Revealing cryptic diversity: vocal and morphological evidence for a new species of Chamaeza (Aves: Formicariidae)

ABSTRACT

Chamaeza campanisona (Lichtenstein, 1823) is a medium-sized, ground-dwelling passerine bird widely distributed in South America, inhabiting Neotropical forests across the Amazon Basin, Atlantic Forest, and Andean regions. Across its range, 12 subspecies are recognized, predominantly distributed along altitudinal and longitudinal gradients in the Andes. Within the Atlantic Forest, two subspecies are currently recognized, although C. c. tshororo Bertoni, 1901 (described from southeastern Paraguay) has been proposed as inseparable from the nominotypic form. Given the limited taxonomic studies on this genus and the demonstrated utility of vocal analysis in suboscine taxonomy, we investigated bioacoustic variation among eastern South American taxa of this species complex. We analyzed 183 recordings representing the full geographic range of these populations, combining traditional bioacoustic methods with machine learning algorithms. Our results revealed three distinct loudsong patterns: a southern variant occurring from southeastern Paraguay and northeastern Argentina north to the state of Bahia, Brazil; an isolated population in the state of Alagoas, northeastern Brazil, represented by a single recording and lacking voucher specimens; and a northern pattern restricted to the state of Ceará, northeastern Brazil. The loudsongs typically consist of a rapid series of closely spaced notes exhibiting population-specific variation, particularly in pace and in the presence of a terminal series of slowed, descending grunting notes. The southern vocal type corresponds to C. campanisona, under which C. c. tshororo should be treated as a junior synonym. The Alagoas population requires additional study given the limited available material. The Ceará population represents a previously undescribed taxon, which we name and describe herein. This new species is morphologically diagnosable, as it lacks the black frontal spot seen in all other populations, as well as vocally distinct. It is restricted to the Baturité Mountains of Ceará, where it persists in relictual montane rainforests surrounded by the seasonally drier Caatinga xerophytic forest in the municipalities of Guaramiranga and Pacoti.

KEYWORDS:
Bioacoustics; Serra de Baturité; Short-tailed Antthrush; species limits; taxonomic review

INTRODUCTION

Chamaeza Vigors, 1825 comprises ground-dwelling suboscine birds specialized for life in the understory of Neotropical forests. These secretive birds occupy forest floors from lowland to mid-elevation montane habitats (Winkler et al. 2020), where they forage primarily in leaf litter and among fallen logs for arthropods. Although they occasionally follow army ant swarms, this behavior occurs less frequently than in other ant-following birds (Willis 1984). Chamaeza species exhibit cryptic plumage adapted for leaf-litter camouflage, with no apparent sexual dimorphism. Their reproductive strategy involves nesting in natural cavities of decaying logs, with biparental care extending through the nestling and fledgling stages. Ecologically, they are solitary foragers with limited dispersal capacity, reflecting their specialization for stable forest interiors (Sick 1997, Studer et al. 2018, Winkler et al. 2020). The genus consists of six species with distinct biogeographic distributions across South America’s forested regions. Chamaeza ruficauda Cabanis & Heine, 1860 and C. meruloides Vigors, 1825 are endemic to the Atlantic Forest; C. nobilis Gould, 1855 is restricted to the Amazon Basin; C. turdina Cabanis & Heine, 1860 and C. mollissima Sclater, 1855 are distributed along the eastern Andean foothills, whereas C. campanisona (Lichtenstein, 1823) is the most widely distributed species, with a circum-Amazonian range (Winkler et al. 2020).

Chamaeza campanisona is a medium-sized antthrush, measuring approximately 20 cm in length and weighing 70 g, on average. The species exhibits characteristic terrestrial adaptations, including long legs and relatively short toes, complemented by a medium-length tail. Its plumage displays olive-brown upperparts with distinctive markings, such as a rufous crown, white supercilium, and small black spots on the white throat. The tail features a diagnostic black terminal band with white to yellowish tips, while the underparts are yellowish white with black streaking (Greeney 2020). Notable geographic variation occurs in Atlantic populations, which possess a distinctive black forehead patch and pale bill; this frontal marking appears absent from the population of the state of Ceará, northeastern Brazil (Willis 1992). The species’ vocal repertoire consists of a prolonged series of monotonous whistles that end in distinctive croaking syllables, a well-documented pattern across its range (Ridgely and Tudor 1994, Sick 1997, Greeney 2020).

Currently, 12 subspecies of C. campanisona are recognized, exhibiting a broad yet fragmented distribution across northern and central South America. These include C. c. fulvescens (Sclater & Godman, 1882) in eastern Venezuela and western Guyana; C. c. obscura (Zimmer & Phelps, 1944) restricted to the tepui region of southeastern Venezuela in the states of Bolívar and southern Amazonas; C. c. huachamacarii (Phelps & Phelps Jr, 1951) endemic to Cerro Huachamacari tepui in southern Venezuela; C. c. yavii (Phelps & Phelps Jr, 1947) endemic to Cerro Yaví tepui in south-central Venezuela; and C. c. venezuelana (Ménégaux & Hellmayr, 1906) distributed across the coast ranges of northern Venezuela from eastern Miranda to western Táchira. Andean subspecies comprise C. c. columbiana (Berlepsch & Stolzmann, 1896) from the eastern slopes of the eastern Andes in Colombia; C. c. punctigula (Chapman, 1924) ranging from eastern Ecuador south to northern Peru; C. c. olivacea (Tschudi, 1844) from central Peru in the department of Junín; C. c. berlepschi (Stolzmann, 1926) from southeastern Peru and likely extreme western Bolivia; and C. c. boliviana (Hellmayr & Seilern, 1912) from the Bolivian Andean foothills in the departments of Santa Cruz to La Paz. The remaining taxa are Atlantic Forest endemics. The nominotypic C. c. campanisona - initially described as Myiothera campanisona Lichtenstein, 1823, with the state of São Paulo as the type locality - ranges from the Serra do Baturité (state of Ceará), Quebrangulo (state of Alagoas), and southern Bahia south to the state of Santa Catarina in eastern Brazil. The last subspecies is C. c. tshororo (Bertoni, 1901), described from Alto Paraná in eastern Paraguay, which extends into southernmost Brazil and northwestern Argentina (Greeney 2020).

Chamaeza remains understudied from a taxonomic perspective, with no comprehensive phylogenetic studies (e.g., Harvey et al. 2020). The most comprehensive phylogenetic hypothesis to date exists only in the gray literature (Oliveira 2016). The taxonomic arrangement of the genus currently relies primarily on morphological and vocal studies, which clarified species boundaries among eastern Brazilian populations and resolved nomenclatural issues concerning Andean taxa (Raposo and Teixeira 1992, Willis 1992).

In suboscine passerines, vocalizations play a central role in species delimitation because they are largely genetically inherited rather than learned, making them reliable indicators of lineage divergence and potential premating reproductive isolation (Tobias et al. 2012, Touchton et al. 2014). Divergence in vocalizations is particularly relevant because it can directly mediate mate recognition and assortative mating, thereby contributing to reproductive isolation (Seddon 2005). Numerous taxonomic revisions in Neotropical suboscines have demonstrated that discrete differences in vocal structure often correspond to strong behavioral barriers to interbreeding, even in the absence of marked plumage differentiation. Consequently, bioacoustic traits are widely regarded as among the most informative phenotypic characters for diagnosing species limits in this group (e.g., Lima et al. 2025, Bocalini et al. 2026). Given the challenges of morphological differentiation in these cryptically plumaged birds and considering the diagnostic value of vocalizations in suboscines (Touchton et al. 2014), we employed bioacoustic analyses to clarify taxonomic relationships among Atlantic populations of C. campanisona.

We combined traditional spectrographic analysis (Isler et al. 1998) with a novel machine learning approach using a deep neural network (DNN) developed for bird sound classification (Kahl et al. 2021, McGinn et al. 2023). In addition, morphological traits were analyzed to complement and validate the vocal groups delineated by bioacoustic analysis. Our results revealed three distinct loudsong types exhibiting discrete geographic distributions, supporting the recognition of two diagnosable species under the Phylogenetic Species Concept (Cracraft 1983).

MATERIAL AND METHODS

Acquisition and selection of vocalizations

We conducted a comprehensive analysis of 183 audio recordings (C. c. campanisona = 75, C. c. tshororo = 108) spanning the complete geographic distribution of the Atlantic populations (Supplementary Table S1). Recordings were sourced from four major repositories: the ornithological platforms WikiAves (https://www.wikiaves.com.br) and Xeno-canto (https://www.xeno-canto.org), along with the institutional collections of the Fonoteca Neotropical Jacques Vielliard (Universidade Estadual de Campinas) and the Macaulay Library (Cornell University). We used recordings with sampling rates of 44.1 kHz and 48 kHz, which were subsequently standardized using Audacity 3.1 (Audacity Team 2021) to 44.1 kHz and 16-bit depth prior to analyses. All recordings were formatted as “.wav” files.

Following established bioacoustic protocols (Isler et al. 1998), we defined notes as continuous sound elements separated by silent intervals in spectrograms (time-frequency representations), with phrases consisting of grouped notes. Quantitative analysis was performed using Raven Pro 1.6.5 (K. Lisa Yang Center for Conservation Bioacoustics 2024) with standardized parameters: a 1000-point window size and 50% overlap hop size. We selected one vocalization per recording, except for recordings where vocalizations were poorly represented, yielding a total of 340 vocalizations analyzed (Table 1). We measured eight key acoustic parameters: (i) lower, (ii) higher, and (iii) peak frequencies, (iv) 90% bandwidth-the difference between the 5% and 95% frequencies-and (v) maximum entropy. In addition, we determined (vi) song duration (s), (vii) number of notes, and (viii) pace (i.e., notes/s), and incorporated a qualitative assessment of note shape. All spectrographic visualizations were generated using the warbleR package (Araya-Salas and Smith-Vidaurre 2017) within the R environment (R Core Team 2023).

Table 1
The mean loudsong characteristics of the Chamaeza campanisona species complex are provided. The letter “N” repre sents the number of loudsongs analyzed, with the values in parentheses indicating the number of vocalizations analyzed.

Vocalization	Population	N	Duration	Low Frequency	High Frequency	Bandwidth	Peak Frequency	Maximum entropy	Number of notes	Pace
Alarm call	southern	1 (1)	2.7	980.9	2529.8	775.2	2325.6	3.8	24	8.9
Call	Ceará	3 (9)	0.1±0	1079.7±174.4	2242.6±149.6	648.6±132.1	1945.7±91.5	2.6±0.5	-	-
	Bahia	2 (6)	0.1±0	978.8±80.3	2162±86.2	689.1±94.4	1866.2±70.3	3.3±0.2	-	-
	southern	31 (31)	0.2±0.1	976.4±169.7	2236.3±149.3	665.7±165.3	1982±141.8	3.2±0.3	-	-
Song	Ceará	33 (33)	9.8±1.3	880.3±77.6	1274.7±66.7	144.8±42.7	1093.8±56.6	1.9±0.2	99.1±14.6	10.1±0.5
	Alagoas	1 (16)	4.4±0.9	956.6±35.4	1236.4±32.9	149.4±46	1130.9±33.7	2.3±0.3	33.8±6.1	7.7±0.6
	Bahia	19 (19)	9.3±1.3	886±72.4	1288.7±83.2	171.2±28.3	1105.1±54.8	2±0.1	54.9±7.9	5.9±0.5
	southern	89 (90)	9.4±1.4	880.2±62.8	1255.8±59.6	161.2±46.5	1080.2±51.1	1.9±0.4	65.4±11.3	6.9±0.6
Croak	Ceará	33 (31)	3.4±1.1	573.8±107.4	1344.8±144	425.1±103.3	1065.6±81.2	3±0.4	33.6±15.2	9.9±2.6
	Alagoas	-	-	-	-	-	-	-	-	-
	Bahia	19 (19)	2.4±0.6	537.6±73	1335.1±89.6	416.5±87.1	1053.6±107.7	3.1±0.3	18.4±4.5	7.7±0.7
	southern	89 (85)	0.8±0.5	599.4±105.2	1312.4±172.3	371.8±125.6	1051.4±83.7	2.9±0.3	5.1±3.4	6.5±1.9

Supervised classification

We employed BirdNET, a pre-trained DNN based on a 157-layer residual network (ResNet) architecture containing 27 million trainable parameters (Kahl et al. 2021, McGinn et al. 2023). The DNN was trained on >500,000 recordings representing >10,000 bird species worldwide, with demonstrated efficacy in identifying 984 Nearctic and Palearctic species (Kahl et al. 2021). This system processes audio inputs and generates 1024-dimensional feature embeddings that capture essential acoustic characteristics for species classification. For feature extraction, we isolated the central three-second segment of each loudsong recording. These segments were processed through BirdNET’s DNN. To visualize population-level vocal variation, we applied Uniform Manifold Approximation and Projection (UMAP; McInnes et al. 2018) to reduce the dimensionality of the embeddings to two dimensions.

Vocalizations were categorized into four geographic populations based on previous inspections: southern (south of the Doce River in the state of Espírito Santo), Bahia (north of the Doce River north to the state of Bahia), Alagoas, and Ceará, both states located north of the São Francisco River in northeastern Brazil (Table 1). To evaluate the statistical distinguishability of populations, we employed a supervised classification using a Support Vector Machine (SVM; Cortes and Vapnik 1995). Its performance was benchmarked against a random classifier implementing the “most frequent” strate gy (Sanchís-Pedregosa et al. 2011), which assigns all vocalizations to the same class. Model performance was assessed via repeated random subsampling validation (100 iterations, 60/40 train-test split; Lakdari et al. 2024), using overall accuracy and class-specific precision, recall, and F1-scores. Precision indicates how reliable a positive identification is. A high precision (e.g., > 0.8) means that when the model predicts a song belongs to a specific population, it is almost always correct. Recall indicates how completely the model identifies songs belonging to a given population. A high recall means that most or all songs that truly belong to that population are correctly identified. The F1-score combines precision and recall into a single metric, providing an overall measure of classification performance for each population. All computational analyses were performed in Python.

Specimen examination

We examined 134 specimens (C. c. campanisona = 48, C. c. tshororo = 86) of both sexes, 54 of which were C. campanisona housed at the Museu de Zoologia da Universidade de São Paulo (MZUSP). In addition, another 80 specimens were examined through photographs. These included the unsexed nominotypic holotype deposited at the Museum für Naturkunde (ZMB) and specimens from Argentina, Paraguay, and the southern Brazilian state of Paraná (Museu de História Natural Capão da Imbuia), as well as the northeastern Brazilian states of Bahia and Ceará, deposited in four other international ornithological collections (Appendix 1).

The coloration of the (i) crown, (ii) forehead, (iii) lores, (iv) eyebrows, (v) auricular region, (vi) upperparts, (vii) throat, (viii) belly, (ix) flanks, (x) primaries, and (xi) tail were determined and compared between specimens. Plumage coloration was analyzed through direct comparison with standardized color charts (Munsell 1994) under controlled lighting conditions, except for the holotypes of C. campanisona and Myioturdus marginatus Wied, 1831, and one Ceará specimen deposited at the Field Museum of Natural History (FMNH), Chicago, USA, which were examined through photographs provided by colleagues.

Vocal statistical analyses

We employed multivariate statistical approaches to quantify differences among population groups. Vocal characteristics were compared using Multivariate Analysis of Variance (MANOVA), with subsequent univariate Analysis of Variance (ANOVA) and Tukey’s HSD post hoc tests to examine specific between-group and within-group differences. We assessed the relationship between pace and latitude using Spearman’s rank correlation coefficient (ρ). All hypothesis tests employed a conservative significance threshold of α = 0.01 to account for multiple comparisons. To evaluate the diagnostic potential of measured traits, we performed Linear Discriminant Analysis (LDA) using the MASS package in R (Venables and Ripley 2002). This analysis incorporated vocal variables to assess population discriminability.

Species delimitation

Our results are discussed in light of the Biological Species Concept (Mayr 2000) and the Phylogenetic Species Concept (Cracraft 1983). Under the first, a species comprises populations whose members actually or potentially interbreed in nature, placing less emphasis on shared similarities. Under the second, a species is defined as the smallest set of organisms sharing a common ancestor and distinguishable from other such clusters. Our taxonomic diagnosis employed a multi-criteria framework that treated qualitative characters (plumage and loudsong structure) and quantitative measurements with equal weight. We applied the standard McKitrick and Zink (1988) threshold, whereby 95-99% of individuals must be assignable by at least one diagnostic trait. For loudsongs, continuous characters required non-overlapping ranges, while categorical characters had to be both unambiguously identifiable aurally and/or visually in spectrograms and free of clinal variation (Isler et al. 1998). Thus, each proposed taxon is diagnosed by consistent, qualitative and quantitative character states that satisfy these criteria.

TAXONOMY

Vocal variations

Our previous analyses of loudsongs revealed no vocal differences attributable to sex. Therefore, given the absence of sexual dimorphism in plumage and because the sex of the vocalizing individuals was not specified in the recordings, all vocalizations were analyzed together.

We identified three vocalization categories: (1) an alarm call (Fig. 1A), recorded only once in Paraguay (XC 13489); (2) isolated notes emitted without a fixed numerical pattern, classified as calls; and (3) longer sequences of notes, classified as loudsongs. The calls showed greater variation in frequency than in temporal parameters, producing a characteristic vertical, slash-like pattern in spectrograms (Fig. 1B). A multivariate analysis of variance revealed significant differences among populations (F1,2 = 4.7; p < 0.001). Call duration differed between the southern and Ceará populations, whereas maximum entropy differed among all populations except the southern and Bahia and Alagoas groups (p < 0.001; Supplementary Tables S2, S3). The median number of calls per recording was 4.0 (range: 4-20) in Ceará, 5.5 (4-7) in Bahia, and 8.0 (3-25) in the southern populations.

Figure 1
Vocalizations of Chamaeza campanisona: (A) Alarm call from San Rafael National Park, Paraguay, XC 13489; (B) calls from Santa Catarina, southern Brazil, XC 43197; (C) loudsong from Misiones, Argentina, XC 52805; (D) loudsong from Boa Nova, Bahia, northeastern Brazil, XC 80407; (E) loudsong from Alagoas, northeastern Brazil, FNJV 3019; (F) loudsong from Serra de Baturité, Ceará, northeastern Brazil, XC 80404.

Loudsongs consisted of a bipartite series of notes, the second part ending in a distinctive croak. For analysis, the two components were treated separately: the initial portion was designated the song, and the terminal portion was designated the croak. Although the internal structure of individual notes was similar among vocalizations, temporal organization allowed recognition of four loudsong patterns.

Loudsong 1 (southern) had a song duration of 9.5 ± 1.4 s (mean ± SD). The song comprised 65.4 ± 11.3 notes delivered at a moderate pace (6.9 ± 0.6 notes/s), followed by a short descending croak (0.8 ± 0.5 s) containing 5.1 ± 3.4 notes delivered at 6.5 ± 1.9 notes/s, yielding a pace/note ratio of 1. Of the 31 loudsongs analyzed, five (16%) lacked the terminal croak. Total vocalization duration was 10.2 ± 1.5 s (Fig. 1C). This vocal type occurs from the Brazilian states of Minas Gerais and Espírito Santo south to Rio Grande do Sul and adjacent Argentina and Paraguay (Supplementary Fig. S1).

Loudsong 2 (Bahia) was structurally similar to Loudsong 1, with a song duration of 9.3 ± 1.3 s. The initial segment contained fewer notes (54.9 ± 7.9) delivered at a slower pace (5.9 ± 0.5 notes/s). In contrast, the croak was substantially longer (2.4 ± 0.6 s), comprising 18.4 ± 4.5 notes delivered at a faster pace (7.7 ± 0.7 notes/s), resulting in a pace/note ratio of 2. All 19 loudsongs analyzed included a terminal croak. Total vocalization duration was 11.7 ± 1.3 s (Fig. 1D). This vocal type occurs north of the Doce River in southern Bahia, northeastern Brazil (Supplementary Fig. S1).

Loudsong 3 (Alagoas) resembled the preceding vocalizations but was markedly shorter, with a song duration of 4.4 ± 0.9 s. None of the 16 loudsongs analyzed (representing at least two individuals, although only a single recording was available) included a terminal croak (Fig. 1E). This vocalization comprised 33.8 ± 6.1 notes delivered at an intermediate pace of 7.7 ± 0.6 notes/s. It is restricted to the state of Alagoas, north of the São Francisco River, northeastern Brazil (Supplementary Fig. S1).

Loudsong 4 (Ceará) retained the characteristic bipartite structure but differed in several respects. With a song duration of 9.8 ± 1.3 s, it contained significantly more notes (99.1 ± 14.6) delivered at the fastest pace recorded (10.1 ± 0.5 notes/s). The terminal croak was present in all but one loudsong and was more elaborate, lasting 3.4 ± 1.1 s and comprising 33.6 ± 15.2 notes delivered at 9.9 ± 2.6 notes/s. This resulted in a pace/note ratio of 3, and a total vocalization duration of 13.0 ± 2.3 s (Fig. 1F). This vocal type is restricted to the state of Ceará, northeastern Brazil (Supplementary Fig. S1).

Total loudsong duration differed significantly among populations (ANOVA; F1,2 = 278.1; p < 0.001). Pairwise comparisons detected differences between all populations except Bahia and Ceará (Supplementary Table S4). Loudsongs from Alagoas (n = 16) were the shortest and showed no overlap in duration with those of the remaining populations (Fig. 2A). Other loudsong components also varied markedly among populations, including both the song (F1,2 = 38.4; p < 0.001) and croak segments (F1,2 = 34.6; p < 0.001). In contrast, frequency parameters were relatively consistent, whereas note number and temporal pace differed significantly in both components (Table 1).

The song segment was significantly shorter in Alagoas (Tukey; p < 0.001). Although song duration overlapped among the remaining populations, both note number and pace differed significantly across populations (Supplementary Table S3). Note number showed non-overlapping distributions between Alagoas and Ceará (Fig. 2B), whereas pace differed significantly and without overlap between the southern-Bahia and Ceará populations (Fig. 2C). Song pace showed a weak but significant positive correlation with latitude (ρ = 0.41; r² = 0.17; p < 0.001). In contrast, note number was only weakly associated with latitude (ρ = 0.21; r² = 0.04; p < 0.010), providing no evidence of a clinal pattern across the 4°-29°S gradient (Fig. 2D).

Figure 2
Violin plots with embedded boxplots depict: (A) Total loudsong duration; (B) number of notes in the first part of the loudsong; (C) pace of the first part of the loudsong; (D) lack of correlation between the number of notes in the first part of the loudsong and latitude.

The croak also differed significantly among populations in both duration and note number (Tukey; p < 0.001 for all comparisons), although the observed ranges overlapped. Croak pace did not differ significantly between the southern and Bahia populations (Supplementary Table S3), which likewise showed overlapping distributions. Both croak note number (ρ = 0.42; r² = 0.17; p < 0.001) and pace (ρ = 0.36; r² = 0.13; p < 0.001) increased significantly with latitude, despite overlap between the southern-Bahia and Ceará populations.

Despite overlap in individual acoustic variables, LDA clearly discriminated population-level variation in loudsongs. The first discriminant function (LD1), explaining 80.4% of the variance, was positively associated with song pace and negatively associated with maximum entropy; for the croak, it was negatively associated with both pace and duration (Supplementary Table S5). The second discriminant function (LD2), explaining 19.2% of the variance, further improved population discrimination. Together, the analyses identified three vocal groups within Atlantic Chamaeza populations: (a) a southern-Bahia cluster with highly similar vocal profiles; (b) the markedly divergent Alagoas population; and (c) the strongly differentiated Ceará population (Fig. 3). Although individual acoustic variables overlapped among groups, the multivariate analysis effectively captured population-level bioacoustic differentiation.

Figure 3
Multivariate ordination plots with linear discriminants (LD1 and LD2) colored according to the two-part loudsong types of Chamaeza campanisona. The shaded ellipses represent the 95% confidence intervals for each subspecies.

Supervised classification

Because BirdNET cannot analyze recordings shorter than 1.0 s, all calls were excluded from the automated analysis. Although some vocalizations appeared superficially similar among populations, quantitative visualization revealed distinct clustering patterns. Song components showed partial overlap between the southern and Bahia populations, whereas Ceará formed a more clearly separated cluster. This distinction was even more pronounced for the croak component (Fig. 4).

Figure 4
Two-dimensional UMAP projections of loudsong embeddings of Chamaeza campanisona. Individual points represent a single vocalization, colored according to their respective loudsong types. Ellipses represent 95% confidence intervals.

The SVM model showed high classification performance, achieving accuracies of 0.92 for songs and 0.90 for croaks, with consistently moderate to high precision, recall, and F1-scores across categories. Classification was perfect (100% accuracy) for both Alagoas and Ceará vocalizations in the test set, whereas performance was lower for the southern and Bahia populations. Croak recall was particularly low for Bahia (43% true positives), accompanied by similarly reduced precision and recall values. By comparison, the random classifier performed substantially worse, with accuracies of 0.50 for songs and 0.45 for croaks. Although this null model correctly classified the most frequent classes, it consistently misclassified all remaining instances (Supplementary Table S6). Overall, the supervised bioacoustic analyses supported the recognition of three vocal groups: (a) the broadly distributed southern-Bahia populations (Loudsongs 1 and 2), (b) the allopatric Alagoas population (Loudsong 3), and (c) the isolated Ceará population (Loudsong 4) of the Short-tailed Antthrush.

Morphology

Examination of specimens at MZUSP revealed no major plumage differences between Ceará and nominotypic (southern-Bahia) populations (Supplementary Fig. S2), with two notable exceptions. First, the throat and chest of the Ceará specimen consistently showed a pale-yellow coloration (2.5Y 8/4), whereas southern-Bahia specimens were typically whitish (10Y 8/1). However, some southern-Bahia specimens (e.g., MZUSP 49978) exhibited coloration similar to the Ceará phenotype. Second, nominotypic birds generally showed darker and more extensive black scaling on the underparts, including both the feather margins and vane interiors, than birds from Ceará.

The most reliable diagnostic character was the presence of a distinct black spot immediately above the bill, associated with yellowish lores (2.5Y 8/8), in specimens south of the São Francisco River. In contrast, all examined Ceará specimens, including the MZUSP material and the adult male from FMNH (122751; Fig. 5), lacked this frontal black spot and had lores concolorous with the crown. Despite individual variation in other plumage traits, this feature consistently distinguished the two populations.

Figure 5
Representation of two Chamaeza campanisona populations. The specimen on the left (MZUSP 33408) is from Ceará, northeastern Brazil, and the specimen on the right (MZUSP 80747) is from São Paulo, southeastern Brazil. (A) Ventral and (B) dorsal views. The throat and chest of the Ceará population are buffier (C), while the south population shows a conspicuous black spot on the forehead, a feature absent in both specimens from northeastern Brazil (FMNH 122751) (D). Scale bar: 15 mm.

Although the Alagoas population possesses a diagnosable loudsong (Loudsong 3), it is currently represented by a single recording (FNJV 3019) made by J. M. Vielliard at Fazenda Riachão, Quebrangulo, on 25 January 1981, documenting at least two individuals. Additional evidence includes three photographic records from the same municipality deposited in WikiAves (WA 18611, 15 February 2008; WA 1216352, 13 January 2014) and observations reported by Studer et al. (2018), all indicating the absence of the black frontal spot. Although the Alagoas population is vocally distinctive because of the absence of the terminal croak and its unique pace, the currently available evidence remains insufficient to support a formal taxonomic change. Additional vocal recordings and museum specimens are needed to adequately evaluate its taxonomic status.

Based on the available bioacoustic and morpholo gical evidence, we propose retaining the southern-Bahia populations as C. campanisona and recognizing the Ceará population as a distinct phylogenetic species because of its consistent vocal and morphological differentiation, including buff ventral plumage and the absence of the frontal black spot. Although the Alagoas population exhibits promising diagnostic characters, additional material is required before formal taxonomic recognition can be proposed.

Taxonomic summary

Chamaeza campanisona (Lichtenstein, 1823)

Turdus brevicaudus Vieillot, 1818

Myiothera campanisona Lichtenstein, 1823

Chamaeza meruloides Vigors, 1825

Myioturdus marginatus Wied, 1831

Myiothera strigilata Pucheran, 1855

Chamaezosa strigilata Burmeister, 1856

Chamaeza brevicauda Sclater, 1858

Grallaria campanisona Reinhardt, 1870

Chamaezosa brevicauda Cabanis, 1874

Chamaeza tshororo Bertoni, 1901

Chamaeza brevicauda brevicauda Ménégaux & Hellmayr, 1906

Chamaesa brevicauda Chubb, 1910

Chamaeza brevicauda tshororo Bertoni, 1914

Short-tailed Antthrush (English name)

Tovaca-campainha (Portuguese name)

Type locality. São Paulo, Brazil.

Type material. Holotype. ZMB 3163. Unsexed specimen collected by F. Sellow and I. von Olfers between 1818-1820 in Brazil, South America.

Description. The species shows no sexual dimorphism in plumage. The crown is uniformly olive-brown (2.5Y 4/4) and bears a distinct black spot immediately above the bill, one of the diagnostic characters highlighted in the origi nal description (Lichtenstein 1823). The upperparts are olive-brown (5Y 4/4). The rectrices are reddish brown (5YR 2.5/2), with a black subterminal band and narrow whitish tips. The lores are yellowish (2.5Y 8/8), the postocular supercilium is short and whitish, and the sides of the neck bear whitish patches (10Y 8/1). The throat is white, whereas the upper breast and crissum are washed with buff (2.5Y 6/6). The breast and flanks are strongly streaked dark olive-brown (2.5Y 3/3), contrasting with the lightly marked central belly. The crissum also shows scattered black barring or spotting (Fig. 6A). The species is further diagnosed by its vocalizations, corresponding to Loudsongs 1 and 2.

Distribution. Endemic to the Atlantic Forest of eastern South America, ranging from southern Bahia, Brazil, south to Rio Grande do Sul, and extending into the interior Atlantic Forest of northeastern Argentina and southeastern Paraguay (Fig. 7).

Figure 6
(A) Chamaeza campanisona; (B) Chamaeza baturitensis sp. nov. Plate by Eduardo P. Brettas.

Figure 7
(A) Range of the Chamaeza campanisona species complex in South America (IUCN 2024). (B) Distribution limits of Chamaeza campanisona and Chamaeza baturitensis sp. nov. in eastern South America. The type locality of C. campanisona is not accurate to be represented in the map. The Alagoas population remains taxonomically uncertain.

Remarks. Cory and Hellmayr (1924) also suggested São Paulo as the type locality. The specimens described by Lichtenstein (1823) were subsequently sold at auction. Because C. campanisona was already represented in the Berlin collection, the original description was based on a type series rather than on a designated holotype. The whereabouts of the remaining specimens are unknown (F. Sylke, in litt.). The consistent vocal diagnosability, together with the examination of the two known Ceará specimens, supports the recognition of a previously unrecognized taxon characterized by a unique combination of vocal and morphological features. Based on this integrative evidence, we propose the following new species.

Chamaeza baturitensis sp. nov.

Chamaeza campanisona campanisona Pinto, 1978

Chamaeza campanisona Willis, 1992

https://zoobank.org/DDA8BC60-7500-4142-89DA-67F2CC874C23

Baturité Antthrush (suggested English name)

Tovaca-de-baturité (suggested Portuguese name)

Type locality. Pacoti, Serra do Baturité, Ceará, Brazil.

Type material. Holotype. MZUSP 33408. Adult male collected by E.G. Holt on 7 February 1941. Paratype. FMNH 122751. Adult male collected by E.G. Holt on 10 February 1941 at Pacoti, Serra do Baturité, Ceará.

Diagnosis. Morphologically, Chamaeza baturitensis sp. nov. is similar to C. campanisona but differs in lacking the black frontal spot and the yellowish lores characteristic of that species. It is further distinguished by its vocalizations, producing Loudsong 4, which has a faster song pace than that of C. campanisona, with no overlap between species.

Description. Morphologically, the new species closely resembles its congener, although examination of the two available specimens revealed subtle but consistent differences. Both specimens have a distinctly buff throat and chest (2.5Y 8/4), whereas the corresponding region is predominantly whitish in C. campanisona. The most consistent diagnostic character is the absence of the black frontal spot present in C. campanisona (Fig. 6B). This feature was already emphasized in the original description of C. campanisona (Lichtenstein 1823) and later noted by Willis (1992), who recognized the distinctiveness of three Chamaeza taxa in eastern Brazil. Acoustically, the new species is readily diagnosed by Loudsong 4, which differs from that of C. campanisona in both duration and temporal pattern. Among the populations examined, it has the longest loudsong and the fastest pace in the initial song segment.

Distribution. The species is restricted to the Baturité Mountains, a highland rainforest complex locally known as Brejos de Altitude, surrounded by the seasonally dry Caatinga, in the municipalities of Guaramiranga and Pacoti, Ceará, northeastern Brazil, north of the São Francisco River (Fig. 7).

Etymology. The feminine name baturitensis refers to the Baturité Mountains, which encompass both the type locality and the known geographic range of the new species.

DISCUSSION

Under the framework proposed by Isler et al. (1998) for thamnophilid antbirds, a minimum of three vocal differences is considered a conservative criterion for delimiting allopatric species. This threshold was established based on the number of vocal differences observed between undisputed syntopic congeneric species pairs, with the aim of minimizing Type I errors (i.e., erroneously elevating populations to species status). The authors emphasized that this guideline is not absolute; fewer vocal differences may be sufficient when accompanied by strong non-vocal differentiation or in cases of parapatry. Applying these criteria, C. baturitensis sp. nov. does not meet the conservative threshold of three non-overlapping vocal differences, differing in one quantitative vocal character (pace of the song segment of the loudsong), together with one consistent plumage character. Under the Biological Species Concept (Mayr 2000), it could therefore be regarded as a subspecies of C. campanisona. In contrast, under the Phylogenetic Species Concept (Cracraft 1983), its diagnostic characters support recognition as a distinct species.

It has long been recognized that Chamaeza requires a formal taxonomic review, with previous studies suggesting that currently recognized species may encompass additional cryptic diversity (Krabbe and Schulenberg 2003). Oliveira (2016) included one C. campanisona specimen from Ceará in a molecular phylogeny, but sequencing failure prevented its inclusion in the final analyses. As a result, our bioacoustic and morphological analyses led to the discovery of a previously unrecognized Chamaeza species, highlighting the importance of fundamental research in documenting Neotropical biodiversity.

Initial assessments of vocal distinctiveness within this species complex focused exclusively on loudsongs, revealing pronounced differentiation among Andean populations, whereas Atlantic Forest populations were treated as a single vocal unit (Boesman 2016). The calls examined here exhibited limited variation, probably reflecting the scarcity of recordings from the Bahia and Ceará populations, which were also represented by fewer vocalizations. Notably, one call type was more frequent in the southern populations, the only group in which alarm calls were documented. Differences in call frequency among populations also agree with patterns reported for disjunct oscine taxa (Cavarzere et al. 2024). Morphological examination revealed a diagnosable character-a black frontal spot-present in all but the two Ceará specimens. Although this feature had previously been mentioned (Lichtenstein 1823, Bertoni 1901, Willis 1992), it had never been formally evaluated from a taxonomic perspective. Interestingly, the original description of Myioturdus marginatus from Bahia makes no reference to this character (Wied 1831). However, examination of one syntype (AMNH 5405) and the only available photographic record from the region (WA 2530821) confirmed the presence of the frontal black spot. The second syntype (AMNH 5406), which lacks the black spot, is a juvenile female (LeCroy and Sloss 2000). By integrating vocal and morphological evidence, we identified diagnostic features that, although previously noted (Willis 1992, Albano and Girão 2008), had been overlooked in subsequent studies (Boesman 2016, Greeney 2020).

Previous molecular studies recovered well-supported phylogenies revealing the paraphyly of Chamaeza (Oliveira 2016, Harvey et al. 2020). Specifically, C. campanisona was recovered as paraphyletic with respect to the Amazonian C. nobilis, with Atlantic Forest taxa genetically distinct from the Andean subspecies (Oliveira 2016, Harvey et al. 2020). A pronounced genetic discontinuity within C. campanisona was also reported, with individuals from southern Bahia (corresponding to our Bahia population) forming a distinct clade separate from the southern Brazilian populations. However, we found no comparable differentiation in our data. Bahia specimens (AMNH 5405, 5406) were indistinguishable in plumage from those of southern Brazil, and Wied himself probably recognized the need to synonymize his own description with the nominotypical form (LeCroy and Sloss 2000).

Southern loudsongs (1 and 2) showed high classification performance (F1-score = 0.92). The slightly lower precision (0.90) than recall (0.93) indicates the occurrence of some false positives. In contrast, Bahia loudsongs showed lower performance (F1-score = 0.60), probably reflecting the small sample size (support = 5) and overlap with other populations. Although SVM performance was lower for this group, both LDA and UMAP indicated a similar pattern, revealing some degree of differentiation but also substantial overlap. Similar results were reported in the taxonomic revision of the Gray Antbird Cercomacra cinerascens species complex (Cavarzere et al. 2026), further supporting the usefulness of supervised machine-learning approaches in avian taxonomy.

Initially, we hypothesized that the Bahia population would produce loudsongs with more notes, particularly in the croak. However, analyses across the entire latitudinal gradient of southern Bahia showed these vocalizations to be indistinguishable from those of the southern populations. In contrast, note number was significantly correlated with latitude, suggesting a possible vocal cline-a pattern previously documented only in Thamnophilidae (Isler et al. 2005). The tendency for Bahia birds to produce croaks with more notes nevertheless represents subtle geographic variation, similar to that reported for another suboscine with a distribution comparable to that of the Short-tailed Antthrush. The Drab-breasted Pygmy-Tyrant Hemitriccus diops exhibits minor vocal differences between populations north and south of the Doce River, where limited mtDNA divergence is accompanied by only slight plumage and vocal differentiation (Acero-Murcia et al. 2021).

The Alagoas population showed high precision (1.00) and recall (0.88), indicating that the model failed to identify only 12% of true Alagoas instances (false negatives). However, because all analyses were based on a single recording, possibly representing only two individuals from one locality, pseudoreplication cannot be ruled out. All analyzed loudsongs lacked the terminal croak, a feature absent in only ~3% of the remaining populations, and exhibited a qualitatively distinct spectrographic pattern. Although the species was described as “very common” during the 1980s (Teixeira et al. 1988), no additional recordings have become available since then. The population was studied for more than two decades, culminating in a detailed reproductive biology study (Studer et al. 2018), although taxonomic evaluation was not its primary focus. Available visual documentation consists of two WikiAves photographs from Quebrangulo (2008, WA 18611, and 2014, WA 1216352) and one photograph in Studer et al. (2018), all showing the frontal region but lacking the diagnostic black frontal spot present in C. campanisona. Because no museum specimens are available (Teixeira et al. 1988), additional material is needed before the taxonomic status of this population can be properly evaluated. This need is reinforced by the Pleistocene environmental stability of the Pernambuco Center of Endemism, which has promoted the evolution of numerous endemic bird taxa (e.g., Feijó et al. 2023, Percequillo and Weksler 2023, Bocalini et al. 2026).

The Ceará population was consistently identified as vocally diagnosable by both LDA and UMAP. Although most vocal parameters differed significantly from those of the remaining populations, the most striking distinction was the non-overlapping pace of the first loudsong segment-a feature considered biologically informative for species delimitation, even in the presence of vocal clines (Isler et al. 2005). Morphological examination further supported this differentiation: Ceará specimens uniquely lacked the black frontal spot, a diagnostic character present in all examined C. campanisona specimens. This marked difference in frontal plumage is noteworthy because this region is often involved in sexual selection and typically shows geographically structured variation. A similar pattern occurs among recognized species of Formicarius (e.g., F. rufifrons versus congeners). Although cryptic diversity has recently been documented within Formicariidae (Areta and Saldívar 2025), variation in frontal plumage has not previously been reported in Chamaeza (see Willis 1992). Our bioacoustic results are therefore interpreted within the framework of innate suboscine loudsongs, using vocal divergence as a proxy for reproductive isolation, consistent with recent systematic and taxonomic studies (e.g., Lima et al. 2025).

Our study identified three distinct loudsong patterns within eastern South American populations of the C. campanisona complex. One pattern is diagnosable but requires additional evidence, including more vocal recordings, museum specimens, and molecular data, before its taxonomic status can be adequately assessed. The remaining two are diagnosable based on both vocal and morphological evidence. The southeastern populations, corresponding to C. campanisona, are distributed throughout the Atlantic Forest from southeastern Paraguay and northeastern Argentina north to southern Bahia, Brazil. In contrast, C. baturitensis sp. nov. is restricted to the montane forest enclaves of Ceará, northeastern Brazil.

The origin of these humid forests has been attributed to Pleistocene climatic fluctuations (Andrade-Lima 1982). During this period, expansions of the Atlantic Forest into the Caatinga were followed by subsequent contractions, leaving isolated forest remnants in favorable microclimatic conditions. These enclaves now function as humid forest islands within the semiarid Caatinga, serving as refugia for Amazonian and Atlantic Forest species and promoting the persistence and evolution of distinct taxa, including a unique fauna and flora (Fernandes-Ferreira et al. 2015, Silvera et al. 2020, Farias et al. 2024, Pires and Sobczak 2024). In Ceará, these forests also harbor several endemic bird and mammal taxa (Albano and Girão 2008, Feijó and Langguth 2013, Percequillo and Weksler 2023, Araújo et al. 2025).

The new Chamaeza species is threatened by poaching (Fernandes-Ferreira et al. 2012) and is currently known from only two localities in the Baturité Mountains (Albano and Girão 2008). Because it has not previously been recognized as a distinct lineage, C. baturitensis sp. nov. is not included in current national or global threat assessments (Brasil 2022, IUCN 2024). Nevertheless, given its estimated extent of occurrence of less than 100 km² and its apparent dependence on well-preserved forest understory (Albano and Girão 2008), the species should currently be treated as Data Deficient (DD) until population studies allow a robust assessment of its conservation status.

ACKNOWLEDGMENTS

We thank the Fonoteca Neotropical Jacques Vielliard and the Macaulay Library for providing sound recordings used in this study. We are grateful to Luís F. Silveira (Museu de Zoologia da Universidade de São Paulo) and to the collection managers Antenor Silva Jr (Museu de História Natural Capão da Imbuia), Ben D. Marks (Field Museum of Natural History), Brett Benz (University of Michigan Museum of Zoology), Juliana S. Patiño (Natural History Museum of Los Angeles County), Pascal Eckhoff (Museum für Naturkunde), and Sahid M.R. Bello (American Museum of Natural History) for providing specimen photographs. We thank Deborah F. N. de Oliveira for relevant information and the reviewers for their valuable comments and suggestions.

LITERATURE CITED

• Acero-Murcia AC, Amaral FR, Barros FC, Ribeiro TS, Myiaki CY, Maldonado-Coelho M (2021) Ecological and evolutionary drivers of geographic variation in songs of a Neotropical suboscine bird: the Drab-breasted Bamboo Tyrant (Hemitriccus diops, Rhynchocyclidae). Ornithology 138(2): 1-15. https://doi.org/10.1093/ornithology/ukab003
  » https://doi.org/10.1093/ornithology/ukab003
• Albano C, Girão W (2008) Aves das matas úmidas das serras de Aratanha, Baturité e Maranguape, Ceará. Revista Brasileira de Ornitologia 16(2): 142-154.
• Andrade-Lima D (1982) Present day forest refuges in Northeastern Brazil. In: Prance GT (Ed.) Biological diversification in the Tropics. Columbia University Press, New York, 245-254.
• Araújo KC, Andrade EB, Brasileiro AC, Santos EM, Ávila RW, Roberto IJ (2025) Knowledge leads to protection: anuran composition and conservation of the ‘Brejos de altitude’ of Northeastern Brazil. Studies on Neotropical Fauna and Environment 60(2): 191-200. https://doi.org/10.1080/01650521.2024.2402614
  » https://doi.org/10.1080/01650521.2024.2402614
• Araya-Salas M, Smith-Vidaurre G (2017) warbleR: an R package to streamline analysis of animal acoustic signals. Methods in Ecology and Evolution 8(2): 184-191. https://doi.org/10.1111/2041-210X.12624
  » https://doi.org/10.1111/2041-210X.12624
• Areta JI, Saldívar MJB (2025) Species limits in the Black-headed Antthrush (Formicarius nigricapillus). Journal of Ornithology 166: 841-851. https://doi.org/10.1007/s10336-025-02265-5
  » https://doi.org/10.1007/s10336-025-02265-5
• Audacity Team (2021) Audacity(R): Free Audio Editor and Recorder. Sony Creative Software Inc., v. 3.0.0. https://www.audacityteam.org/ [Accessed: 03/02/2025]
  » https://www.audacityteam.org/
• Bertoni AW (1901) Aves nuevas del Paraguay. Continuación á Azara. Talleres Nacionales de H. Kraus, Asunción, 216 pp.
• Bocalini F, Bolívar-Leguizamón SD, Vaz RV, Silveira LF (2026) Integrating genomic and vocal evidence to reassess species’ limits and the biogeography of a widely distributed Neotropical passerine, the white-throated spadebill, Platyrinchus mystaceus (Aves: Platyrhinchidae). Zoological Journal of the Linnean Society 206(1): zlaf197. https://doi.org/10.1093/zoolinnean/zlaf197
  » https://doi.org/10.1093/zoolinnean/zlaf197
• Boesman P (2016) Notes on the vocalizations of Short-tailed Antthrush (Chamaeza campanisona). In: Del Hoyo J, Elliott A, Christie DA (Eds) Handbook of the birds of the world alive. https://doi.org/10.2173/bow-on.100075
  » https://doi.org/10.2173/bow-on.100075
• Brasil (2022) Portaria MMA nº 148, de 7 de junho de 2022. Altera os Anexos da Portaria no 44, de 17 de dezembro de 2014, da Portaria no 444, de 17 de dezembro de 2014, e da Portaria no 445, de 17 de dezembro de 2014, referentes à atualização da Lista Nacional de Espécies Ameaçadas de Extinção. Diário Oficial da União 148: 1-89.
• Cavarzere V, Breviglieri EL, Silveira LF (2026) Integrative taxonomy of the Cercomacra cinerascens (Sclater, 1857) (Aves: Thamnophilidae) species complex with description of two new species. Vertebrate Zoology 76: 73-91. https://doi.org/10.3897/vz.76.e171834
  » https://doi.org/10.3897/vz.76.e171834
• Cavarzere V, Costa TVV, Cabanne GS, Trujillo-Arias N, Marcondes RS, Silveira LF (2024) A new species of tanager (Aves: Thraupidae) from the eastern slopes of the Andes. Zootaxa 5468(3): 541-556. https://doi.org/10.11646/zootaxa.5468.3.7
  » https://doi.org/10.11646/zootaxa.5468.3.7
• K. Lisa Yang Center for Conservation Bioacoustics (2024) Raven Pro: Interactive Sound Analysis Software. Cornell Lab of Ornithology, Ithaca. https://www.ravensoundsoftware.com/software/raven-pro/
  » https://www.ravensoundsoftware.com/software/raven-pro/
• Cortes C, Vapnik V (1995) Support-vector networks. Machine Learning 20: 273-297. https://doi.org/10.1007/BF00994018
  » https://doi.org/10.1007/BF00994018
• Cory C, Hellmayr C (1924) Catalogue of the birds of the Americas. Part 3. Pteroptochidae, Conopophagidae, Formicariidae. Field Museum of Natural History Publication 223: 1-369.
• Cracraft J (1983) Species concepts and speciation analysis. In: Johnston RF (Ed.) Current ornithology. Springer, New York, 159-187. https://doi.org/10.1007/978-1-4615-6781-3_6
  » https://doi.org/10.1007/978-1-4615-6781-3_6
• Farias ES, Souza CRL, Costa RCM, Carvalho-Fernandes SP, Santana EWP, et al. (2024) Diversity of biting midges Culicoides (Diptera: Ceratopogonidae) in Pacoti municipality, Baturité Highland, Ceará state, northeast of Brazil. Biota Neotropica 24: e20241668. https://doi.org/10.1590/1676-0611-BN-2024-1668
  » https://doi.org/10.1590/1676-0611-BN-2024-1668
• Feijó A, Beltrão M, Costa-Pinto AL, Rocha PA, Freitas MA, et al. (2023) Mammals of the Pernambuco Endemism Center: Diversity, Biogeography, Research Gaps, and Conservation Concerns. In: Pereira Filho G, França FGR, Alves RRN, Vasconcellos A (Eds) Animal biodiversity and conservation in Brazil’s northern Atlantic Forest. Springer, Cham. https://doi.org/10.1007/978-3-031-21287-1_13
  » https://doi.org/10.1007/978-3-031-21287-1_13
• Feijó A, Langguth A (2013) Mamíferos de médio e grande porte do nordeste do Brasil: distribuição e taxonomia, com descrição de novas espécies. Revista Nordestina de Biologia 22: 223-225.
• Fernandes-Ferreira H, Gurgel-Filho NM, Feijó JA, Mendonça SV, Alves RRN, Langguth A (2015) Non-volant mammals from Baturité Ridge, Ceará state, Northeast Brazil. Check List 11: 1-7. https://doi.org/10.15560/11.3.1630
  » https://doi.org/10.15560/11.3.1630
• Fernandes-Ferreira H, Mendonça SV, Albano C, Ferreira FS, Alves RRN (2012) Hunting, use and conservation of birds in Northeast Brazil. Biodiversity and Conservation 21: 221-244. https://doi.org/10.1007/s10531-011-0179-9
  » https://doi.org/10.1007/s10531-011-0179-9
• Greeney HF (2020) Short-tailed Antthrush (Chamaeza campanisona). In: Schulenberg TS (Ed.) Birds of the World. Cornell Lab of Ornithology, Ithaca . https://doi.org/10.2173/bow.shtant1.01
  » https://doi.org/10.2173/bow.shtant1.01
• Harvey MC, Bravo GA, Claramunt S, Cuervo AM, Derryberry GE, et al. (2020) The evolution of a tropical biodiversity hotspot. Science 370(6522): 1343-1348. https://doi.org/10.1126/science.aaz6970
  » https://doi.org/10.1126/science.aaz6970
• Isler ML, Isler PR, Brumfield RT (2005) Clinal variation in vocalizations of an Antbird (Thamnophilidae) and implications for defining species limits. Auk 122(2): 433-444. https://doi.org/10.1093/auk/122.2.433
  » https://doi.org/10.1093/auk/122.2.433
• Isler ML, Isler PR, Whitney BM (1998) Use of vocalizations to establish species limits in antbirds (Passeriformes: Thamnophilidae). Auk 115(3): 577-590. https://doi.org/10.2307/4089407
  » https://doi.org/10.2307/4089407
• IUCN (2024) The IUCN Red List of Threatened Species. International Union for Conservation of Nature, Cambridge. https://www.iucnredlist.org [Accessed: 04/06/2025]
• Kahl S, Wood CM, Eibl M, Klinck H (2021) BirdNET: A deep learning solution for avian diversity monitoring. Ecological Informatics 61: 101236. https://doi.org/10.1016/j.ecoinf.2021.101236
  » https://doi.org/10.1016/j.ecoinf.2021.101236
• Krabbe NK, Schulenberg TS (2003) Family Formicariidae (ground antbirds). In: Del Hoyo J, Elliott A, Christie DA (Eds) Handbook of the Birds of the World. Broadbills to Tapaculos. Lynx Edicions, Barcelona, 682-731.
• Lakdari MW, Ahmad AH, Sethi S, Bohn GA, Clink DJ (2024) Mel-frequency cepstral coefficients outperform embeddings from pre-trained convolutional neural networks under noisy conditions for discrimination tasks of individual gibbons. Ecological Informatics 80: 102457. https://doi.org/10.1016/j.ecoinf.2023.102457
  » https://doi.org/10.1016/j.ecoinf.2023.102457
• LeCroy M, Sloss R (2000) Type specimens of birds in the American Museum of Natural History. Part 3. Passeriformes: Eurylaimidae, Dendrocolaptidae, Furnariidae, Formicariidae, Conopophagidae, and Rhinocryptidae. Bulletin of the American Museum of Natural History 257(1): 1-88. https://doi.org/10.1206/0003-0090(2000)257<0001:TSOBIT>2.0.CO;2
• Lichtenstein H (1823) Doubletten Zoologischen Museums. Commission bei T. Trautwein, Berlin, 118 pp.
• Lima RD, Bolívar-Leguizamón SD, Bocalini F, Marcondes RS, Brumfield RT, Silveira LF (2025) Geographic variation, population structure, and taxonomy of the Variable Antshrike (Thamnophilus caerulescens). Zoological Journal of the Linnean Society 205(2): zlaf129. https://doi.org/10.1093/zoolinnean/zlaf129
  » https://doi.org/10.1093/zoolinnean/zlaf129
• Mayr E (2000) The biological species concept. In: Wheeler QD, Meier R (Eds) Species concepts and phylogenetic theory: A debate. Columbia University Press, New York, 17-29.
• McGinn K, Kahl S, Peery MZ, Klink H, Wood CM (2023) Feature embeddings from the BirdNET algorithm provide insights into avian ecology. Ecological Informatics 74: 101995. https://doi.org/10.1016/j.ecoinf.2023.101995
  » https://doi.org/10.1016/j.ecoinf.2023.101995
• McInnes L, Healy J, Melville J (2018) Umap: Uniform manifold approximation and projection for dimension reduction. arXiv Preprint. https://doi.org/10.48550/arXiv.1802.03426
  » https://doi.org/10.48550/arXiv.1802.03426
• McKitrick MC, Zink RM (1988) Species concepts in ornithology. The Condor 90: 1-14. https://doi.org/10.2307/1368426
  » https://doi.org/10.2307/1368426
• Munsell AH (1994) Soil, Color Charts, Revised Edition. Kollmorgan Instruments Co., New York, 31 pp.
• Oliveira DFN (2016) Filogenia molecular de Chamaeza Vigors, 1825 (Aves, Formicariidae) e implicações para a história evolutiva das florestas da América do Sul. MSc Thesis, Universidade Federal de São Paulo, Diadema, Brazil, 42 pp. https://repositorio.unifesp.br/bitstreams/3d825efa-3d16-4be3-9c17-87922fc36909/download
  » https://repositorio.unifesp.br/bitstreams/3d825efa-3d16-4be3-9c17-87922fc36909/download
• Percequillo AR, Weksler M (2023) Systematics of the genus Euryoryzomys Weksler et al., 2006 (Rodentia: Cricetidae): integrative analysis reveals a new species from north-eastern Brazil. Zoological Journal of the Linnean Society 199: 594-632. https://doi.org/10.1093/zoolinnean/zlad048
  » https://doi.org/10.1093/zoolinnean/zlad048
• Pires JC, Sobczak JF (2024) The high biodiversity of the Serra de Baturité Environmental Protection Area, Ceará, Brazil. Biota Neotropica 24: e20241644. https://doi.org/10.1590/1676-0611-BN-2024-1644
  » https://doi.org/10.1590/1676-0611-BN-2024-1644
• R Core Team (2023) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna. http://www.R-project.org
  » http://www.R-project.org
• Raposo M, Teixeira D (1992) Revalidação de Chamaeza meruloides Vigors, 1825 (Aves, Formicariidae). Boletim do Museu Nacional 350: 1-11.
• Ridgely RS, Tudor G (1994) The Birds of South America. The University of Texas Press, Austin, vol. 2, 940 pp.
• Sanchís-Pedregosa C, Machuca JAD, González-Zamora M (2011) Logistics outsourcing: performance models and financial and operational indicators. Mechanical and Industrial Engineering 5: 1311-1316.
• Seddon N (2005) Ecological adaptation and species recognition drives vocal evolution in Neotropical suboscine birds. International Journal of Organic Evolution 59: 200-215. https://doi.org/10.1111/j.0014-3820.2005.tb00906.x
  » https://doi.org/10.1111/j.0014-3820.2005.tb00906.x
• Sick H (1997) Ornitologia Brasileira. Nova Fronteira, Rio de Janeiro, 912 pp.
• Silvera AP, Loiola MIB, Gomes VS, Lima-Verde LW, Oliveira TS, et al. (2020) Flora of Baturité, Ceará: a Wet Island in the Brazilian Semiarid. Floresta e Ambiente 27: e20180320. https://doi.org/10.1590/2179-8087.032018
  » https://doi.org/10.1590/2179-8087.032018
• Studer A, Sousa MC, Goyena BB (2018) The breeding biology and nest success of the Short-tailed Antthrush Chamaeza campanisona (Aves: Formicariidae) in the Atlantic rainforest of northeastern Brazil. Zoologia 35(1): e12906. https://doi.org/10.3897/zoologia.35.e12906
  » https://doi.org/10.3897/zoologia.35.e12906
• Teixeira DM, Nacinovic JB, Luigi G (1988) Notes on some birds of northeastern Brazil (3). Bulletin of the British Ornithologists’ Club 108: 75-79.
• Tobias JA, Brawn JD, Brumfield RT, Derryberry EP, Kirschel AN, Seddon N (2012) The importance of suboscine birds as study systems in ecology and evolution. Ornitologia Neotropical 23: 261-274.
• Touchton JM, Seddon N, Tobias JA (2014) Captive rearing experiments confirm song development without learning in a Tracheophone Suboscine bird. PLOS One 9: e95746. https://doi.org/10.1371/journal.pone.0095746
  » https://doi.org/10.1371/journal.pone.0095746
• Venables WN, Ripley BD (2002) Modern applied statistics with S. Springer, New York, 455 pp.
• Wied M (1831) Beiträge zur Naturgeschichte von Brasilien. Landes-Industrie-Comptoirs, Weimar, 1277 pp.
• Willis EO (1984) Antthrushes, antpittas, and gnateaters (Aves, Formicariidae) as army ant followers. Revista Brasileira de Zoologia 2(7): 443-448. https://doi.org/10.1590/S0101-81751984000300005
  » https://doi.org/10.1590/S0101-81751984000300005
• Willis EO (1992) Three Chamaeza Antthrushes in eastern Brazil (Formicariidae). Condor 94(1): 110-116. https://doi.org/10.2307/1368801
  » https://doi.org/10.2307/1368801
• Winkler DW, Billerman SM, Lovette IJ (2020) Antthrushes (Formicariidae). In: Billerman SM, Keeney BK, Rodewald PG, Schulenberg TS (Eds) Birds of the World . Cornell Lab of Ornithology, Ithaca . https://doi.org/10.2173/bow.formic2.01
  » https://doi.org/10.2173/bow.formic2.01

• ZooBank register
  https://zoobank.org/D8EBBA1A-9A01-4105-B75E-9C999FF1C6F6
• Data Availability Statement
  All data generated and/or analyzed are included in this article.
• Funding Statement
  This study was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP; grant 2023/09512-6) and Universidade de São Paulo through institutional resources and logistical support.
• Ethical Statement
  Not applicable. This study was based exclusively on specimens deposited in scientific collections and did not involve field collection or handling of live vertebrate animals.
• AI Statement
  Artificial intelligence tools were used solely to assist with language editing and grammar.
• How to cite this article
  Romero L, Breviglieri EL, Cavarzere V (2026) Revealing cryptic diversity: vocal and morphological evidence for a new species of Chamaeza (Aves: Formicariidae). Zoologia 43: e25083. https://doi.org/10.1590/S1984-4689.v43.e25083
• Published by
  Sociedade Brasileira de Zoologia at Scientific Electronic Library Online - https://www.scielo.br/zool

APPENDIX

Appendix 1. Specimens of Chamaeza campanisona examined in the following institutions: American Museum of Natural History (AMNH), Field Museum of Natural History (FMNH), Museu de História Natural Capão da Imbuia (MHNCI), Museu de Zoologia da Universidade de São Paulo (MZUSP), Museum für Naturkunde (ZMB), Natural History Museum of Los Angeles County (NHMLAC) and University of Michigan Museum of Zoology (UMMZ). Specimens examined through photographs are indicated separately, and includes photographs of live individuals obtained from Wikiaves (WA).

Specimens personally examined

Chamaeza campanisona campanisona. Brazil, CE, Pacoti, Serra do Baturité (1♂, MZUSP-33408); ES, Chaves, Santa Leopoldina (3♀, MZUSP-28274 -28275 -28276; 1 unsexed, MZUSP-28273); MG, Rio Doce, Baixo Suassuna (1♂, MZUSP-25677); Alto Rio Doce (1♂, MZUSP-25678); RJ, Terezópolis, Fazenda C. Guiles (1♂, MZUSP-33411); SP, Alto da Serra (1♂, MZUSP-854); Anhembi, Barreiro Rico (1♂, MZUSP-43228); Barro Branco, Rio Guaraú (1♂, MZUSP-51421); BR-115, km 94 (1♂, MZUSP-103416); Cananéia, Tabatinguera (1♂, MZUSP-15004); Cantareira (1 unsexed, MZUSP-24484); Caraguatatuba (1♀, MZUSP-43725); Caucaia do Alto (1♀, MZUSP-78875); Cotia, Reserva Florestal do Morro Grande (1♂, MZUSP-80747); Estação Ferraz (3♀, MZUSP-55074 -550750 -56859; 1 unsexed, MZUSP-55076); Ilha de São Sebastião (1♀, MZUSP-5951); Ipiranga (1♂, MZUSP-2832); Itapura (2♂, MZUSP-5006 -23964); Itaquera (1♀, MZUSP-11680); Onça Parda (♂, MZUSP-47977; 1 unsexed, MZUSP-47978); Primeiro Morro (1♀, MZUSP-49975); Ribeirão Fundo (2♀, MZUSP-49976 -49985); Rio Ipiranga, Tamanduá (2♂, MZUSP-47973 -47974; 1 unsexed, MZUSP-47976); Rocha (1♀, MZUSP-49979; 1♂, MZUSP-49978); Salesópolis, Estação Biológica de Boraceia (2♀, MZUSP-70596 -83481); São Paulo (1♂, MZUSP-103415); Serra da Cantareira (2♂, MZUSP-24485 -60815); Ubatuba (1♂, MZUSP-5509).

Specimens examined through photographs

Chamaeza campanisona campanisona. Brazil, AL, Quebrangulo (2 unsexed, WA-18611 -1216352); BA, Camacan (1 unsexed, WA-2530821), Vitória da Conquista (1♀, AMNH-5405; 1♂, AMNH-5405); CE, Pacoti, Serra do Baturité (♂, FMNH-122751); RJ, Parati, Pedra Branca (2♀, NHMLAC-27934 -27935); SP, São Paulo (1 unsexed, ZMB-3163). Chamaeza campanisona tshororo. Argentina, Misiones, Tobunas (18♀, NHMLAC-48634 -48640 -48641 -48642 -48649 -48651 - 48652 -48654 -48658 -48659 -48660 -48666 -48667 -48668 -48670 -48672 -48673 -48674; 26♂, NHMLAC-48635 -48636 -48637 -48638 -48639 -48643 -48644 -48645 -48646 -48647 -48648 -48650 -48653 -48655 -48656 -48657 -48661 -48662 -48663 -48664 -48665 -48671 -48675 -48676 -48677 -48678 -48679 -54409; 1 unsexed, NHMLAC-48669); km 10; Arroyo Uruguai (2♂, NHMLAC-48678 -48679); Brazil, PR, Adrianópolis, Mato Limpo, Fazenda Primavera (1♀, MHNCI-5538); Campina Grande do Sul, PCH Novo Horizonte (2♀, MHNCI-6696 -6722); Guaratuba, Baía de Guaratuba, Rio do Meio (1♀, MHNCI-1140; 2♂, MHNCI-1139 -1141), Limeira, Serra da Prata (1♂, MHNCI-4412; 1 unsexed, MHNCI-4622), Rio Cubatão (1♂, MHNCI-1069); Pinhão, Foz do Ribeirão Capoteiro (1♂, MHNCI-4221); Piraquara, Manaciais da Serra (1 unsexed, MHNCI-4358); Quatro Barras, Corvo, Morro Caranguejeira (1♀, MHNCI-3191); Reserva do Iguaçu, Rio da Divisa (1 unsexed, MHNCI-5112); Reserva Florestal Santa Cruz (1♀, MHNCI-577; 1 unsexed, MHNCI-515); SP, Iguape, Taquaruçu (1♀, NHMLAC-28805; 2♂, NHMLAC-28802 -28803); Paraguay, Amambay, Capitan Bado, 40 km WSW (2♀, UMMZ-101064 -101791; 2♂, UMMZ-101063 -101790); Itapua, San Rafael, 3.5 km E (1♂, UMMZ-200817); Puerto Gibaja, Puerto Bertoni, 7 km N (1♀, UMMZ-108835; 3♂, UMMZ-108832 -108833 -108834).

SUPPLEMENTARY MATERIAL

Figure S1

Distribution of the loudsong types of the populations of Chamaeza campanisona in South America.

Figure S2

Specimens of Chamaeza campanisona exhibit no clinal variation in plumage coloration across their latitudinal range (4°-29°S), as evidenced by both A ventral and B dorsal views. Representative specimens shown from north to south (left to right) include Ceará (MZUSP 33407), Minas Gerais (MZUSP 25678), Espírito Santo (MZUSP 28274), Rio de Janeiro (MZUSP 33411), São Paulo (MZUSP 49978), Paraná (MZUSP 6968) and Rio Grande do Sul (MZUSP 38655). Scale bar: 15 mm.

Table S1

Recordings of Chamaeza campanisona obtained from the following databases: Fonoteca Neotropical Jacques Vielliard (FNJV), Macaulay Library (ML), Wikiaves (WA) and Xeno-canto (XC).

Table S2

Results from morphometric Analysis of variance (ANOVA) between Chamaeza campanisona populations. Degrees of freedom are within parentheses after the vocalization types. Sum of squares (Sum Sq), mean square values (Mean Sq) are reported.

Table S3

Results of the Tukey post hoc tests for morphometry within Chamaeza campanisona populations. The differences between groups (diff) along with their lower (lwr) and upper (upr) bounds are reported.

Table S4

Results of the Tukey post hoc tests for the entire loudsong within Chamaeza campanisona. The differences between groups (Diff) along with their lower (lwr) and upper (upr) bounds are reported.

Table S5

Coefficients of the linear discriminants for loudsongs within the Chamaeza campanisona species complex.

Table S6

Precision, recall and F1-Score of the Support vector machine and the Random Classifier to tell subspecies contained in the recordings.

Authors: Romero L, Breviglieri EL, Cavarzere V.

Data type: Species data.

Copyright notice: This dataset is made available under the Open Database License - ODBbL (https://opendatacommons.org/licenses/odbl/1.0/). The ODbL is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited.

Link: https://doi.org/10.1590/S1984-4689.v43.e25083

Edited by

• Editorial responsibility
  Vinicius Tonetti

Data availability

All data generated and/or analyzed are included in this article.

Publication Dates

• Publication in this collection
  03 Aug 2026
• Date of issue
  2026

History

• Received
  19 Sept 2025
• Accepted
  20 Apr 2026