Proposal (566) to South American Classification Committee
Treat Geotrygon purpurata as a separate species from G. saphirina
Proposal: This proposal, if it passes, would result in G. purpurata being split from G. saphirina, a treatment accepted by several other authorities, including several major publications on pigeons over the last decade and others listed in Donegan & Salaman (2012).
This split was rejected in proposal 105. This was back in 2004, before sources like xeno-canto were really available or had good samples and consistent with the SACC's then approach to other "field guide splits" around this time. Most committee members cited a lack of vocal data supporting the split, and no dissenting comments in support of the split are evident. There are two new publications relevant to this issue published in the last couple of years, and a further relevant paper was overlooked in 2004.
Brumfield & Capparella (1996: http://www.museum.lsu.edu/brumfield/pubs/brum96.pdf), using three samples of each group, found unusually high divergence between purpurata and saphirina for lowland conspecifics with a Chocó / Amazonian distribution. More recently, Johnson & Weckstein (2011) studied a single Ecuadorian purpurata and single Peruvian saphirina in a phylogeny including good sampling of new world pigeons. They found strong support for a sister relationship and moderate (modeled c.1.2 million years) differentiation, consistent with that observed between samples of Russet-crowned Quail-Dove Geotrygon goldmani & Chiriquí or Rufous-breasted Ground-Dove G. chiriquensis and those between nominate Grey-fronted Dove Leptotila rufaxilla & Yungas Dove L. megalura, but also consistent with intraspecific variation in widespread Leptotila verreauxi. One of their BEAST chronogram-based phylogenies is set out below. Note these two Geotrygon species at the bottom of the tree, and the relative depth of the branch.
Neither studies reject monophyly for the combined group (and the latter study found strong support for this). But both studies show a relatively deep division, consistent with Chocó/Amazonia splits generally and those between several Neotropical pigeon species.
Donegan & Salaman (2012: link below) published various data on these pigeon species' natural history, voice and conservation, and included a rare photograph of purpurata in life by Juan Carlos Luna. We measured two acoustic variables and found no overlap for the length of the main note in songs between saphirina and purpurata (see figure reproduced below). The differences are statistically significant, but despite there being no observed overlap, the data did not meet the Isler et al. '97.5% using t-distributions' diagnosability test. Acoustic frequency also varies between the taxa, with statistical significance found, but not diagnosability. G. saphirina (in some but not all recordings) varies the frequency of songs but G. purpurata has flat notes in all recordings. It is this variability of frequency that past transcriptions of vocal differences in field guides seem to have concentrated on as a difference, but this feature is non-diagnostic: saphirina also gives flatter calls. Song length does however separate out the sample fully based on recorded data. It would be speculative to conclude whether, with a greater sample size, the ideal 97.5% diagnosability test would be met or not. The vocal differences are, however, very real and substantial, and have gone largely overlooked or misunderstood in the literature until Salaman & Donegan (2012). See our figure 3, reproduced below (song length on x axis, frequency on y axis; p=purpurata; s=saphirina).
As discussed in our paper, given the way in which these songs are made, by birds inhaling a lot of air, expanding the chest, and then exhaling through their nostrils without bill movements (see this video: http://ibc.lynxeds.com/video/sapphire-quail-dove-geotrygon-saphirina/bird-singing-forest-while-perched-eye-level), such vocal differences seem likely to be constrained by physiological factors and not a result of learning.
The two species seem to have different elevational ranges, and, hence one assumes, ecological requirements. In particular, purpurata is a foothill bird that has not been recorded on the Chocó "floor" whilst saphirina is found in western Amazonia.
The molecular and vocal data seem consistent with other pigeon splits. The split passes Tobias et al.'s "species scoring tests" also. A subsequent communication with Nigel Collar revealed that their team last year independently assessed purpurata as (just) meeting the "species scoring test", disregarding vocal differentiation data (which were not available to them). We were more conservative in assessing plumage scores and had purpurata 'crossing the line' as a result of 2 points out of 3 for the highly differentiated and statistically significant (but not statistically diagnosable) vocal differences. Neither the paper nor this proposal is intended as an endorsement of this score system, but it is a useful further indication for allopatric birds like these. On the basis of all the above, my own view, and that of all our team who work on the Colombian checklist, is that these molecular and vocal data tip the balance in favour of treating purpurata separately under a conservative BSC approach. We appreciate that really hardcore Petersian lumpers could legitimately take another view given monophyly and failure to meet the toughest statistical tests of diagnosability.
We also discussed the conservation status of purpurata. Although not relevant to taxonomic determinations, it is perhaps of note that (whether lumped or split), these are both threatened forest-dependent birds, VU when lumped based on the recent papers on Amazonian deforestation rates; and with G. purpurata recommended in our paper for EN treatment if split. This decision is, therefore, not solely of relevance to listers or bird checklist accuracy fanatics.
Vernacular names: Quoting from our paper: "Hellmayr & Conover (1942) used the name Purple Quail-Dove, which is appropriate given that this is the most purple Geotrygon. However, this name seems to have been overlooked in the recent literature in favour of Indigo-crowned (e.g. Ridgely & Greenfield 2001, Restall et al. 2006)." "Purple" for purpurata is a nice name and transliteration, which mirrors "Sapphire". Perhaps a separate proposal on that issue is needed if this passes? We would suggest retaining saphirina's existing vernacular name (Sapphire) given its appropriateness in light of plumage and the scientific name, relative sizes of distributions (this name would still apply to by far the greater part of the old combined range) and history of usage in other sources.
Brumfield, R. T. & Capparella, A. P. 1996. Historical divergence of birds in north-western South America: a molecular perspective on the role of vicariant events. Evolution 50(4): 1607-1624.
Donegan, T.M. & Salaman, P.G.W. 2012. Vocal differentiation and conservation of Indigo-crowned Quail-Dove Geotrygon purpurata. Conservación Colombiana 17: 15-19. http://www.proaves.org/proaves/images/RCC/Con_Col_17_15-19_Geotrygon.pdf
Johnson, K.P. & Weckstein, J. 2011. The Central American land bridge as an engine of diversification in new world doves. Journal of Biogeography 38: 1069-1076.
Other papers mentioned are cited in the above.
Thomas Donegan, November 2012
Comments from Robbins: “YES, given the plumage and genetic (especially when compared to other recognized Geotrygon species) data. The relatively minor differences in vocalizations do not concern me, as most Geotrygon primary vocalizations sound very similar.”
Comments from Stiles: “YES, given the genetic and biogeographical information; the vocal data are also suggestive, if not wholly conclusive.”
Comments from Pacheco: “YES, from the new genetic and vocal data.”
Comments from Zimmer: “YES. We now have some genetic data plus some more suggestive (if not 100% conclusive) vocal data to hang our hats on – something that was lacking when voting on the earlier proposal.”
Comments from Pérez-Emán: “YES. I think genetic and vocal data, when analyzed within the genus context, tip the evidence in favor of recognizing these two species. It also makes sense biogeographically.”
Comments from Remsen: YES, barely. Using comparative genetic distance as a yardstick for species ranking is inherently flawed, despite its seductive appeal. Subjectively, however, plumage and vocal differences between the taxa are consistent with the slight differences in these features between close relatives in this lineage of doves (which includes Zenaida and Leptotila). Finally, these were treated as separate species by Hellmayr & Conover (1942), and I convincing rationale for their merger was never presented as far as I know.”
“Concerning English names, I favor Donegan’s proposed names, i.e. Purple and Sapphire, for the reasons he mentioned. This is not the case of a new split requiring two new names – this is a “re-split” of two taxa treated as separate species previously, when ‘Sapphire Quail-Dove’ was restricted to G. saphirina. Proposals for alternatives are welcomed as always.”
Comments from Nores: “YES, barely. Plumage differences are not so noticeable and neither are the genetic differences.”