Revised Proposal (589) to South American Classification Committee
Split Epinecrophylla haematonota into four species
Effect on SACC: If adopted, three species would be added to the South American checklist by splitting Epinecrophylla haematonota into four species as recommended by Whitney, Isler, Bravo, Aristizábal, Schunck, Silveira, and Piacentini (2013). Following a comment to the original proposal and the ensuing discussion amended herein, this revision is split into two parts for the committee's consideration as follows:
A. E. amazonica (Ihering, 1904),E. haematonota (Sclater, 1857), E. pyrrhonota (Sclater and Salvin, 1873), and
B. the newly described E. dentei.
Background: The Stipple-throated Antwren is a widespread denizen of western Amazonia, occurring both south and north of the Amazon. During survey work in the Aripuanã-Machado interfluvium, Amazonas, Brazil, in 2000-2003, Bret Whitney noted that the vocalizations of E. haematonota antwrens were audibly distinct from those on the opposite side (left bank) of the Rio Madeira.
Newly Published information
A. To provide a foundation for evaluating the taxonomic status of the Aripuanã-Machado population, molecular and vocal analysis of currently recognized subspecies of E.haematonota was undertaken, Maximum-likelihood and Bayesian analyses of mtDNA showed that the Aripuanã-Machado populationwas embedded within a large well-supported clade containing four distinctive lineages comprising taxa currently known as E. spodionota, E. h. haematonota (including E. fjeldsaai), E. h. pyrrhonota, and a fourth lineage including E. h. amazonica and the Aripuanã-Machado population. The relationships among the four lineages are not well resolved, and they are ~5–6 % divergent. E. spodionota has previously been recognized as a distinct species, and differences in loudsong pace support the acknowledgement of the remaining lineages as distinct species. The slowest paced loudsongs are delivered by amazonica, the fastest by nominate haematonota and pyrrhonota, and the pace of the loudsong of the Aripuanã-Machado populationis intermediate. All pace comparisons were significant statistically except for between h. haematonotaand pyrrhonota. The latter two taxa, however, are distinguished morphologically, especially as the female plumage of pyrrhonota is most distinct in the complex. Given vocal and morphological differences, the extent of genetic diversity among the lineages and the inability to resolve their relationships, even though not a direct basis for species consideration, indicates that the evolution of the four clades, spodionota, pyrrhonota, haematonota, and the clade containing amazonica and the Aripuanã-Machado population has proceeded to the species level.
B. The phylogeny showed that the Aripuanã-Machado population was sister to E. h. amazonica, its neighbor to the west on the opposite bank of the Rio Madeira. Differences in vocalizations and morphology support its description as a new species. The new species, E. dentei, differs vocally by the significantly faster pace of its loudsong and by a diagnosable distinction in note shape which produces loudsongs of a subtlety different quality to the human ear. Adult females of dentei are readily distinguished from those of amazonica by color of the throat and of the belly. Maximum-likelihood and Bayesian analysis showed that dentei and amazonica were ~3% divergent. Phenotypic differences of vocalizations and plumage supported by genetic results meet previously established guidelines for species consideration in thamnophilids in parapatry (Isler et al. 1998).
Recommendation: I recommend a "YES" vote on accepting all four species, including Epinecrophylla dentei, in the list, based on morphology, vocalizations, and genetic distinctions. The status of fjeldsaai is considered in a separate recommendation.
Literature cited:
Isler, M. L., P. R. Isler, and B. M. Whitney. 1998. Use of vocalizations to establish species limits in antbirds (Passeriformes; Thamnophilidae). Auk 115:577–590.
Whitney, B. M., M. L. Isler, G. A. Bravo, N. Aristizábal, F. Schunck, L. F. Silveira, and V. de Q. Piacentini. 2013. A new species of Epinecrophylla antwren from the Aripuanã-Machado interfluvium in central Amazonian Brazil with revision of the “stipple-throated antwren” complex. In: del Hoyo, J., Elliot, A., Sargatal, J., & Christie, D.A. (Eds.), Handbook of the Birds of the World. Special Volume: New Species and Global Index. Lynx Edicions, Barcelona, Spain, pp. 263–267.
Morton Isler, November 2013
======
Further comments:
Comments from Thomas Donegan: Our team who works on the Colombian checklist is presently considering the various HBW splits and new species, such as this. Proposal 589 refers to splitting E. haematonota into 4 species, i.e. treating pyrrhonota, dentei and amazonica each as species when they were previously considered subspecies or undescribed. I had some queries with the vocal data supporting this approach which have been discussed offline with Mort Isler and Bret Whitney, and we agreed to include an edited version of this discussion here:
At the outset, I should note great appreciation for the work that Bret Whitney and colleagues put into getting all these Amazonian species "out there" in such an informative volume; and in particular for this interesting paper on ex-Myrmotherulas. I apologise for what may be perceived as a negative tone of commentary. One of the most depressing things about publication is the (generally only) negative feedback one tends to receive in return. To redress that here: congratulations for such a mammoth and impressive effort and piece of work, which will surely put the taxonomy of Amazonian birds on a surer footing; and in particular for a very interesting review of these antwrens. I would though raise the following issues on this proposal:
1. The only claimed differences in voice between the four proposed species in this group are in song speed. The populations postulated for species rank (haematonota, pyrrhonota, amazonica and dentei) are claimed to differ from one another "significantly" in song speed (except pyrrhonota vs. haematonota). However, the wording used is confusing: “Isler et al. (1998, 2007)” (who published a test of diagnosability) are cited in the introduction to the section, as a test that two species “differed significantly”. Then when discussing song speed, the paper states that the “average pace of the four populations differ significantly”. This wording suggests usage of a Student's t-test or similar (tests of statistical significance) rather than diagnosability. The authors cited only means for a single variable, with no standard deviation data or information about upper and lower bounds of this variable presented and no information about other variables, so it is not possible to work out whether diagnosis is likely. Whitney et al. (2013) did not state any other quantitative vocal differences (except near-diagnostic note interval length differences between amazonica and dentei). Subjective differences in note shape are also cited but were considered diagnostic only “almost” (i.e. not entirely?) perfectly.
2. If the single variable cited is diagnostic between some populations, this level of differentiation falls below that recommended for diagnosing antbird species – where three diagnosable differences (not a single one) in variables for calls or songs is the usual benchmark for sympatric species (Isler et al. 1998). (Calls were not studied here.)
3. Any splitting then relies heavily on molecular data. The molecular study concludes impressive (>5%) differentiation between various groups in a polytomy: spodionota; amazonica+dentei; haematonota+fjeldsaai; and pyrrhonota. However,splitting based onmolecular data does not adopt a minimalistic approach to resolving the polytomy, because dentei groups with vocally similar amazonica and is not in any separate part of the polytomy but is proposed here for species rank.
There is then a highly favourable factor towards most of the splits advocated in this paper and proposal, which is almost entirely overlooked. Two members of the polytomy, spodionota and pyrrhonota, are fully sympatric in the East Andes of Colombia. Salaman et al. (2002) noted as follows:
"Myrmotherula spodionota, FOOTHILL ANTWREN: Ten birds were caught (one collected) in both primary and secondary growth at Alto Río Hornoyaco; five were captured and collected at Fragua (1000-1400 m); one female and 3 males were caught and collected at 1000-1460 m on the Río Rumiyaco. These are the first specimen records from Colombia; Willis (1988) reported sightings at El Paraíso, Dpto Huila in Apr 1962. The species is otherwise known from the Andean East Slope of Ecuador and Peru from 600-1300 m (Hilty & Brown 1986). Interestingly, extensive data from Ecuador indicate an elevational segregation of M. spodionota from its close relative, the Stipple-throated Antwren M. haematonota (Krabbe et al. 1999). However, both were captured in the same mist-nets at 1100 m at Alto Río Hornoyaco although they were not observed foraging together."
As part of the team that captured and processed these two species following simultaneous capture in the same mist-net, I can confirm the accuracy of the notes in this paper! Interestingly, some Colombian East slope specimens referred to by Salaman et al. (2002) from the contact zone were among those sampled for mtDNA by Whitney et al. (2013). Consistent with observations at sites of sympatry and a lack of noted intergradation in the above study, the molecular differentiation between these forms was supported. (E. spodionota is already recognised as a separate species by the SACC.)
Lumping all of members of the polytomy is not a viable proposition, due to spodionota and pyrrhonota being sympatric and demonstrably good species. The logical next step to analyse species limits would be to determine vocal differentiation between the two sympatric members of the polytomy, and compare these differences to pairwise differences between other taxa in the polytomy. However, the authors did not do this: no vocal data on spodionota is included. We therefore do not know if this polytomy shows atypically low vocal variation but in a way consistent with the vocal variation between sympatric related antbirds.
Reading the data as its stands, splitting the polytomy does not have basis in the vocal data presented using the Isler et al. (1998) model. Splitting of the polytomy is, however, supported by molecular results, assuming that recognising monophyletic species is virtuous, because lumping sympatric spodionota with haematonota is not viable. New taxon dentei then emerges as a weaker candidate for species rank, as a result of it being grouped with amazonica in molecular data (and not an independent part of the polytomy), parapatric in its distribution with respect to amazonica (these replacing one another on different sides of an Amazonian river, a typical subspecies distribution pattern) and not demonstrating the level of vocal variation from amazonica as is demonstrated by sympatric antbirds.
I would therefore like to propose splitting this proposal into Part A (splitting haematonota into three species: (haematonota, pyrrhonota, and amazonica), for which I see no reasonable alternative even based on the limited vocal data in Whitney et al. (2013), due to the molecular data and sympatry between two forms in the polytomy and the similar divergence of all of them. Part B (splitting dentei from amazonica) should be considered separately and does not seem at all clear-cut based on the data that are currently available.
Additional literature cited:
Isler, M. L., P. R. Isler, and B. M. Whitney. 1998. Use of vocalizations to establish species limits in antbirds (Passeriformes; Thamnophilidae). Auk 115:577–590.
Salaman PGW, Stiles FG, Bohórquez CI, Alvarez M, Umaña AM, Donegan TM & Cuervo AM. 2002. New and noteworthy records from the east slope of the Andes of Colombia. Caldasia 24(1): 157-189.”
Comments from Bret Whitney: “I just read [the comments above] and it makes a lot of sense to look into each of the arguments raised. E. dentei is not a strongly differentiated species, it's a weakly differentiated species. It shows weak diagnosability morphologically and vocally (note structure being "almost" diagnostic means one sample failed perfect taxon allocation), genetic divergence is not huge, and it possesses some ecological characteristics that appear to be unique, but the fact remains that the population was well-sampled (relative to many other phylogeographic studies) and shows near-perfect consistency in all of the above character sets throughout its range. As explained in the paper, one taxonomic treatment option would be to recognize haematonota, spodionota, pyrrhonota, and amazonica as species with dentei as a subspecies of amazonica. If the committees decide that's the way to go, so be it, you will get no further explanation from me. That said, I continue to believe that the best course, the one most indicative of speciation in this group of birds, be it of long-standing or relatively incipient, is to call them all separate species. It is not possible to make a truly strong case for either species or subspecies rank in this case.
Comments from Mort Isler: “As the author who was responsible for the vocal analysis recommending that dentei be considered distinct at the species level, I will respond to Thomas Donegan's comments. Before I do, let me say that I thought Donegan's comments to be well thought out and considered, and I greatly appreciate the time and energy that they involved. It is great to have a community of individuals, like Thomas and SACC members, who carefully examine species-level recommendations.
In response to Donegan comment 1, I can confirm that tests of diagnosability were performed on the data. Unfortunately, space limitations of the HBW volume did not allow the inclusion of the description of the methodology employed nor the results of the test as we have done in all of our previous papers. Pace, a ratio, is not normally distributed, and consequently our test (using the t distribution) for likelihood of overlap with larger samples could not be employed. The statistical test employed was a non-parametric bootstrap simulation (see Isler and Whitney 2011, p. 4 for more description). The Difference Between Means (DBM) of dentei and amazonica resulting from this test was five standard deviations, which is well above an acceptable level for considering the difference to be diagnostic. Results of other pairwise comparisons of taxa in the group were diagnostic at a similar level except for between pyrrhonota and haematonota as noted in the paper.
In response to Donegan comments 2 and 3, the following recommendation in Isler et al. 1998 (p. 586) is sometimes overlooked: "However, three diagnosable differences in vocal characters, as defined herein, should be considered a point of reference, not a requirement----for parapatric taxa, fewer than three diagnosable vocal characters might be considered an appropriate threshold for species recognition." With regard to dentei and amazonica, a question can be raised as to whether they are parapatric and gene flow between them is possible or whether the rivers that appear to separate them geographically are a sufficient barrier. The answer to that question is not known, but we do know that these headwaters rivers provide a less formidable obstacle than those downstream; how much of an obstacle is the issue. Other considerations supporting the recommendation for species status include the almost perfect distinction in note shape (see Whitney comments), ecological distinctions, and genetic diversity. Although I agree with the Donegan comment on the last named, I think it appropriate to keep genetic distance in mind when the decision is borderline, as it is with the current example.
For all the reasons expressed by Donegan and expanded above, when the paper was being written, I struggled with the recommendation, and (obviously) came down on the side of a species recommendation for dentei). But Bret Whitney, in his response, put it perfectly: although dentei is a clearly differentiated population, arguments can be made for either species or subspecies rank; and although we came down on the species side, and whatever the committee decides will be acceptable.
On the second issue (Part A in the Donegan statement) regarding haematonota, pyrrhonota, and amazonica, I would only add one observation. The geographic pattern provided by loudsong pace is informative. The slower paced loudsongs of amazonica are juxtaposed in between the fastest paced loudsongs of haematonota (and pyrrhonota) and the intermediate paced ones of dentei. This pattern clearly obviates the possibility of west to east clinality, and although it is not directly relevant to the amazonica/dentei species rank issue, it seems to add support that haematonota and amazonica be considered specifically distinct.
Additional literature cited:
Isler, M. L., and B. M. Whitney. 2011. Species limits in antbirds (Thamnophilidae): the Scale-backed Antbird (Willisornis poecilinotus) complex. Wilson Journal of Ornithology 123: 1–14.”
Additional Comment from Thomas Donegan: As to falling back to lack of gene flow, supported by c3% mtDNA variation and good sampling, many diagnostic subspecies could be subjected to this sort of argumentation. If subspecies are to be a valid category based on diagnosability but small differences, then one has to draw the line somewhere between those showing smaller and larger differences. I have found Isler et al.'s "point of reference" a good basis for this in suboscine birds. E. dentei is borderline in this respect, but falls the wrong side of the standard suggested. I am not particularly trying to push subspecies treatment versus species as cases like this end up being pretty subjective, but as the other proposed splits within the proposal are much more certain, I do feel they should at least better be unbundled in this committee process.
Additional Comment from Bret Whitney: The yardstick suggested by Isler et al. 1998 continues to be a conservative standard, but as Mort pointed out, we never imagined that it would become a requirement, and we included statements to that effect. To the contrary, we knew full well that some well-differentiated, universally recognized groups with very simple vocalizations (e.g. Grallaricula flavirostris complex) were definitely not going to display three diagnostic vocal characters. In Hypocnemis, we documented fairly extensive syntopy of subflava and peruviana in Brazil and Peru, but there is only one diagnostic character separating the vocal repertoires of these birds (that fact will no doubt come into play in the SACC proposal dealing with newly described H. rondoni).”
Additional Comment from Mort Isler: I hesitate to extend the discussion further, but I think it should be emphasized that the three-character yardstick continues to provide a valuable insight into whether the vocalizations of allopatric thamnophilid populations have diverged sufficiently to facilitate reproductive isolation. After many years of use, the consistency of results derived from its use with the results of molecular studies is especially informative. But it remains a yardstick, to be used in conjunction with other considerations.