Use of vocalizations to establish species limits in antbirds (Passeriformes: Thamnophilidae)

MORTON L. ISLER,13 PHYLLIS R. ISLER,1 AND BRET M. WHITNEY2

ABSTRACT.-We introduce an empirically derived methodology for the definition, measurement, and application of vocal characters in assessing species limits in the numerically important Neotropical family Thamnophilidae. On the basis of available evidence, we assume that vocalizations are innate and that all vocal characters have a role in maintenance of species integrity in thamnophilids. Vocalizations of eight syntopic pairs, the members of which resemble one another in both plumage and voice, were analyzed to identify diagnosable vocal characters. To be diagnosable, characters had to distinguish members of a pair unambiguously. Three of the eight pairs were diagnosable by three vocal characters, four pairs by four characters, and one pair by seven characters. Vocalizations differed most consistently by characters reflecting loudsong note structure, loudsong pace, and note structure of calls. In establishing species limits for allopatric antbird populations, we suggest that three vocal characters, the minimum number distinguishing the syntopic pairs, be used as a point of reference. Keeping this point of reference in mind, we recommend that multiple diagnosable vocal characters be present when vocalizations are a major factor in determining species limits in thamnophilid antbirds. To the extent that populations differ strongly in other characters (e.g. morphological, molecular, other behavioral), fewer vocal characters may suffice. The identification of multiple vocal characters as a point of reference, as opposed to individual characters (e.g. note shape) or vocalization types (e.g. the loudsong), allows for greater latitude in the diagnosis. A focus on the number of characters is appropriate given the possibility that the role and importance of vocalization types in species' repertoires may differ across groups of taxa. The methodology and results may also have application in the study of other avian groups in which vocalizations are innate, especially the suboscines. Received 16 September 1997, accepted 2 February 1998.

THE CONSISTENT DEFINITION OF SPECIES iS a necessary foundation for systematic, zoogeographic, ecological, and conservation research. The Neotropics are home to more species of birds than any other region and support the most diverse ecosystems on earth (see Stotz et al. 1996). Among Neotropical birds, recent studies (Robbins and Ridgely 1992, Willis 1992, Prum 1994, Whitney et al. 1995, Bierregaard et al. 1997, Isler et al. 1997, Krabbe and Schulenberg 1997, Zimmer 1997) exemplify a growing concern that many taxa presently considered subspecies are more appropriately recognized at the species level. Additionally, undescribed but diagnosable populations are known to exist (Fjeldsi and Krabbe 1990, Whitney 1994, Brumfield and Remsen 1996). One of the numerically most important families of Neotropical birds is the Thamnophilidae (sensu AOU 1997), represented by 35 to 50 sympatric species in Amazonian forest localities. As with other Neotropical birds, their current ranking into species and subspecies was established by Hellmayr (Cory and Hellmayr 1924), Zimmer (1931 and subsequent papers), and more recent authors solely on assumptions of the significance of differences and similarities in plumage and external measurements. Furthermore, as Mayr (1982) noted, "every isolated 'species' was . . . scrutinized for the possibility that it was simply a geographic representative of some other species, in which case it was reduced to the rank of subspecies." The resulting judgments were unavoidably inconsistent and sometimes highly arbitrary.

Since these decisions were made, and especially over the past 20 years, knowledge of Neotropical birds has been augmented greatly. New specimens clarify our understanding of morphological variation and distribution. Molecular evidence regarding genetic variation among thamnophilid populations also is beginning to emerge (Capparella 1987, Hackett and Rosenberg 1990, Hackett 1993, Bates 1995, Brumfield and Capparella 1996). Thanks to a growing body of new information from the field, we now have the opportunity to consider vocalizations, behavior, and ecology together with morphology and patterns of distribution in assessing taxonomic rank and systematic relationships. Of all the newly emerging data sets complementing the study of specimens, that of vocalizations is the most complete. We now have in hand an inventory of nearly 9,000 recordings of thamnophilid antbirds, contributed by many field workers.

The use of vocalizations in avian systematics studies has been reviewed by Becker (1982) and Payne (1986). One potential problem with the use of vocal characters as a guide to species relationships is the possible effect of song learning. In this respect, it is important to distinguish between oscines, among which several elements of song learning have been documented, and suboscines, among which there is no evidence of song learning (Kroodsma and Konishi 1991). Experimental evidence that vocalizations are innate in one family of suboscines, the Tyrannidae, is substantial and convincing (Kroodsma 1984,1985,1989). In addition, neurological evidence indicates that a motor pathway in the anterior brain is essential for song learning (Brenowitz and Kroodsma 1996), and that the cell clusters that control song production in oscine forebrains are absent in at least one tyrannid studied (Kroodsma and Konishi 1991). Evidence for the lack of vocal learning in suboscines is supplemented by a finding of geographic congruence between genetic and vocal differentiation of a tyrannid complex (Johnson 1980, Johnson and Marten 1988) and by field observations, prominent among which are studies demonstrating the lack of geographic variation in tyrannid songs (e.g. Lanyon 1978).