SPECIES LIMITS IN ANTBIRDS (THAMNOPHILIDAE): THE WARBLING ANTBIRD (HYPOCNEMIS CANTATOR) COMPLEX

Author(s): Morton L. Isler, Phyllis R. Isler, Bret M. Whitney Source: The Auk, 124(1):11-28. 2007. Published By: The American Ornithologists' Union DOI: http://dx.doi.org/10.1642/0004-8038(2007)124[11:SLIATT]2.0.CO;2 URL: http://www.bioone.org/doi/full/10.1642/0004-8038%282007%29124%5B11%3ASLIATT %5D2.0.CO%3B2

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOnes Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research.

The Auk 124(1):1128, 2007 The American Ornithologists Union, 2007. Printed in USA.

SPECIES LIMITS IN ANTBIRDS (THAMNOPHILIDAE): THE WARBLING ANTBIRD (HYPOCNEMIS CANTATOR) COMPLEX
M L. I,1,3 P R. I,1 B M. W2
Department of Vertebrate ZoologyBirds, National Museum of Natural History, Smithsonian Institution, P.O. Box 37012, Washington, D.C. 20013, USA; and 2 Museum of Natural Science, Louisiana State University, 119 Foster Hall, Baton Rouge, Louisiana 70803, USA
1

A.Six populations of Warbling Antbird (Hypocnemis cantator) currently considered subspecies are more appropriately recognized as species, given the multiplicity of vocal dierences that distinguishes them. These vocal dierences are as great or greater than vocal dierences documented between syntopic species-pairs in the family Thamnophilidae (Isler et al. 1998). Evidence is also provided that two forms currently considered subspecies are syntopic and that two others are parapatric without apparent physical barriers. Vocally, the pair of syntopic taxa were distinguished dramatically, but solely, by common calls. The nding that common calls were as diversied as songs in this species-group suggests that calls may be as important as songs in reproductive isolation. Future eld studies should therefore focus on clarifying the function of dierent types of thamnophilid vocalizations and elucidating their role in speciation. This is the rst contribution to a multifaceted analysis of the vocalizations and genetic structure of typical antbirds (Thamnophilidae) in Amazonia. In future work, results of the vocal analysis will be compared with a parallel genetic study, aer which the two analyses will be integrated to suggest a phylogeny. The present study and preliminary studies of genetic dierentiation (e.g., Bates et al. 1999) make it clear that the H. cantator complex has had a long evolutionary history that produced greater species-level diversity than was recognized previously. Received 15 January 2005, accepted 12 February 2006. Key words: biodiversity, Hypocnemis, Neotropics, speciation, systematics, Thamnophilidae, vocalizations.

Lmites de Especies en Thamnophilidae: El Complejo de Hypocnemis cantator

R.Teniendo en cuenta las mltiples diferencias en vocalizaciones que distinguen a seis poblaciones de Hypocnemis cantator que actualmente son consideradas como subespecies, es ms adecuado reconocerlas como especies diferentes. Estas diferencias en vocalizaciones son iguales o mayores que las documentadas entre pares de especies sintpicas en la familia Thamnophilidae (Isler et al. 1998). Tambin se provee evidencia de que dos formas actualmente consideradas subespecies son sintpicas, y que otras dos son paraptricas, sin barreras fsicas evidentes. Vocalmente, el par de taxones sintpicos se diferenciaron dramticamente, pero nicamente en sus llamadas comunes. El hallazgo de que las llamadas comunes estn tan diferenciadas como los cantos en este grupo de especies sugiere que las llamadas pueden ser tan importantes como los cantos en el aislamiento reproductivo. Por lo tanto, estudios de campo futuros deberan enfocarse en aclarar la funcin de los distintos tipos de vocalizaciones de los Thamnophilidae, y en elucidar el papel que stos juegan en la especiacin. Esta es la primera contribucin de un anlisis con mltiples facetas de

E-mail: antbird@cox.net

11

12

I, I, W

[Auk, Vol. 124

las vocalizaciones y la estructura gentica de los Thamnophilidae en la Amazona. En estudios futuros, los resultados de los anlisis de vocalizaciones sern comparados con los de un estudio gentico paralelo, para luego integrar los dos anlisis y sugerir una logenia. El presente estudio, junto con anlisis preliminares de variacin gentica (e.g., Bates et al. 1999) indican claramente que el complejo de H. cantator ha tenido una historia evolutiva prolongada que produjo una mayor diversidad de especies que la que ha sido reconocida hasta ahora.

V a relevant yardstick (sensu Mayr and Ashlock 1991) for estimating reproductive isolation and species status of sympatric and allopatric populations of suboscine passerines (Isler et al. 1998, Johnson et al. 1999, Baptista and Kroodsma 2001, Helbig et al. 2002, Remsen 2005), though cautions regarding individual variation and clinality apply (Isler et al. 2005). Recent studies of vocalizations of the suboscine family Thamnophilidae have documented extensive, geographically based dierentiation within some species (e.g., Isler et al. 1997, 1999, 2001; Whitney et al. 2000; Isler and Isler 2003). Parallel to these analyses, molecular studies of antbirds have revealed higher-than-expected levels of intraspecic genetic structure (Capparella 1988; Hacke and Rosenberg 1990; Hacke 1993; Brumeld and Capparella 1996; Bates et al. 1999; Bates 2000, 2002). Questions remain, however, about the extent to which species diversity as expressed by vocal characters is consistent with genetic dierentiation. Over a period of nearly two decades, we collected recordings of vocalizations of the Warbling Antbird (Hypocnemis cantator) from diverse parts of its extensive range and noted substantial geographically related dierences among vocalizations. More recently, exploratory molecular studies (Bates et al. 1999; Bates 2000, 2002) have revealed high levels of regional genetic dierentiation among some populations of H. cantator. General concordance in these results led to the inclusion of H. cantator in an ongoing study of thamnophilid differentiation in Amazonia in which vocal and molecular character sets are being developed for populations of ve widespread speciesgroups. Vocal and molecular characters will be employed independently to derive estimates of the most appropriate taxonomy of each group. These results will be compared, an eort will be made to understand the basis for any discrepancies between them, and an estimate of

phylogeny will be constructed from both data sets. Ultimately, the results of the ve studies will be integrated to provide tests of hypotheses of historical processes leading to avian diversication in the Amazon basin. Although an initial step within this long-term program, the results of the vocal analysis of the H. cantator complex are suciently robust to recommend modication of species limits at this time. The analysis builds on the methodology for considering taxonomic limits in allopatric antbird populations established in studies cited above. The criteria are conservative, and our conclusions may be revised aer incorporation of the molecular results. Additionally, because the geographic distribution of available tissue samples for H. cantator is much more restricted than the inventory of tape recordings, we expect that determinations of distributional limits of most populations and, by extension, taxonomic recommendations, will depend on the results of analysis of the more robust sample of recordings. We expect the molecular analysis to occupy a more dominant role in constructing a phylogeny, the balance depending on the concordance of the vocal and molecular results and consideration of morphological distinctions among populations. Given the accelerating anthropogenic alteration of the landscape in the Neotropics, empirically based re-evaluations of species denitions are not only vital for studies of speciation and phylogenetics, but are an urgent necessity for ecological and conservation research. M Hypocnemis cantator is distributed widely in Amazonia and the Guianan region (Zimmer and Isler 2003). Eleven subspecies were accepted by Peters (1951), and a 12th was described subsequently by Pinto (1966). All were diagnosed by plumage characters based on specimen comparisons. We accepted existing

January 2007]

Warbling Antbird Species Limits

13

subspecies and morphological distinctions to dene a baseline taxonomy and, to simplify exposition, we employed subspecies names without initials as to genus or species and abbreviated subspecies names to the rst three leers in maps and tables. Because of the possibility, however, that existing, plumage-based subspecies denitions may not fully reect distinctions among independently evolving populations as reected in their vocalizations, we further disaggregated the inventory of vocal recordings to search for possible diagnostic vocal dierences within the geographic range of named subspecies. Twenty-six geographic clusters (Fig. 1), identied in the text as subpopulations, were dened on the basis of existing knowledge of geographic limits of thamnophilid antbirds. For example, if a major river bisecting the range of an H. cantator subspecies was known to demarcate the ranges of sister populations of other thamnophilid taxa, recordings obtained on opposite sides of the river were analyzed as subpopulations. Initially, pairwise vocal comparisons were made between subpopulations of currently described subspecies (e.g., a1 was compared with a2). Then, data for subpopulations whose vocal data did not dier diagnostically were aggregated, and pairwise comparisons were made of the vocal characteristics of all subspecies and vocally diagnostic subpopulations. The nal results provided the basis for taxonomic recommendations, including a test of the extent of geographic congruence with plumage-based subspecies. Tape recordings were compiled from our own inventories, from currently unarchived contributions by other individuals, and from the Macaulay Library (ML, at the Cornell Laboratory of Ornithology), the National Sound Archive (NSA, at The British Library), and the Florida State Museum (FSM, at the University of Florida). Appendix 1 (online; see Acknowledgments) provides a list of recordings employed in the study (774 recordings from 196 localities) with locality names, recordists, and archival locations. No recordings were available from within the range of H. c. perava. We reviewed every recording to identify the number and sex of individuals vocalizing and to label every vocalization as to type. We identied ve vocalization types that we termed male loudsong, female loudsong, common call, chirr, and chit call. Loudsongs (following Willis 1967)

refer to the ringing series of notes delivered in a consistent paern oen described simply as song. Common calls are also multinote vocalizations delivered in a consistent paern but are much briefer than and structurally distinct from loudsongs. Common calls are sometimes subject to individual variation (e.g., in duration or frequency structure), but atypical common calls are almost always interspersed within lengthy series of typical ones. Chirrs are single, vibrant, harsh notes, and chit calls are abrupt notes that are typically repeated rapidly in clusters of two to four. Stereotypy of placement of spectral elements within these vocal types (relative position, following Remane [1952]) allowed us to place vocalizations into these categories with a relatively high degree of certainty that they were homologous. CANARY, version 1.2 (Bioacoustics Research Program, Cornell Laboratory of Ornithology), was employed to make a spectrogram of every vocalization type delivered by each individual of either sex on every recording. Visual characters were obtained by examining printed copies of all clearly delineated spectrograms (i.e., sampling was not employed). We considered a character to be diagnostic visually when examination of the character completely distinguished every spectrogram of one population from another. If there was any uncertainty in this regard, we conducted blind tests in which spectrograms were stripped of any identication except a randomly selected code number and sorted visually into groups according to perceived dierences. The groups were then considered to dier diagnosably if they proved to be separated geographically. We employed the following terms in visual descriptions of note shape. Downslurred means that the note drops in frequency (i.e., the spectrogram trace goes from a higher to a lower frequency). An upslurred note goes in the opposite direction. A at note shows lile or no change in frequency (a horizontal line on the spectrogram). A rectangular note is a at note with enough band width that the note appears as a rectangle on a spectrogram at the scale employed here. Frequency rises and falls in an inverted U note, and the spectrogram trace is distinctly rounded near the point of peak frequency. A chevron is like an inverted U but shows a sharp peak at the highest frequency, resembling an inverted V. Sometimes, inverted

14

I, I, W

[Auk, Vol. 124

F. 1. Geographic distribution of recordings and definitions of hypothetical populations analyzed in Stage 1. Recording localities are assigned to geographic sectors using methods described by Isler (1997). Solid circles = recording locations. Heavy lines connect recording locations included in each population and encircle other recording locations in the population. Areas outside the heavy lines contain populations for which we had no recordings. Currently defined subspecies: aff = H. c. affinis, can = H. c. cantator, col = H. c. collinsi, fla = H. c. flavescens, imp = H. c. implicata, not = H. c. notaea, och = H. c. ochrogyna, per = H. c. peruviana, sat = H. c. saturata, str = H. c. striata, and sub = H. c. subflava. No recordings were available for H. c. perflava. Numbers after abbreviations identify hypothesized populations within subspecies ranges. Vocalizations of per3 and col or sub have been recorded at sites in the hatched area, and labels for these populations are repeated on each side of the hatched area to locate where the population occurs. Both per3 and sub have also been taperecorded at the location marked A on the map. Loudsong recordings (n = 2) from Ilha de Maraj, Par, Brazil (identified in the figure as aff2), were omitted from the aggregated analysis because of uncertainty regarding the population to which they should be referred. U notes or chevrons are not symmetrical; one side of the note may be more intense (as expressed by wider band width on the spectrogram), in which case we refer to the le-hand or right-hand side of the note. Lengthy downward extensions to lower frequencies are called le- or right-hand tails. Note shape paern describes how the shapes of notes, including note duration, change throughout a multinote vocalization. Clear notes are represented on a spectrogram by distinct edges, whereas a note is said to become raspy when a

January 2007]

Warbling Antbird Species Limits

15

noticeable part of its edges dissolve into noise. Most raspy notes cover a wide band width and may or may not contain a distinct solid area of high intensity in their core. Terminal raspy notes in loudsongs were not included in descriptions of note shape paern because individuals oen vary the number of such notes in a series of songs. Visual determinations of change in peak frequency were made in reference to a horizontal line drawn through the peak frequencies of a spectrogram of a multinote vocalization. Peak frequency refers to the highest frequency aained by the most intense trace of an individual note. Rise, at, and fall refer to changes of peak frequencies among the notes and signify an increase, no change, and decrease, respectively. Quantitative measures of continuous vocal characters were restricted to a sample of recordings. Sample sizes (reecting number of individual birds recorded, not number of vocalizations measured) are provided in the Tables 25 (online; see Acknowledgments). If more than six suitable recordings were available, we sought to distribute the sample throughout the region dened for the population. In the nal stage, aer populations were aggregated, data for more than six individuals were available for most populations. To obtain quantitative measures, spectrograms were projected on the screen of a Macintosh G4 computer using default seings of CANARY (Charif et al. 1995), except that the display was set to smooth, overlap was adjusted from 50% to 93.75% depending on recording quality, and contrast was adjusted according to recording intensity, with care taken to retain all elements of the vocalization. Cursor measurements were made by PRI, typically at scales of 0.3 s inch1 and 4.0 kHz inch1. Spectrograms shown in the gures were made by exporting CANARY les into CANVAS, version 9.0.4 (ACD Systems, Victoria, British Columbia). Whenever possible, we measured three vocalizations of each vocalization type for six individuals of every population identied in Figure 1. It was not always possible to achieve this goal, because of an inadequate number of recordings for some populations and because a high level of recording quality was required for many of the measurements. As in the case of visual descriptions, terminal raspy notes were not included in measurements. Vocal characters were dened

on two levels. General characters included (1) number of notes, (2) duration, (3) pace, (4) change of pace, (5) note shape, (6) change in note shape, (7) note duration, (8) change in note duration, (9) interval duration, (10) change in interval duration, (11) peak frequency, and (12) change in peak frequency. Specic measures (24) were assigned to general characters and included the number of notes; duration; overall pace (measured from the beginning of the second note to the beginning of the rst raspy note); durations of rst, middle, and last clear notes and the intervals (spaces) following them; ratios between note and interval durations; and peak frequencies of rst, middle, and last clear note and ratios between them. Measures are described more fully in earlier papers (Isler et al. 1998, 1999), and measurement units are provided in table legends. In dening vocal characters that relate to species dierences in antbirds, we are interested in identifying distinct character states that have the potential for unambiguous signal recognition. Thus, rather than using standard statistical tests, which would only test for dierence of means, we apply more stringent criteria that require diagnostic dierences in vocal characters to represent discrete, non-overlapping character states, as described in detail in earlier studies (Isler et al. 1998, 1999). In the case of continuous variables, ranges of sample values cannot overlap, and the means (x) and standard deviations (SD) of the population with the smaller set of measurements (a) and the population with the larger set of measurements (b) had to meet the requirement: xa + taSDa xb tbSDb (1)

where ti = t-score at the 97.5 percentile of the t distribution for n 1 degrees of freedom (except for ratios, for which this statistical test is inappropriate). Visual and mensural characters were selected to distinguish populations of each pair only if they were considered independent. This required careful judgments of whether variation in one character could occur without resulting in a dierent outcome for another character. If two characters appeared to be highly dependent, one was selected that appeared most appropriate to the vocalization type. For example, because common calls comprised only a few notes that aorded a

16

I, I, W

[Auk, Vol. 124

limited number of comparable measurements, visually dened characters were deemed more informative. On the basis of vocal distinctions among consolidated populations, we assessed their taxonomic status, employing guidelines developed in previous studies. We recommended species status under the Biological Species Concept for populations that diered diagnostically in both vocalizations and morphology (Johnson et al. 1999). Using a study of vocalizations of syntopic, congeneric species-pairs of antbirds and subsequent applications (Isler et al. 1998, 1999, 2001) as a point of reference, vocal dierences between allopatric populations were considered diagnostic at the species level if they were reected in at least three independent vocal characters. As noted above, current subspecies denitions were accepted as representing diagnostic morphological dierences. Existing taxa that did not meet vocal requirements for consideration as a distinct species were maintained as subspecies pending results of the parallel molecular study. Future studies will describe the results of the molecular analysis, review the results of the vocal analysis in this context, and, eventually, estimate a phylogeny for the H. cantator complex based on both data sets. R Diagnostic dierences were identied in male loudsongs, female loudsongs, common calls, and chit calls. No dierences in chirrs were found between populations. Chirrs were recorded for 17 subpopulations but not for saturata2, peruviana1, peruviana2, peruviana4, peruviana5, subava, collinsi, and anis2. Given that chirrs tended to be the least frequently recorded vocalization type, their absence from the sample of some populations repertories may be a sampling artifact. However, the lack of chirrs in the recordings of subava and collinsi, for which there were large samples, may prove to be taxonomically relevant. No diagnostic vocal dierences were identied in pairwise comparisons of subpopulations (Fig. 1), with one exception: a dierence in common calls was found between implicata2 and the remaining subpopulations of implicata. Some other subpopulation comparisons, however, were constrained by small sample sizes, and although there were other pairwise comparisons in which

ranges of values did not overlap, these did not meet our statistical test. These dierences will be re-examined in the future, when the results of the ongoing genetic study are available. As a consequence of the pairwise comparisons of subpopulations, implicata vocal data were maintained in two populations, but data for all other subpopulations were consolidated into ten existing subspecies (no recordings were available for perava) in the following round of analysis. Although vocalizations of most subspecies differed diagnostically in multiple characters, pairwise comparisons revealed no diagnostic vocal dierences between notaea and cantator, saturata and peruviana, collinsi and subava, or implicata (excluding implicata2) and anis. It is especially noteworthy that neither loudsongs nor common calls distinguished vocalizations between members of these four pairs of taxa. In the nal stage, subspecies for which there were no diagnosable vocal dierences were consolidated, resulting in nine taxa, which, using the principle of priority, were designated as: cantator (including notaea), avescens (including perava, on the basis of range and plumage characteristics), peruviana (including saturata), subava (including collinsi), ochrogyna, implicata, taxon novum (= implicata2), striata, and anis. Each taxon was distinguished from every other by at least two vocal characters, except for implicata and anis, which were maintained as taxa because their geographic ranges appeared to be disjunct, separated by striata. Geographic ranges are depicted in Figure 2. Range boundaries were based on specimen as well as on vocalization information and reect resolution of some ambiguities or inaccuracies (e.g., contrary to Peters [1951], the Rio Madeira and Rio Tapajs delimited the western and eastern limits, respectively, of implicata). Table 1 provides the aggregated numbers of diagnostic vocal characters distinguishing pairs of taxa. Vocal characters underlying these numbers are provided in Appendix 2 (online; see Acknowledgments), along with supporting data. The following descriptive summary points out salient, primarily visual, dierences in diagnostic characters. It uses groupings of taxa to facilitate comparisons. These groupings do not necessarily reect estimates of phylogeny, which will be developed subsequently. Male loudsongs.All populations delivered a series of clear notes ending with one or more

January 2007]

Warbling Antbird Species Limits

17

F. 2. Geographic distributions of taxa found to differ in at least one vocal character from all other populations; vocalizations of implicata and affinis were not distinguishable. Open circles = specimen locations; open stars = recording locations, most of which are also specimen locations; 1 = cantator, 2 = flavescens, 3 = peruviana, 4 = subflava, 5 = ochrogyna, 6a = taxon novum, 6b = implicata, 6c = striata, and 6d = affinis. Ranges of subflava and peruviana overlap in the hatched area and near the location labeled A where these taxa have been observed in sympatry; double lines to the north and south of A represent apparent altitudinal parapatry. Parapatry without any major physical barrier also appears to occur between flavescens and cantator in the southeastern corner of Venezuela, located by letter B on map. Question marks identify regions lacking data where a form in the complex may occur. raspy notes (which were omied occasionally). Measurements were conned to the series of clear notes because of intra-individual inconsistency in number and structure of raspy notes. Male loudsongs (Fig. 3) fell into three groups according to change in pace. Male loudsongs of ve taxa (cantator, taxon novum, implicata, striata, and anis) accelerated (except for one individual each of cantator and implicata, which had aberrant songs); those of three taxa (avescens, peruviana, and subava) decelerated; and ochrogyna had relatively evenly paced loudsongs. Male loudsongs of taxon novum, implicata, striata, and anis of southeast Amazonia began with a long downslurred note followed by an accelerating series of abrupt notes. These rapidly delivered vocalizations almost always consisted of at least 7 clear notes, and usually 9 or 10. Notes following the initial note were downslurred or boxy. Within the group, only

18

I, I, W

[Auk, Vol. 124

male loudsongs of striata diered diagnostically in having more abrupt notes whose frequency peaks fell toward the end of the series. Dierences between male loudsongs of cantator (the h population whose notes accelerated) and the above southeastern Amazonian taxa were evident visually. Notes in the cantator loudsong could be placed in three groups: a short initial note, two to six notes of similar shape and peak frequency but at a lower peak frequency than the initial note, and an abrupt shi to a third group of one to four similar notes at an even higher peak frequency than the initial note. Male loudsongs of peruviana and subava were not distinguishable from one another. The decelerating loudsongs of these two taxa diered from the accelerating loudsongs of taxa in southeastern Amazonia in multiple characters, but were closest to those of cantator. In contrast to cantator notes, which abruptly changed shape and peak frequency, note changes in peruviana and subava loudsongs were gradual, broadening in approximately even steps. In uncommon examples of peruviana and subava that had sudden shis in note shape, only a single shi occurred, rather than two as in cantator, and subsequent notes widened gradually as in typical songs. Like those of peruviana and subava, male loudsongs of avescens decelerated, but shapes of initial avescens notes were boxy, with a narrow frequency range. Through the middle portion of the vocalization, notes became broader, frequency-modulated, and downslurred, ending with distinctive long, burry, and downslurred notes. Note peaks remained at in frequency or gradually rose slightly while the boom of the notes extended downward in frequency. The nal population, ochrogyna, showed the least change of pace in male loudsong. It was readily distinguished from all other taxa by frequency paern, with the peaks of its initial four to six notes increasing gradually. Except for southeastern Amazonian populations, loudsongs of ochrogyna also diered by having an initial note that was substantially longer than the clear notes that followed. It was distinguished from the southeastern Amazonian populations by its U-shaped notes. Female loudsongs.Female loudsongs (Fig. 4) of all populations were similar, a descending series of 6 to 9 (extremes 412) notes, easily distinguished from male loudsongs. They were oen delivered in response to loudsongs of

T. 1. Number of vocal characters distinguishing populations identied as distinct in Stage 1. Initial number is the total number of independent diagnostic characters and is followed by number of diagnostic characters for each vocalization type. M = male loudsongs; F = female loudsongs; C = calls, including both common calls and chit calls. If a leer is omied, no diagnostic dierences in that vocalization type were found. Characters are identied in Appendix 2 (online; see Acknowledgments).

peruviana avescens

subava

cantator avescens peruviana subava ochrogyna taxon novum implicata striata

5/M3C2

3/M1C2 3/M1C2

4/M1F1C2 5/M2C3 3/C3

6/M2F2C2 8/M2F4C2 4/M3F1 9/M4F2C3

ochropyga

7/M4F1C2 9/M4F3C2 11/M7F2C2 11/M8C3 6/M3F1C2

taxon nov.

8/M4F2C2 6/M1F3C2 7/M3F2C2 6/M3C3 4/M2C2 2/C2

implicata

9/M5F2C2 13M6F5C2 12/M8F2C2 13/M7F3C3 9/M5F2C2 4/M2C2 3/M2C1

striata

6/M4C2 6/M2F2C2 10/M6F2C2 9/M5F1C3 4/M2C2 2/C2 0 2/M1C1

anis

January 2007] Warbling Antbird Species Limits

F. 3. Male loudsongs of the Hypocnemis cantator complex. Arrows identify the first notes that were considered raspy (see text). (A) cantator 90 km north of Manaus, Amazonas, Brazil (Whitney ISL-BMW.189: 24). (B) flavescens Frente de la Isla Cigarrn, Amazonas, Venezuela (Schwartz ML 62045). (C) peruviana Santa Rita, Loreto, Peru (Whitney ISL-BMW.187:02). (D) subflava Sanamarca, Ayacucho, Peru (WiddowsonSterling ISL-BW&JS.1:06). (E) ochrogyna Vila Bela da Santssima Trindade, MT, Brazil (Whitney ISL-BMW.190:10). (F) implicata Borba, Amazonas, Brazil (Whitney ISL-BMW.188:12). (G) taxon novum Vila do Carmo, Amazonas, Brazil (Whitney ISL-BMW.197:17). (H) striata 15 km south of Altamira, Par, Brazil (Whitney ISL-BMW.178:06). (I) affinis Serra dos Carajs, Par, Brazil (Whitney ISL.BMW.6:12).

19

20
I, I, W [Auk, Vol. 124

F. 4. Female loudsongs of the Hypocnemis cantator complex. Arrows identify the first notes that were considered raspy (see text). (A) cantator Brownsberg Nature Reserve, Suriname (Whitney ISL- BMW.119:26). (B) flavescens Frente de la Isla Cigarrn, Amazonas, Venezuela (Schwartz ML 62045). (C) peruviana Yarapa Reserve, Loreto, Peru (Whitney ISL-BMW.163:05). (D) subflava Sanamarca, Ayacucho, Peru (WiddowsonSterling ISL-BW&JS.1:06). (E) ochrogyna Serra dos Pacas Novos, Rondnia, Brazil (Whittaker ISL-AW.7:23). (F) implicata Caima, Par, Brazil (Whitney ISLBMW.158:17). (G) taxon novum Vila do Carmo, Amazonas, Brazil (Whitney ISL-BMW.197:14). (H) striata Rurpolis, Par, Brazil (P. Isler ML 47910). (I) affinis Caxiuan, Par, Brazil (Zimmer ISL-KJZ.85:22).

January 2007]

Warbling Antbird Species Limits

21

their mates and typically started immediately before or as the male loudsong ended. Note shapes were similar in all populations; initial notes were usually downslurred and shaded gradually into terminal notes, which were shaped like an inverted U. Note shapes of avescens were most divergent. These were typically rectangular in shape, as shown in the gure, but a few individuals delivered notes resembling those of other populations, and, consequently, dierences were not considered diagnostic. Diagnostic dierences in female loudsongs were fewer than those found in males: 44 pairwise comparisons as compared to 132. The most prevalent character distinguishing female loudsongs (Appendix 2) was a shi in note duration, expressed by a ratio between durations of initial and terminal clear notes. As is evident in the spectrograms, females of the southeastern Amazonian taxa (taxon novum, implicata, striata, and anis) and, to a lesser extent, neighboring ochrogyna, initiated their loudsongs with a long note and ended with an abrupt note (setting aside terminal raspy notes). Dierences in ratios between rst and last notes in cantator, peruviana, and subava were diagnostically smaller and, at the extreme, notes of avescens loudsongs increased in duration. Comparisons of note duration and overall pace also provided diagnostic dierences, but they were rarer given large within-taxa variances. Note that the intervals between notes diminished at a similar rate in all taxa, and a dierence in interval duration was diagnostic in only one pairwise comparison. With regard to frequency paern, peak frequencies gradually declined in female loudsongs of all taxa, though peaks of the rst two notes were constant or nearly so for most taxa. However, peaks of the rst three notes of ochrogyna ascended and those of cantator descended in frequency. Loudsongs of anis were lowest in frequency, diagnostically lower in initial notes than cantator and subava. We documented no diagnostic dierences in pairwise comparisons of female loudsongs among cantator, avescens, peruviana, and subava. Similarly, no diagnostic dierences were found among female loudsongs of taxon novum, implicata, striata, and anis. Common call.Observing Warbling Antbirds in the wild over many years has revealed that the common call (Fig. 5) is an important and frequently delivered vocalization that may be

employed by both sexes and all age classes in various intraspecic contexts, such as maintenance of auditory contact between members of pairs or families and encounters between neighbors. It was typically a short (usually 0.51.0 s) series of two to ve notes. Although female calls of some taxa tended to contain fewer notes, common calls showed no other sex-linked dierentiation, and samples from both sexes were combined in the analyses. Despite the variety of contexts in which an individual may deliver this call and variation in the number of notes, the great majority of common calls were highly consistent within populations. Vocal characters were dened visually rather than measured because common calls contained few notes. Except in pairwise comparisons of peruviana and ochrogyna and implicata and anis, common calls of all populations diered diagnostically in either note shape or frequency paern, and both were diagnostic in 31 of 36 pairwise comparisons. Thus, of all vocalization types, common calls revealed the most consistent vocal dierences among taxa. Common calls of cantator consisted of a brief repetitive series of two to four (most oen three) similar, clear notes having essentially level peak frequencies (Fig. 5A). Final notes oen became raspy, but maintained the width and spacing of clear notes (Fig. 5B). Notes sounded upslurred because their intensity was greatest at the beginning of the note. Calls of peruviana, like those of cantator, typically consisted of two to three notes (rarely four to six) of approximately the same note-duration and spacing. Unlike cantator, however, the initial two notes diered in shape, the second starting at a higher frequency than the rst and becoming more downslurred (Fig. 5E). When present, the third note was usually identical to the second (Fig. 5F), but the third note sometimes began at an even higher frequency and became more downslurred than the second note (Fig. 5G). Occasionally, second and third notes were rounded like cantator, but the emphasis remained on the downward side of the note (Fig. 5H). Rarely, the initial note was replicated in the second note. In addition, end notes of peruviana calls rarely became raspy (2 of 70 recordings), whereas cantator calls typically ended in raspy notes (22 of 29 recordings). Calls of ochrogyna (Fig. 5KL) t the paern of peruviana in having notes following the initial note becoming more downslurred. However, notes of peruviana calls maintained the same

22

I, I, W

[Auk, Vol. 124

F. 5. Calls of the Hypocnemis cantator complex. A through S are common calls and variants. The latter illustrate the range of variation and are almost always delivered within series of common calls. T and U exemplify chirrs, which were not found to distinguish taxa. V is an example of a chit call, which was not found to differ among taxa except for subflava (W). Caption continued on next page.

January 2007]

Warbling Antbird Species Limits

23

width and were almost never raspy, whereas equivalent notes of ochrogyna shortened and usually became raspy (Fig. 5K; 14 of 20 individuals). We suspect that ochrogyna calls without raspy endings (Fig. 5L), whose notes shortened like the raspy calls, may have been given by females, although notes lengthened in one aberrant call identied as given by a female. Without more knowledge of the sources of possibly overlapping calls of peruviana and ochrogyna, we do not consider that calls of these taxa dier diagnosably at this time. There were also no diagnosable dierences between calls of implicata (Fig. 5MO) and anis (Fig. 5S). They shared a common call consisting of a relatively long whistled note that was at or nearly at in frequency followed by a distinct interval and an clear, abrupt, downslurred note starting at a higher frequency than the rst. The second note was followed instantaneously by a sharply descending, raspy note which typically began at an even higher frequency (Fig. 5M, S). The abrupt, clear note before the terminal raspy note usually had no overtones. Occasionally, the raspy note was omied (Fig. 5N), and sometimes a downslurred note was inserted, producing a more distinctly three-noted call (Fig. 5O). Common calls of striata (Fig. 5QR) were similar

to those of implicata1 and anis, though separable in 100% of blind trials. The rst note was short and usually in the shape of a shallow inverted U (sometimes aened), and the following note was raspy and complex (Fig. 5Q). The second and third notes (if present) were almost always raspy; occasionally, the second note was intermediate between a clear and raspy note, consisting of a short clear segment and overlapping harmonics in a broad frequency band (Fig. 5R). The common call of avescens consisted of a medium-to-long, clear note followed by three to nine abrupt, sharply downslurred notes usually starting at a higher frequency than the rst note (Fig. 5C). The abrupt notes typically maintained the same shape and frequency range, but shortened in length as the series accelerated. In longer calls of four or more notes, the abrupt notes also usually dropped in intensity toward the end. Calls of males appeared to be more rapid than those of females, but sample sizes were too small to conrm this dierence. Occasionally, female calls were shortened to three or even two notes and slowed to the point where they were similar to those of other populations (Fig. 5D), but these aberrant calls were always given in series with more typical calls. No recorded calls of avescens included raspy terminal notes.

F. 5. Continued. Examples: (A) cantator common call without raspy ending; 90 km north of Manaus, Amazonas, Brazil (Bierregaard ML 42840). (B) cantator common call with raspy ending; Iwokrama Forest Reserve, Guyana (Whitney ISL-BMW.145:31). (C) flavescens common call; Cao La Urbana, Bolvar, Venezuela (Schwartz ML 62038). (D) flavescens variant given by female; P. N. N. Chiribiquete, Caquet, Colombia (M. Alvarez ISL-MAR.4:36). (E) peruviana common two-noted call; Humait, Amazonas, Brazil (Whitney ISL-BMW.195:05). (F) peruviana common three-noted call; Limoncocha, Napo, Ecuador (Coffey ISL-BBC.1:24). (G) peruviana variant three-noted call; Catuaba Reserva, Acre, Brazil (Whitney ISL-BMW.141:12). (H) peruviana variant three-noted call; El Dorado, Loreto, Peru (Whitney ISL-BMW.185:10). (I) subflava common call; Amazonia Lodge, Madre de Dios, Peru (Zimmer ISL-KJZ.109:14). (J) subflava common call; Catuaba, Acre, Brazil (Whitney ISL-BMW.141:20) (It is unclear whether this call or I is more typical.) (K) ochrogyna common call; 26 km east of Vila Bela da Santssima Trindade, Mato Grosso, Brazil (Whitney ISL-BMW.190:14). (L) ochrogyna variant without raspy ending; 26 km east of Vila Bela da Santssima Trindade, Mato Grosso, Brazil (Whitney ISL-BMW.190:14). (M) implicata common call; 25 km west, by road, of Boim, Par, Brazil (Whitney ISL-BMW.107:30). (N) implicata variant without raspy ending; Prainha Nova, Amazonas, Brazil (Whitney ISL-BMW.195:30). (O) implicata three-note variant without raspy ending; Prainha Nova, Amazonas, Brazil (Whitney ISL-BMW.196:08). (P) taxon novum common call; Manicor, AM, Brazil (Whitney ISL-BMW.187:21). (Q) striata common call with raspy chirr; Porto do Meio, Par, Brazil (Whitney ISL-BMW.175:11). (R) striata common call with raspy overtones; So Benedito, Par, Brazil (Whitney ISL-BMW.173:24). (S) affinis common call; Caxiuan, Par, Brazil (Zimmer ISL-KJZ.83:04). (T) ochrogyna example of a chirr; Florida, Santa Cruz, Bolivia (Whitney ISL-BMW.199:16). (U) cantator example of a chirr; Foengoe Eiland, Suriname (Whitney ISL-BMW.120: 06). (V) peruviana example of chit call; Quebrada Sucusari, Loreto, Peru (Parker ML 29101). (W) subflava example of chit call; Humait Reserva, Acre, Brazil (Whitney ISL-BMW.142:25).

24

I, I, W

[Auk, Vol. 124

Common calls of the remaining two populations were most unique. That of taxon novum typically consisted of four, less oen three or ve, notes (Fig. 5P). The rst note was short but complex (embellished with overtones), rose in frequency, and had a screechy quality. The following two, three, or four short clear notes were delivered at successively higher frequencies, though the nal two notes were sometimes at the same frequency. Additionally, pace was the most rapid of any population, and taxon novum apparently never ends its call with raspy notes. The common call of subava (Fig. 5IJ) also differed dramatically from those of other populations, comprising two (rarely three) broad-band raspy notes, consisting of noise from which a line tracing could not be made. The notes were closest to chirrs of other subspecies (Fig. 5TU) but were always given in pairs or triplets (chirrs are delivered more erratically) and were longer in duration, and the center of intensity maintained a level frequency or dried upward (that of chirrs usually dried downward). Chit calls.Abrupt chit calls (Fig. 5) were recorded for all taxa. Sexual identication was available on too few recordings of chit calls to permit quantitative analysis, but we saw no obvious dierences in calls between males and females. Most oen repeated in pairs, but also in longer groups of up to ve notes and occasionally given singly, chit notes were vibrant and unclear (e.g., Fig. 5V), except for one taxon. The chit calls of subava (n = 6, all from range of collinsi) were clear notes composed of a base and an apparent harmonic (Fig. 5W), a diagnostic dierence. Sympatry of peruviana and subava.The throat and breast of most subspecies in the H. cantator complex are white, but some are yellowish or yellow. Yellow-breasted forms have been regarded at times (e.g., Cory and Hellmayr 1924) as a distinct species, H. avescens, with two disjunct populations, one in northwestern Brazil and southern Venezuela, the other ranging from central Peru to central Bolivia. In his critical morphological examination of relationships and distribution of the H. cantator complex, the only such study to date, Zimmer (1932) considered the two disjunct, yellow-breasted forms (northern avescens and southern subavacollinsi) to have an obvious close association linked by the intervening, white-breasted (but weakly yellow-bellied)

peruviana, and merged all forms as subspecies of cantator. There has been continuing concern (Sibley and Monroe 1990, Ridgely and Tudor 1994), however, that southern yellow-breasted populations could be specically distinct, the laer authors citing an abrupt and striking shi in southern Peru from the nominate type (ssp. peruviana) to subava and collinsi and a report by T. Schulenberg of a sighting of an apparent mixed pair at the Tambopata Reserve in Madre de Dios, Peru (Ridgely and Tudor 1994). In April 1997, B.M.W. made several tape recordings of peruviana and subava (as collinsi) at Reserva Catuaba (1004S, 6737W), Acre, Brazil, providing the rst documented evidence of the sympatry of these taxa. Over an area of ~400 ha of terra rme forest surveyed for the presence of both taxa, it appeared that subava outnumbered peruviana 3:1 or 4:1. Most encounters involved mated pairs, and there was no indication of mixed pairing. Soil type and vegetation structure in the forest understory across the survey area at Reserva Catuaba was homogeneous, with numerous and scaered light gaps contributing the major element of variation on a local scale. Including the diversity aributable to light gaps, plant species composition seemed essentially uniform. A large, Guadua sp. bamboo was abundant and locally quite dense. Despite careful aention to the microhabitats in which pairs of Hypocnemis were located, and particular scrutiny of the relatively few pairs of peruviana encountered, B.M.W. could discern no clear distinctions in habitat or foraging behavior for either peruviana or subava; the two taxa occupied adjacent territories. Sympatry of these taxa has also been documented with tape recordings at the Manu Wildlife Center (1222S, 7043W) in Madre de Dios, Peru, in August 2000 (K. Zimmer pers. comm.; also recorded at this site by A. Whiaker [pers. comm.]). There, however, peruviana and subava were found in similar numbers but in dierent microhabitats, and no pairs were found side-by-side in the forest. Zimmer found that subava was restricted to extensive stands of Guadua sp. bamboo in oodplain forest (though not all such stands appeared to be occupied) and peruviana was present only in terra rme without bamboo. A similar habitat division was found in the same Peruvian department at Los Amigos Research Center (Peruvian acronym CICRA; 1234S, 7006W)

January 2007]

Warbling Antbird Species Limits

25

by D. Lebbin and W. Tori (pers. comm.), who, in studies of the avifauna of Guadua bamboo, need 27 individuals of subava inside patches of bamboo, whereas net lines (maintained for fewer hours) in nearby forest outside the bamboo need ve peruviana and no subava. However, further study by N. Seddon and J. Tobias (pers. comm.) at this site found the habitat distinction to be less strict. For example, they found peruviana territories in the understory of palm swamps and in areas containing bamboo, and subava territories in tangled riverine vegetation and in areas lacking bamboo; and they found that the two forms occasionally inhabited adjacent territories. Thus, distinctions in habitat preferences between the two forms remain unclearan important subject for future eld work. Syntopy of peruviana and subava has also been reported in Peru from near Pozuzo (1004S, 7532W), Pasco (T. Schulenberg pers. comm.), and at Cuzco Amazonica Lodge (1233S, 6903W), Madre de Dios (T. J. Davis pers. comm.), but habitat distinctions were not studied at these sites. Whether they inhabit adjacent territories or not, it appears that reproductive-isolation mechanisms between peruviana and subava have remained operative following secondary contact over a fairly extensive area in the southwestern corner of Amazonia. Given this range overlap, and the close similarities in loudsongs and behavior of these taxa, we expect that some very limited hybridization between peruviana and subava eventually may be documented, as suspected in the mixed pair observed by Schulenberg, cited above. Phenotypic distinctions we have been able to identify between them in the eld are white versus yellow throat and breast, leg color (usually bluish-gray in peruviana, greenish-yellow in subava), the highly dierentiated common call (Fig. 5, EH vs. IJ), the dierence in chit call, and the apparent absence of a chirr call in the vocal repertoire of subava. Parapatry of cantator and avescens. Evidence of parapatry between these taxa in southeastern Venezuela was provided by specimens deposited in the Coleccin Ornitologia Phelps (COP). Taxon identication was substantiated by the authors using specimen photographs. Both taxa have been collected on or adjacent to a plateau (known as the Macizo de Chimant) situated between

the southwestern corner of La Gran Sabana and the Sierra Pacaraima in the headwaters of the Ro Caron (region identied by B in Fig. 2). Specimens of cantator were collected on the western and northern edge of this plateau at Ro Caruai (= Kariay), Ro Apcara, and Chimanttepu. Specimens of avescens were obtained at Salto Arebuchi (= Aripichi?), Acopn-tepu, Ro Icabar, Santa Elena de Uairn, and Paraitepu. In particular, the Ro Caruai (location for cantator) adjoins Acopn-tepu (location for avescens) at its base. Although precise geographic coordinates are unavailable for most of these localities, they do not appear to be isolated by major topographic features, though they may represent dierent ecosystems. Additional eld work is needed to understand the biogeography of cantator and avescens in this region. D C The number of vocal characters distinguishing most pairs of H. cantator subspecies was substantial, in most cases greater than those documented for syntopic pairs of congeneric thamnophilid species in an earlier study (Isler et al. 1998). All three principal types of vocalizations (male loudsongs, female loudsongs, and common calls) diered diagnosably in 19 of 36 pairwise comparisons (Table 1). Most surprising was the nding that calls diered diagnosably in all but two pairwise comparisons, and a specic result of particular interest was that the vocalizations of the only pair of subspecies now known to exist in syntopy (peruviana and subava) diered diagnostically only in the structure of their calls. Whether these ndings reect that calls rather than songs function in mate choice or species recognition is a question for future eld studies using experimental designs, such as mate-removal experiments, necessary for understanding vocal behavior and evolution in the Thamnophilidae. On the basis of vocal dierences described herein and morphological dierences described by earlier authors, we recommend that the H. cantator complex be considered at this time to consist of six species: Hypocnemis cantator (Boddaert) Guianan Warbling-Antbird H. c. notaea Hellmayr H. c. cantator (Boddaert)

26

I, I, W

[Auk, Vol. 124

Hypocnemis avescens Sclater Imeri Warbling-Antbird H. f. avescens Sclater H. f. perava Pinto Hypocnemis peruviana Taczanowski Peruvian Warbling-Antbird H. p. saturata Carriker H. p. peruviana Taczanowski Hypocnemis subava Cabanis Yellow-breasted Warbling-Antbird H. s. subava Cabanis H. s. collinsi Cherrie Hypocnemis ochrogyna Zimmer Rondonia Warbling-Antbird Hypocnemis striata (Spix) Spixs Warbling-Antbird H. s. taxon novum H. s. implicata Zimmer H. s. striata (Spix) H. s. anis Zimmer From the perspective of their vocalizations, we nd no reason to maintain notaea, saturata, and collinsi as taxa. However, given the possibility that diagnosable dierences may be found between them and cantator, peruviana, and subava, respectively, in other suites of characters, we recommend maintaining these taxa as subspecies, awaiting the completion of further molecular, morphological, and behavioral studies. We are unable to shed light on the validity of perava, given the absence of vocal recordings. We also recognize the possibility that further vocal distinctions will be found within the ranges of proposed species when additional recordings provide improved geographic coverage. Although recommendations regarding cantator, avescens, peruviana, subava, and ochrogyna are straightforward, the treatment of striata and its four constituent populations is more problematic. A paradox is posed in the relationship between implicata and anis, whose vocalizations could not be dierentiated but whose ranges appear to be separated by that of striata, whose vocalizations diered from both, though striata was found to dier vocally from implicata by three characters and from anis by two, fewer than found in most pairwise comparisons of taxa recommended for species

status. The importance of this vocal result is buressed by parallel morphological ndings of Zimmer (1932), who discovered minimal plumage dierences between implicata and anis, even though both diered in plumage from the geographically intervening striata. The most likely explanation for this paern is that striata has become genetically isolated relatively recently, and that there remains or recently has been gene ow between implicata and anis in the essentially unstudied region south of the range of striata (identied by a question mark in Fig. 2). In the interest of maintaining a conservative stance, we recommend that implicata and anis be maintained as subspecies of striata until further information is available from the region and the results of molecular analysis can be added to the data set. A second diculty concerns assessing the status of taxon novum. Although dierences were not identied in other vocalizations, differences in the calls of taxon novum and other populations within striata (including anis and the remainder of implicata) were substantial and comparable to the only known vocal dierences between the syntopic peruviana and subava. We maintain taxon novum as a subspecies until it is formally described. It is noteworthy that taxon novum and striata (sensu stricto) diered in four vocal characters found in male loudsongs as well as calls. Further recommendations regarding status of all taxa within the Warbling Antbird complex and estimates of its phylogeny within the Thamnophilidae await molecular analyses now underway. It is now apparent, however, through conservative analysis of vocalizations, that the H. cantator complex has had a long and dynamic evolutionary history, and we are condent that this initial revision beer reects the taxonomic status of populations in the complex than current classications. A Supplementary data for this study are available, as Appendices 1 and 2 and Tables 25, at hp://www.aou.org/suppl/auk_124_1_isler.pdf. We deeply appreciate the continuing support of the Macaulay Library, Cornell Laboratory of Ornithology (G. Budney); National Sound Archive, The British Library (R. Rant); and Florida State Museum. The many recordists

January 2007]

Warbling Antbird Species Limits

27

who contributed to the eort are identied in Appendix 1, and we are indebted to them all. T. J. Davis, D. Lebbin, N. Seddon, T. Schulenberg, J. Tobias, A. Whiaker, and K. Zimmer contributed unpublished information regarding distribution and areas of sympatry of peruviana and subava in Peru, and Whiaker also provided information on the distribution of taxon novum in Brazil. We appreciate the eorts of C. Rodner and C. Milensky to provide photographs of specimens and related data at Coleccin Ornitologa Phelps. J. Bates and K. Zimmer kindly reviewed earlier dras of the manuscript, and the nal manuscript was thoughtfully reviewed by N. Seddon and an anonymous reviewer. National Science Foundation grant no. DEB 9974104 to J. M. Bates and S. J. Hacke at the Field Museum of Natural History helped provide computer equipment to perform the vocal analyses. We thank Field Guides, Inc., for support of B.M.W.s eld work. L C B, L. F., D. E. K. 2001. Avian bioacoustics. Pages 1152 in Handbook of the Birds of the World, vol. 6: Mousebirds to Hornbills (J. del Hoyo, A. Ellio, and J. Sargatal, Eds.). Lynx Edicions, Barcelona, Spain. B, J. M. 2000. Allozymic genetic structure and natural habitat fragmentation: Data for ve species of Amazonian forest birds. Condor 102:770783. B, J. M. 2002. The genetic eects of forest fragmentation on ve species of Amazonian birds. Journal of Avian Biology 33:276294. B, J. M., S. J. H, J. M. G. 1999. High levels of mitochondrial DNA dierentiation in two lineages of antbirds (Drymophila and Hypocnemis). Auk 116: 10931106. B, R. T., A. P. C. 1996. Historical diversication of birds in northwestern South America: A molecular perspective on the role of vicariant events. Evolution 50:16071624. C, A. P. 1988. Genetic variation in Neotropical birds: Implications for the speciation process. Pages 16581664 in Acta XIX Congressus Internationalis Ornithologici (H. Ouellet, Ed.). National Museum of Natural Sciences, University of Oawa Press, Oawa.

C, R. A., S. M, C. W. C. 1995. CANARY 1.2 Users Manual. Cornell Laboratory of Ornithology, Ithaca, New York. C, C. B., C. E. H. 1924. Catalogue of birds of the Americas and the adjacent islands. PteroptochidaeConopophagidae Formicariidae. Field Museum of Natural History Zoological Series, no. 13, part III. H, S. J. 1993. Phylogenetic and biogeographic relationships in the Neotropical genus Gymnopithys (Formicariidae). Wilson Bulletin 105:301315. H, S. J., K. V. R. 1990. Comparison of phenotypic and genetic differentiation in South American antwrens (Formicariidae). Auk 107:473489. H, A. J., A. G. K, D. T. P, G. S, M. C. 2002. Guidelines for assigning species rank. Ibis 144:518525. I, M. L. 1997. A sector-based ornithological geographic information system for the Neotropics. Pages 345354 in Studies in Neotropical Ornithology Honoring Ted Parker (J. V. Remsen, Jr., Ed.). Ornithological Monographs, no. 48. I, M. L., J. A. A, P. R. I, B. M. W. 2001. A new species of Percnostola antbird (Passeriformes: Thamnophilidae) from Amazonian Peru, and an analysis of species limits within Percnostola rufrons. Wilson Bulletin 113:164176. I, M. L., P. R. I. 2003. Species limits in the Pygmy Antwren (Myrmotherula brachyura) complex (Aves: Passeriformes: Thamnophilidae): 1. The taxonomic status of Myrmotherula brachyura ignota. Proceedings of the Biological Society of Washington 116: 2328. I, M. L., P. R. I, R. T. B. 2005. Clinal variation in vocalizations of an antbird (Thamnophilidae) and implications for dening species limits. Auk 122:433444. I, M. L., P. R. I, B. M. W. 1997. Biogeography and systematics of the Thamnophilus punctatus (Thamnophilidae) complex. Pages 355381 in Studies in Neotropical Ornithology Honoring Ted Parker (J. V. Remsen, Jr., Ed.). Ornithological Monographs, no. 48. I, M. L., P. R. I, B. M. W. 1998. Use of vocalizations to establish

28

I, I, W

[Auk, Vol. 124

species limits in antbirds (Passeriformes; Thamnophilidae). Auk 115:577590. I, M. L., P. R. I, B. M. W. 1999. Species limits in antbirds (Passeriformes; Thamnophilidae): The Myrmotherula surinamensis complex. Auk 116:8396. J, N. K., J. V. R, J., C. C. 1999. Resolution of the debate over species concepts in ornithology: A new comprehensive biological species concept. Pages 1470 1482 in Acta XXII Congressus Internationalis Ornithologici (N. J. Adams and R. H. Slotow, Eds.). BirdLife South Africa, Johannesburg. M, E., P. D. A. 1991. Principles of Systematic Zoology. McGraw-Hill, New York. P, J. L. 1951. Check-list of Birds of the World, vol. 7. Museum of Comparative Zoology, Harvard University, Cambridge, Massachuses. P, O. M. O. 1966. Estudo crtico e catlogo remissivo das aves de Territrio Federal de Roraima. Cadernos da Amaznia, Instituto Nacional de Pesquisas da Amaznia 8: 1176. R, A. 1952. Die Grundlagen des natrlichen System der vergleichenden Anatomie und der Phylogenetik. Geest und Portig K.-G., Leipzig, Germany. R, J. V., Jr. 2005. Paern, process, and rigor meet classication. Auk 122:403413.

R, R. S., G. T. 1994. The Birds of South America, vol. II: The Suboscine Passerines. University of Texas Press, Austin. S, C. G., B. L. M, J. 1990. Distribution and Taxonomy of the Birds of the World. Yale University Press, New Haven, Connecticut. W, B. M., J. F. P, D. R. C. B, R. P. 2000. Systematic revision and biogeography of the Herpsilochmus pileatus complex, with description of a new species from northeastern Brazil. Auk 117: 869891. W, E. O. 1967. The behavior of bicolored antbirds. University of California Publications in Zoology, no. 79. Z, J. T. 1932. Studies of Peruvian birds. V. The genera Herpsilochmus, Microrhopias, Formicivora, Hypocnemis, Hypocnemoides, and Myrmochanes. American Museum Novitates 538:127. Z, K. J., M. L. I. 2003. Family Thamnophilidae (Typical Antbirds). Pages 448681 in Handbook of the Birds of the World, vol. 8: Broadbills to Tapaculos (J. del Hoyo, A. Ellio, and D. Christie, Eds.). Lynx Edicions, Barcelona, Spain. Associate Editor: K. Yasukawa