Proposal (945) to South American Classification Committee

Split Piranga flava complex into multiple species

Effect on SACC classification: This would split our Piranga flava into two or more species.

Background: This is a well-known problem in species limits that has been dealt with differently by different authors for at least 120 years. Our current note reads as follows:

Zimmer (1929) was the first to treat all members of the P. flava group as a single species, and this has been followed by most subsequent authors, although AOU (1983, 1998) recognized three subspecies groups, and Isler & Isler (1987) suggested that each might be better treated as separate species. Meyer de Schauensee (1966) and Ridgely & Tudor (1989) also proposed that this species probably consists of two or three separate species. Two of these occur in South America: nominate flava of southern and eastern South America, and the lutea group of the Andes region (and also Panama and Costa Rica). See Zimmer (1929) concerning earlier claims of sympatry between flava and lutea. Burns (1998) proposed that the three subspecies groups should be treated as three phylogenetic species, and possibly biological species, based on comparative genetic distance data within Piranga. Ridgely & Greenfield (2001) and Hilty (2011) treated the three groups as separate species. Haverschmidt and Mees (1994) treated the subspecies haemalea of the Tepuis as a separate species from P. flava based on habitat differences. Manthey et al. (2016) found that haemalea was not part of the lutea group. SACC proposal needed.

Ridgway (1902) treated the complex as 2 species, as follows:

1. Piranga hepatica (Hepatic Tanager): SW USA to Guatemala

2. Piranga testacea: (Brick-red Tanager: Nicaragua to Bolivia (also with pine-lands subspecies “Belize Tanager” P. t. figlina from Guatemala to Honduras)

Although many of the Neotropical subspecies were not yet described, haemalea of the Tepui region was described in 1883, but was not mentioned by Ridgway, nor was any member of the even older lowland flava=azarae group of south-central South America. This might suggest that Ridgway did not consider them part of this group.

Zimmer’s (1929) 50+ page monograph on the group is the basis of our current classification. However, this monograph was written in an era when vocalizations were not taken into account – all taxonomy was based on external morphology combined with distributional considerations (e.g., sympatry/parapatry vs. allopatry). Here is Zimmer’s synopsis, and you can see that similarities between the extreme northern and southern taxa strongly influenced his reasoning for treating them all as conspecific:

“Examination of numerous specimens of these forms and certain of their unquestioned allies has fostered the belief that all are races of a single species whose distribution extends from eastern Argentina to southwestern United States with little interruption in continuity, though with lateral extensions into Brazil and the Guianas and into Venezuela and the Guianas in two lines of development which meet at their outward extremities. Throughout this extensive group, the color, general pattern, size, shape of bill, and other major characters are substantially identical or are subject to variability which largely overcomes the individual differences. There are certain features of plumage and molt which seem to be present in all the forms under consideration but which are different from those of the other congeneric groups. The various forms replace each other geographically in all parts of the range. Finally, at the northern and southern extremities forms are produced which are strikingly alike in racial characters that are not shared by the intervening subspecies.

These considerations together present a volume of evidence that is more than circumstantial. There is no question that all the forms under discussion are of common phylogenetic origin. Some of them are more strongly differentiated than others, some distinctly intergrade with adjacent forms, while others are separated from their nearest allies by so slight a gap in proportion to the individual variation in that direction that the relationship is not seriously impaired. In the following treatment, therefore, I have considered as races of P. flava all the forms under discussion”.

Zimmer’s study was perhaps the most thorough study of a single widespread Neotropical bird of that era. So, this situation is about as different from an unjustified Peters lump as you can get.

Subsequent classifications all followed Zimmer on this, including Hellmayr, Storer in Peters, Sibley & Monroe, AOU, Dickinson & Christidis, etc., except for those mentioned in the SACC note above and the recent HBW/BLI classification.

Isler & Isler (1987) summarized the qualitative differences in songs and calls for the three groups, as well as the rather exceptional range of habitats for a single passerine species (e.g. from arid pinyon-juniper scrub to edges of cloud-forest). The habitat differences were remarked upon and noted as exceptional by Zimmer himself. Ridgely & Tudor (1989) proposed that at least 2 and maybe 3 species were involved.

Here is the breakdown from Isler & Isler (1987), with distribution refinements from Dickinson & Christidis (2014). Because this is a complex proposal, I suspect these details might come in handy

(A) hepatica group (extralimital)

1.nominate hepatica: highlands of SE California, Arizona, and w. New Mexico S in western Mexico to Oaxaca

2. dextra: highlands of e. New Mexico [presumably this species also in Las Animas Co., CO] and SW Texas south on Caribbean slope to Chiapas

3. albifacies: in highlands of w. Guatemala to n. Nicaragua

4. figlina: lowland pine savannahs of Belize and e. Guatemala

5. savannarum: lowland pine savannahs of extreme e. Honduras and NE Nicaragua

(B) lutea group

6. testacea: highlands of n. Costa Rica to e. Panama (Darién)

7. faceta: Santa Marta to highlands of Venezuela; Trinidad

8. haemalea: tepui region from c. Venezuela, n. Brazil, c. Guyana, c. Suriname

9. toddi: two spots in n. Andes of Colombia

10. desidiosa: W. Andes of Colombia (Antioquia to Cauca)

11. nominate lutea: Andes from Nariño S through Ecuador and Peru to c. Bolivia (Cochabamba)

(C) flava group

12. macconnelli: lowlands of n. Brazil (Roraima) and the southern Guianas.

13. saira: lowlands of e. Brazil from Amapá south (patchily) to Mato Grosso and Rio Grande do Sul

14. rosacea: lowlands of se. Bolivia e. Santa Cruz)

15. nominate flava: foothills of E Bolivia from Cochabamba and w. Santa Cruz, S in lowlands to Paraguay, n. Argentina and Uruguay [note that if the Bolivian records pertain to breeding birds, then this is not strictly a lowland taxon; however, because this is an austral migrant, I would not be surprised if these foothill records are wintering birds only – this needs to be sorted out. ]

Here is the HBW plate by H. Burn that illustrates 5 of the subspecies:

Burns (1998) proposed that the three subspecies group be treated as separate phylogenetic and perhaps biological species based on comparative genetic distance data (cyt-b); however, this was based on just seven specimens, only one from the lutea group. Burns concluded: “The DNA data of this study add to the morphological, distributional, and ecological evidence that suggest that the three subspecies groups of P. flava represent different phylogenetic, if not biological species.”

New information (since the original SACC classification): There really isn’t much in the way of new quantitative data.

Ridgely and Greenfield (2001; Ecuador book) cited Burns (1998) for their treatment of Andean lutea group as a separate species (Highland Hepatic-Tanager) from lowland flava group (Lowland Hepatic-Tanager) as well as the northern hepatica group (Northern Hepatic-Tanager), as did Hilty (2003; Birds of Venezuela) and Restall et al. (2006; Birds of Northern South America. Vol. 1); however, note that Burns was rightfully hesitant in calling them biological species.

Manthey et al. (2016) used Piranga to compare different genetic techniques (UCEs vs. RAD-seq). They used only 6 individuals of the flava complex, two from each subspecies group. Although taxon-sampling was weak, they corroborated what had been concluded by Zimmer nearly a century earlier, i.e. that the three groups formed a monophyletic unit. However, they found that lutea, was paraphyletic: the Andean sample (N. Peru, ergo nominate lutea) was sister to the two samples of hepatica (from El Salvador, ergo albifacies), but the sample from Guyana (haemalea) was sister to the two samples of lowland flava (one from Guyana, ergo presumably macconnelli and one from e. Bolivia, ergo likely rosacea), both with strong support.

A problem here is that both KU samples from Guyana are listed as from the same locality (“Upper Takutu”), which is a lowland savanna locality (fide M. Robbins). For a second, I thought I’d found evidence of sympatry of lutea and flava groups, and thus automatic species rank for both.

But because Manthey et al. didn’t point that out, I suspected an error in their table. So I checked, and sure enough, SACC’s Mark Robbins, who collected both specimens (and who obviously was not asked to go over the paper), told me that this turns out to be an undetected copy-paste error. From Mark:

“90809: 1.5 km S of Karaudanawa, in the upper Rupununi, i.e., lowland site; ID as macconnelli.

“89864: Acai Mts, in the extreme south [of Guyana]! ID as haemalea”

Nonetheless, it still remains that, as Manthey et al. pointed out, the lutea group as currently constructed is not monophyletic. The tepui subspecies haemalea groups with the lowland flava group even though in plumage and elevation, it fits with the Andean lutea group. Therefore, either the haemalea group should be transferred to the flava group, which would not make sense from plumage, or be treated as its own group. Given that Haverschmidt and Mees (1994) treated haemalea as a separate species, and that haemalea does not fit with flava in plumage or elevation, this should be at least treated as a separate, fourth group in analyses. Finally, that Robbins’ two specimens are only about 150 km from each other, but have such strong genetic and phenotypic differences, the evidence for treating haemalea as a separate species from flava, much less lutea, ironically could be considered stronger than that for any of the other proposed splits even though it’s never been recognized as a separate group.

Time for a mini-rant. The above is yet another vivid demonstration of the importance of broad taxon-sampling. In terms of genetic sampling, one cannot just assume that a group of subspecies are a monophyletic group unless all the taxa are sampled. For all we know, the isolated lutea subspecies in the north group with haemalea rather than nominate lutea, or form their own group, or each forms a separate group, or testacea actually groups with hepatica, or …. .

With respect to Manthey et al., with such limited taxon-sampling (only 5 of 15 subspecies represented), I find it hard to extract any firm evidence for species rank either way. I personally don’t think the use of comparative branch lengths and genetic distance can be used as a metric for species limits, but even so the longest branch between the two clusters of Summer Tanager (P. rubra) samples is longer than any between any of the Hepatic Tanager groups despite the enormously greater geographic distances among samples of the latter.

By the way, Manthey et al. is cited as evidence for species rank by HBW/BLI and “IOC”.

As for vocalizations, Boesman (2016) presented sonograms of songs and calls for each. The N for song sonograms was 3 for flava group , 3 for lutea group, and 2 for flava group, with location/subspecies not specified. The N for calls was 4 hepatica, 3 for lutea, and 5 for flava, also with location/subspecies not specified except for one for testacea in the lutea group, which appears to be the most different of all.

Here’s what my impressions are from the sonograms. In terms of song, I see more differences within the hepatica and lutea groups as I do between them. As for flava, the notes themselves do seem on average more complex, as noted by Boesman: “Song of flava group seems to have the most complex-shaped notes, lacking any simpler-shaped notes. Many notes are complex underslurred, and apparently there is little variation in note shape (few different note shapes), unlike other races.” As for call notes, those of hepatica and flava look very similar to me, but those of flava look more complex, as noted by Boesman: “Call of flava group is also clearly different, having an upslurred ending, while both other groups are about identical, having a very sharp upturned V-shape.“

I suspect Peter Boesman would be the first to tell you that this sort of sampling and qualitative comparisons has its problems. I will also repeat the same mini-rant that on taxon-sampling that I did on the genetic data. Boesman’s studies of this and many other groups are valuable for pointing out potential issues (note his strange recording of testacea mentioned above) and can serve as launching pads for more thorough studies. But until all the taxa are sampled, with sufficient N and careful attention to homology, their use as determinants of species limits is perilous. I think we should be grateful that he has set the table for the more detailed analyses needed to really sort things out. Lots of potential here, but also lots of sampling gaps in terms of subspecies and geography. One problem that could be fixed quickly is that the location and subspecies need to be given for each recording; for example, three lutea songs are represented, each looking fairly different. But “which” lutea? They could be from testacea from Costa Rica, nominate lutea from Bolivia, or any of the other four subspecies in the group, including even haemalea from the tepui region, which we now know is not a member of the group. Or they could all three be from different individuals within any one of the 6 subspecies.

Just to give you an idea what the call notes are like, I grabbed a few links:

• A call note from nominate hepatica from AZ (Richard Webster): https://xeno-canto.org/678668

• A call note from testacea from Panama (Peter Boesman): https://xeno-canto.org/271527

• a call note from nominate lutea from Peru (Fabrice Schmitt): https://xeno-canto.org/102734

• A call note from nominate flava from Paraguay (Fabrice Schmitt): https://xeno-canto.org/616311

• A call note from haemalea from Guyana (Ted Parker): https://search.macaulaylibrary.org/catalog?taxonCode=heptan&tag=call&regionCode=GY

What stands out to me from superficial browsing of these and others on xeno-canto is (1) as Boesman (2016) noted about testacea, that twittered, crossbill-like call seems to be standard, and if that is the case, then testacea has by far the most distinctive call; (2) hepatica, lutea, and haemalea calls sound fairly similar; (3) flava calls are higher-pitched and slightly inflected, but in the background of the Paraguay recording, I think I hear another bird giving a lower-pitched call more like northern birds. Obviously, a quantitative analysis of all of the is needed, with larger N of presumed homologous calls. My gut impression is that when all the data are analyzed, they will support testacea as a separate species, with the obvious caveat that this is just from a brief perusal of what is easily available.

Piranga songs are fairly complex, and so the final analysis will not be easy. Note that the vocal pattern does not fit the plumage pattern, e.g. lutea and hepatica are evidently the most similar vocally, whereas flava and hepatica are most similar plumage-wise (which is at the core of Zimmer’s rational for conspecificity

Discussion and Recommendation:

This issue is a real problem. On the one hand, I share the instinctive feeling of just about everyone who has commented on the group post-Zimmer that more than one species is involved, but I would bump it up to at least 5 potential species If this really is to be treated as one biological species, then the breadth of habitat types covered would be exceptional, ranging from arid, rocky, pinyon-juniper slopes in the north to pine savanna to the margins of cloud forest and semi-humid montane forest to tropical dry forest; the only common denominator might be that the structure of the habitat is open woodland and edge. (House Wren and Squirrel Cuckoo are the only species of woodland birds that I can think of offhand that could compete with Hepatic Tanager for habitat breadth.) But it’s hard to find conclusive evidence for treating them as separate species. The plumages differ, but not outside the range of variation for many polytypic species. The genetic data are too weak to be interpreted either way, especially with the sampling gaps. The vocal differences, even qualitatively, between the flava group vs. the other two suggest multiple separate species, but are the data officially strong enough for a split?

This complex is begging for dissertation-level research. Geographic areas in need of careful sampling are:

(1) the potential contact areas between haemalea of the tepuis and macconnelli of the lowlands. They come pretty close in Guyana and Suriname, but are likely separated by unsuitable habitat: tall tropical forest. If there is no sign of gene flow, then that’s all you need to argue for species rank in my opinion, even if not precisely parapatric.

(2) central Bolivia, where it is unclear how close lutea and breeding flava come to each other. The sampling in our Bolivia book (Herzog et al. 2016) shows a continuous distribution from the Andes of La Paz and Cochabamba through the foothills of Santa Cruz and Chuquisaca to the lowlands of Santa Cruz and Tarija. If these refer to resident populations, then there is a contact zone somewhere in there, as Hellmayr (1929) noted Specimens from w. Santa Cruz and Chuquisaca are all assigned to flava despite their montane distribution. But we also noted that flava is a partial austral migrant. A first pass through the dates, elevations, and subspecies identification would likely clear much of this up (but I’m out of time/energy to do that within a SACC proposal). Hellmayr (1929) did not pick up on the possibility of austral migrants messing up the distribution. Follow-up fieldwork in the region might be highly productive. This is the region where lowland subspecies of Thamnophilus caerulescens meet Andean subspecies, with connecting populations with intermediate phenotypes and genotypes (Brumfield papers), and where the same thing appears to be happening in other taxa that have similar distributions (e.g. Pyriglena leuconota).

The sampling and analysis of songs and calls needs to be done rigorously, with all taxa sampled. The Piranga I know, including northern Hepatic, respond vigorously to playback, so careful playback experiments might be illuminating. Eyal Shy’s dissertation on North American Piranga songs needs to be read for guidance, as well as his several subsequently published papers on geographic variation in song within Summer (P. rubra) and Scarlet (P. olivacea) tanagers. The complexity of the structure of Piranga songs requires quantitative analyses of their differences with a large sample size.

I went into this issue fairly confident that I would find sufficient anecdotal evidence that in aggregate would make a case for elevating two or three of these subspecies groups to species rank, but was unable to do so. The deeper I went, the more complexity was unveiled. Although I am certain that once all the data are available, we will have evidence for multiple species in the group, I would be extremely reluctant to change current taxonomy without having a firm foundation..

Let’s break down the voting on this proposal as follows:

A YES vote means you are in favor of splitting up Piranga flava into 2, 3, 4 (or more?) species, with the precise breakdown to be determined in subsequent voting round. Note that any vote on extralimital hepatica would be strictly advisory to NACC.

A NO vote means leave as is for now.

I recommend a NO vote. This is a juicy project waiting for a thorough analysis. I’m not opposed to piecemeal taxonomy, but I really can’t find any convincing evidence for any of the splits, as outlined in the details. I see no immediate rush to resolve this one using fragmentary, unsatisfactory evidence. Our current taxonomy certainly masks species-level diversity, but any changes to it could be considered just as misleading if not fortified by data. Tough decision.

English names: Many classifications that recognize 3 species use Northern Hepatic-Tanager, Highland Hepatic-Tanager, and Lowland Hepatic-Tanager for the three groups. They have had traction in the literature since Isler and Isler (1987). But they strike me as particularly ugly, for some reason. If this proposal passes, then I recommend a separate proposal on English names, which would also give us time for other choices to emerge. For example, I would consider retaining Hepatic Tanager for the northern group, contrary to our usual policy for parent-daughter splits, and using Tooth-billed and Red for the other two groups, just because they have historical precedent, albeit inconsistent (and not because they are particularly appropriate), and this is the way IOC/xeno-canto do it.

Literature (partial)

BOESMAN, P. Notes on the vocalizations of Northern Hepatic-tanager (Piranga hepatica), Highland Hepatic-tanager (Piranga lutea) and Lowland Hepatic-tanager (Piranga flava). https://birdsoftheworld.org/bow/ornith-notes/JN100386

BURNS, K. J. 1998. Molecular phylogenetics of the genus Piranga: implications for biogeography and the evolution of morphology and behavior. Auk 115: 621–634.

HERZOG, S. K, R. S. TERRILL. A. E. JAHN, J. V. REMSEN, J V, O. Z. MAILLARD, V. H.GARCÍA-SOLÍZ, R. MACLEOD, A. MACCORMICK, J. Q. VIDOZ, C. C. TOFTE, H. SLONGO, O. TINTAYA, M., AND J. FJELDSÅ, J. 2016. Birds of Bolivia : field guide. Asociación Armonía, Santa Cruz de la Sierra, Bolivia.

ISLER, M., AND P. ISLER. 1987. The Tanagers, Natural History, Distribution, and Identification. Smithsonian Institution, Washington, D.C.

MANTHEY, J.D., L. C. CAMPILLO, K. J. BURNS, AND R. G. MOYLE. 2016. Comparison of target-capture and restriction-site associated DNA sequencing for phylogenomics: a test in cardinalid tanagers (Aves, genus: Piranga). Systematic Biology 65: 640–650,

ZIMMER, J. T. 1929. A study of the Tooth-billed Red Tanager Piranga flava. Field Museum Natural History, Zoological Series 17 (5): 169-219.

Van Remsen, June 2022

_______________________________________________________________________________

Comments from Stiles: “ Definitely vote NO on this one: this is a particularly glaring example of the problems arising from insufficient taxon sampling, with apparently conflicting results from plumage, genetics, and vocalizations producing an undecipherable conundrum.”

Comments from Robbins: “NO. You have distilled a complex problem down to where one can at least begin to understand the issues across multiple lines of data. As you point out, much more data across the entire complex is needed before we can start elevating taxa to species level. At this point, I concur with your recommendation of a NO vote.”

Comments from Claramunt: “NO. The new data is just too sparse to be informative regarding species limits in this widespread and variable species.”

Comments from Areta: “NO. I think that Van´s recommendation is spot on: we know the currently taxonomy is defective and masking species-level differentiation within P. flava. But we lack solid data to make informed decisions as to how the different species should be circumscribed. The main problems are clear, but the data at hand falls short to solve them.”

Comments from Pacheco: “NO. I agree with Van's recommendation. The information available for the splits in Piranga – suggested by several sources – is insufficient. Genetic and vocal analyzes have a very low N and unsatisfactory geographic coverage.”

Comments from Bonaccorso: “NO. The geographic, taxonomic, and vocal sampling available is too sparse to make an informed decision.”

Comments from Zimmer: “NO, somewhat reluctantly, because I think there’s enough smoke swirling around this complex, that there has to be a fire someplace! My field experience has told me for decades now that there has to be more than one species of “Hepatic Tanager”, but, clearly, the evidence is just too fragmented, and the sampling (be it genetic, vocal or even morphological) is too weak, with too many gaps, so that basically, we don’t know what we don’t know. Complex problems usually defy simple fixes, and I think that’s the case here. As Van states, this complex is “begging for dissertation level research.” A couple of observations that I would just throw out there: 1) To me, testacea is very different vocally, particularly from nominate hepatica, both in terms of the unique quality of its common twittering call, and, in the relative frequency with which it delivers its various vocalizations, as compared to other taxa in the complex, again, particularly, when compared to nominate hepatica. I hear that latter taxon singing more than I hear it delivering its husky, single-noted “CHUCK” calls. With testacea, the case is a polar opposite, and I hear the unique calls far more frequently (and when delivered, given more repetitively, in rapid-fire sequence) than I hear them give the occasional song. To me, it’s somewhat analogous to situations we’ve commented on before, such as the Lophotriccus galeatus complex, or the Zimmerius gracilipes/acer complex, where populations on the N and S banks of the Amazon may not differ appreciably in vocalizations, but they do differ in syntax and the relative frequency with which different vocalizations are given. 2) With respect to the subtle plumage distinctions between the various subspecies-groups, and Zimmer’s quoted belief that individual variation swamped any perceived geographic structure to plumage variation: I don’t have access to the HBW plate that Van partially reproduced in the proposal, and that partial reproduction doesn’t include illustrations of extralimital (for SACC) nominate hepatica, but, it has always been my experience that NONE of the members of either the lutea-group (including haemalea) or the flava-group with which I am personally familiar approaches the distinctly dark-cheeked look of male & female nominate hepatica. That character, by itself, sets hepatica apart, at least morphologically, from all of the others, in my experience. These types of distinctions in facial patterns are often obscured by specimen preparation, causing them to be overlooked or downplayed by someone used to working with skins (as opposed to seeing the birds in life) and/or by illustrators relying on skins for reference material.”

Comments from Jaramillo: “YES – This proposal has already gone down, but I will go in here with a yes vote perhaps to encourage someone out there to take this on as a project? I think we all agree that there is almost certainly more than one species here, and likely multiple. The uncertainty is how do divide them up because the data are still lacking. But Yes, I do think there are multiple species, and I would be willing to chop this up in some way…even if piecemeal and incomplete. I don’t like to wait for the full story because that story sometimes never comes. But we do have part of the story.”