells of ravens encode information about food. Free-ranging ravens foraging daily in a game park were shown food of different qualities and quantities at a wild boar enclosure before it was provided to both the wild boars and ravens. Focusing on the timing of the calling response, we investigated whether certain types of yells were given at the sight of food or just when the food became available. In addition, we investigated whether the yelling rates were linked to the type and the amount of food. METHODS
Study Site The Cumberland game park, in the valley of the river Alm in the northern part of the Austrian Alps, is the year-round focus of activity of a group of 20–120 nonbreeding ravens (XSE=351). Together with some adult breeders from adjacent territories, they use the park for foraging (Drack & Kotrschal 1995). At the time of the study, 25 of these ravens had been captured in drop-in traps baited with meat (Stiehl 1978). They were marked with individually distinct patagial wing tags made of Saflag (45 cm; Kochert 1973), with a combination of two or three coloured metal leg bands and a numbered aluminium ring of the Vogelwarte Radolfzell, Germany (Huber 1991). We have never observed any harmful effects of this marking on the ravens. Juvenile ravens could be distinguished from subadults and adults by the coloration of their eyes, oral cavity and feathers (Heinrich
1994a, b). Monitoring of marked individuals indicated that ravens join this foraging group at the park repeatedly (median every second week), but do not stay for long (median 6 days; unpublished data). Visitors are allowed to enter the park from 0900 hours. Prior to this, the zoo animals are fed. During feedings, ravens snatch food at various enclosures, although they concentrate particularly at the wolves, Canis lupus, brown bears, Ursus arctos, and wild boars. As the feeding of wild boars (eight adults, six juveniles) varies between days with respect to both type and amount of food, we chose this site for our experiment. In contrast to wolves, which may kill ravens at their feeding site, wild boars do not pose a threat to foraging ravens (Bugnyar & Kotrschal, in press). Owing to long-term observations (mainly by T. B.) the ravens and the wild boars are well habituated to the presence of certain humans which makes close observation (5–10 m) possible without the necessity to hide or camouflage observers or video equipment.
Experiments From 27 May to 27 August 1998 we conducted experiments over 42 morning sessions, each lasting 40 min. Between sessions we varied both the type (quality) and amount (quantity) of food for the wild boars. Three visually distinct types of food, meat (mainly beef, cut into pieces of 100–300 g), kitchen leftovers from a nearby restaurant (containing ca. one-third meat and two-thirds vegetables), and wild boar chow (Fixkraft Erhaltungsfutter), were offered in quantities of either one or three buckets (each bucket containing 10 litres). When meat or kitchen leftovers were fed, a bucket of chow was routinely added. This additional bucket of chow was not shown to the ravens before feeding. Both ravens and wild boars were habituated to these kinds of food. All six variations (three food types; two quantities) were presented in seven replications in random alternation. Each of the experimental sessions consisted of four consecutive observation phases, each lasting 10 min. The first 10 min were used as a control period (baseline phase). During this period, we recorded calls 80 m from the wild boar enclosure, from next to the zoo entrance, to control for calling rates that were the result of a possible conditioned association between the presence of observers and a feeding event. During observation periods 2–4, we recorded at the wild boar enclosure. At the beginning of the second 10-min phase, we visually exposed buckets containing food beside the fence in front of the enclosure (presentation phase). In this period, ravens passing overhead could see the quantity and contents of the buckets. The food, however, remained inaccessible to them because they were reluctant to take food from the road when humans where present (see also Heinrich 1988). The third 10-min phase started when we threw the food over the fence into the enclosure and both the wild boars and the ravens could start eating (phase of food availability). Finally, a second control period started 5 min after the end of the feeding phase, when most of the food had been consumed and ravens were still present (control phase after feeding).
BUGNYAR ET AL.: FOOD CALLING IN RAVENS
Frequency (kHz)
3
2
1
3 ms Chii call
2 ms Haa call
1 ms Who call
1 ms Appeasement call
2 ms
3 ms Intimidation call
2 ms
Figure 1. Spectograms of the three types of yells (juvenile chii calls, haa calls and who calls), an appeasement call and one type of intimidation call. Spectograms were made in S–Tools (version 3.55). Sampling rate 8 kHz, frame length 14.25 ms, overlap 50%, filter band width 70 Hz. Grey scale represents amplitude range (dB).
Ravens foraging in the game park readily responded to the sight of food buckets by approaching the feeding sites and with increased calling activity, which suggests that an association between food buckets and feeding had already been formed (Bugnyar & Kotrschal, in press). Pilot studies on six birds in aviaries revealed a preference for meat over kitchen leftovers and wild boar chow, and a preference for leftovers over chow (unpublished data). Wild boars usually gathered at the feeding site as soon as humans approached with food. However, the boars did not appear to behave differently to the types or amounts of food presented in buckets outside their enclosure (unpublished data). Also, on five occasions, none of the wild boars was present at the feeding site at the beginning of the presentation phase, but the ravens responded with calling. Thus, it seems unlikely that the ravens’ response was merely due to the behaviour of wild boars rather than to the food itself.
Data Collection and Analysis During the experimental phases we recorded four distinct calls (three types of yells and the appeasement call) and another category of five calls (intimidation calls). All the calls were structurally distinct (Fig. 1) and could be easily identified by ear. For the entire experiment, they were directly coded in the field by M. K., who was then blind to the hypothesis tested. To judge the reliability of categorizing calls by ear we tested 10 control observers, with only limited prior knowledge of raven
communication, with randomly assembled sequences of taped calls. After watching a videotape of a feeding event with a commentary on the different call types, control observers correctly coded 23.70.5 (XSE) out of a standardized series of 25 calls. They distinguished call types with a reliability of 95%. Three of the recorded calls were variations of the yell vocalization (Fig. 1). (1) Haa calls (long yells) are mostly given at food sources by vagrant subadults and adults (Heinrich 1988). (2) Who calls (short yells) are primarily given by adult dominants as they approach food (Heinrich & Marzluff 1991). (3) Chii calls (high-pitched, long yells) are given by juveniles and develop into the (sub)adult haa yelling within the first year of life (Heinrich & Marzluff 1991; see also Gwinner 1964). For a quantitative sound analysis of the three types of yell, we measured call length, basal frequency and main frequency of taped calls, using the software S–Tools developed by the Research Laboratory of Acoustics at the Austrian Academy of Sciences. Yell types differed highly significantly with respect to all of these parameters (ANOVA: call length: F2,18 =289.1, P
