Redefine species limits in Pyrrhura picta and leucotis complexes

Proposal (#306) to South American Classification Committee

Redefine species limits in Pyrrhura picta and leucotis complexes

Effect on South American CL: this would recognize a number of new species within Pyrrhura picta and P. leucotis and would reverse an earlier decision (Proposal 181) to recognize P. griseipectus as distinct from P. leucotis.

Background: In proposal 181, SACC voted to recognize P. p. pfrimeri and P. griseipectus as species distinct from P. leucotis, based on the work of Joseph (2000, 2002) and Olmos et al (1998). These taxa are all allopatric and within eastern and central Brazil. The distantly disjunct populations in northern Venezuela (emma) were not treated by Olmos et al. directly, and not considered in proposal 181. At the time, Robbins called for holding off until a molecular treatment of the group was available. That treatment now exists (Ribas et al. 2006).

New information: Ribas et al (2006) did an extensive molecular phylogeny of picta and leucotis based on mitochondrial sequences from cytochrome b and the control region, totaling 2500+ base pairs. They did not have material from some taxa including luciani (west Amazonian Brazil), subandina, and caeruleiceps (both in Northern Andes of Colombia and Venezuela. Nonetheless, most taxa were sampled and clear results obtained. They found that the leucotis and picta complexes formed a monophyletic unit, but leucotis and picta, even as defined by SACC, were not reciprocally monophyletic. They identified seven lineages which they considered well-supported and morphologically distinct (their figure 2). These lineages would seem like the appropriate starting point for defining species taxa.

1 roseifrons (includes peruviana) - far-western Amazonia
2 amazonum (includes snethlageae) - southcentral and southeastern Amazonia
3 leucotis (includes griseipectus) - Atlantic forest of Brazil
4 pfrimeri - cerrado region of Brazil
5 eisenmanni - extralimital in Panama
6 picta - northeastern Amazonia
7 emma - northern Venezuela; presumably includes auricularis, a weakly defined subspecies

Where the taxa that were not sampled fit into this scheme is complicated. P. luciani is a taxon in south Amazonia, geographically situation between P. roseifrons and P. amazonum. Unfortunately it is the oldest name in this group. Joseph (2002) suggested treating it as specifically distinct based on a variety of small morphological characters. Because of the nomenclatural issues and the lack of any compelling reason to unite it with either roseifrons or amazonum, I am inclined to treat in as distinct in the short-term, until it is added to the molecular framework or other data supports associating it with one of the other taxa. The northern Andean taxa (caeruleiceps and subandina) presumably belong with eisenmanni, emma, and picta, which form a poorly supported clade. Morphologically they fit and geographically they fit. Because we have no clear idea how they would fit into this group, I see a couple of options: one is to recognize 1 species that includes eisenmanni, emma, picta, caeruleiceps and subandina, and wait for additional data that might potentially subdivide this group; or recognize 4 species with eisenmanni, emma, picta and caeruleiceps (including subandina; Joseph and Stockwell [2002] argued for splitting these) or recognize all 5 as distinct species.

Recommendation:
This is a very complicated situation. Several pieces need to be considered in turn:

A) Treat all taxa in the leucotis and picta complexes as subspecies of a single very complex species. This would put a lot of molecular, ecological and morphological diversity into a single species. I recommend a NO vote on this, although it would deal with the uncertainty that incomplete molecular sampling brings about. Assuming that we won't sweep everything under the rug, there are a long series of subdecisions that need to be made.

B) Sink griseipectus back into leucotis. They are closely related, similar in plumage, and more or less weakly differentiated ecologically. Olmos et al describd them as more distinct morphologically than are leucotis and pfrimeri, which personally I don't see. Depends on which characters you emphasize, I suppose. Given that they are not reciprocally monophyletic in the mitochondrial DNA analysis (not required of species but a nice thing to find), I recommend returning to the treatment of griseipectus as a subspecies of leucotis.

C) Remove emma from leucotis. It doesn't come close to leucotis in the tree, and has never made any sense biogeographically. I recommend this change.

D) Separate amazonum (including snethlageae) from picta. This seems straightforward, and I recommend this change. Picta seems to have its relationships with the northern taxa, not with the other Amazonian forms.
D1) If this change is made, then split amazonum and snethlageae. This would follow the treatment of Joseph (2002), but he was not applying a biological species concept. The two taxa are wildly not monophyletic in their DNA ,and Ribas et al provided some evidence of morphological intergradation. I recommend a NO vote.

E) Separate roseifrons (including peruviana) from picta. Roseifrons is very distinctive with its rose-colored head, and seems like it should be clearly split. I recommend a YES vote.
E1) If this change is made, then split roseifrons and peruviana. As noted above, roseifrons is very distinctive morphologically, whereas peruviana is more similar in appearance to the other southern Amazonian taxa. It would seem on morphological grounds that roseifrons should be separated from the other taxa. However, the molecular data does not find roseifrons to be monophyletic with respect to peruviana. Additionally, an unsampled disjunct population is morphologically peruviana, and morphologically roseifrons populations are also disjunct from one another (although both were sampled). Just to add to the complication is a morphologically distinctive population from northern Peru that is in between the northern population of peruviana and the northern roseifrons (Group 6 of Joseph 2000). I recommend a NO vote, leaving roseifrons and peruviana together until the geography and relationships among these populations is clarified. But I could understand a yes vote.

F) Recognize luciani as a distinct species, splitting from picta. There is no good evidence on this taxon, with no molecular material. Where it should go with respect to amazonum and roseifrons is not clear. For a while, luciani was treated as an immature plumage of roseifrons. Because it had priority as a name, roseifrons was a junior synonym. I recommend treating luciani as a distinct species, recognizing that with more information we might associate it with roseifrons, or amazonum, at which point its name would take precedence over the name of the other taxa.

G) lump emma with picta
H) split eisenmanni from picta
I) split caeruleiceps (with subandina) from picta
I1) split subandina from caeruleiceps
These are all changes are related. Based on the geography and the molecular topology, eisenmanni would be recognized as a distinct species before emma. Ribas et al. recognized emma and eisenmanni as distinct units from picta, but they form a weakly defined clade in figure 2. Caeruleiceps and subandina may form a unit with eisenmanni (based on morphology), but without molecular data it is hard to know what to do with them. The realistic options would be split all 5 (NO on G, and YES on the other 3 proposals), recognize 4 (NO on G and I1, but YES on the others), or leave them all together awaiting more data (YES on G no on the other 3 subproposals). I recommend the last option, lumping this whole set.

Overall recommendation: The status quo is not a viable option: picta and leucotis as we define them are intertwined with each other, so there need to be some yes votes. I recommend voting yes on subproposals B, C, D, E, F, and G and no on A, D1, E1, H, I, and I1. Voting as I suggest would recognize 5 species in the complex: picta, leucotis, and pfrimeri (currently recognized by SACC), plus amazonum and roseifrons. It would sink griseipectus back into leucotis, and transfer emma from leucotis to picta.

Two reasonable alternatives in my view would be to split roseifrons and peruviana (so yes on E1) and to split up the northern taxa that I would lump under picta (No on G, and Yes on H, I, and I1).

Literature cited

Joseph, L. 2000. Beginning an end to 63 years of uncertainty: The Neotropical parakeets known as Pyrrhura picta and P. leucotis comprise more than two species. Proceedings of the Academy of Natural Sciences of Philadelphia 150:279-292.

Joseph, L. 2002. Geographical variation, taxonomy and distribution of some Amazonian Pyrrhura parakeets. Ornitología Neotropical 13:337-363.

Joseph, L. and D. Stockwell. 2002. Climatic modeling of the distribution of some Pyrrhura parakeets of northwestern South America with notes on their systematics and special reference to Pyrrhura caeruleiceps. Ornitología Neotropical 13: 1-8.

Olmos, F., P. Martuscelli, and R. Silva e. Silva. 1998. Ecology and habitat of Pfrimer's Conure Pyrrhura pfrimeri, with a reappraisal of Brazilian Pyrrhura leucotis. Ornitología Neotropical 8: 121-132.

Ribas, C. C., L. Joseph, and C. R. Miyaki. 2006. Molecular systematics and patterns of diversification in Pyrrhura (Psittacidae) with special reference to the picta-leucotis complex. Auk:123: 660-680.

Doug Stotz, September 2007

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

Comments from Remsen: "Recognizing that additional work is badly needed in this messy group, especially in potential contact zones, I think that Doug's overall recommendation is the best solution given current information, so ... "YES on subproposals B, C, D, E, F, and G and NO on A, D1, E1, H, I, and I1."

Comments from Stiles: "After sone head-scratching, I would agree that Doug's proposal best reflects the current imperfect state of our knowledge.. I am a bit ambiguous about lumping all the N taxa under picta, but at least as far as plumage patterns go, they all do seem to be variants of a common theme; many are very poorly known and no genetic, behavioral or vocal information is available for several, hence they may well be more split as knowledge increases but for the moment, putting all the eggs in one basket might be best (as we did for Momotus a while back). Hence- YES on B,C,D,E,F and G and NO on A, D1, E1, H and I1."

Comments from Robbins: "Whew, is this complicated. I believe Doug has done a great job of distilling things down to a level that makes sense given the information that is currently available. Thus, I follow his recommendation of voting "yes" for B, C, D, E, F, and G and no to A, D1, E1, H, I, and I1."

Comments from Jaramillo: "This is certainly the most confusing and complex of all proposals I can recall, they are also birds away from my region of expertise. I have gone over it a couple of times, and it does make sense given the available data to accept Doug's recommendations. So "YES on subproposals B, C, D, E, F, and G and NO on A, D1, E1, H, I, and I1."

Comments from Zimmer: "This one is a huge mess, and Doug is to be commended for taking it on. First off, it seems to me that there is a discrepancy between Doug's recommendations and what he says would be the resulting taxonomy (although no one else noted this, so perhaps I'm just reading it wrong). I quote:

'Overall recommendation: The status quo is not a viable option: picta and leucotis as we define them are intertwined with each other, so there need to be some yes votes. I recommend voting yes on subproposals B, C, D, E, F, and G and no on A, D1, E1, H, I, and I1. Voting as I suggest would recognize 5 species in the complex: picta, leucotis, and pfrimeri (currently recognized by SACC), plus amazonum and roseifrons. It would sink griseipectus back into leucotis, and transfer emma from leucotis to picta.'

"Subproposal F recommends a YES vote on recognizing luciani as a distinct species, so it seems to me that the taxonomy resulting from the adoption of all of these proposals would be to recognize 6 species: picta, leucotis, pfrimeri, luciani, amazonum, & roseifrons.

"I would agree with most of Doug's subproposals. I think Gary is correct in expressing some queasiness about folding all of the northern taxa into picta, but then again, on current published data, that is probably the best choice.

"I disagree with Doug's recommendation on subproposal B. I agree that pfrimeri is morphologically and ecologically more distinct from leucotis and griseipectus than either is from one another. However, the morphological and ecological differences between griseipectus and leucotis are at least on a par with those between various members of the picta group that are here being recognized as or recommended for separate species status. When this yardstick is applied, in combination with the recognition of the formidable disjunction between the ranges of griseipectus and leucotis - which effectively means that the two forms are on independent evolutionary trajectories - it seems that recognition as separate species is warranted. I am not troubled by the lack of genetic evidence supporting this move - see Gary's comments under Proposal #181, which nicely sum up the arguments against giving too much weight to genetic evidence in matters of resolving species limits. So, I would vote NO on B.

"The other subproposal on which I have some ambivalence is D1 (split amazonum and snethlageae). Morphologically, these two forms are very different from one another - in fact, snethlageae is about as different from amazonum and all other taxa (with Amazonian distributions) in the complex as is any single taxon, including the highly distinctive roseifrons. Also, if you map the distributions of snethlageae versus amazonum, the breaks correspond with those between closely related species-pairs in many other families, so it makes sense biogeographically. A problem arises however, in the published distributions of the two forms. Joseph (2002), restricted the distribution of snethlageae to the Rio Madeira drainage, and listed amazonum as being on both banks of the lower Amazon, south to Alta Floresta and the Rio Teles Pires (northern Mato Grosso). I examined a series of MPEG specimens from Alta Floresta (specifically the left bank of the Rio Teles Pires) that would seem to refute these distributions. The specimens that I examined from that region appeared typical of Joseph's "group 3" morphotypes, i.e. = snethlageae, and are very different from south bank amazonum/microtera from east of the Rio Xingu. From what I can tell, the break looks pretty sharp, and given the degree of morphological difference, I would favor recognition of snethlageae as distinct from amazonum. One could advance the argument that we still don't have the distributions of these two forms worked out properly, and that until we do, we should keep the two as one species. I could live with that, while suggesting that more data will make the case for splitting. However, I will vote YES on D1.

"The rest of Doug's recommendations make perfect sense to me, so, in summary, my votes would be: YES on C, D, D1, E, F and G, and NO on A, B, E1, H, I, and I1. Whew!!"

Comments from Pacheco: "Após avaliar atentamente esta complexa proposição mas também as opiniões dos demais membros, sobretudo Kevin os meus votos são: YES on C, D, D1, E, F and G, and NO on A, B, E1, H, I, and I1."

Comments from Schulenberg:

"306A (treat all taxa in the /leucotis/ and /picta /complexes as subspecies of a single, polytypic species): No

306B (sink /griseipectus/ back into /leucotis): /No (although in some ways it would make our lives easier to do so). It's a little disheartening that these two did not prove to be reciprocally monophyletic with respect to the sampled DNA, but lineage sorting is a messy business. I'm swayed by Kevin's arguments regarding the distinctiveness of griseipectus.

306C (remove emma from leucotis): Yes.

306D (separate amazonum, including snethlageae, from picta): Yes.

306D1 (split amazonum and snethlageae): Yes

306E (separate roseifrons from picta): Yes.

306E1 (split peruviana from roseifrons): I'll go out on a limb and recognize this. I don't understand what's going in between roseifrons and all the allopatric and parapatric, brown-headed taxa; and no one else does either. But recognizing peruviana as a species to me better reflects the geographic pattern and the morphological differences between these two.

306F (recognize luciani as a species): Yes.

306G (lump emma with picta): No. I'm happy to recognize emma. It is, I'm told, a realistic option.

306H (split eisenmanni from picta): Yes. Makes sense to me.

306I (split caeruleiceps, with subandina, from picta): Yes.

306I1 (that's hard to read) (split subandina from caeruleiceps): Yes."