Proposal (922) to South American Classification Committee

Treat Turdus daguae as a separate species from Turdus assimilis

Effect on SACC: This would treat the subspecies daguae, our sole representative of otherwise Middle American Turdus assimilis as a separate species. Note: this is a slightly modified version of a proposal submitted to NACC.

Background: This taxon is currently treated by NACC and SACC as a subspecies of Turdus assimilis, but as a result of anecdotal comments, NACC and Dickinson & Christidis (2014) treated it as a subspecies “group”. Here’s the text from AOU (1998):

“Groups: T. assimilis [White-throated Thrush] and T. daguae Berlepsch, 1897 [Dagua Thrush]. Turdus assimilis and the South American T .albicollis Vieillot, 1818 [White-necked Thrush], constitute a superspecies (Sibley and Monroe 1990). Many authors (e.g., Wetmore 1957, Wetmore et al. 1984, Ripley in Mayr and Paynter 1964) consider them conspecific but see Monroe (1968) and Ridgely and Tudor (1989).”

I included the part about albicollis to explain why some older literature treats daguae under Middle American Turdus albicollis; also, albicollis becomes part of the problem, as you’ll see below.

Here is our SACC Note on this:

12a. The subspecies daguae of the Chocó region has been treated as a subspecies of T. assimilis, and most recent authors have treated them as conspecific (e.g., Hellmayr 1934, Ripley 1964, Meyer de Schauensee 1970, Ridgely & Tudor 1989, AOU 1983, 1998, Sibley & Monroe 1990, Clement 2000, Dickinson & Christidis 2014). Ridgely & Greenfield (2001), however, considered daguae to be a separate species, in part because its voice resembles that of T. albicollis more than that of T. assimilis. Nuñez-Zapata et al. (2016) presented evidence that daguae should be treated as a separate species from Turdus assimilis. Del Hoyo & Collar (2016) treated as a subspecies of T. albicollis, not T. assimilis, based on Boesman’s (2016) assessment of songs. SACC proposal badly needed.

History of taxonomic treatments: Berlepsch described daguae as a species in Turdus in 1897 from lowland western Colombia, with the Río Dagua as part of the type locality, which is south of Buenaventura and NW of Cali (dpto. Cauca). Ridgway (1907) did not mention the taxon, so I assume it was not recorded in Panama until sometime after 1907. However, Hellmayr (1911; PZSL), Bangs & Barbour (1922; Bull. MCZ), and Chapman (1926: Birds of Ecuador) soon treated daguae as a subspecies of Turdus assimilis (at that time known as Turdus tristis).

Hellmayr (1934) treated it as a subspecies of T. assimilis with the following explicit rationale:

“Turdus assimilis daguae Berlepsch: Differs from T. a. cnephosa [the adjacent subspecies in central Panama] in smaller size, shorter bill, much darker (bister brown) upper parts, and very much darker, nearly sepia brown color of the chest, sides, and flanks. ….

“In coloration, this race comes nearest to T. a. rubicundus, but is still much more intensely colored. Although its much smaller dimensions and its shorter, entirely dusky bill serve to distinguish it without difficulty, yet the close similarity to the west Guatemalan form seems to afford sufficient evidence for its association with the assimilis group, which, as suggested by Miller and Griscom, may ultimately prove to be conspecific with albicollis.”

This treatment was followed by essentially all subsequent authors, including Ripley in “Peters” (1964), Meyer de Schauensee (1966), and the AOU (1983), which stated: “The populations of T. assimilis from eastern Panama (eastern Darien) south to Ecuador are sometimes considered a distinct species, T. daguae Berlepsch, 1897 [Dagua Robin].” I’m actually uncertain where this statement “sometimes” came from because I can’t find a treatment from the 1900s on that did treat it as a separate species. [Anecdote: the frustration that hundreds of examples like this in AOU 1983 caused me to start lobbying when I joined the Committee in 1984 for citations for all such statements in future AOU Checklists; Burt Monroe, who wrote almost all those Notes in the 1983 Checklist often could not remember the source of many of the statements].

Subsequently, Ridgely & Tudor (1989) treated daguae as a subspecies of T. assimilis and mentioned only that it was not well known. Clement (2000; Thrushes; Princeton U. Press) listed it as the southern subspecies of the 10 that he recognized for T. assimilis; he described the plumage differences but did not illustrate it separately.

Then, Ridgely & Greenfield with the collaboration of Robbins and Coopmans (2001; The Birds of Ecuador Vol. 1) treated daguae as a separate species (Dagua Thrush) from T. assimilis, with the following text --- note that no actual data are presented:

“Daguae has usually been treated as a subspecies of T. assimilis …. Now that more information is available regarding the voice of daguae – it is distinctly different from that of T. assimilis of Middle America – we consider it more appropriate to treat T. daguae as a separate monotypic species, differing not only in voice but also in several morphological features. Daguae’s voice actually more closely resembles that of cis-Andean T. albicollis, suggesting that daguae may be more closely related to that species.”

Collar in HBW (2005) not only treated them as conspecific but also treated all of assimilis as conspecific with South American T. albicollis. He mentioned that daguae had been proposed as a separate species. Here is the relevant section of the plate:

Dickinson & Christidis (2014) treated daguae as conspecific with T. assimilis but placed daguae in its own subspecies group, as per AOU (1998), and cited Ridgely & Greenfield for the possible split. Angehr and Dean (2010; The Birds of Panama. A Field Guide) treated them as conspecific; their range maps nicely illustrate the substantial gap in their distributions in the lowlands of central Panama

Del Hoyo & Collar (2016) treated daguae as conspecific with …. drumroll … not T. assimilis but with cis-Andean T. albicollis. This was based on Boesman (2016), who actually used the Collar (2005; HBW) species limits, i.e. broadly defined T. albicollis, not the two species treatment in Del Hoyo & Collar (2016). His analysis, however, shows that T. assimilis is clearly different from T. albicollis in song features, and that the song of daguae is actually difficult to distinguish from the eastern group of T. albicollis subspecies. Although Boesman would be the first to tell you that more in-depth analysis is needed, he does establish that if voice is a reliable indicator, then daguae belongs with T. albicollis. Recall also that Ridgely & Greenfield (2001) noted the similarity of daguae song to that of albicollis. That trans-Andean Chocó and cis-Andean populations are sisters is a common biogeographic pattern in Neotropical birds.

Genetic data

As for genetic data, as argued in previous proposals, I think they are of dubious value for determining taxon rank of allopatric populations, although perhaps our best estimates of divergence times.

The only study that specifically addressed the daguae issue was that of Núñez-Zapata et al. (2016), which focused on the T. assimilis group. Using 2 mitochondrial genes (cyt-b, ND2) and 25 individuals, their tree shows that daguae is strongly separated, with strong support, from the other subspecies and populations of assimilis:

Of interest is that daguae is sister to assimilis, not the albicollis group, which provides evidence against inclusion of daguae in T. albicollis as in Del Hoyo and Collar (2016). However, this result could be a “gene tree/species tree” or ILS problem given the limited gene sampling, and so hopefully subsequent studies will include nuclear or genomic data. Meanwhile, this places burden-of-proof, in my opinion on moving daguae to albicollis.

Núñez-Zapata et al. (2016) considered the genetic data as indicating species rank for daguae based on genetic distance and reciprocal monophyly. However, they themselves noted the problems with using comparative genetic distances. Also, with N=6 daguae samples, all from three localities in NW Ecuador near the southern extreme of its range, a claim of reciprocal monophyly seems premature. All such claims are one additional sample away from being refuted. (Would someone please write a paper on the problems with “reciprocal monophyly” with respect to N and geographic sampling; certainly, some minimum number of specimens would seem required to make such a claim, depending on genes sampled, as well as some consideration of the geography of sampling given that the probability of detecting shared alleles should decrease to some degree with distance from the former contact zone.)

Batista et al. (2020) used UCEs in their broad study of Turdus but unfortunately did not include daguae.

Discussion: Voice is the key indicator of species limits in Turdus and relatives, as outlined in my proposal on Turdus [m.] confinis. Between xeno-canto and Macaulay there are a sufficiently large number of recordings from throughout the range of the assimilis-albicollis group that someone could pick up where Boesman left off and do a formal analysis of songs and calls to analyze species limits. Until that is done, I do not see how we can change the status quo. On the one hand, a preliminary inspection of songs suggests daguae is closer to albicollis than to where we have it at present, in contrast to a genetic data set that suggests that it is close to assimilis. The weaknesses in both analyses make it unwise, in my opinion, to change current classification for now, either in terms of taxon rank or relationships.

Recommendation: Too much uncertainty remains, in my view, to make any changes, and so I recommend a NO on this one.

Note on English names: Dagua Thrush has a track record and is the only English name associated with the taxon. Although Río Dagua was an important early collecting locality, it is nonetheless a pretty obscure river in the greater scheme of things. Whether it’s worth changing to something like Choco Thrush is at least worth considering. This taxon’s range corresponds almost perfectly to the Chocó biogeographic region and its plumage, darkest brown of any on the assimilis-albicollis group, also reflects a prevailing Gloger’s Rule color trend shown by taxa endemic to the region. On the other hand, there are already 9 “Choco Somethings”, so at least Dagua Thrush is novel.

Van Remsen, October 2021

Comments from Robbins: “NO. Without further analyses of both genetic and vocal characters it would be premature to treat daguae as a separate species.”

Comments from Lane: “NO. It is frustrating when an otherwise good study makes one big blunder that severely weakens its punchline. In Nuñez-Zapata et al (2916), it was the choice of taxa from Turdus albicollis to include in the study. The samples of albicollis they used were one of nominate from the Atlantic Forest and two of contemptus from La Paz, Bolivia. Lowland Amazonian phaeopygius was not included, and this is the taxon that sounds most like daguae from what I can tell. Furthermore, whereas albicollis and contemptus are very similar in voice and plumage, they differ strongly from phaeopygius (to the point that I'd say it's a no-brainer that the former two should be separated from the latter at the species level... the question is: where to fit the remaining northern South American taxa?) the latter is the population that could well be sister to daguae. The oversight of not including phaeopygius in the study badly damages the conclusion that daguae should be separated as its own species.

Comments from Pacheco: “NO. Until a comparative vocal analysis of assimilis/daguae/albicollis with appropriate geographic coverage is done, I consider it equally premature to alter the status quo.”

Comments from Claramunt: “YES. Evidence of reciprocal monophyly is not strong but combined with clear phenotypic differences in plumage and song, I see a strong case for separating daguae from assimilis. I totally agree with Dan that the big picture may be more complex and Amazonian albicollis may be closer to daguae than to the true albicollis from the south. But I think separating daguae is a step forward. Once we figure out the affinities of Amazonian albicollis we can revisit this problem, but I think we don’t need to wait until we know it all to improve our taxonomy of this complex.”

Comments from Areta: “YES. Nuñez-Zapata et al. (2016) have provided convincing evidence that daguae is not part of albicollis (I applaud that they have sampled nominate albicollis), and even if their dataset is only of mitochondrial DNA it seems highly unlikely that this pattern will be caused through introgression. Certainly more sampling of ssp. of albicollis and a proper analysis of vocalizations can shed more light. But the burden of proof is now on those wanting to treat daguae as a ssp. of albicollis or of assimilis. Vocally, daguae sounds different to any ssp. in the albicollis group (at least to my ear), and the genetic divergence from assimilis (0.051; 1.35 Ma ago [0.95 to 2.08]) seems to place it within expected distances between good Turdus species. Biogeographically, it also makes sense to have this Chocó species being more related to a Central American counterpart. So, although evidence is not perfect, I prefer to treat daguae as a species on its own.

“Some bits of information are still missing, but I think a fairly strong case can be made to recognize Turdus daguae for the following reasons: 1) Reciprocal monophyly despite extensive sampling of assimilis, including geographically distant samples [unfortunately, there are no samples of daguae closer to the range of the southern subspecies of assimilis, but I don´t think this will change the picture given the degree of divergence] and 2) distinctive vocalizations: the song of albicollis is remarkably constant across its geography (at least to my ears) and I have never had trouble in identifying this species from Venezuela through NW and NE Argentina and Brazil, this holds true to the east of the Andes in Ecuador (T. albicollis spodiolaemus), differing remarkably from the higher-pitched, slower and differently patterned song of daguae to the west of the Andes in Ecuador. In this subjective and express analysis, I agree with Boesman in that songs of assimilis are the most distinctive, but daguae seems to me consistently different from albicollis as well; a thing that always bothers me on these HBW vocal assessments is their lack of attention to detail and the impossibility of tracking down the recordings used, which seriously undermines their use as a source of arguments. I wonder why the apparently distinctive ssp. atrotinctus was not sampled in the genetic analyses, it does look remarkably different in plumage! Finally, it is clear that more thorough sampling of albicollis is needed (sampling in Nuñez-Zapata et al. 2016 is very poor, and samples come only from the southern range of albicollis), but I don´t think that this will change the placement of daguae or the degree of differentiation with respect to assimilis (note that ssp. phaeopygus and spodiolaemus of albicollis in Batista et al. (2020) are found as sister to assimilis, which is relevant to the status of daguae).

Comments from Bonaccorso: “YES. Although the genetic evidence is only mitochondrial, the differentiation is tremendous compared to all other forms within T. assimilis, and I agree with Nacho that it is very unlikely that the pattern is contradicted by additional genetic evidence. Even though the T. a. daguae samples are from Ecuador, I would not think sampling specimens from northern Colombia would make a huge difference. I am no expert on vocalizations, but the songs of T. a. daguae from Xenocanto sound very different to the forms from Panama and Costa Rica. It is also interesting that Xenocanto is treating T. daguae separately from T. assimilis (I guess that is just IOC taxonomy, but it makes sense to me). It seems that plumage color is not a very good indicator of species status in this group of very similar taxa. Still, that is what one would expect from a process of “cryptic” speciation, right? Finally, this split makes perfect biogeographic sense.”

Comments from Stiles: “NO. Although I tend to agree with Nacho and Santiago, I agree that there is still sufficient uncertainty regarding both vocal and genetic data to reach a definitive conclusion. No genetic data from the albicollis populations closest to daguae, and while I suspect that tissue samples from Colombia might exist (Check with Daniel Cadena and Andrés Cuervo). If so, these should be sequenced. Likewise, need recordings of T. assimilis phaeopygus … so, NO pending additional data.”

Comments from Jaramillo: “YES – Sure, more data are necessary, and some things are unclear. But it seems to me that as Elisa, Nacho, and Santiago comment, there is enough here to change the status quo. The status quo appears to be off; we may not quite know how it is off, but I would rather make this change and adjust at a later date than not do anything given the data we have.”