Proposal (421) to South American Classification Committee


Split (A) Grallaricula cumanensis and (B) G. kukenamensis from G. nana


Effect of Proposal: A Yes vote on both parts of this proposal would result in a split of Grallaricula nana into three species.


Donegan (2008) studied plumage, biometrics and voice of Grallaricula nana in connection with the description of two new Andean subspecies in the group.  Given that such a wealth of data became available in the course of this project, species limits were also considered.  The methods for determining species limits are similar to those applied by Isler et al. to assess antbird species limits.  Taxa were recommended for species rank where vocal, biometric and plumage differences were equivalent to those between sympatric Grallaricula species.  (A study of vocal differences between a sample of various sympatric or parapatric Grallaricula species pairs showed at least three diagnostic differences in song, in the studied variables, to exist.)  The split species would be (i) G. nana, a widespread tropical Andean species; (ii) G. cumanensis of the Sucre peninsula and Paria Mountains of Venezuela; (iii) G. kukenamensis of the tepuis.  These three taxonomic groups postulated for species rank are each separated by 200-350 km of unsuitable lowland habitat and are restricted to well-known endemic regions.  The distances between these populations are likely to prove substantial for terrestrial, montane birds with poor flight such as Grallaricula.

G. cumanensis ("Sucre Antpitta") of the Paria and Caripe mountains of Venezuela (including subspecies pariae) differs vocally from G. nana (subspecies: nana, occidentalis, hallsi subsp. nov., nanitaea subsp. nov. and olivascens) in both its song and call.  The differences are equivalent to, and, in the case of call, greater than, differences between sympatric or parapatric Grallaricula species.  Three such differences were shown between each of cumanensis group, nana group and kukenamensis.  The call of the cumanensis group is also very different from other G. nana.  Bob Planque in an online article ( recently considered the call of this taxon to be unique, not only within G. nana but also within the genus Grallaricula.  G. cumanensis also shows substantial differences in biometrics and bill morphology from all other members of the G. nana group.  Ridgely & Tudor (1994) had previously noted that this population likely should be treated as a different species based on its call, song and plumage.

The biometrics and plumage of the disjunct population G. kukenamensis ("Guianan Antpitta") of the tepuis are diagnosably distinct, as for G. cumanensis.  Turning to voice, a single recording “80%” identified as G. kukenamensis (Boesman 2003) was the only one available.  A number of regional experts were consulted on this recording and everyone seemed to agree that there was nothing else it could be.  This recording differs from G. nana and G. cumanensis to the same extent as sympatric or parapatric Grallaricula species, being much longer in length and having a note-shape more like Hooded Antpitta G. cucullata.  Again, three diagnostic differences in song were found both with respect to the cumanensis group and nana group, applying various highly conservative assumptions about the available recording.  Further details of the statistical tests involved are set out in Donegan (2008).  Further personal communications suggest that this population has a further, distinctive call but recordings of this other call were not available for the study.  Hopefully, those recordings can be published in due course but that should not prevent a decision being made available based on the available data.  In any event, the treatment of kukenamensis as a species is supported by diagnostic biometric differences, based on a good sample of specimens for all taxa, as well as notable differences in bill morphology (illustrated in Donegan 2008) and plumage.

For committee members who are strongly wed to status quo treatments and Peters or who otherwise typically reject my proposals for whatever reason, one could at a stretch reject this proposal on the basis that a better and more certain vocal sample for kukenamensis is needed.  Such an approach would in my view represent a missed opportunity.  Additional vocal data for this particular population would be welcome.  However, the split of kukenamensis is strongly supported by other data in addition to voice.  Furthermore, if cumanensis is split, a step which is strongly supported, then it would be difficult to recommend which of nana or cumanensis the taxon kukenamensis should be placed in, given that it is different from both in biometrics (more similar to cumanensis), bill morphology (not closer to either), plumage (more similar to nana), as well as in voice (apparently not closer to either) and it is the most geographically isolated of all three proposed species groups.  One might also argue that molecular data would be helpful, which is also true.  However, the three populations postulated for species rank are all fully diagnosable by measures from multiple sources (plumage, voice, biometrics, bill morphology) so must be genetically different in some way, at least mutually monophyletic on some of the genes that code for plumage, voice, biometrics and bill morphology, one supposes.  SACC has recently accepted splits in other groups without molecular data, e.g. the recent Momotus proposal.  A discussion of other approaches is set out in Donegan (2008) but none of them are recommended.

The differences between each of the three G. nana groups, together with full data on biometrics and vocal variables, are set out in some gory detail, subspecies by subspecies, in the appendices to the paper.  (Please note that an erratum was published in the next issue of Bull BOC to correct errors in the legends for one of the recordings and one of the specimen images in the paper.)

The proposed three-way split of G. nana is itself a conservative approach.  The "G. nana group" (after cumanensis and kukenamensis are split) itself comprises at least 4 phylogenetic species (fully diagnosable, allopatric populations with small differences).  Authorities that adopt other species concepts might split wish to split it further.  However, such approaches were not proposed by Donegan (2008) nor are they proposed here, because differences between phylogenetic species in the nana group (as redefined) are not equivalent to those between sympatric Grallaricula species.

English Names:

The names "Sucre Antpitta" for G. cumanensis and "Guianan Antpitta" for G. kukenamensis were proposed.  “Paria Antpitta” was previously suggested by Ridgely & Tudor (1994) for cumanensis, but this appears to be based on the subspecies name pariae, which is junior, and the cumanensis group has a wider range than the Paria peninsula, being near-endemic to Sucre department.  This proposal is provisionally to adopt the names suggested by Donegan (2008).  However, if this proposal passes, I will raise a separate proposal on the English name for cumanensis.


Donegan, TM. 2008. Geographical variation in Slate-crowned Antpitta Grallaricula nana, with descriptions of two subspecies, from Colombia and Venezuela. Bull Brit. Orn. Club. 128(3): 150-178.

Other references cited therein.

I have also added a discussion of G. nana recordings on xeno-canto which does not include sonograms of recordings for all taxa but which may be helpful to committee members:

Recommendation: A Yes vote, to split this species into three for the reasons set out above.  To accommodate potential differences in voting on the cumanensis group and on kukenamensis, consider the former to be part A and the latter part B of this proposal.

Thomas Donegan, March 2010.



Comments from Robbins: “A = YES.  I do support recognizing cumanensis/pariae as a species separate from other nana based on the relatively pronounced vocal and plumage differences.

B = NO.  Given the lack of definitive song and whether described calls are truly analogous (hopefully the reference in Donegan [2008] to an upcoming paper by D. Ascanio will clarify this), it would be premature to elevate kukenamensis to species level.  Obviously, genetic data would further clarify relationships in this group.”


Comments from Remsen:  “A = YES.  Evidence for splitting cumanensis is strong. 

B = NO.  I would hold off on kukenamensis until comparable vocal data are obtained.  That kukenamensis is diagnosable in terms of plumage and morphology to the same degree as cumanensis is from nana is suggestive.  It’s not a lot to ask to wait for a few more recordings of unequivocal kukenamensis.  This is not a “stretch,” contra the proposal, but a logical position given the disparity in data quality.


“The proposal portrays a NO vote as indicating ‘strongly wed to status quo treatments and Peters or who otherwise typically reject my proposals for whatever reason.’  First, I suspect there is no group of 10 people more aware of the shortcomings of the status quo, whether Peters or otherwise, than those on SACC, all of whom are keenly aware that status quo species limits in Neotropical birds badly under-represent true species-level diversity under a BSC framework.  If it were up to us as individuals to realign current taxonomy to reflect what we know or suspect is the case based on vocal differences without publishing the documentation, then we could add hundreds of species overnight.  It is often frustrating and certainly time-consuming to have to go through the process of publication.  Nonetheless, rather than being wedded to status quo, what we ARE “wedded” to is a process of change than involves a degree of rigor that was missing from the extinction-by-penstroke approach to species limits by Peters, Hellmayr, and others, namely publishing data and explicit rationale for change.  To do otherwise only repeats the philosophical mistakes of the Peters era by perpetuating low standards for change.  I’m sure Peters and his soul mates were just as convinced that their decisions were correct as we are convinced that our own, unpublished views are correct.  As for Donegan’s implication that votes against some of his proposals are rejected ad hominem, that 27 have passed suggests that this perception is unwarranted; and most of those that have failed are proposals on English names, hyphens, or status of introduced species that represent differences in taste or philosophy.”


Additional comments from Donegan: “Those who have voted have an interesting concept of the scientific rigour of different publications.  Peters and Hellmayr did an excellent job in formulating a rational and consistent taxonomy for a large number of species based on the data available to them.  However, it is worth comparing their material and methods with those adopted here and in other recent studies.  Peters, Hellmayr and others of their age based their taxonomies on studies of a few collections they were able to visit and on hand-written notes of specimens in different museums (possibly aided by black-and-white photography).  Their compendia treated large numbers of taxa, so attention to any one species was probably limited to a few days at most, probably more often a few hours.  Only a handful of specimens were studied by Hellmayr (1917) in the case of Grallaricula nana, many of which had vague locality data (e.g. “Bogota”, “East Ecuador”).  There many substantial distributional gaps in the specimen record compared to what we have now; and no data on voice was available.

With this group in particular, Hellmayr essentially concluded that the material available to him was inadequate to assess the status of many populations, such as those in the Venezuelan Andes (see passages quoted in Donegan 2008).  Peters did not have much to say about this group and simply followed earlier treatments.  Since Hellmayr (1917), five subspecies, as well as a host of major range extensions for described populations, have been discovered / described.  In Donegan (2008), I measured over 150 G. nana specimens, the vast majority of which had good locality data.  I collated data on approximately a hundred more specimens (through inspecting them or photographs), studied 224 sound recordings and applied a series of statistical tests to analyse the diagnosability of all known populations based on 6 biometric, 8 vocal and various plumage variables as well as bill morphology against various postulates tests for species and subspecies rank.  Some fieldwork was carried out in Colombia.  GIS analysis was carried out to map distributions of each subspecies based on these data and data from other sources (photographs, sound recordings, other specimens).  The study was pretty exhaustive of available (non-molecular) resources and took 3 and a half years to complete from its inception to publication.  I fail to see where the scientific rigour is in rejecting the conclusions of this in-depth study in favour of a treatment suggested almost 100 years ago, based on a small fraction of the material available today and only some of the subspecies known today; or in favour of the recommendation of a footnote by Ridgley in a field guide. 

A short note on the other calls of kukenamensis would be helpful, but based on personal communications, such a note would seem likely only to support the taxonomic recommendations in Donegan (2008), which were reached independently.  It would also be nice to have a molecular study, but I don’t see that being used as a reason to reject other proposals based on studies using traditional methods.

It’s moderately comforting that there have been positive comments on splitting cumanensis.  Ridgley & Tudor (1994) thought that should be split too, so adopting such an approach does not require a view to be taken on the merits of Donegan (2008).  I have asked Van Remsen to split this proposal into part A to split cumanensis; with Part B to split kukenamensis.  The first option would be somewhat a case of two steps forward, one step back and result in a species with a bizarre, possibly unique, distribution that may be paraphyletic.  However, the resulting treatment is likely to be ‘less paraphyletic’ and less offensive to the BSC than the current treatment and, as a result, it would be an improvement compared to doing nothing.

(A PDF of the paper has been posted (unauthorised) on line.  If anyone wants to see a PDF of the short erratum, please let me know.)


Response from Remsen:  It doesn’t matter how many specimens you’ve measured, what stats were used, etc., in terms of determining the taxon rank of kukenamensis if those data do not include the data critical to assigning species limits.  That kukenamensis is a diagnosable unit was never disputed as far as I know.  Donegan’s data show conclusively, much more so than Hellmayr-Peters, that kukenamensis is a valid taxonomic unit, i.e., should be ranked at least as a subspecies under BSC.  [A paper I have in press proposes that properly described subspecies in the BSC are equivalent to PSC species.]   Your best argument for species rank for kukenamensis is that including it in nana might make that species paraphyletic with respect to cumanensis, but we won’t know that for use until we have genetic data.  As for species rank for kukenamensis, the proposal sates: “Turning to voice, a single recording “80%” identified as G. kukenamensis (Boesman 2003) was the only one available.”  Donegan, I’m sure, would support vocal comparisons as the best way rank allopatric populations in terms of species vs. subspecies.  So, how then can Donegan consider it rigorous to change the rank of kukenamensis based on one recording of uncertain identification?????

            “As for molecular data and whether this is required to make such a decision, many are under the spell of the widespread misconception that molecular data “solve” ranking of allopatric populations; in other words, all we might need to assign species rank is sequence data on a couple of genes to “measure” whether it’s a species or subspecies.  This is incorrect.  Genetic distance among known good species is as low as 0% (Galapagos finches), whereas genetic distances among populations that do not differ in any phenotypic way ranges up to 11% (Ben Marks dissertation).  Even within a closely related group of birds, say the nana complex, genetic distance probably is the best predictor of time-since-isolation, but it is also influenced by things like effective population size (all else being equal a small population would differentiate more rapidly than a large one); further, the genes used are explicitly assumed to be neutral, i.e., unrelated to something relevant to reproductive isolation.”

            “As for Donegan’s point that not ranking kukenamensis as a species retaining kukenamensis in nana until we have more data would result in a species with a bizarre, possibly unique, distribution that may be paraphyletic’, this evidently refers to a distribution that would place a tepui bird with nana is in the Venezuelan Andes rather than the geographically closer cumanensis in the Coast Range.  However, note that mostly Andean Doryfera johannae has a tepui subspecies without even a representative in the Venezuelan Andes much less Coastal Range.  At least G. nana is present in the Venezuelan Andes (nominate nana).  Similarly, Aulacorhynchus derbianus has two endemic subspecies in the tepuis but has no representative in either the Venezuelan Andes or Coastal Range.  Macroagelaius has a similar distribution:  Colombian Andes + tepuis.  The point is that tepui species’ closest relatives do not necessarily occur in the Coastal Range or even the Venezuelan Andes.  So, the distribution pattern created by including kukenamensis in nana is not bizarre.  Whether it is unique or not would depend on having sufficient N of species with populations in all three places (Andes, Coastal Range, and Tepuis) and finding that in no other case was the Andean taxon sister to the Tepui taxon.”

Comments from Stiles:  YES for recognizing cumanensis and for calling it “Sucre Antpitta”.  For now, I vote NO on recognizing kukemanensis, pending a better analysis of vocalizations: that given for the Pseudocolopteryx citreola proposal should serve as a model for such cases.  It seems a bit unrealistic to call this a “missed opportunity” when the evidence is incomplete – it’s really a “golden opportunity” for someone to do it right, now that attention has been called to this case.”


Comments from Nores:


“A = YES.  Yo estoy de acuerdo en reconocer cumanensis/pariae como una especie separada de nana, ya que tiene importantes diferencias en coloración y vocalizaciones.

“B = NO. No me parece apropiado elevar a kukenamensis a nivel de especie ya que las diferencias en vocalización sugeridas no son lo suficientemente seguras como para tomar esta decisión. Generalmente las subespecies tienen diferencias biométricas con la típica y no es razón para considerarlas especies. Prefiero esperar hasta que haya grabaciones seguras de kukenamensis o análisis moleculares que apoyen esta separación.”


Comments from Stotz:


“A = YES.

“B = NO. I believe it likely that kukenamensis is a good biological species, but the equivocal vocal evidence makes it hard for me to vote to recognize it at this point.  What does it mean that a recording is 80% certain that it belongs to a particular taxon?  The plumage resemblance for kukenamensis to nana makes me comfortable leaving it in nana for the time being.”


Comments from Cadena:


“A = YES

“B = NO, for reasons clearly stated by Remsen. Donegan's paper is a nice piece involving careful study of museum specimens, some of which were collected by the author himself.”