Many of the signals that animals use to communicate transmit relatively large distances and therefore encompass several potential signallers and receivers. This observation challenges the common characterization of animal communication systems as consisting of one signaller and one receiver. Furthermore, it suggests that the evolution of communication behaviour must be considered as occurring in the context of communication networks rather than dyads. Although considerations of selection pressures acting upon signallers in the context of communication networks have rarely been expressed in such terms, it has been noted that many signals exchanged during aggressive interactions will transmit far further than required for information transfer between the individuals directly involved, suggesting that these signals have been designed to be received by other, more distant, individuals. Here we consider the potential for receivers in communication networks to gather information, one aspect of which has been termed eavesdropping. We show that male Betta splendens monitor aggressive interactions between neighbouring conspeci¢cs and use the information on relative ¢ghting ability in subsequent aggressive interactions with the males they have observed.

Male singing behaviour correlates with extra-pair success in several passerine birds. Singing interactions during territorial contests provide relative information on the males involved. Such information may be important in female extra-pair behaviour and eavesdropping on singing interactions among males may allow females to make such relative assessments. We used interactive playback to instigate singing contests with male great tits during the peak fertile period of their mate in an attempt to alter females' assessment of mates' quality relative to neighbours (potential extra-pair partners). We escalated a contest to one male (by overlapping his songs) and then subsequently de-escalated a contest (by alternating) to a neighbour. Intrusions onto neighbouring territories by females mated to either treatment male were then monitored. Females mated to escalation treatment males were more likely to intrude following playbacks than females mated to de-escalation treatment males. Although the absolute song output of males did not di¡er between treatments, males produced more song relative to playback in de-escalation treatments and relative song output was positively correlated with female intrusions. Therefore, female great tits eavesdrop on singing interactions and change their visitation rates to neighbouring territories according to their mate's singing performance relative to neighbours.

No abstract

Communication and social behaviour are inextricably linked, with communication mediating important social behaviours such as resource defence and mate attraction. However, the social environment in which communication occurs is often ignored in discussions of communication behaviour. We argue that networks of several individuals are the common social environment for communication behaviour. The consequences for receivers and signallers of communicating in a network environment are the main subjects of this review. Eavesdropping is a receiving behaviour that is only possible in the environment of a network and therefore we concentrate on this behaviour. The main effect of communication networks on signallers is to create competition with other signallers for receiver attention. We discuss the consequences of such competition. To conclude, we explore the role of signals and signalling interactions as sources of information that animals exploit to direct their behaviour. IntroductionCommunication mediates virtually all important social behaviours. For communication to occur, at least two individuals, a signaller and a receiver, must be involved. Therefore communication can itself be considered a social behaviour. Given such obvious and strong links between communication and social behaviour, it is surprising how often the social environment of communication is ignored. For example, theoretical considerations of communication deal with a dyad of one signaller and one receiver (recent exceptions are Johnstone 1998, 2000; Hasson 2000) -social environment has literally been left out of the equation. The purpose of this review is to identify the major social environment in which communication occurs and to give a variety of examples of how this environment influences communication behaviour. The network environmentThe social environment in which much, if not all, communication occurs is a network of several signallers and receivers. For a long-range signal this assertion follows almost by definition, since such a signal will encompass several individuals during the course of transmission. The area encompassed is sometimes referred to as the signal's active space (e.g. Brenowitz 1982). This obvious association between long-range and large active space has perhaps deflected attention from the fact that it is the distance travelled by the signal in relation to the average spacing between individuals that is central to the idea of communication networks, rather than distance per se. Since many social groupings of animals are closely spaced, it is likely that many signals that are shortrange in an absolute sense can also encompass several signallers and receivers if these individuals occur sufficiently close together. Thus the commonest social environment in which communication occurs is a network (Dabelsteen 1992;McGregor 1993;McGregor and Dabelsteen 1996). Whether networks are ubiquitous still remains to be seen. Receivers and networksThe network environment creates costs and benefits for receivers in addition to those usua...