Have you ever lain awake at night wondering how, when, and why birds sing? If so, Bird Song may be the book for you. But no one should imagine that it is about the charming and euphonious (or sometimes cacophonous) things birds say to one another. Catchpole and Slater present a competent and logically organized synthesis of the vast scientific literature on bird song, concentrating almost entirely on experimental investigations by researchers. “Bird song” is defined here in the narrowest sense as “long, complex vocalizations produced by males in the breeding season.” All avian utterances that do not conform to this narrow definition, and there are many, are excluded from consideration. The text has citations to nearly seven hundred scientific sources, and its dry prose makes liberal use of technical jargon. It begins with a concise review of the neurophysiology of sound production and sound reception in birds, discusses how songs are learned, and concludes with an inquiry into the evolutionary function of song and its variability in nature. It tells, in fact, a story of remarkable scientific advances.

Surprisingly little was known about the functional and acoustic complexity of bird song until the second half of this century. Although birds are eminently accessible to study, being found even in the centers of our busiest cities, rigorous scientific inquiry into their vocalizations awaited the invention of high-fidelity tape recorders and speakers. These are the basic tools that both field and laboratory workers use to record bird sounds and later to play them back to the birds used in tests. Since the earliest playback experiments, conducted in the 1950s, continuing technological advances in both sound and information processing have stimulated ever deeper inquiries into the structure of song and its function.

Birds have been the main object of studies of animal vocal communication because they are both ubiquitous and relatively cheap and easy to maintain in captivity. The birds under discussion here are members of the great “passerine” order, comprising nearly half of all birds, and distinguished anatomically from non-passerine birds by their grasping feet, with the first toe turned backward. But birds are by no means the only animals to “sing.” Myriad frogs and insects produce sounds to attract mates or to proclaim territories. There are even singing mammals, including a number of monkeys, the South American bamboo rat, and, of course, whales.

Birds produce sound in a structure called the syrinx, located deep in the throat at the juncture of the bronchi, the two main branches of the windpipe. A membrane on each side regulates the passage of air. It was first thought that sound emanated from vibrations of the membranes, but it is now acknowledged that the source is air turbulence, produced much in the manner of a whistle. The truly remarkable feature of sound production by birds is that the two sides of the syrinx can act independently, so that some birds are literally able to produce a chord. The song of the common cat-bird, for example, consists of a sequence of phrases, some of which arise on the left side of the syrinx, some on the right, and some on both sides working in concert. It took some very clever experimentation to discover this. Tiny thermistors sensitive to pressure had to be implanted in the middle of each bronchial passage and monitored simultaneously on a time scale of milliseconds.

Tones, once produced, are then modified by changes in the length and shape of the vocal tract, just as an opera singer is able to vary the quality of sound from mellow to edgy by relaxing or tightening the throat, and varying the size and shape of the cheeks and mouth (the “buccal cavity”). Birds perform analogous operations by lengthening and shortening the neck and opening and closing the bill. The highly coordinated activity that generates song does not, however, originate in the syrinx itself, but in as many as nine neural centers in the brain. Separate but interconnected neural pathways are involved in producing sounds and in learning to make sounds.

With many songbirds the males sing and the females don’t. Such marked asymmetry of behavior led Rocke-feller University scientist Fernando Nottebohm and others to examine the avian brain, a significant portion of which is devoted to sound production and processing. Nottebohm’s investigations led to the first demonstration of a distinct difference in vertebrate brains between males and females of the same species; the neurological centers for sound production were revealed to be three times as large in male canaries as in females. Since Nottebohm published his results, individual differences in the structure of human brains (as between males and females or homosexuals and heterosexuals) have become one of the hottest topics in human neurobiology.

Nottebohm’s studies of the neurological basis of song in the canary yielded another major discovery, one that may also have important medical repercussions. Song in canaries is seasonal, as it is for most other birds that breed in the temperate regions of the world. It has long been known that seasonal variations in hormone levels, mediated by the changing length of days, were associated with differences in vocal activity and reproductive behavior. What had not been appreciated was the extent to which the regulation of seasonal behavior was reflected in the structure of the brain itself. It seemed reasonable to think that varying hormonal levels could have an effect on preexisting neurological circuits so as to turn on and off seasonally appropriate behavior.


Instead, Nottebohm discovered that the principal center for song production in the canary brain dramatically increases in size as the birds enter the spring breeding period. Precise analysis of brain tissue with radioactive tracers provided evidence that enlargement of the song center was accompanied by cell divisions and an increase in the number of neurons making up the center. Until then, it had been accepted among neurobiologists that new neurons could not form in the adults of higher vertebrates. Nottebohm’s discovery has led to a flurry of new research into the factors that regulate the proliferation of neurons in brains and other neurological structures.

For a bird to sing, it has to know what to sing. There are two ways that birds acquire such knowledge, and the distinction divides songbirds of the passerine order into two major divisions. In the so-called suboscines, including particularly the New World flycatchers, song is genetically encoded, so that birds raised in acoustic isolation are capable, when mature, of emitting the characteristic vocalizations of the species. Among the oscines (including all other songbirds, such as warblers, finches, wrens, and many more), song must be learned. Auditorily deprived birds are only able to generate disorganized sounds or, at most, slurred and garbled versions of the typical song.

Much research has been invested in studying how oscine birds learn to sing. When do they learn, and what instructs them? Although the precise answers to these questions vary with the species, one universal finding is that birds are not indiscriminate learners. Birds exposed to inappropriate sounds, including the songs of closely related species, do not learn how to sing. Nor do they imitate irrelevant sounds. They must hear members of their own species, which shows that song learning involves a complex interaction between experience and instinct. Instinct guides what is and what isn’t learned, but in oscine birds instinct alone does not suffice to pass on the appropriate behavior.

Most species have a “learning window,” a time in their first year of life when they acquire the necessary experience. Often this takes place during the fledgling and post-fledgling months when the birds themselves are silent. The details of the song are memorized, forming an “auditory template” that then serves as a basic reference the following spring when rising hormone levels induce the first attempts at song.

The first utterances, termed subsong, may come months after the crucial learning experience. Subsong tends to be weak and variable, often containing only elements or rudiments of the final product. With practice, the efforts become more structured and less variable, until “crystallization” occurs. By this time the bird is singing in full voice, and producing a highly structured and precisely repeated rendition of the song typical of its species.

Learning may or may not continue after crystallization of the song. Some species gradually develop elaborate repertoires of dozens or even hundreds of variants of a basic song, in which elements are added, deleted, or recombined. A familiar example is the mockingbird, renowned for its ability to imitate other species. Mockingbirds borrow bits and pieces of the songs of many other species, and weave them into an exuberant and highly varied outpouring. But mockingbirds living in urban or agricultural environments, where there are few models to imitate, sing relatively barren and uninteresting songs.

The world record holder for the largest variety of songs may be the brown thrasher, a relative of the mockingbird, whose song consists of repeated phrases, A-A, B-B, C-C, etc. What is remarkable is that some brown thrashers have been found to include well over one thousand different phrases in their repertoires. In variability and individual flavor, the songs of most passerine birds fall somewhere in between the highly stereotyped utterances of suboscines and the nearly limitless versatility of the brown thrasher.

Why mimics such as the mocking-bird of North America, the lyrebird of Australia, and the nightingale of Europe imitate other species is still an unanswered question. But the related question of why individual birds elaborate their own distinctive variants of the species’s basic song has received a great deal of attention, and some clear answers have emerged.


The basic answer is that songs serve multiple purposes. But this now widely accepted conclusion was a long time in coming. Romantics and poets held that birds sang for the sheer joy of being alive, or in rapture over the long-awaited arrival of spring. But Darwin, in keeping with his theory of natural selection, proposed the more practical view that males sing to attract females. It took fifty years before E. H. Howard put forward, in 1920, the alternative theory that male birds sing to advertise their dominion over a patch of space, a territory. The two views are really quite different but in practice they are difficult to disentangle. In the first case, males direct their exertions at females; in the second, at rival males. Which is it? In reality, it is both—and more.

It has taken decades of experiment to show that there is no simple answer to the question “What is the function of bird song?” Birds have several things to communicate—their species, territory location, sex, marital status (whether paired or not), and even individual identity. All this information can be, and often is, coded into the songs we are so pleased to hear each spring.

Early experiments were directed to supporting or refuting the hypotheses of Darwin or Howard. Redwing blackbirds that had been experimentally muted by puncturing the interclavicular air sac were vulnerable to increased rates of territorial incursion by other males; they engaged in more fights, and, in some cases, lost their territories. Muted birds partly compensated by resorting more to visual displays of their red epaulets. When the muted birds recovered their ability to sing, moreover, they rapidly regained lost territory, clearly demonstrating that song was effective in competition with neighboring males.

As straightforward as this experiment might seem, there remained the difficulty of distinguishing the effects of song from those of other characteristics of the experimental males, such as their size, age, luster of plumage, and degree of dominance. In experiments conducted in England by John Krebs, males were replaced by pure song in the form of recordings played in sequence over several speakers placed in the Wytham Woods, near Oxford. Krebs found that male great tits delayed in setting up territories where speakers were playing songs peculiar to their species, but they readily occupied adjacent areas. Eventually, however, the experimental areas were invaded and occupied as well, indicating that pure song was initially effective in guarding territory, but that males ceased to be deterred when the song was not backed up by the assertions of a live resident. These and other experiments clearly indicate an important role of song in competitive interactions among males, as Howard had surmised. But what about Darwin? He was seldom wrong in his insights.

It took a clever experiment to demonstrate the role of song in attracting females, which tend, as the authors put it, to be “elusive and retiring” denizens of male territories. Researchers in Scandinavia took advantage of the scarcity of natural tree cavities in secondary woodlands to attract flycatchers by using artificial nesting boxes. Half the boxes were outfitted with a dummy male; the other half of the boxes also had dummy males, but, in addition, were equipped with speakers that played the male song. The boxes were booby-trapped with spring-loaded nets, so that prospecting females would be caught if they perched at the entrance. The result was definitive. Nine times as many females were captured at the boxes from which song emanated. Clearly then, male song is heard and responded to by females as well as by other males, and plays important roles in communication between both sexes.

We can still ask why do mocking-birds, brown thrashers, and members of many other species elaborate their own individual refrains by modifying elements of a more or less standard song. We can conjecture that it may be advantageous for a particular bird to be known to its neighbors or its mate. And, in turn, it may be advantageous for a bird to be able to distinguish its neighbors from strangers—potential intruders into its territory.

The classic experiment designed to investigate this question was conducted on ovenbirds, and later on white-throated sparrows, by three Americans, J. Falls, J. Weeden, and R. Brooks. They selected birds that had established territories and directed at them recorded songs both from neighbors and from unfamiliar males. The songs were broadcast from locations on the territorial boundaries. Male birds responded aggressively to the threat of a territorial intruder, frequently rushing up to the loudspeaker in a state of visible agitation. However, in this experiment, the responses to neighbors were subdued in comparison to those accorded the songs of unfamiliar males.

In a variant of this experiment, Brooks and Falls compared the different reactions to the songs of neighbors when the songs were played along the shared territorial boundary and when they were played along the boundary on the opposite side of the territory. In the latter case, the neighbor’s song was coming from an unexpected direction and elicited a response as strong as that of a stranger. The birds remained calm when the usual geographical relation to neighbors was respected; but they reacted with alarm at signals suggesting that something was awry. By knowing its neighbors and what to expect from them, a bird saves time and energy by avoiding unnecessary aggression.

An extreme case of a benefit derived from individual recognition of vocalizations may be found in some seabird colonies, whose inhabitants can number in the thousands. The young of many seabirds are precocial, that is, they hatch with a covering of feathers and begin to walk around long before they are able to fly. How does an adult returning with food find its chick among hundreds of others? In experiments with guillemots, young birds were played the calls of their parents and other adult members of the colony. The results were decisive. The young birds responded only to sounds produced by their parents, and ignored or avoided all others. In further experiments, it was found that young guillemots learn their parent’s calls while still in the egg, so important is it for them to make the right connection upon hatching.

In view of the crucial importance to both males and females of recognizing the songs of their own species, investigators began to wonder what exactly it was that was being recognized, the whole song or just certain features of the song. The pioneer in this field was a gifted Frenchman, J.-C. Bremond, who combined a passion for birds with an advanced knowledge of electronics and computers.

Bremond worked with European robins, a species that sings an elaborate and varied song. He showed in 1968 that the song was constructed in accord with a set of “syntactical rules,” such as “all phrases within a song are different,” “all songs in a bout [a linked series of songs] are different,” and “successive phrases alternate between high and low pitch.” A knowledge of these rules then made it possible to modify artificially generated song in ways that violated one, but not the rest, of the rules. For example, when reconstructed songs consisting entirely of high or low phrases were presented to male robins, only half of them responded, instead of the usual 90 percent that responded to unaltered songs. In further experiments, Bremond used sound generators to synthesize completely artificial songs composed of simple sine waves, which oscillate in regular patterns. Modulated sine waves that alternated between high and low pitches elicited no response from male robins, but when separate elements following the other systematic rules were included, the response jumped to nearly 70 percent.

Bremond’s ground-breaking work led to a major growth industry in the analysis of bird song. One finding to emerge from two decades of research with artificially modified or constructed songs is that the critical features differ between species. Syntax, the structure of the different elements of the song, timing, and frequency can each be important. In most species, the consistency of several features assures fail-safe recognition among members of the same species.

It is now established that in the songs of many species there are invariant as well as variable features. Information peculiar to the species is carried in the invariant features, while information about the individuality of the singer is carried in the variable features. Songs are thus far more complex than anyone imagined a generation ago; they carry multiple messages directed at different groups of potential listeners.

So far we have considered the function of song mainly from a male perspective. Less research has been done on females, because the females of many species in the temperate zones don’t sing, and because their responses are so much more difficult to elicit and evaluate than are those of males. Nevertheless, it is crucially important for females to recognize males of their own species if they are to avoid the heavy cost to their own reproductive success of producing sterile hybrids; such recognition is particularly important for naive birds that are breeding for the first time.

One simple but ingenious experiment showed conclusively that females possess superior ability to distinguish the song of their own species. Two recordings were played to both male and female redwing blackbirds. The first contained song produced by male redwings; the second was a mockingbird’s imitation of a redwing’s song. Male redwings responded strongly and indiscriminately to both recordings. In contrast, females reacted four times more frequently to the real thing than they did to the imitation, showing that the imprecisions of the mockingbird’s rendition, subtle as they were, were noticed by and, moreover, mattered to the females.

Female redwings not only listen to males but sing themselves. Unlike the females of most songbirds, female redwings often live in territories where the males have more than one mate, and each female defends an individual sub-territory against other females within the larger territory of the resident male. Female redwings are also atypical in producing two types of songs, one in the presence of a male, or when exposed to playback of male song, and a second when other females appear. When they are confronted with playback of female song of either type, females produce only the second kind of song. The second song type thus appears to be the counterpart of male song, in that it is used in defending territory against rival females.

The authors state quite categorically that the female members of only a “small number of species” usually sing. Here they reveal both a bias in favor of the temperate zones and a deep ignorance of tropical birds. In the tropics (where a large majority of the world’s 9,600 bird species live), mates commonly remain paired and resident on the same breeding territory throughout the year. Among tropical birds, it is common for females both to sing and to exclude other females from the territory; they defend their reproductive prerogatives just as males do. The entire subject has so far been little investigated, but among Amazonian species, females may sing in a manner indistinguishable from males (at least to my ear), or they may produce songs that are quite noticeably different.

As the authors rightly point out, vocal duets are common among tropical birds. Such duets can be “antiphonal,” in which one bird sings phrases in alternation with its mate, or “contrapuntal,” in which two independent melodies are sung together; or they can be more loosely coordinated. But the female part is usually distinctive. Unlike the male song of temperatezone birds, a duet carries the message that both residents of the territory are paired, and that no intruders of either sex are welcome.

At the field station in Peru where I spend several months each year, I have observed a common wood quail that lives, like most quail, in family units. Each morning and evening, mated pairs sing a sonorous, rollicking antiphonal duet composed of fournote phrases—ter-weet-chur-wee, ter-weet-chur-wee, etc. But occasionally I will hear a half-duet, either ter-weet, (pause) ter-weet, or chur-wee, (pause) chur-wee, indicating that a breeding bird has died, or that a young bird is attempting to attract a mate into a new territory. In the tropics, where territories are difficult to acquire and often held for the lifetime of the occupants, the singing of duets has evolved to advertise the status of the territory and the availability of vacancies.

The duets of the Amazonian wood quails are a prime example of what students of animal communication have termed “truth in advertising.” The highly coordinated roles of the two sexes provide accurate information about territorial occupancy by sex, and perhaps even about the physical condition of the occupants. May there also be circumstances in which birds “deliberately” try to deceive one another by sending misleading information? The whole question of deception in animal behavior is highly controversial, and therefore a subject of active research.

Professor John Krebs of Oxford has put forward the “Beau Geste” hypothesis concerning deception. He wondered why in so many bird species the males possess a whole repertoire of song variants. Is it because they are trying to give the impression that the territory is hopelessly overstocked with males and that competing males should stay away? The hypothesis is named for the fictitious French Legionnaire who single-handedly defended a fort against an overwhelming force by propping up dead comrades all around the ramparts and firing furiously from behind their backs. He put up such an impressive show that the demoralized enemy slunk off in retreat.

Krebs himself has put the Beau Geste hypothesis to a test. Broadcasting the songs of great tits, he found that recordings of birds with large repertoires were more effective at deterring would-be territory seekers than recordings of variants of single songs. But it must be said that this outcome was also the more natural one, since male great tits commonly use several song types, and the more experienced (and perhaps dominant) male birds tend to have larger repertoires.

Krebs’s experiment therefore could not be considered decisive, but, in my view, another one was. Captive great tits were conditioned to peck in response to the recorded vocalizations of only one of two males. However, the subjects were exposed only to a sample of the song types in the repertoires of the two males. When recordings of the same two males were played, now singing song types the subjects had never heard before, they correctly discriminated between them. Surely, birds in the wild are not going to be fooled into thinking that a large repertoire must emanate from several individuals. In all likelihood they recognize individual voices just as surely as any of us recognizes a friend over the telephone from a simple hello.

If we put aside the Beau Geste hypothesis, is there any solid evidence for deception in the vocal communication of birds? The evidence brought forward so far is scanty indeed, but it may not reflect the occurrence of deception in nature because, if effective, the ruse may fool the human observer as well as the intended victim.

One particularly delightful example was exposed by Luis Baptista in his long experience with white-crowned sparrows. During the fall, winter, and early spring, the females sing a truncated version of the male song, but they lapse into silence as the breeding season begins. Nevertheless, their ability to produce song does not subside, as is occasionally revealed when the male is overdue in taking his turn to sit on eggs before they hatch. At such times the female can produce a full song, indistinguishable from that of a male; this suggests that she is feigning the amorous advances of an intruder in order to bring her mate back to the nest.

The complexity and sophistication of animal communication are made clear in Bird Song, but the story is far from complete. Every species is distinct in some way or another. But more than that, the authors do not address entire topics in their highly concentrated work. What, for example, do birds say to one another, apart from the male songs considered here? Surely many things, because birds produce myriad vocalizations—assorted notes, calls, and other signals. Song, as has been seen, can communicate information about species, sex, territorial location, and individual identity, and, through its repertoire, it can provide clues to age and social status. But these are not the only things birds and other animals can communicate. Among the others are danger (through the sounding of alarm), sexual readiness, fear, hunger, aggression, scolding (particularly of predators), contentment, irritation, and location. Bird Song barely mentions alarm calls, and doesn’t touch upon any of the rest.

One might therefore imagine a companion volume on vocalizations apart from song. But at this stage in the investigation of avian communication, it would be a short one. Vocalizations in forms other than song are much more difficult to investigate, because birds don’t respond to them in such clear and identifiable ways as they do to playback of primary song. Finding effective ways of studying what is in fact the main part of the avian vocabulary remains a major challenge to the investigator. Science has made a very satisfactory beginning in its understanding of bird communication—but it is only a beginning.

This Issue

January 11, 1996