Integrative taxonomy of Cercomacroides serva (Sclater, 1858) demonstrates the validity of C. hypomelaena (Sclater, 1890) comb. nov. (Aves: Thamnophilidae)

validity


Introduction
Cercomacroides Tello & Raposo, 2014 was once classified within Cercomacra Sclater, 1858, which has traditionally comprised two clades: the nigricans and tyrannina groups (Silva 1992;Tello et al. 2014).A well-supported molecular phylogeny showed that Cercomacra (sensu Dickinson 2003) is a paraphyletic assemblage, and a new generic name for the tyrannina clade was warranted (Tello et al. 2014).As a result, Cercomacroides currently has six species (C. laeta, C. parkeri, C. tyrannina, C. serva, C. nigrescens, and C. fuscicauda;Gill et al. 2023).The latter was treated as a subspecies of C. nigrescens but was subsequently shown to be vocally distinct from other subspecies of C. nigrescens (Mayer et al. 2014).
Pyriglena serva (= Cercomacroides serva) was described by Sclater (1858) based on specimens collected near the Napo River, in Eastern Ecuador.Cercomacra hypomelaena (= Cercomacroides hypomelaena), from southeastern Peru, was originally described as a full species a few decades later (Sclater 1890), but was later synonymized with C. serva due to the similarity of the plumage between these topotypical males and males of C. hypomelaena from Peru (Cory andHellmayr 1924, Dickinson andChristidis 2014).This taxon was treated as a valid subspecies, C. s. hypomelaena, after an evaluation of a larger series of specimens (Zimmer 1931).This arrangement was followed by Peters (1951) and maintained by Zimmer and Isler (2003), who suggested that hypomelaena represents paler plumaged birds on a cline found in the southernmost portion of the distribution area of the species (Zimmer and Isler 2003).Dickinson and Christidis (2014) referred to Zimmer and Isler (2003) to justify synonymizing hypomelaena with serva, a recommendation in accordance with current authorities (Gill et al. 2023).None of these authors considered C. tyranina (sic) atrogularis, Lletget, 1918, a neglected synonym of C. s. serva (Lletget 1918).Because these taxonomic and nomenclatural histories have not been properly evaluated in modern times, and since there is only incipient genetic information that cannot properly address the taxonomy of multiple lineages (Tello et al. 2014), we conducted a taxonomic review of the Cercomacroides serva complex.We based our evaluations on morphological and vocal characteristics, considering all available names, and analyzing specimens from the entire range of its taxa.

Morphometrics
Morphometric characters were measured to the nearest 0.05 mm using a digital caliper and a 1 mm precision ruler, and included (i) culmen length (BNdist, from the distal edge of the nostril to the tip of the bill), (ii) bill height (Bd, at the distal edge of the nostril), (iii) bill width (BWd, at the distal edge of the nostril), (iv) tail length (T1, from between the innermost rectrices to the tip of the longest feather), (v) tail graduation (Tgrad, from the tip of the longest to the shortest feather), (vi) tarsus length (Tar1, from the joint between the tarsus and toes to the intertarsal joint), and (vii) wing chord (Wchord, distance from the carpal joint to the tip of the wing).All birds were measured by one person and on the left side of specimens in a standardized manner (Eck et al. 2011).

Vocalizations
A total of 145 recordings were analyzed (Appendix 2).Notes were considered as continuous lines separated by silence on a spectrogram, while phrases comprised one or a group of notes (Isler et al. 1998).Vocal characteristics included (i) highest frequency, (ii) lowest frequency, (iii) maximum frequency, (iv) bandwidth, (v) song duration, (vi) pace 1 (pace of the first half of a phrase), (vii) pace 2 (pace of the second half of a phrase), (viii) change in pace (pace 1/pace 2, indicating whether loudsongs accelerate or deaccelerate), (ix) peak time, and (x) number of notes.Highest frequency is the upper frequency bound of the selection, and the maximum frequency is defined as the frequency at which maximum power occurs within the selection.Loudsongs were analyzed with Raven 1.4 (Charif et al. 2010), and spectrograms were created with the warbleR package (Araya-Salas and Smith-Vidaurre 2017) with window length = 600 and overlap = 90.
We considered diagnostic morphological and vocal features as an indication of full species rank, following the Phylogenetic Species Concept (Cracraft 1983).Since vocal learning is absent from most Thamnophilidae, their vocalization is said to be innate and evidence for species recognition (Isler et al. 1998).Thus, differences in songs are likely evidence of reproductive isolation, the criterion for species' recognition under the Biological Species Concept (Mayr 2000).

Analyses
Morphometrics and vocal characteristics were compared between groups with Multivariate analysis of variance (MANOVA), and differences within groups were compared with Analysis of variance (ANOVA).Principal component analysis (PCA) was used to reduce the dimensionality of several characters into two axes, eliminating correlated variables.Potential discrimination between populations according to morphometrics and vocalizations was tested with a Linear discriminant analysis (LDA), which was performed by using the same criteria as for the PCA.All analyses were conducted within the R environment (R Core Team 2022).
New specimens (n = 16) from Rondônia, Brazil, were collected under license #25589-6, issued by Instituto Chico Mendes de Conservação da Biodiversidade (ICM-Bio), which was also approved by an Ethics Committee at the Museu de Zoologia da Universidade de São Paulo (MZUSP), and were deposited in the same institution.Localities were mapped following ornithological gazetteers (Stephens and Traylor 1983, Paynter and Traylor 1991, Paynter Jr. 1993, Paynter Jr. 1997).Lists of synonymies were prepared according to Dubois (2000).

Plumage coloration
Two geographically cohesive, geographically adjacent populations were diagnosed based on female plumage color, one from northwestern Amazonia and the other from southern Amazonia.Male specimens of the northwestern Amazonian population were entirely black (black N 2.5/) with sickle-shaped white fimbriae on the outer wing coverts.These showed no white coloration on the inner wing coverts, and had white interscapular patches and shoulders.The females had an orange (7.5YR 5/8) belly, eyebrows and auriculars, while the forehead was concolorous with the crown (olive brown 2.5Y 4/3).Their upperparts were grayish brown (very dark gray 10YR 3/1), tails were dark brown (very dark brown 10YR 2/2), and primaries are yellowish-brown (dark yellowish brown 10YR 3/6).
Male specimens of the southern Amazonian population were virtually identical to those of the northwestern population.For this reason, the holotype of C. tyranina (sic) atrogularis, a male, could only be morphologically evaluated based on its geographic range (see Discussion).Females of the southern Amazonian population, however, had brown backs (dark yellowish brown 10YR 3/6), brown tails (dark brown 7.5Y3/4) and reddish-brown primaries (dark reddish brown 5YR 3/4).The orange (7.5YR 5/8) of the eyebrows, auriculars and bellies was quite similar to that of females of the northwestern population (Fig. 1).The plumage coloration of the specimens from a significant part of the species' ranges do not show intergradation (Fig. 1C, D).

Morphometrics
Morphometrically, females did not differ significantly between populations (MANOVA, F = 1.1, p = 0.388, d.f.= 1), but northwestern males had a slightly, but significantly (MANOVA, F = 4.0, p = 0.001, d.f.= 1), longer tails and tarsus than southern males (Table 1).Seven principal components accumulated 100% of the total variance of quantitative traits (Table S1).Eigenvalues and PC loadings are given in Table S2.The PCA demonstrated a wide overlap in the distribution of morphometric measurements of males and females from northwestern and southwestern populations, including 95% confidence interval ellipses, meaning that differences between those populations are not significant (Fig. 2).The linear combination of predictor variables that were used to form the decision rule of the LDA model (accuracy = 0.77) for males was 0.56*culmen -0.17*height + 0.01*width

Vocalizations
Visual and metric evaluations allowed us to distinguish between two types of male loudsongs (Fig. 3), distinct from the females given their higher frequencies.Of the 145 recordings, there were five (8%) recordings in which female loudsongs could be analyzed without overlapping male loudsongs.Therefore, we only analyzed male loudsongs in our analyses.
Loudsong 1 (N = 55).Characterized by the presence of downslurred-upslurred notes.The number of notes was 6.5 ± 1.8, with a mean duration of 2.1 ± 0.5 s.The pace changed from the middle of the loudsong onwards, meaning that the loudsong accelerates.The respective females emit lower-pitched notes of similar shape, but with 7.4 ± 1.4 notes and a duration of 2.1 ± 0.3 s (Table 2).
Loudsong 2 (N = 90) did not display downslurred-upslurred notes, as they were less angled and had a concave lower boundary (corresponding to the minimum frequency).There were 6.3 ± 1.6 notes lasting for 2.1 ± 0.5 sec.The pace changed from the middle of the loudsong onwards, meaning that the loudsong accelerates.The song of the respective females was similar to that of loudsong 1, but consisted of 6.0 ± 0.4 notes, lasting for 1.6 ± 0.2 s (Table 2).
Loudsong 1 was restricted to the right (south) bank of the Putumayo River, on the border between Colombia and Ecuador, with a southern limit on the left (north) bank of the Marañon River, in northern Peru.Loudsong 2 was distributed on the right bank of the Marañon-Solimões-Amazonas River systems, with an eastern limit of distribution on the left bank of the Madeira River, in Brazil.
The values for the spectrogram measurements correspond to spontaneous vocalizations (Table 2).Comparisons between different loudsongs indicated significant differences between northwestern and southwestern male loudsongs (MANOVA, F = 13.9, p = 0.000, d.f.= 5).All measurements showed overlap between populations, except for the lower frequencies, which were significantly higher in loudsong 2 than in loudsong 1 (Fig. S1).Since highest frequency, pace 2 and change in pace were highly correlated with other vocal features, seven parameters were used for the PCA.Thus, seven functions accumulated 100% of the total variance (Table S1).Eigenvalues and PC loadings are included in Table S3.The PCA also distin-  S3.Pink squares represent centroids, and the 95% confidence intervals are shown by the ellipses.
guished differences in vocal measurements between populations, as seen from the 95% ellipses confidence intervals, indicating the categories of the supplementary variable are significantly different from each other (Fig. 4).The linear combination of predictor variables that were used to form the decision rule of the LDA model (accuracy = 0.67) for males was 0.18*maximum frequency + 0.59*lowest frequency -0.09*peak time -0.01*bandwidth −1.34*number of notes + 1.50*song duration + 0.07*pace.

Discussion
Mitochondrial DNA distances slightly vary (0-0.3%)among individuals within the same population in the Thamnophilidae (e.g., Fernandes et al. 2014), but preliminary genetic data only indicated the p-dist divergence value based on two individuals from Ecuador and Bolivia, which is insufficient evidence of divergence at species level between Cercomacroides serva taxa (Tello et al. 2014).Thus, further evidence from population genetic analyses is needed.However, morphological and vocal characteristics diagnose two lineages within the Cercomacroides serva species complex, one in northwestern (C.serva) and another in southwestern (C.hypomelaena comb.nov.)Amazonia.The latter, as suggested by loudsong recordings, may also occur on the left bank of the Mamoré River, in Bolivia.Plumage coloration was informative for females only, which was expected, due to the genus inherent heterogynism, when specific identification relies on female plumage characters (Hellmayr 1929).
Male loudsongs, on the other hand, were informative in diagnosing species.Based on differences in plumage coloration, loudsongs, and lack of evidence for intergradation, we suggest that two species are best treated independently.Cercomacroides serva hypomelaena (Sclater, 1890) is morphologically and vocally distinct from C. serva serva (Sclater, 1858), and the two species show no evidence of intergradation.We interpret this as evidence for full species rank.Although males are virtually identical, females can be distinguished by plumage coloration, and male loudsongs are distinguished based on the shape of their notes (Table 3).The combination of these characters is constant, and defines two lineages with cohesive geographical distribution (Fig. 5).Syntypes.BMNH 1889.9.20.449(male) and 1889.9.20.450 (female).Napo, Eastern Ecuador.
Diagnosis.Males are entirely black, except for the white interscapular patch and fimbriae on the outer wing coverts.
Females with grayish brown upperparts, which are concolorous with the eyebrows, a white interscapular patch, and a dark brown tail, and yellowish-brown primaries, an orange belly, concolorous with the auriculars and fimbriae on the outer wing coverts.The loudsong is Type 1.

Distribution.
It is distributed north of the Marañon River in Peru, north to eastern Ecuador and southern Colombia, with the Andes serving as its westernmost range.
Diagnosis.Males indistinct from C. serva.Females with brown upperparts, which are concolorous with the eyebrows, a white interscapular patch, brown tail, reddish brown primaries, and an orange belly, concolorous with the auriculars and fimbriae on the outer wing coverts.The loudsong is Type 2.
Distribution.It can be found on the southern bank of the Marañon River and on both banks of the Ucayali River, in Peru.Its westernmost limit is the base of the Andes.It also ranges south of the Amazon River east to the Madeira River, in Brazil, and its southernmost range is in northwestern Bolivia.
Although the Ucayali is an important barrier to several taxa, the upper and middle sections of the Ucayali do not segregate the C. serva and C. hypomelaena comb.nov.This was already noted in a comprehensive biogeographic study, which documented only hypomelaena on both banks of that river (Harvey et al. 2014).Cercomacroides hypomelaena comb.nov.occurs on both banks of the Yavarí (Javari) River, being limited to the north by the Marañon-Amazonas River complex.We found no indication of clinal variations of female plumage coloration across the ranges of C. serva and C. hypomelaena comb.nov.(Fig. 1C, D) in contrast to what has been suggested (Zimmer and Isler 2003).The coloration of the upperparts was constant in females of each species, and from Colombia, south to Peru and Brazil, the female upperparts did not gradually fade into the lighter southwestern form (Zimmer and Isler 2003).This is especially valid when considering specimens that are separated by approximately 100 km in northwestern Peru.Female MUSM 10208 from Rio Cenepa is perfectly distinguished from female AMNH 240187 from Santa Rosa.These populations are probably not in contact, since the Marañon River apparently acts as a barrier, even in its narrower upper region (Fig. 5B).It has been shown that the width of a river near the headwater may not act as a barrier (Harvey et al. 2014), and the fact that those females are clearly distinguishable corroborates their specific status.Furthermore, the recordings from the northern (XC335224) and southern (XC89140) banks of the Amazon River on the borders of Colombia and Brazil, which are ~30 km apart (Fig. 5C), further corroborate the absence of intergradation.Via spectrogram analyses, both male loudsongs can be diagnosed as C. serva and C. hypomelaena comb.nov., respectively, indicating that the river is a geographic barrier.From these few contact areas where we could assess the identify of Cercomacroides species, gene flow may be reduced or absent, reaffirming their specific status under the Biological Species Concept (Mayr 2000).
Our study indicates that Cercomacra tyranina (sic) atrogularis Lletget, 1918 is not a valid taxon.The holotype, a male from Archidona, Ecuador, was collected by Marcos Jiménez de la Espada during his visits from Guayaquil to Tabatinga between 1862-1866 (J.Barreiro in litt.2013).We evaluated this specimen via photographs, but due to the similarity of male plumages of C. serva and C. hypomelaena comb.nov.we could not distinguish this specimen from C. serva on plumage alone.Lletget's type specimen originated from within the distribution of C. serva.Thus, we suggest that, pending molecular analysis, atrogularis is best synonymized with C. serva (Sclater, 1858).
The type locality of C. serva, Napo, is imprecise, but the banks of this river do not act as barriers to other species of Thamnophilidae (e.g., Cavarzere et al. 2012).Records from Ecuador merit a few comments.The presence of the species in the Ecuadorian Chocó is based on only two specimens (MNHN 1936n117, 1936n118) collected by Carlos Olalla and sons, and this information is neglected without explanation in some references (Zimmer and Isler 2003).There is a great deal of discussion about some of the specimens collected by the Olalla family and kept in the American Museum of Natural History (AMNH), as well as in other museums (Wiley 2010).Few errors in locations can be attributed to the Olalla family, especially Alfonso, who contributed thousands of bird and mammal specimens, which are currently kept in dozens of institutions.His work significantly shaped the field of zoological studies of neotropical fauna (Wiley 2010).It is possible that these two specimens (as well as two other specimens of C. cinerascens, from the same locations and from similar dates) might be the result of a location error.This is partly because the species is restricted to the Amazon Basin, but also because among all the specimens examined in this study, only these two came from this location.For C. Vaurie, the former curator of ornithology at the AMNH, some specimens sold by Olalla exclusively to the Natural History Museum in Stockholm, then curated by N. Gyldenstolpe, were also traded with other buyers, which had incorrect information on their labels (Wiley 2010).Some of these specimens were found in the Muséum National d'Histoire Naturelle, Paris, and may include the examples cited here.P. E. Vanzolini, then curator of herpetology at the Museu de Zoologia da Universidade de São Paulo (MZUSP), was confident of the Olalla locations, and in a letter to Vaurie, dated 1965, explained that Olalla had more employees that collected on opposite riverbanks, at least for the Amazonian cases.This would explain why the collection includes taxa that inhabit different localities collected on the same day (Wiley 2010).
The altitude of Carondelet, in the province of Esmeraldas, Ecuador, was questioned, since the amphibian species collected there normally occur at much higher elevations and further west than where this location was said to be, according to the gazetteers (Hoogmoed 1989, Paynter Jr. 1993).Hoogmoed (1989: 15) refrained from contradicting the locality itself, because Olalla specimens are generally reliable.These two specimens were collected by Manuel Olalla in February 1952, today held at MZUSP.For bats of the genus Sturnira, which inhabit western Ecuador and Colombia, nothing unusual has been reported for rio Cachabi, the location for S. ludovici, where a male of this species was collected by Carlos Olalla on 9 August 1935 (McCarthy et al. 2006).This date coincides with two specimens of C. serva (1 and 6 August 1935) and another two C. cinerascens (27 July and 5 August 1935).
There is a record of a young male C. hypomelaena comb.nov.from the right bank of the Madeira River (MZUSP 109098) collected on 8 November 2010.At the time of this collection in Porto Velho, Rondônia state, Brazil, all collected specimens were processed on the same day by a person accustomed to specimen tagging.This strongly suggests that there is no location error (E.Machado, pers. comm. 2015).There were no other records available for this species on the right bank of the Madeira River.Specimens and recordings from the left (southern) bank of the Marañon River near its headwater, and from the right (eastern) bank of the Ucayali River are warranted to further elucidate if those rivers impose barriers or whether the two taxa might form a hybridization zone.

Figure 2 .
Figure 2. Principal component analysis of six morphometric variables of males (A) and females (B) of the Cercomacroides serva complex, depicting the northwestern (black) and southwestern Amazonian (orange) populations.Cumulative proportion of the total variance explained by function 1 (culmen length, bill height and tarsus length) and function 2 (bill width, tail length and graduation and wing chord) are in TableS2.Pink squares represent centroids, and the 95% confidence intervals are shown by the ellipses.

Figure 5 .
Figure 5. (A) Distribution of specimens (triangles), and recordings (circles) of Cercomacroides serva (black), and Cercomacroides hypomelaena comb.nov.(orange), examined in this study.A diamond in SE Peru represents the type-locality of C. hypomelaena comb.nov.; there is no precise type locality for C. serva (see text).Insets indicate potential contact areas in which species are separated by the Marañon (B) or the Amazon (C) River.Range map (light grey) from IUCN (2023).

sefina
Barreiro kindly sent precious photographs and information about Lletget's holotype in Museo Nacional de Ciencias Naturales (Madrid, Spain).A special thanks to all museum curators and staff, as well as museum curators who loaned hundreds of specimens within the United States.We especially thank naturalists and collectors, and ornithologists and birders whose Cercomacroides recordings were made available through online repositories.The Macaulay Library of Natural Sounds kindly provided all recordings of Cercomacroides.Earlier drafts of this text benefitted from several suggestions and critiques by two reviewers and the editor, who considerably improved the quality of the final version of this manuscript.We would like to thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, #2010/11798-5) for funding.We are grateful to the American Museum of Natural History for a Collection Study Grant to VC, and to Idea Wild for financial support.Universidade Tecnológica Federal do Paraná assisted with language editing services.

Table 1 .
Descriptive statistics (mean, standard deviation [SD], sample size [N], and minimum and maximum range) of the morphometrics of the Cercomacroides serva complex.* p ≤ 0.010 (MANOVA) for males.

Table 3 .
A combination of qualitative characters which diagnoses Cercomacroides serva and C. hypomelaena comb.nov.Plumage colors refer to female specimens.