Just as the colors and patterns of the feathers that birds wear show tremendous variation, so, too, do the songs that they broadcast--but much more so. Songs may be absent, or they may range from a few simple genetically encoded notes endlessly repeated, to virtuosos of variety resulting from copying and learning, and even to seemingly endless improvisation. In The Singing Life of Birds, Donald E. Kroodsma, an emeritus professor of biology at the University of Massachusetts Amherst, celebrates the diversity through carefully chosen examples, one for each of the 30 years that he has studied birdsong.

The book is best described by its subtitle, The Art and Science of Listening to Birdsong. Kroodsma shares his secrets--solid, practical advice on how to record bird sounds and how to "see" the sounds in sonagrams, visual representations of the recordings of songs. A compact disc that accompanies the text aids readers in this task. He concludes: "There's no longer any mystique to what I have done all these years. Anyone can do this kind of stuff. And anyone should."

His infatuation started with a single male Bewick's wren in his backyard in Oregon. Kroodsma discovered that this one wren sang 16 different songs, and in any singing bout it poured forth 40 to 50 renditions of one of them before switching to another, and then to another, and on and on. Meanwhile neighboring wrens hearing the song replied with the same one, while distant males sang other songs. Why?

The proximal answers to why birds sing and what they sing run from the trivial to the fascinating: they enjoy it, they are primed by hormones that activate neuronal pathways, they respond to neighbors. But the ultimate, evolutionary question of why they sing and what they sing can be answered only by the comparative study of many species.

Sometimes the anomalies provide a clue. For example, most individual wrens of different species learn many songs, and neighboring birds have similar songs--that is, they have dialects. The sedge wren of North America is an exception, however. Unlike other wrens and the sedge wrens of Central and South America, it has lost the ability to learn songs; it can only improvise on songs that are inscribed on its DNA. It is therefore unable to "match" the songs of its neighbors, and no dialects are found.

So what is different about the North American sedge wrens in respect to other wrens? They are nomads that live in unpredictable habitat--meadows that can quickly dry up. As a consequence, these birds can never predict who their neighbors will be from one season to the next; hence, learning songs as youngsters for later use in song matching is pointless. Contrast this to the bellbird, a long-lived tropical bird in which individuals come to know one another well. These birds listen to one another all year long and learn the changes in others' songs throughout life. The young birds learn the latest of these variations, and the dialect of the population changes from year to year.

Kroodsma takes us repeatedly into the field, into the birds' world. He shares an all-night vigil with a whip-poor-will, tallying 20,898 identical repetitions of its one song for the entire night. He describes a brown thrasher that in one two-hour session sang 4,654 songs, 1,800 of them different (many borrowed from neighbors of other species). We enter the mind of the researcher as he tries to penetrate the mind of the bird.

As much as we humans may enjoy the spectacle of birds flaunting their gaudy garb to the accompaniment of vocalizations and dancelike antics, the show is meant primarily to attract females. It is about sex--about who will be the father of the female's chicks. The males presumably enjoy putting on their show, but whatever else it may do for them (such as serving as a territorial marker), it is the females who have shaped the performance by their tastes and preferences, and these are as various as the 10,000 or so species of birds.

Kroodsma emphasizes that we know little about why one or another bird has a specific repertoire. Yet despite the dazzling variety, it appears to me that all birdsongs have general requirements and constraints, and I believe that these shared characteristics may in themselves shed some light on the enigma. The primary requirement of a species' display song is that it must stand out from environmental noise--that is, it must carry--and it must be distinct from competing voices on the stage. Once females reward a specific song type with mating, then success breeds success, and whatever it is that attracts, the male that has more of it enjoys a huge advantage.

But singing is not cheap: the performers are conspicuous to predators, and the displays are so costly in time and energy that the performers may appear to handicap themselves. I doubt, however, that it is the flaunting of handicap as such that attracts the females ("I am so strong and healthy that I have energy to waste on singing"). The singer must cater to the females' taste. As in our own fashions of clothing and music, there is not necessarily rhyme or reason in the specifically chosen attribute, except the most important one--it works.

Konrad Lorenz reputedly said that birdsong is "more beautiful than necessary." It seems to me that it is just as likely that the flamboyant displays of song and dance, of feathers and, in the bowerbirds, of decorated love shacks are indeed necessary, because females compare, and they are picky. Arbitrary though their criteria of choice may be, it is significant that we humans also find many of the same displays beautiful.