Proposal (1054) to South American Classification Committee

Revise species limits in Rhynchocyclus olivaceus: A. Treat R. aequinoctialis as a separate species; B. Treat R. guianensis as a separate species; C. Treat R. cryptus as a separate species.

Effect on SACC: This would split widespread Rhynchocyclus olivaceus into as many as four species, including newly described Rhynchocyclus cryptus.

Background: Our current Note is as follows:

73a. Boesman (2016) proposed that Rhynchocyclus olivaceus be treated as two species based on vocal differences, and this was followed by del Hoyo & Collar (2016). Simões et al. (2022) found that Rhynchocyclus olivaceus as currently defined is paraphyletic with respect to R. fulvipectus, and also proposed that Amazonian populations of R. olivaceus consists of four species: R. olivaceus of the Atlantic Forest, R. guianensis of the Guianas and eastern Amazonia, R. aequinoctialis of e. Panama, northwestern Colombia, and northwestern Amazonia, and newly described R. cryptus of southwestern Amazonia. SACC proposal badly needed.

Birds of the World/Clements has instituted a 2-way split: https://birdsoftheworld.org/bow/species/olifla3/cur/introduction?login#sys

New information: Simões et al. (2022) densely sampled R. olivaceus and found some fascinating remarkable results, including strong vocal differences among several of the groups and sympatry between two of the main vocal groups. This is a classic example of speciation that would be difficult to detect without genetic and bioacoustic techniques, with parallels to the amazing case of Turdus sanchezorum, as well as specimen labels that include habitat type.

Genetics: Simões et al. (2022) sampled 83 specimens from five of the recognized subspecies of R. olivaceus (bardus of e. Panama, nw. Colombia, aequinoctialis of w. Amazonia, guianensis of the Guianan Shield, sordidus of e. Amazonian Brazil S of Amazon, and nominate olivaceus of the Atlantic forest region of Brazil. They did not have samples of four others, all from n. Colombia and n. Venezuela): jelambianus (ne. Venezuela), tamborensis (nw. Santander), flavus (n. Colombia and nw. Venezuela), and mirus (Atrato Valley). They sequenced 3 mitochondrial and 2 nuclear genes.

Their tree (below) showed that R. olivaceus separates into two broad clades, one of which is sister to R. fulvipectus. One clade consists of aequinoctialis and bardus, but aequinoctialis is not monophyletic within that clade. The other clade consists of nominate olivaceus, guianensis, sordidus, and a different set of aequinoctialis populations. (You will need good resolution of this graphic to understand what follows, so if this screen shot doesn’t accomplish that, then refer to the original paper – see link in Literature Cited.)

A preliminary point to note is that, as noted by Simões et al. (2002), their tree conflicts with the results of Harvey et al.’s (2020) genomic analysis, which showed that pacificus is not sister to all the others but sister to brevirostris, as expected on phenotypic grounds because these two were considered conspecific until recently. This raises a warning for me in terms of interpretations of the topology of the Simões et al. tree because of differences in genetic sampling (DNA sequence data for 5 markers vs. genomic data with 2K+ genomic regions). With a focus on an entirely different taxonomic level, Harvey et al. (2020) had only one sample of R. olivaceus (which was sister to R. fulvipectus), and so no other conflicts can be detected. On the other hand, also note that with the exception of aequinoctialis, which appears in at least 5 clades, the DNA sequence data are otherwise generally consistent with subspecies boundaries.

Morphology: They examined 113 study skins from four of the five subspecies from which they had tissue samples (missing bardus). They used Smithe (1975) to characterize color in 8 plumage regions. They took standard morphometric measurements from the 52 specimens for which sequence data were obtained.

They were unable to find any diagnostic plumage characters for any of the five subspecies analyzed, although guianensis were generally brighter yellow below than the other clades. According to Cory & Hellmayr (1927), the subspecies can be separated by minor differences, and although not mentioned directly by Simões et al., their sampling covered the regions used by Cory & Hellmayr in describing the differences among the taxa. (I wish that this had been discussed in more detail with respect to each subspecies, if only in supplemental material.)

In terms of morphometrics, none of the lineages was diagnosable, but nominate olivaceus stood out in PCA spaced as being larger overall than any of the others.

Vocalizations: They examined 84 song recordings, distributed among all the subspecies except mirus (including recordings by Dan, Mark, and Gary among current SACC voters as well as former voting members Bret Whitney, Fernando Pacheco: see SI https://onlinelibrary.wiley.com/action/downloadSupplement?doi=10.1111%2Fzsc.12519&file=zsc12519-sup-0006-Supinfo.pdf). Because of concerns for homology in vocalizations, analyses were restricted to what were interpreted as loudsongs, although a few distinctive call types were also mentioned. They wrote:

“Unlike many other suboscine passerines, Rhynchocyclus are usually quiet most of the time and representativeness of sound recordings in databases is an important bottle- neck for robust quantitative vocal analyses. Nonetheless, we evaluated at least one type of song of all species of Rhynchocyclus and most of described subspecies and clades recovered in R. olivaceus, except R. o. mirus and clade E birds, which were unavailable

They found distinct, diagnosable differences for the currently recognized species R. pacificus, R. brevirostris, and R. fulvipectus. It is great to see these differences quantified and described in terms of those quantitative data.

Within broadly defined R. olivaceus, their general results are as follows:

Within the remaining clades (D–N) and taxa making part of the polytypic and paraphyletic R. olivaceus, two highly variable loudsong and call types could be identified, each corresponding to reciprocally monophyletic groups. One of these vocal groups includes clades D, F, G, H, I, J and K, whereas the other clusters clades L–N, plus subspecies R. o. tamborensis and R. o. flavus, which were not sampled genetically (Figures 3 and 4).

In other words, referring to the figure above, the two main genetic lineages shown in the tree had notable differences in songs and calls, which they referred to as Vocal Groups 1 and 2. These correspond to the two main groups found by Boesman (2016). Below is a sonogram of the loudsong of guianensis:

And below is the sonogram for nominate olivaceus. Note both of these are in the same general lineage (D-L):

And below is the loudsong of clade N (aequinoctialis group, but later revealed to be the new species, cryptus):

See Simões et al. for detailed descriptions of the songs of all the clades – the details are too lengthy to repeat here. A PCA analysis of the measurements from the sonograms produced the following result:

“Four vocal measurements presented significant statistical differences across clades and taxa (PF, PF1, PF2 and NN), but with no single group being unequivocally diagnosed by any of the characters measured, given overlap in measurements between two or more groups (Table S10). Nevertheless, a PCA analysis retaining 81% of the total variance in two first principal components (PC1: 53% and PC2: 28%), recovered along PC2 the two main vocal groups reported above, one including clades D, H, I, J and K (hereafter called vocal group 1—VG1) and an- other containing clades L, M and N, plus the samples attributed to R. o. tamborensis and R. o. flavus (hereafter called vocal group 2—VG2; Figure S4).”

Here their Figure S4:

The most unexpected finding in this project was that representatives of two of the vocal groups are locally sympatric:

“Also, our results demonstrated that clade N (belonging to VG2) overlaps broadly in distribution with distantly related clades F and G (both belonging to VG1) in different parts of south-western Amazonia (Figure 1), clearly indicating that these lineages may have already acquired a significant level of reproductive isolation consistent with their paraphyletic status, and current sympatry. This sympatry in south-western Amazonia between different clades of R. o. aequinoctialis is apparently mediated by occupancy of distinct habitat types. Out of the 25 specimens sequenced attributed to R. o. aequinoctialis and belonging to clades F, G and N, a clear pattern emerged: specimens from closely related clades F + G (N = 9) were collected mostly in upland terra-firme forest (N = 5), transitional forest (N = 2) and bamboo patches (n = 2), whereas those in distantly related clade N (N = 16) were obtained from várzea seasonally flooded forest (N = 12), riparian forest (N = 2) and secondary forest (N = 1) (Table S2). Our sampling revealed that sequenced specimens of distantly related clades G and N are found within a few kilometres from each other in suitable habitats in the Rio Branco capital area in the Brazilian state of Acre, with birds in clade G associated with bamboo patches, and those in clade N with seasonally flooded forests with varying levels of disturbance along the banks of the Rio Acre (Figure 1; Table S2). We anticipate that a similar pattern of local ecological replacement across upland and seasonally flooded habitats may also occur between birds in clades F and N in the westernmost part of the Amazon (Figure 1; Tables S2 and S3). These genetically, vocally and ecologically divergent lineages of the Olivaceous Flatbill lineages overlapping in south-western Amazonia are apparently undistinguishable from a morphological perspective, when both plumage and morphometric variation are considered (Table S8, Figures S2 and S3), highlighting another remarkable case of cryptic diversification in the Neotropics.”

”

Simões et al. then interpreted all this in a taxonomic framework. First, they reasoned that the two major lineages should be considered separate species, minimally, with the oldest name for lineages D through K being olivaceus, and lineages L-N, aequinoctialis (from Traylor “Peters” 1979: Rhynchocyclus olivaceus aequinoctialis (Sclater) from Cyclorhynchus aequinoctialis Sclater, 1858, Proc. Zool. Soc. London, 26, p. 70, type loc. Rio Napo, Ecuador. Then they proposed recognizing an additional two species:

“However, our data and analyses also clearly show that within these two major R. olivaceus groups, there is still significant molecular and diagnostic vocal variation allowing for the delimitation of additional species occupying distinct biomes and river drainages across the Neotropics (Figures 1–3; Figure S4; Tables S5 and S11). Namely, vocal and genetic patterns of variation allow for the split of the major R. olivaceus and R. aequinoctialis groups into at least two additional species level taxa each.”

Based on that variation, they recommended the following treatment:

1. R. olivaceus for clades H, I, J, K (Atlantic Forest region from Pernambuco to Rio; also allopatrically in se. Amazonian Brazil E of the R. Tapajós. Includes sordidus.

2. R. guianensis for clades D, E, F, G (Guianan Shield, N of the Amazon in Peru and Ecuador; S of the Amazon from R. Jutai in Amazonas, Brazil west to Loreto, San Martín, and Pasco, Peru; south of the Amazon from the R. Tapajós west to the R. Purús in Pará, Amazonas, and Acre, Brazil. Note that this includes several populations traditionally assigned to aequinoctialis.

3. R. aequinoctialis for clades L and M only (e. Panama and n. Colombia; nw. Amazonia in ne. Ecuador; presumably also in e. Colombia. Includes bardus and, provisionally, the subspecies mirus, tamborensis, flavus, and jelambianus.

4. R. cryptus sp. nov. for clade N

The latter is formally described as a new species in the paper, which see for all the details, including registry in ZooBank and deposition of the type specimen, from Acre, at Museu Goeldi. The new species is a varzea specialist that like some other varzea species occurs also in human-degraded habitat. It is widespread in the Inambari area of endemism (south of Amazon from e. Peru east to the Madeira in Brazil, and south to northern Bolivia); where sympatric with guianensis, occurs in varzea forest whereas guianensis is in adjacent terra firme forest.

I attempted to map these four distributions using the localities mapped for each clade in the map above on the map below. Apologies in advance to Simões et al. for any mistakes – corrections welcomed. Note that the situation in e. Colombia, the R. Negro region of Brazil, and Amazonian Venezuela is unclear to me.

Discussion

I spent more time on this proposal than any in 25 years of writing proposals for SACC. This is a complex situation with surprising and perhaps anomalous results. Simões et al. did a very good job of putting this all together, including helpful Supplemental Information, and a huge amount of work went into this project.

That there are more than one species hidden in a widespread Neotropical tyrannid is of course no surprise. Nonetheless, I have some concerns. First, the biogeographic pattern is unique, as far as I can tell, especially for the four disjunct populations grouped under aequinoctialis. Some of that may be due to incomplete sampling. Second, the taxa are essentially undiagnosable in terms of plumage and morphology. Third, although the vocalizations look to be diagnostic, the samples of loudsongs are small and the PCA analysis seems to show that they overlap or nearly so. Fourth, given that the species tree typology conflicts with the Harvey et al. topology that has much better genetic sampling, how then are we supposed to trust the finding that R. rufipectus is embedded to the olivaceus clade? If the latter were correct, then of course division of traditional olivaceus into at least two species would be required.

For voting purposes let’s break this down as follows, with YES/NO votes on the following

A. Treat R. aequinoctialis as circumscribed by Simões et al. a separate species. Even if the position of R. rufipectus is questionable, a case can be made for the split based on voice.

B. Treat R. guianensis as a separate species.

C. Treat R. cryptus as a separate species. This would be required by the sympatry of cryptus and guianensis in sw. Amazonia, but whether cryptus should be treated as a separate species from R. aequinoctialis needs to be considered.

I do not have any firm recommendations. It seems to me that Simões et al. have discovered a fascinating system, and I look forward to comments by others.

Literature Cited:

BOESMAN, P. 2016p . Notes on the vocalizations of Olivaceous Flatbill (Rhynchocyclus olivaceus). HBW Alive Ornithological Note 120. In: Handbook of the Birds of the World Alive. Lynx Edicions, Barcelona. https://doi.org/10.2173/bow-on.100120

SIMÕES, C. C., P. V. CERQUEIRA, P. PELOSO, AND A. ALEIXO. 2022. Integrative taxonomy of flatbill flycatchers (Tyrannidae) reveals a new species from the Amazonian lowlands. Zoologica Scripta 51: 41-57.

Note on English names: BOW uses Western Olivaceus Flatbill for aequinoctialis and Eastern Olivaceous Flatbill for olivaceus. Simões et al. retained parental Olivaceus Flatbill for olivaceus, and used Guianan Flatbill for guianensis, Equinoctial Flatbill for aequinoctialis, and Cryptic Flatbill for cryptus. We will have to do a separate SACC proposal on English names depending on which taxonomy, if any, we adopt.

Van Remsen, May 2025

Voting Chart: https://www.museum.lsu.edu/~Remsen/SACCPropChart1044+.htm

Comments from Areta: “This was my AviList/WGAC vote:

‘These papers with a mixture of scientific names and letters for clades are so exasperating that I want to commit arson.

‘I don´t think that the 2-species arrangement is the best possible one, but I also see that there is room for nuanced taxonomic variants here.

‘The bioacustic analyses are far from convincing. However, in examining the spectrograms visually, I must say that clades DEFG and HIJK seem to differ markedly in vocalizations (the former having a descending series of overslurred notes, the latter a series of upswept notes). This suggests that R. guianensis may also be split from R. olivaceus. Their PCA provides some support for this, although not overwhelming. Note also that the Amazon and Tapajos rivers separate the guianensis from the olivaceus groups. The deep split between clades G (of guianensis) and adjacent K (from olivaceus) contrasts markedly with the shallow differentiation between the widely allopatric populations of the Atlantic forest (clade H) and the south-east Amazonian populations (IJK).

‘On the other hand, the species-level status of cryptus (clade N) seems more debatable given the vocal similarities to the relatively deeply split from aequinoctialis (clades LM) and is to me a borderline case that needs more refinement, but which may be correct.

‘I vote for the following: A) Split R. aequinoctialis (with cryptus as a subspecies, unless someone finds a compelling reason to also split cryptus), and B) Split R. guianensis.

‘Check the bioacoustic PCA and genetic distances between clades in the following suppl mat:

zsc12519-sup-0004-figs4.docx
zsc12519-sup-0006-supinfo.pdf

"In terms of SACC voting this would be:

A) YES, recognize R. aequinoctialis (clades LMN), including cryptus as a subspecies

B) YES, recognize R. guianensis (clades DEFG; descending series of overslurred notes)

C) NO, include cryptus (clade N) as a subspecies of aequinoctialis (but if someone provides good arguments to recognize it, I am willing to reconsider)

D) This would mean that I also support a newly circumscribed R. olivaceus (clades HIJK; series of upswept notes)”

Comments from Stiles: “YES to all splits, but I agree with Nacho that on present evidence, cryptus may best be considered a subspecies of aequatorialis.”

Comments from Robbins: “After spending nearly two hours over a two day period reviewing data associated with the Simões et al. (2022) publication, I fully understand why Van stated that he spent more time on this proposal than any other (!) AND why he did not make a recommendation. I also share Nacho’s frustration with aligning and digesting data sets based on a letter code for each clade. Moreover, if you also do a deep dive into the vocalizations, i.e., going to Xeno-canto and Macaulay Library, you almost certainly will reiterate Van’s remarks.

“It is very impressive the amount of work that Simoes et al. put into elucidating this very cryptic complex that is not dissimilar to Tolmomyias. Major kudos to them for drastically increasing our understanding of this group. Clearly, there is more than one species in the currently recognized olivaceus complex; however, as with any study that involves such a difficult group there are a few issues that still need to be addressed. Given that plumage morphology does not offer insight to species limits within this complex, my comments are focused on genetics and vocalizations.

“1. Genetics: As Van pointed out, it is of concern that the results of the Harvey et al. (2020) contradict that of Simões et al. with the relationship between pacificus and brevirostris. But because the Harvey et al. study only included one sample of olivaceus (nominate from southeastern Brazil) nothing more can be concluded comparing the two data sets. Despite the relatively large uncorrected distances in the mitochondrial genes among taxa, a whole genomics perspective is needed including sequencing the holotype of aequatorialis (see below).

“2. Vocalizations. Naturally, since the publication of Simões et al. (2022) there are more vocalizations available online. Without going into too much detail, please note that in Figure 4 the example of clade N, i.e., what they are calling the new species, cryptus, is ML 75277 (also see a complete list of vocal samples in their Supplementary material). That recording and others allocated to clade N are striking similar to a more recent recording (https://ebird.org/checklist/S197298853) from Reserva Munay Suyu, at the base of the Andes in Prov. Napo, Ecuador, i.e., that locality appears to be in terra firme.

“So, this raises a couple of questions. If the Napo bird isn’t an example of cryptus then the loudsong of aequatorialis is quite similar to cryptus. If the Napo recording is of cryptus and given the holotype of aequatorialis (sensu stricto) is from the Upper Río Napo, Prov. Napo, Ecuador, and Simões et al. apparently didn’t sample the holotype (see Supplemental material), then the question needs to be asked, could the new taxon actually be aequatorialis? That would result in the taxon flavus (it has priority) to be applied to cis- and trans-Andean populations that Simões et al. consider subspecies under aequatorialis. Regardless, the holotype of aequatorialis needs to be sequenced to clarify what name should be applied to what group.

“The above along with the truly unusual biogeographic pattern of proposed species limits by Simões et al. makes me question just how many species are involved. Again, I have no doubt that more than one species is involved, but because of what I have highlighted above, at this point, I can only support part B, elevating guianensis (there is no issue with that name being applied to at least birds north of the Amazon) to species level because of the unique loudsong. Finally, if aequatorialis and cryptus are considered the same species, I would also vote for recognition of a third species while waiting for information that clarifies what I have outlined above.”

Comments from Bonaccorso: “NO to all. Definitively a fascinating system, but very complex. I am not sure we have the necessary data to make a decision now.”