Species limits in Arremon torquatus


Proposal (468) to South American Classification Committee



Background: Species limits in Arremon (formerly Buarremon and Atlapetes) torquatus have been controversial over the years, as reviewed by Remsen and Graves (1995) and more recently by Cadena and Cuervo (2010). The complex consists of 14 forms, currently treated as subspecies, ranging from Costa Rica to Argentina; 12 of these occur in the area of SACC coverage (13 if one counts the form tacarcunae from eastern Panama). The difficulty in establishing species limits in the group despite the existence of marked geographic variation has been, as usual for many groups, the allopatric distribution of nearly all of these forms.


New analysis: Cadena and Cuervo (2010[CC1] ) examined species limits in the group by using a molecular phylogenetic analysis (Cadena et al. 2007) as a basis to examine patterns of phenotypic, vocal, and ecological variation in the group. In short, they found that forms assimilis from high elevations of the northern Andes and atricapillus from lower elevations of Panama and Colombia can be considered geographically sympatric because their favored environments are intermingled over a large geographic area where they are not separated by any physical barriers. Despite their sympatry, the two forms are strongly differentiated genetically (they are not even sister taxa and differ by more than 6% in mtDNA distances), morphologically (in plumage color and pattern and bill size and shape), and vocally. Therefore, these two taxa are clearly two different species. Cadena and Cuervo also documented that two other forms, poliophrys and torquatus, seem to come together at a presumably narrow contact zone over relatively continuous environments near the Peru-Bolivia border, where hybridization is likely limited. These two taxa represent strongly differentiated non-sister mtDNA clades and differ vocally and phenotypically (plumage and morphometrics), so they are good candidates for species status as well, but information on patterns of variation in all the relevant traits near the contact zone was sparse.


After determining that assimilis and atricapillus (and possibly poliophrys and torquatus) are different species, the difficult part of the analysis was deciding what to do with the rest of the complex based on incomplete information and considering the allopatric distributions of all other forms. I note that, although incomplete, the available information was nonetheless sufficient to show that different sets of traits do not vary in parallel, so using divergence between the two sympatric taxa as a yardstick to gauge the degree of reproductive isolation between other taxa would be extremely difficult and problematic (see the paper for a -long- discussion of this issue, including other more general problems that may apply to other situations in which we have applied this comparative yardstick approach). Being more of a conceptual and methodological paper focused on the challenges of species delimitation in difficult groups than an empirical analysis, our discussion focused on contemplating several taxonomic possibilities and then concluded that none was truly satisfactory.


But we needed to decide something. We thus proposed splitting the complex into eight species, which are admittedly supported largely (and, in all fairness, sometimes only) by mtDNA variation. The other alternative would have been to treat atricapillus as a different species from assimilis and leave everything else unchanged, but this created two difficulties. First, to which of the two species should one assign all other allopatric populations? We were unable to answer this in several cases and when we could make some educated guesses, invariably we were faced by the second difficulty: the recognition of non-monophyletic species. Non-monophyletic species are not at odds with the BSC, so this is not a problem per se, but it was simply too hard to find arguments for supporting the existence of paraphyletic units one would call species in this group given all the variation (i.e. it would not be entirely clear which form goes where).


The eight proposed species, seven of which occur in South America would be:


(1) A. costaricensis from Costa Rica and western Panama

(2) A. atricapillus from central and eastern Panama and the Colombian Andes (includes atricapillus and tacarcunae)

(3) A. basilicus from the Sierra Nevada de Santa Marta, northern Colombia

(4) A. perijanus from the Serranía del Perijá, northeast Colombia and northwest Venezuela

(5) A. assimilis from the Andes of Venezuela, Colombia, Ecuador, and most of Peru (includes larensis, assimilis, nigrifrons, and poliophrys)

(6) A. torquatus from the Andes of extreme southern Peru, Bolivia, and Argentina (includes torquatus, fimbriatus, and borelli)

(7) A. phaeopleurus from the Cordillera de la Costa, northern Venezuela

(8) A. phygas from the Cordillera de la Costa Oriental, northeast Venezuela.



Recommendation: With all the above being said, and trying to be consistent with what we concluded in the paper, I recommend voting YES to accept our proposed classification. I note, however, that I am the first to admit that this classification is quite far from being perfect (again, see detailed discussion in the paper). However, I think that this proposed change is far better than doing nothing, considering we already have very strong evidence for species status for two (and maybe four) of the forms. I suggest we take this as a provisional classification we may end up modifying soon, accepting that while being imperfect it is better than the one we have followed thus far.


Literature Cited

Cadena, C. D., and A. M. Cuervo. 2010. Molecules, ecology, morphology, and songs in concert: How many species is “Arremon torquatus” (Aves, Emberizidae)? . Biological Journal of the Linnean Society 99:152-176.


Cadena, C. D., J. Klicka, and R. E. Ricklefs. 2007. Evolutionary differentiation in the Neotropical montane region: molecular phylogenetics and phylogeography of Buarremon brush-finches (Aves, Emberizidae). Molecular Phylogenetics and Evolution 44:993-1016.


Remsen, J. V., Jr., and W. S. Graves. 1995. Distribution patterns of Buarremon Brush-finches (Emberizinae) and interspecific competition in Andean birds. Auk 112:225-236.


C. D. Cadena, September 2010




Comments from Remsen: “YES.  I’ve reviewed Daniel and Andrés’ paper, and this is a near-and-dear-to-me group.  I’m convinced that this is the best way to carve up the taxa into species units given the limitations we have on available data, as outlined clearly and objectively by the authors.”


Comments from Stiles:  YES – after rereading Cadena & Cuero’s paper, I can see no better way to recognize the complex patterns of variation in this decidedly complex group: it certainly represents a major step forward.”


Comments from Nores:  “NO.  Although very probably several of the eight forms were really different species, for example atricapillus, I do not think that they can be elevated to whole species. Neither do I see logic in eight new species being created because the authors have the need to decide something based on a classification that is far from perfect. I think that the proposal should be separated into lesser proposals or have several parts for voting. For example, in the molecular analysis of Cadena et al. (2007) the subspecies perijanus and phaeopleurus are not included, so they should not be included in the eight species proposed. Moreover, polyophrys and larensis could also have been proposed as species because they are in a similar situation to that of assimilis.”


Comments from Robbins:  “YES.  Cadena and Cuervo have presented a thorough and very objective approach to dealing with this confusing complex. Given the current understanding, their recognition of eight species appears to be the best recourse.”


Additional comments from Cadena: “Manuel mentions one could split our treatment into several subproposals, but I do not get how one could do this. Let's take the fact that atricapillus and assimilis are different species. If that were the only proposal, then what should one do with all of the remaining 12 forms? In which of these two species should one put them? I simply do not know, which is one of the reasons why we went for a whole set of splits at once. If someone has any suggestions for alternatives, I would be more than happy to consider them (I am the first to admit that our proposal is not free of problems). Note also that "subspecies" perijanus *was* indeed included in the analyses in the 2007 paper. Subspecies phaeopleurus was not included back then, but as we described in the 2010 paper that forms the basis for the proposal, we already have sequence data for this taxon. Indeed, the paper documenting this has just been published online:


Cadena, C.D., Z. A. Cheviron & W. C. Funk. 2010. Testing the molecular and evolutionary causes of a “leapfrog” pattern of geographical variation in coloration. Journal of Evolutionary Biology: doi: 10.1111/j.1420-9101.2010.02175.x"


Comments from Zimmer: “YES.  This proposed classification, although imperfect, is a clear improvement on the current classification, as supported by the data presented by Cadena and Cuervo 2010.  As far as I can tell, it also makes sense on biogeographic grounds, and advances our understanding of the complex, even if further work reveals the need to make some changes.  I am very sympathetic to the problems of where to place the bulk of the taxa in the event of a simple two-way split (assimilis and atricapillus)  – it would seem that any such attempt would be entirely arbitrary, and would only further confuse the issue.”


Comments from Pacheco:  YES.  Da minha parte, também considero apropriada a solućčo sugerida por Cadena. Em vista das informaćões disponíveis acerca do complexo, sobretudo a ausźncia de monofilia nos táxons agora subordinados a torquatus, seria contraproducente dispor os oito táxons assimilis e atricapillus, sem implicar em arbitrariedade extra.”