The Evolution of Animal Communication: Reliability and Deception in Signaling Systems

The twin concepts of reliability and deception would appear to present an evolutionary paradox for animal communication systems. On one hand, signals should be reliable indicators of a signaler's intentions or quality; otherwise, receivers should ignore these signals and over evolutionary time unreliable signals would disappear. On the other hand, signalers should benefit in many contexts by masking their true intentions or exaggerating their quality, leading to selection for deceptive signals.

This paradox has been reflected over the years in major changes in the way signal reliability has been viewed by those studying animal behavior. In the early days of ethology, communication was often considered a cooperative endeavor in which both signalers and receivers benefited from the exchange of information. This viewpoint predicted selection for clear, unambiguous, and reliable signals to maximize the transfer of information between communicants. A major difficulty with this notion is that it relies on selection operating at the group level, a largely discredited concept. In sharp contrast, Dawkins and Krebs (1978) suggested communicative exchanges should be regarded as attempts by a signaler to manipulate the receiver's behavior for its own individual benefit. Under this scenario, deceit would be rife in animal signals and receivers would be under strong selection to discern the true intentions of the signaler. Neglected here was the critical fact that if signals were truly dishonest and essentially uninformative, then selection should favor receivers that ignore them entirely, leading to the eventual collapse of a signaling system.

A third way to think about signaling was provided by Zahavi's (1975) handicap concept. The handicap hypothesis is based on two assumptions: 1) that the production of some signals entails a cost that increases with increasing signal intensity, and 2) that this cost is more burdensome to low quality signalers than to high quality signalers. Zahavi (1975) suggested that under these conditions, only high quality individuals would be able to sustain the costs of high intensity signaling, and low quality individuals would signal at lower intensity or not at all. Thus, costly signals can lead to an evolutionary equilibrium in which signal intensity is an accurate measure of signaler quality. Receivers would sustain this equilibrium by attending only to signals that were reliable indicators. Although initial responses to Zahavi's verbal formulation of this idea were lukewarm at best, subsequent mathematical models, especially those of Grafen (1990a, 1990b), gradually convinced many behavioral ecologists that costly signals could work as a mechanism to enforce signal honesty. The critical questions then became just what costs signals might impose, how high these costs might be, and whether these costs compelled absolute reliability or permitted some level of deceptive signaling.

The thirty years following Zahavi's initial proposal of the handicap mechanism have seen great growth in theoretical and empirical studies of signal costs, reliability, and deception. The time is ripe for a critical review of these advances, and William Searcy and Stephen Nowicki have obliged with their book The Evolution of Animal Communication: Reliability and Deception in Signaling Systems. The result is an exceptionally clear and well-organized volume that systematically evaluates the fit of empirical data to theory, highlights areas of disagreement and uncertainty, and provides a road map for future research into animal signaling.

This book has many strengths; chief among them is its logical organization. It begins with an introductory chapter reviewing the central problem of signal reliability and explaining relevant theory. To enhance accessibility to the theoretical foundation of this work, the authors utilize graphical versions of the mathematical models that were instrumental in demonstrating that signal costs could impose signal honesty. They introduce the thorny question of just what form these costs might take, a theme that is developed more fully in subsequent chapters that discuss empirical examples of signals. Also introduced briefly are alternatives to the handicap mechanism for maintaining signal reliability; these include an absence of conflict of interest between signalers, mechanistic constraints on signal production, and memory-based individual reputations (here awkwardly termed “individually directed skepticism”). This chapter is the clearest treatment of these topics I have encountered and one I will draw on heavily for my next lectures on the subject.

Chapter 2 discusses signaling when signaler and receiver have overlapping interests. A number of models are presented, most of which focus on signaling between two relatives. The first type of signal discussed is chick begging in birds, and evidence is presented showing that parents do respond to chick begging and that begging rates generally (but not always) are reliable signals of chick need. The question of signal costs is revealed to be a more difficult one, with measures of metabolic activity consistently showing positive but small increases in energy expenditure during begging relative to baseline levels. Likewise, the potential cost of predator attraction to begging chicks has equivocal support, with some evidence showing increased predator attraction for ground-nesting, but not tree-nesting, birds. Despite the low level of costs thought to be associated with begging, there is relatively little evidence of deceitful exaggeration of need. Thus, the first example in the book highlights a mismatch between theory (which requires costly signaling for honesty) and data (in which signals appear generally reliable despite low costs). In this case the resolution may lie in alternative forms of the handicap theory that rely not on differential costs for different classes of signalers, but on different benefits for signalers depending on their level of need. The problem of quantifying signal costs is a recurrent theme in the book, though, and emerges as one of the most pressing challenges for theoreticians and empiricists alike.

The book proceeds to discuss signal exchanges in which the interests of signalers and receivers are increasingly divergent. The format introduced in Chapter 2 is maintained in these chapters, each of which begins with a general discussion of relevant theory followed by a detailed discussion of specific types of signals. For each example there are critical evaluations of the empirical evidence for receiver response, signal reliability, signal costs, and incidence of deceit. The consistent structure of these chapters readily facilitates comparisons among them and the identification of broad trends in the literature.

Chapter 3 discusses mating signals, in which senders and receivers each share an interest in creating viable offspring but may well disagree on whether the other is the most suitable individual with whom to achieve that goal. This is the context in which Zahavi (1975) first formulated the handicap hypothesis, and it has been a fertile area for research. Signal examples include carotenoid pigments, song in oscine birds, and tail length in birds. Chapter 4 discusses signaling in aggressive conflicts, in which the interests of signaler and receiver are in opposition. Signal examples include avian postural displays, avian badges of status, territorial calling in frogs and toads, and weapons displays in crustaceans. It is the latter example that provides the best-documented case of deceptive signaling. Territorial stomatopods give a meral spread display to intruders that indicates their willingness to attack with their club-like raptorial appendages. When challenged by intruders, newly molted stomatopods will give meral spread displays despite their inability to follow through with an effective attack. These bluffs are often effective, presumably because bluffing is relatively rare and intruders who mistakenly call the bluff of an honest signaler would pay a high cost.

Evident in these chapters is the intuitively pleasing trend that as the interests of signalers and receivers become more divergent, the costs required to maintain reliable signal systems become higher. Also evident is that even systems with high-cost signals are not immune to some level of bluff or exaggeration, although determining exactly when dishonest signaling has occurred remains difficult in most systems.

Chapter 5 expands beyond dyadic exchanges between a sender and a single receiver to discuss communication networks. Such networks are formed when third parties intercept signals exchanged between senders and receivers. Eavesdropping occurs when a third party gains information from these signals and alters its behavior accordingly; such eavesdropping may have important implications for the evolution of honest signals. In some cases eavesdroppers may impose an additional cost on signalers that could help maintain signal honesty. One example may be aggressive responses by eavesdroppers to signals of high status given to other receivers. In other cases, eavesdropping may accelerate the formation of reputations for individual signalers that may promote honest signaling. It is clear, however, that signaling in communication networks remains a relatively unexplored topic, with much interesting work to be done.

Chapter 6 summarizes the evidence for the handicap hypothesis and alternative mechanisms for enforcing signal reliability. After this thorough review, can we conclude that animal signals are on the whole honest or is the animal world rife with deceit? The answer according to Searcy and Nowicki is ‘yes’ and ‘yes.’ There appear to be many systems in which animal signals serve as reliable indicators of quality or intention, at least on average. In some cases this reliability is enforced through differential signal costs as envisaged by the handicap mechanism, while in other cases mechanisms such as differential benefits to receivers, signaler reputations, or production constraints may be critical. The fact that signals in a given system are generally reliable, though, allows some fraction of signalers to exploit the system with exaggerated or dishonest signals. Thus, honesty and dishonesty are linked not just axiomatically, but through the costs and benefits of signal exchange within a population of individuals.

The Evolution of Animal Communication leads the reader through these complex issues with exceptional clarity. The language is lucid throughout and care is taken to clearly define terms and to explain how these definitions may differ from those used by previous authors. The figures are of good quality and judiciously chosen to illustrate examples in the text. The critical theoretical models are translated into English, with the mathematical details confined to supplementary boxes. Perhaps most importantly, throughout the book both models and data are never taken at face value, but are evaluated with careful consideration of underlying assumptions, limitations, and alternative explanations.

The book is not without some limitations, however. Chief among them is that it introduces relatively little in the way of new concepts or theories. While much care is given to dissecting the ideas of others, there is relatively little expansion on these ideas other than to point out areas that merit further investigation. The exception to this trend is the detailed exposition of the authors' own “developmental-stress hypothesis.” This intriguing hypothesis suggests that many aspects of oscine bird song, including song complexity, repertoire size, and local convergence, can be viewed as reliable indicators of the degree of nutritional stress endured by young males during critical periods of song learning. Developmental stress is in turn linked to aspects of male fitness of interest to females. The description of this novel hypothesis given here is the most comprehensive to date, but there are many aspects remaining to be explored. For example, to what extent is the development of the brain and male song directly linked to aspects of male quality important for females? Do females really prefer well-spoken males? How does the developmental stress hypothesis apply to those species that exhibit seasonal plasticity in learning and the volume of song learning centers in the brain? Does developmental stress play a role in the development of duets and calls in females? Does it apply at all to open-ended learners, like parrots, that appear to be able to modify their calls throughout life? How can we better quantify the costs and benefits of vocal learning? These are exciting questions, the answers to which may shed light on the origins of vocal learning in birds.

One area that receives little attention in this volume, despite the use of the word “evolution” in the title, is the evolutionary origin of signals. The questions of why signals take the form that they do and why they vary so greatly among different species were early foci of ethologists (e.g., Lorenz 1971), but have received less attention in the last few decades from behavioral ecologists. Two recent trends suggest that a resurgence of interest in the evolutionary origin of signals may be in the offing. One is the appearance of increasing numbers of phylogenetically based comparative studies of signal structure (e.g., Price and Lanyon 2002, Seddon 2005). The other is a growing interest in the links among sensory abilities, the environment, and the form of signals (Boughman 2002). Hopefully within another decade or so this area will be sufficiently developed to merit the same sort of thorough review that Searcy and Nowicki have given to the topic of signal reliability in this volume.

Despite these minor caveats, ornithologists with an interest in animal behavior will want The Evolution of Animal Communication on their shelves. Many will also want to read it. A majority of the examples are drawn from studies on birds, and the themes of reliability and deception are central to an understanding of the communication signals that enable social behavior. The book would provide an excellent springboard for a graduate seminar, especially in combination with Maynard Smith and Harper's (2003)Animal Signals, which includes a discussion of signal reliability within a more general treatment of animal communication. Perhaps most importantly, Searcy and Nowicki have provided an excellent model for how to distill the essence of a topic that is as complex as any in modern biology.

Literature Cited