The primary goal of the Banded Wren study is to determine how males use their repertoire of song types in vocal exchanges with rival territorial males. The project is directed by Professor Sandra Vehrencamp, in collaboration with postdoctoral associates John Burt, Michelle Hall, and Selvino deKort, former students Laura Molles, Paula Trillo, and Anya Illes, the engineering staff of the Bioacoustics Research Program, and numerous undergraduate assistants.

Male birds use their melodious songs to both attract females and defend their territories against rival males. How can a single vocal signal serve such disparate functions? Most oscine songbird species actually sing several to many different songs, which we call song types, and the set of song types used by a given male comprises his song-type repertoire. Each male learns these song types through exposure to the songs of other adult males. Species of oscine songbirds differ in their ability to learn early versus later in life, in their propensity to copy whole song types versus invent new song types combined from learned elements, and in the size of their final repertoire of song types or elements. These learning strategies determine the way in which males can use their songs to communicate with females and other males.

Our study focuses on the male-male communication function of complex singing behavior. Why is this an important issue? Acoustic research by many investigators on a wide variety of avian species has turned up a number of good answers to the question of how song types are used to attract mates. Females have been found to prefer to mate with those males having larger repertoires in every songbird species examined to date. A female may obtain several kinds of benefits by choosing a male with a large repertoire. In species that continue to learn new song types well into adult life, repertoire size is an index of male age.

Older males have proven their ability to survive, and they may possess better territories and offer females more breeding experience. In species with a rapid learning period early in life, the repertoire size of a male may be determined by his nutritional condition, health, and/or foraging ability. In species with extreme sexual selection for large repertoire size, males often mimic the sounds of other species as well as environmental noises, using these vocal elements as raw material for the invention of highly diverse and unique songs that are attractive to females. Female choice for diverse and complex singing behavior is probably the driving force behind the evolution of song learning and the intricate syrinx during the early radiation of oscine songbirds.

By contrast, in those species that continue to sing after mating and clearly use song for territory defense as well, a high diversity of sounds is usually not an important feature. Such species often have small to medium-sized repertoires of song types. The songs themselves are relatively short and separated by significant silent gaps for listening to song responses by other males. We frequently find that males copy some of the song types of their adjacent neighbors. When males share some but not all of their song types with neighbors, they can potentially use shared versus non-shared song types to indicate different levels of aggressive arousal. In particular, they can reply to a song of a rival with a song of the same type, called type-matching, which seems to be a very threatening signal in several species. Song types can be used in other structured ways to encode different kinds of information. For example, synchronized switching could be used to point to a particular singing rival, and songs with different structures, frequencies, durations, or amplitudes could encode different levels of aggressive motivation. The question we pose is: how can different singing strategies serve as effective threat signals and remain honest despite the temptation to bluff? If certain ways of singing are reliably associated with different levels of aggressive motivation or fighting ability, then countersinging interactions may resolve many boundary disputes without the necessity of escalating to a physical fight, to the benefit of both interacting males.

A Banded Wren, Thryothorus pleurostictus, tending its nest

The Banded Wren (Thryothorus pleurostictus) is a common and vocally active species that inhabits the tropical dry deciduous forest of the Pacific slope of Central America. It breeds only during the rainy season but remains resident and paired on the same territory during the dry season. Each male has a repertoire of 20 to 25 song types, which are relatively easy to distinguish with the human ear.

Young males tend to copy whole song types from nearby males and generally do not disperse very far from their natal territory, so established adjacent neighbors share between 50 to 90% of their song-type repertoire. (To see and hear some examples of shared song types, click here.) Males typically switch to a new song type after every song, and during the non-aggressive dawn chorus period they deliver most of their different song types. During mid-morning countersinging interactions between males, birds preferentially use their shared song types, they reduce their song-type diversity, and they often repeat the same song type several times. We believe these subtle changes in singing patterns signal important information about tendencies to approach or withdraw from aggressive encounters. Males frequently attempt to extend their boundaries into the territories of their neighbors, especially after a nest predation forces them to renest in a new location, so boundary disputes are common.

We are using two relatively new and sophisticated techniques to determine the relationships between different singing patterns and the aggressive behaviors of both the singer and the receiver. One technique is multiple microphone array recording. With simultaneously-recording microphones spread throughout a neighborhood of adjacent territories, we not only obtain a record of the song types sung by each of several males, but we can also triangulate the location of each male at the time the song was sung. All countersinging interactions during the recording session can then be analyzed carefully. We will be able to determine which singing patterns are associated with periods of approaching versus retreating from a boundary with a singing rival male. Moreover, interactions that were successfully resolved with song can be compared to interactions that ended in a physical fight. The array system can also be used to monitor the singing and movement behaviors of focal birds and neighbors during experimental manipulations, such as temporary male removals, speaker replacement experiments in territories with removed males, and song playback experiments.

Sandra Vehrencamp operating the digital recording station at the Costa Rica research site.

The second technique we have been using in the banded wren study is interactive playback of song stimuli. Song playback is a useful technique for quantifying the meaning of, and aggressive responses to, different singing strategies. Traditional tape loop playback experiments cannot adequately simulate the types of vocal interactions that occur during countersinging because it is not possible to change either the song type delivered or the precise timing of the song playback in response to the focal bird's songs.

When birds interact vocally, the time delay between each rival's songs appears to provide important information. In the banded wren, neighbors initially avoid overlapping each other's songs and alternate in a regular way. However, if the vocal conflict escalates, birds may start to overlap each other, and one bird may even precisely finish the song that is begun by the other if they share the song type. We use interactive playback to control (or systematically vary) both song-type diversity and song overlap. The only way to investigate the meaning of song-type matching is with interactive playback. We use laptop computers on which a variety of different prepared song stimuli have been stored. A specific song can be broadcast through a battery-operated speaker at the stroke of a key. The software keeps a record of the songs played and other events that the investigator might wish to note. We have both a Macintosh-based software program called Singit!, and a PC-Windows-based program called Syrinx.

During the course of this study we will be monitoring the long-term success of color-marked males who differ in repertoire size and percentage of song types shared with adjacent neighbors. We are particularly interested in following the dispersal success and learning process of young birds. Males that have larger repertoires, or that share more songs with their neighbors, may have fewer aggressive interactions with their neighbors, longer periods of territory tenure, and/or greater probabilities of attracting and maintaining mates. We are developing a computer simulation model of song learning that generates testable predictions about the learning strategies and ecological factors that affect sharing patterns.

Finally, a long-term goal of this research program is to compare the use of song types for regulating male-male interactions in this species, the Banded Wren, to the use of song types in other avian species with different learning strategies. For example, the Song Sparrow has a very narrow window of song learning that is restricted to the first few months of independent life, compared to the approximately nine-month window of learning for the Banded Wren. Song Sparrows possess smaller song-type repertoires than Banded Wrens (i.e., 8 to 10 song types per male) and song-sharing levels are much lower (0 to 40% sharing depending on the population). We have recently determined that Song Sparrows, like Banded Wrens, use type-matching to indicate strong aggressive intentions, and they also preferentially use shared song types in interactions with established neighbors. However, some adjacent Song Sparrow males share no songs and can't employ these strategies. The song-sharing level of a territorial bird with its neighbors may be a better indicator of a male's dispersal distance and fighting ability in a species such as the Song Sparrow than in the Banded Wren. Other species such as wood-warblers have very long learning windows and the ability to copy whole song types from new neighbors. We plan to investigate the meaning of shared versus non-shared song types, and of type-matching, in such species using techniques similar to the ones described here.

Research papers published on this work