Proposal (505) to South American Classification Committee

Split Thrush-like Manakin Schiffornis turdina into any of two to seven species

Background: In Proposal 327, S. turdina was proposed to be split into five species, based on Nyári (2007). Although several committee members voted in favour of this treatment (and all were in favour of some splitting), the proposal was rejected, there has been no follow up proposal, and S. turdina remains a single species on the AOU-SACC list. A large number of archived and published sound recordings are now available of greater turdina in published recording compilations and online resources such as xeno-canto. In Donegan et al. (2011), we re-examined Nyári (2007)’s recommendations for purposes of the Colombian checklist and recent Spanish language field guide (McMullan et al. 2011) in light of the additional sound recordings available today. Photographs of birds representing the three different Colombian populations were also presented. This proposal gives the AOU-SACC the opportunity to reconsider this over-lumped group once again.

Summary of Proposal: Nyári (2007) presented molecular data and some sonograms, as discussed in proposal 327. The present species ‘turdina’ is a complex group obviously constituting several biological species, with several distinctive vocal types and some instances of sympatry. In Donegan et al. (2011), we studied geographic variation in voice, the type localities of various of the names and priority issues. Committee members who wish to consider the issues in more detail may wish to consult the paper, alongside Nyári (2007)’s maps and phylogeny. We presented a series of sonograms and studied recordings with a broader geographical sample for recognised subspecies. We differed from Nyári (2007) in a handful of aspects, but the bulk of Nyári (2007)’s proposals were supported by additional materials now available.

The turdina group needs splitting even under the most conservative of species concepts. Although this aspect was not highlighted by Nyári (2007), two of the taxa in the turdina group are sympatric in Central Panama (Ridgely & Gwynne 1989) and two others of them replace one another by elevation in the Amazon-Andes interface of Ecuador and Peru (Ridgely & Greenfield 2001, Krabbe & Nilsson 2003). Moreover, various other populations are vocally distinct. This proposal should therefore involve a discussion of the extent and manner in which one ought to go about cutting off various limbs in this group so as to produce a set of species which are vocally cohesive or, for those interested in such matters, monophyletic (rather than a question of whether sub-division is warranted at all).  In Donegan et al. (2011), we proposed recognising the following species (using Nyári 2007’s vernacular names):

1. Thrush-like Schiffornis S. turdina (provisionally including subspecies steinbachi, amazonum, wallacii and intermedia) of the Amazon region and Atlantic forest, including the Amazonian region of Colombia.

2. Slender-billed Schiffornis S. stenorhyncha (including panamensis) of the Tacarcuna region of Panama and Colombia, Magdalena valley and Central, East and Merida Andes of Colombia and north-western Venezuela.

3. Brown Schiffornis S. veraepacis (including dumicola, rosenbergi, “buckleyi” and acrophites) of the Chocó from northernmost Peru through Ecuador to Colombia and Central America from northern/western Panamá northwards.

4. Foothill Schiffornis S. aenea of the western Amazon region of Ecuador and Peru.

5. Olivaceous Schiffornis S. olivacea of the Guianan shield.

The table below summarises the range of named populations and other proposed treatments:

Population name	Range	Nyári Vocal group	Nyári’s molecular group	Nyári’s PSC approach	Nyári’s BSC approach	Nyári’s SACC proposal	Donegan et al. 2011 BSC approach
veraepacis, dumicola	Central America (S Mexico and Belize to N/W Panama)	A	1	veraepacis	veraepacis	veraepacis	veraepacis
rosenbergi / buckleyi / acrophites	Chocó of Colombia and N Ecuador	A	3	rosenbergi  [/ veraepacis]	veraepacis	veraepacis	veraepacis
olivacea	Guianan shield of Venezuela to Suriname, N. Brazil	A	7	olivacea	veraepacis	olivacea	olivacea
aenea	West Amazon in Ecuador and Perú	B	4	aenea	aenea	aenea	aenea
stenorhyncha / panamensis	Northern Colombia, Venezuela, S/E Panama	E	2	stenorhyncha	stenorhyncha	stenorhyncha	stenorhyncha
turdina, intermedia	Atlantic forest region of Brazil	D	6	turdina	turdina	turdina	turdina
steinbachi	Southern Amazonia in Peru, Bolivia, Brazil	C	6	amazonum	amazonum	amazonum	turdina
wallacii	Para, Brazil and surrounding region	A/C	6/7	amazona / olivacea	amazonum / veraepacis	amazonum / olivacea	turdina
amazonum	Northern Amazonia in Colombia, Venezuela, Peru, Bolivia, Brazil.	C	5	amazonum	amazonum	amazonum / turdina	turdina

Sub-proposals: This proposal set is split into various cumulative sub-proposals, such that if the SACC disagrees with Donegan et al. (2011)’s treatment or wishes to adopt some other arrangement, it can stop at an earlier or later stage of limb separation and still come up with a new taxonomy.

A: Split veraepacis (with dumicola, rosenbergi, buckleyi, acrophites, aenea and olivacea), from turdina (with wallacii, amazonum, intermedia, steinbachi, stenorhyncha and panamensis).  Taxa dumicola (of the veraepacis group) and stenorhyncha (sister to eastern races for which the name turdina is senior) are sympatric in Central Panama (Ridgely & Gwynne 1989) and differ in their voice and plumage. Recordings of the two vocal types that these names represent by Ken Allaire, Mike Nelson and others from Panama (of which sonograms were presented in Donegan et al. 2011) overlap geographically, confirming sympatry or elevational parapatry. According to recordist notes, the two do not respond to playback of one another. Separately, subspecies aenea (east slope Ecuador) replaces the Amazonian lowland groups (amazonum) by elevation in the Andes-Amazon interface. This split is an easy one under a BSC approach, mandated by two instances of sympatry, but it produces two vocally non-cohesive species with weird distributions, one of which is polyphyletic. It is strongly recommended that this proposal be accepted as a starter, in the context of the next following sub-proposals.

B.  Split stenorhyncha (with panamensis) from turdina (with wallacii, amazonum, steinbachi and intermedia). These populations are allopatric sister populations.  Nonetheless, this is a straightforward split in light of them having perhaps the strongest pairwise vocal differences of any two groups, which are equivalent to those between sympatric Schiffornis. This proposal, if accepted, clears up some of the strange distribution caused by A and deals with vocal non-cohesiveness of the eastern group. The plumage differences between the three species split by proposals A and B together are also very strong, as illustrated by photographs of examples of the three taxa from Colombia in Donegan et al. (2011). Daniel Cadena fairly criticised Nyári (2007)’s split of this taxon on account of it not being based on recordings from different biogeographic regions within the range of stenorhyncha (including ‘panamensis’).  In Donegan et al. (2011), we presented or studied sonograms of birds from near the type locality of stenorhyncha (Falcon, Venezuela), the Merida Andes (including near the southernmost extent of the East slope population), the Magdalena valley, Serranía de San Lucas (Cauca valley border, it does not go further south in this region) and eastern Panama (near the type locality of “panamensis”), confirming the consistency of the song throughout the range of Nyári (2007)’s proposed species stenorhyncha.

--- These first two proposals come very highly recommended for acceptance and should be regarded as uncontroversial under any species concept.  There now follow two further splits of allopatric populations which Nyári proposed making in his SACC proposal and with which Donegan et al. (2011) agreed. ---

C.  Split aenea (monotypic) from veraepacis (with dumicola, rosenbergi, buckleyi, acrolophites; and olivacea if D fails).  S. aenea is the Andean East slope population in Ecuador and Peru.  It may also extend in range into southern Colombia, although there are no records there yet.  The two populations subject to proposal C are apparently allopatric sister populations which straddle the Andes. When split, they appear as mutually monophyletic (c.3% mtDNA difference). Rejecting this split does not therefore produce polyphyly or paraphyly. S. aenea is however vocally rather different from veraepacis, leading to various authors such as Ridgely & Tudor (2001) and Krabbe & Nilsson (2003) noting that more than one species may be involved (see sonograms in Donegan et al. 2011). The differences are not as great as those shown by stenorhyncha (Proposal B), with various of the note shapes of the aenea song having equivalents in a different order in the songs given by the veraepacis group, but the differences in note shape or order of note shape are consistent and diagnosable. We preferred to spit aenea, based on vocal differences, molecular data, and their distributions - which straddle the Andes in a very high part of the mountain range.  Taken together, these factors put the burden of proof on those who would have these two lumped.

D.  Split olivacea (monotypic) from veraepacis (with dumicola, rosenbergi, buckleyi and acrolophites; and aenea if C fails). This is the troubling Guianan shield population.  This proposed split was subject to a differing treatment by Nyári (2007) in his BSC interpretation (lumped) versus his SACC proposal and PSC interpretation (split). Olivacea is vocally very similar to the veraepacis group, with no differences elucidated by Nyári (2007).  In Donegan et al. (2011), we noted small differences in the extent of the upturn of the main note.  In the previous proposal, there was some speculation as to whether the vocal similarities here are a result of limited divergence or convergence, although most committee members seemed in favour of splitting this taxon. It is basal to all other current turdina in the Nyári (2007) phylogeny, showing 9%+ mtDNA differentiation from all other taxa.  In Donegan et al. (2011), we presented additional information suggesting that olivacea is indeed the correct name for this population, as provisionally treated by Nyári (2007).  Not splitting olivacea produces a veraepacis group which is polyphyletic and which has a strange distribution. Because Proposal A should not be regarded as optional (and B and C are strongly recommended), it would be reasonable also to make this split.

--- We went this far in Donegan et al. (2011) and did not adopt any further splits. ----

E. Split amazonum (with wallacii) from turdina (with intermedia and steinbachi). This is the “North Amazon vs. South Amazon and Atlantic forest” split that Nyári (2007) proposed.  There is clearly vocal variation in the southern part of the greater ‘turdina’ range, but this split raises various difficulties, some of which were discussed by Doug Stotz and Van Remsen as being unfavourable factors towards adopting the treatment in the previous proposal. Further research indicates that turdina in the subspecies sense (South Atlantic forest) resembles amazonum (North Amazon) vocally; whilst intermedia (North Atlantic forest) generally resembles assumed steinbachi (south Amazon of Bolivia to Peru) vocally.  All these eastern and southern populations are monophyletic when taken together.  Their songs are all generally comprised of longer notes than the other species, differing among one another in the shapes of up or down-turns at the start of end of particular notes (see sonograms in Donegan et al. 2011).  These vocal differences exceed those shown by olivacea but do not reach the differentiation shown by aenea. Given the scope of our paper (Colombia), a detailed examination of the Bolivian and Brazilian types, their localities and sound recordings on different sides of major Amazonian rivers was out of scope, but could be recommended for further research. This split or other possible treatments for the southern populations may be warranted but lumping them does not cause paraphyly or polyphyly, nor does it produce a vocally uncohesive group or preclude further studies from taking place.  We did not recommend adopting this split for the time being but further research is clearly warranted.

--- Nyári (2007) went this far and did not adopt any further splits. ---

F. A further possible alternative for the Eastern taxa based on tentative vocal data would be to split turdina (with wallacii and amazonum) from intermedia (with steinbachi). This produces two species of strange distribution and it is unknown how this would hang with molecular data, owing to the lack of sampled individuals from very close to the type localities of some of these names in Nyári (2007).  We did not take this step, but SACC members with greater familiarity with southern Amazonian and Atlantic forest birds may wish to comment on this option or consider it, for completeness, as an alternative to E.

G. Split veraepacis (with dumicola) from rosenbergi (with buckleyi and acrophites). Proposals A-C, if accepted, result in veraepacis including two disjunct populations, one in the Chocó-Tumbes and another broadly in Central America.  The ranges of the two are bisected in the Tacarcuna region to southern/eastern Panama by that of stenorhyncha. The veraepacis and rosenbergi groups are apparently mutually monophyletic, although with low molecular differentiation (0.8% mtDNA) and with small vocal differences.  This would be a possible split under some species concepts (e.g. PSC) but we did not adopt it.  Tentative differences in secondary calls (based on only a single Central American recording of the secondary song) should be regarded as a matter for further research.  As with the possible Amazonian splits, not adopting this treatment does not preclude further studies of these birds.

Recommendation: A resolute YES to A and B; YES to C and D for the reasons set out in Donegan et al. (2011) and Nyári (2007). NO to E, F and G for the time being, with a note that further research could shed light on variation among these populations and that other splits may be warranted in the future.

English names: If any of these proposals pass, then the English names suggested by Arpad Nyári in Proposal 327 (see above) would be adopted.  If anyone prefers a different vernacular name for a narrower turdina, then they can raise a separate proposal on that issue.

References:

Donegan, T.M., Quevedo, A., McMullan, M. & Salaman, P. 2011.  Revision of the status of bird species occurring or reported in Colombia 2011.  Conservación -Colombiana 15: 4-21.

Nyári, Á. S. 2007. Phylogeographic patterns, molecular and vocal differentiation, and species limits in Schiffornis turdina (Aves). Molecular Phylogenetics & Evolution 44: 154-164.

Other references are cited in these papers.

Thomas Donegan, October 2011

=========================================================

Comments from Robbins: “YES to subproposals A (veraepacis), B (stenorhyncha), C (aenea), and D (olivacea). Nyari’s genetic and preliminary vocal data clearly established that these taxa should be elevated to species level; Donegan et al.’s (2011) more in-depth vocal analyses corroborated those results.  Undoubtedly, recognition of additional species will be warranted when more detailed information becomes available.”

Comments from Stiles: “This proposal would split Schiffornis turdina into at least four species, based primarily upon data from the genetic analysis of Nyári and a more detailed analysis of vocalizations by Donegan.  The latter are a definite step forward and incline me to agree with the proposal, although given the abundant material available, I would have felt more comfortable if statistical analyses of the vocalizations had been performed.  However, with the data at hand I agree that the burden of proof has now shifted onto those who would retain a single species. I have now had a chance to examine the Colombian material here and have obtained information on the plumage differences of at least the Colombian taxa, and this leaves me with strong doubts regarding the English names proposed by Nyári – given the rather subtle differences between most taxa and the uncertainty regarding the limits of some distributions, I feel that appropriate and descriptive English names may be important, and will make a series of suggestions below (which might warrant a separate proposal). For now, regarding the splits proposed by Donegan:

A. Split the veraepacis group from a broad turdina: YES. The sympatry of the two groups in central Panamá, plus differences in vocalizations and evidence that the two do not respond to each other’s songs makes this split mandatory.

B. Split the eastern amazonum group from the northern stenorhyncha group: YES. Both the genetic data and information from vocalizations, as well as biogeography, make this split logical and desirable.

C. Split aenea from the veraepacis group:  YES. Although the genetic data are not quite as definitive, both vocalizations and biogeography seem better addressed by splitting.

D. Split olivacea of the Guyana Shield from the veraepacis group: YES. This split will avoid massive polyphyly in the genus as a whole and makes good biogeographical sense as well.

“I agree with Donegan that although some further splits could be made, especially in the amazonum group, more data are required. The split of rosenbergi-acrolphites from veraepacis has no genetic backing and mainly reflects the gap in distributions, which might be better explained by historical factors. For now, the four splits suggested are sufficient and desirable.

“Now, for the messy part: English names. Because the splits have as yet not been “officially” adopted, it seems appropriate to make these suggestions now.  To begin with, I feel strongly that the name “Brown Schiffornis” for the veraepacis group (species) is singularly inappropriate. The two Colombian races (acrolophites and rosenbergi) are much the greenest of all the taxa.  By far the brownest of all is stenorhyncha (which is also not the “slenderest-billed”; that distinction belongs to amazonum). Hence, I would propose that the name “Brown Schiffornis” be applied to stenorhyncha (or “Brownish Schiffornis” if one wishes to avoid confusion with Nyari’s names). “Greenish Schiffornis” could then be applied to veraepacis. I make this latter suggestion a bit tentatively, as I do not have material of the northern races of the latter for direct comparison, although my description and the plate in the Costa Rican guide (as well as the descriptions in Ridgway) emphasize olive-green to olive-brown tones. (The name “olivaceous” would also be appropriate, but is perhaps best reserved for olivacea of the Guyana Shield – although for the latter “Guianan Schiffornis” would also be appropriate).  Aside from its overall brownish coloration, the most trenchant plumage characters of stenorhyncha are the decidedly rufescent color of the wings and the sharp division of the brownish to olive-brown breast band and the grayish-olive lower breast and belly; it is also the largest taxon. Hence “Rufous-winged” or “Grayish-bellied” would not be inappropriate for this taxon.  I note here that none of our series of stenorhyncha show the clear gray belly of the bird in Donegan’s photo: given the yellowish coloration of the basal tomia and gape, I suspect that his bird was young – I suspect that young birds in this genus in general are brighter and more contrasty than adults, though few of our specimens are reliably aged.  Actually, the belly in amazonum is also grayish olive and in some is grayer than in most stenorhyncha, although the contrast with the breast is not nearly so sharp. The wing in amazonum is a darker, duller brown than in stenorhyncha so “Brown-winged Schiffornis” would emphasize this difference, but I think that “Amazonian Schiffornis” is certainly simpler and appropriate, and could be kept for amazonum should this group be split further - although such a split (or splits) seem problematic for the present.  

Comments from Stotz: “YES on subproposals A, B, C, and D. NO on subproposals E, F, and G.  I am much more comfortable with this proposal than the previous proposal.  We will probably have to split other taxa farther down the line, but this is a reasonable first step on this complicated group.  I agree with Gary that we need to think about English names.  However, his suggestion of “Greenish” for veraepacis won’t work. Greenish Schiffornis is the name of Schiffornis virescensof SE Brazil.  I think that it would be a mistake to use Olivaceous for veraepacis with S. olivaceus being one of the names.  So I don’t have a good alternative in mind, but I agree with Gary that Brown would not be a great choice.  I would go for Brownish for stenorhyncha, and while I am okay with Olivaceous for S. olivaceus, I think we would be better served to adopt a geographic modifier Guianan for that species.  One further issue is a name for the reduced turdina.  Given the complications in that group, and how much of the turdina (sensu lato) has been carved off from it.  I think we absolutely have to have an alternate English name for this new turdina. Unfortunately, I have not come up with a great answer.  These are really dull birds.  My only thought is Southern Schiffornis.”

Comments from Pérez: “YES to subproposals A, B, C, and D. I think molecular, vocal and distributional data support this awaited treatment for S. turdina. NO to subproposals E, F and G; more in-depth treatments, as suggested by Donegan et al. (2011), would likely provide information for further splitting.”

Comments from Jaramillo: “YES – accept A, B, C, and D. I am not comfortable going further with it, particularly as vocal data are relatively common and further separations require a higher level of scrutiny. Having said this, once we determine which forms the committee has decided to split off, or not, I think we need a separate simple proposal on the English names.”

Comments from Nores: “YES to subproposals A (veraepacis), B (stenorhyncha), C (aenea), and D (olivacea). I consider that the molecular and vocal data support this treatment. However, I did not find either in the proposal or the Nyári´s paper a clear relationship between the new species and the regions 1-7 of the phylogeographic tree of Nyári. How much easier would have been to interpret the Nyári´s tree if he would have put the name of the subspecies or species next to the name of the region.

Comments from Remsen: “YES to subproposals A (veraepacis), B (stenorhyncha), C (aenea), and D (olivacea).  Echoing the comments of others, there is good evidence for at least 5 species, and more will likely come from additional analyses.  By the way, I strongly agree with Manuel’s comment on the poor labeling in Nyári´s tree – just one more example of how poorly edited MPE is.

“English names are a real problem.  Ridgway and Hellmayr both used “Olivaceous” for S. olivacea, so I would favor retaining that one.  Gary’s comment on “Brown” for veraepacis is correct – “Brown” is perhaps the worst name possible for this species, even though used by Ridgway and Hellmayr. Ridgway used “Russet” for stenorhyncha, but I do think Gary’s name is better. Hellmayr used “Slender-billed”, but no reason to perpetuate that if inaccurate.  Also, Doug’s comments are also correct – “Greenish” is not available, and we really cannot retain “Thrush-like” for a dramatically diminished S. turdina (and besides, it’s a poor name, species epithet not withstanding). So, I am installing some temporary English names based on Gary’s recommendations and will be appointing someone to make a formal proposal to examine carefully the English names before they get any traction.”