Recognize Cracidae subfamilies Cracinae and Penelopinae and Change linear order of Cracidae genera

Proposal (#285) to South American Classification Committee

285 A. Recognize Cracidae subfamilies Cracinae and Penelopinae

285 B. Change linear order of Cracidae genera

Summary: These two proposals are to adopt a new linear order for the Cracidae, based on a number of recently-published phylogenetic studies. It also gives an opportunity to SACC to consider whether to recognise the proposed subfamilies Cracinae and Penelopinae.

Background: To give some background to the morphological issues behind the current linear order, I quote from the footnote on the SACC list:

"Vaurie (1968) recognized three major divisions within the family based on morphological criteria: the guans and chachalacas (tribe Penelopini), the curassows (Cracini), and the (extralimital) Horned Guan (Oreophasini). Delacour & Amadon (1973) considered the latter to be part of the chachalaca-guan group and recognized only two major divisions, (a) the curassows and (b) everything else. Del Hoyo (1994) recognized two subfamilies, Cracinae for the four genera of curassows and Penelopinae for everything else."

The current linear order of Cracid genera on SACC is as follows. No sub-families are recognised.

Ortalis
Penelope
Pipile
Aburria
Chamaepetes
Nothocrax
Mitu
Pauxi
Crax

Recent studies:

(a) Subfamilies

Three molecular studies (Pereira et al. 1992; Crowe et al. 2006; Hoeflich et al. 2007) have suggested that the Cracidae fall into two broad groups: (i) Oreophasis (extralimital), Pauxi, Mitu, Nothocrax, Crax and (Curassows and Horned Guan); (ii) Chamaepetes, Penelopina (extralimital), Penelope and Pipile/Aburria (all other Guans). The position of Ortalis (Chachalacas) in one or the other group is controversial. All molecular studies hold it to be related to the Curassows (Pereira et al. 1992; Crowe et al. 2006; Hoeflich et al. 2007). All past morphological studies (Vaurie 1968; Delacour & Amadon 1973) held it to be related to the Guans, as does a recent study of morphological / behavioural and combined morphological / behavioural / molecular data (Hoeflich et al. 2007). The group including the Curassows has been termed the "Cracinae" with the group including the Guans the "Penelopinae".

The Pereira et al. study found relatively weak support for a Chachalaca-Curassow relationship, but strong support for rejecting a Guan-Chachalaca relationship. The Crowe et al. study showed strong support at the Chachalaca-Curassow node. Each of Pereira et al. and Crowe et al. involved analysis of over 4000 base pairs from both mitochondrial and nuclear markers. Frank-Hoeflich et al. (2007) considered morphological / behavioural data and combined data sets in addition to molecular data. Although their molecular data produced similar results to the other studies (and were based on much of the same data from the other studies), analyses of morphological and combined data sets held Ortalis to be more closely related to the Guans than to the Curassows, although without very strong support.

In neither paper is there a detailed discussion of whether the Ortalis (or Oreophasis) should be given their own subfamily. Unhelpfully, however, the name Ortalinae is preoccupied by some Diptera genera and Ortaliinae by some Coleoptera. Ortalidainae may be an appropriate name if a subfamily were to be recognised for the Chachalacas [this is not a nomenclatural act].
Per the SACC introductory text, "Most traditional subfamilies are omitted unless supported by multiple independent data sets that mark major, deep branches within a family." Given that the position of Ortalis is not consistent between the various studies, it is perhaps a stretch to call the division between Cracids/Horned Guan vs. Guans/Chachalacas or Cracids/Horned Guan/Chachalacas vs. Guans "supported by multiple independent data sets" or involving "deep branches". Frank-Hoeflich et al. (2007) found the relevant subfamily clades to have relatively low posterior support values (0.78 / 0.82). However, interestingly, when Ortalis is excluded, 1.00 support for each of the Curassow and Guan clades is produced.

The following paragraph is based on some interesting comments from Sergio Pereira: "Re Cracinae and Penelopinae, I have no doubt about Cracidae being subdivided into these two groups. [The position of Ortalis, whether with the Curassows or Guans] is controversial for sure. But I believe that more analysis can be done when Frank-Hoeflich's matrix becomes available online. I'm planning to do some work on this, joining my molecular data set with their behavioural/anatomical data set for each of the 11(12) genera. In my Systematic Biology paper published in 2002, we performed a parametric bootstrap test to evaluate the differences between our trees (Ortalis and Oreophasis with curassows) and that of Vaurie (Ortalis and Oreophasis with guans). In brief, we simulated molecular characters to test the differences between the molecular and morphological trees. Undoubtedly, Vaurie's tree could be strongly rejected based on the molecular data (over 10,000 analyzable sites - which still is one if the biggest data set gathered for a group of birds so far). Unfortunately, the same test cannot be applied the other way around: simulate the morphological data to test the molecular tree. To summarize, although I am more inclined to suggest Ortalis and Oreophasis with curassows, I recognize that more analyses are needed on existing data sets."

Although the statistical analyses give only weak support for recognition of the traditional subfamilies within Cracinae, morphological considerations give some pause for thought. The Cracinae have robust or hooked bills often with ornaments; are generally larger and heavier; have elaborate ground display courtship; and have generally pied or rufous-brown plumage. Penelopinae have thinner, straighter bills with no ornaments; smaller bodies; shorter incubation periods; plumage is more heterogeneous; and various species have wattles or naked throat patches (Frank-Hoeflich et al. 2007). Ortalis sits more comfortably with the Penelopinae in the characters described above, though vocally is different to each of Cracinae and Penelopinae; and ecologically, Ortalis is less forest-dependent than the guans or curassows.

(b) Linear Order of Genera

(i) Curassows

The order for genera within the Curassows is somewhat muddled as Mitu and Pauxi were found to be paraphyletic by each of Crowe et al., Pereira et al., Frank-Hoeflich et al. and a further study (Pereira & Baker 2004). Certain differences in placement of curassow genera in phylogenies are apparently influenced by a past hybridisation event involving Mitu/Pauxi (Pereira & Baker 2004).
Pereira et al. found the relations between Curassow genera to be as follows: Crax (Nothocrax (Mitu / Pauxi morass)), although with only weak support at the Crax / Nothocrax node. Crowe et al. presented a series of trees based on different analyses in which the position of various Curassow groups varied with respect to one another. Basal to the other taxa were either Pauxi pauxi or Nothocrax and, in one case, Crax. Frank-Hoeflich et al. proposed Nothocrax (Crax (Mitu /Pauxi morass)) based on each of their molecular, morphological/behavioural and combined data sets, which is consistent with certain of Crowe et al.'s hypotheses and only a small tweak from the Pereira et al. study. The latter two genera have recently been proposed for merger (Frank-Hoeflich et al. 2007) due to paraphyly, which is a separate and difficult issue not subject of these proposals.

The position of Nothocrax at the start of the SACC order is, whether through accident or design, consistent with various of Crowe et al. (2006)'s trees, Pereira & Baker (2004)'s phylogenies and all three of Frank-Hoeflich et al.'s published trees (although not with those phylogenies mentioned above that place Crax before the other Curassows). I would suggest retaining Nothocrax's position in the absence of any strong reason to change it. However, in order to place the more basal genera first, Crax should be moved to above Mitu and Pauxi.

(ii) Guans

All molecular studies referred to above hold Chamaepetes to be basal to the other Guan taxa on the SACC list, with extralimital Penelopina either basal to all other Penelopinae or sister to Chamaepetes. Penelope and [Aburria/Pipile] were found to be sisters by each of these studies, as well as a third molecular study (Grau et al. 2005). Analyses using only morphological / behavioural characters, contrary to the others, found Ortalis to be embedded within the Guans, sister to either Chamaepetes or Ortalis (Frank-Hoeflich et al. 2007). The order Penelope, Pipile, Aburria in the current baseline is consistent with the results of all three molecular studies and the Frank-Hoeflich et al. combined morphological / behavioural / molecular tree. However, Chamaepetes should more sensibly go before the other genera, given that it is the most basal Guan taxa in the overwhelming majority of phylogenies.

(iii) Chachalacas

As foreshadowed above, the linear placement of the Chachalacas Ortalis requires re-evaluation. According to three phylogenies based on molecular studies (Pereira et al. 2002; Crowe et al. 2006; Frank-Hoeflich et al. 2007), Ortalis is sister to the Curassows - the relevant nodes scoring 97-100 in certain jack-knife and Bayesian analyses (Crowe et al. 2006). A Chachalaca-Guan relationship was also strongly rejected by Pereira et al. (2002). However, according to morphological / behavioural analyses, Ortalis is embedded in the Guans, sister to either Penelope or Chamaepetes and per analyses combining morphological and molecular results, is sister to all the other Neotropical Guans, although with rather weak support (Frank-Hoeflich et al. 2007). Ortalis is currently at the start of the SACC linear order adjacent to the guans. Among all the studies discussed above, SACC treatment is consistent only with some (but not all) of the Frank-Hoeflich et al. morphological phylogenies and with one of their two combined phylogenies. The current placement of Ortalis is not consistent with any of the phylogenies based on molecular data, some of which include studies involving thousands of base pairs.

The most sensible approach, consistent with almost all published studies, would be to move Ortalis in the linear order to between the Guans and the Curassows. This could be criticised in that the Frank-Hoeflich phylogeny involving combined data holds a basal relationship of Ortalis to the Guans, which could mandate retention of Ortalis at the start of the order. However, as noted in Sergio Pereira's comments above, Ortalis' relation to the guans is rejected strongly by molecular data. The proposed change here would render the linear order not inconsistent with either a Curassow-Chachalaca or Guan-Chachalaca relationship being correct - whilst the current order only "works" if Ortalis is related to the Guans.

Conclusion on Proposal A: A Yes vote on Proposal A would lead to the Cracinae and Penelopinae being recognised. It is clear from published analyses that the Curassows and Guans form two separate groups that meet the SACC requirement for subfamilies. If those were the only Cracids, subfamily recognition would be an easy question. However, the position of the Chachalacas clouds issues considerably. A "yes" vote on this proposal would result in treatment of Ortalis as incertae sedis as to subfamily position. If a vote for recognition of the subfamilies passes, a separate three-way proposal or set of proposals can be considered as to where Ortalis should be placed among the various options (i.e. in Cracinae, Penelopinae or in its own subfamily / left as incertae sedis). I make no strong recommendation either way on this proposal.

Conclusion on Proposal B: A Yes vote on Proposal B would result in changes to the positions of Ortalis, Crax and Chamaepetes in the linear order, as described above. The proposed new linear order is set out below in two versions, one with subfamilies and one without. Although linear orders are not the best way of presenting phylogenies, this new order is considerably more informative as to Cracid relationships than the previous linear order and therefore comes strongly recommended.

The version of the proposed linear with subfamilies included is as follows:

Penelopinae

Chamaepetes
Penelope
Pipile (if recognised)
Aburria

Subfamily uncertain: pending another proposal

Ortalis

Cracinae

Nothocrax
Crax
Mitu (if recognised)
Pauxi

The version of the proposed linear order without subfamilies is as follows:

Chamaepetes
Penelope
Pipile (if recognised)
Aburria
Ortalis
Nothocrax
Crax
Mitu (if recognised)
Pauxi

References:
CROWE, T.M., BOWIE, R.C.K., BLOOMER, P., MANDIWANA, T., HEDDERSON, T., RANDI, E., PEREIRA, S.L., & WAKELING , J. (2006). Phylogenetics and biogeography of, and character evolution in gamebirds (Aves: Galliformes): effects of character exclusion, partitioning and missing data. Cladistics 22: 495-532. http://individual.utoronto.ca/sergiolp/pdf/Cladistics2006.pdf

FRANK-HOEFLICH, K., SILVEIRA, L.F., ESTUDILLO-LOPEZ, J., GARCIA-KOCH. A.M., ONGAY-LARIOS, L. & PINERO, D. 2007. Increased taxon and character sampling reveals novel intergeneric relationships in the Cracidae (Aves: Galliformes). J. Zool. Syst. Evol. Res. In press. http://dx.doi.org/10.1111/j.1439-0469.2007.00396.x

GRAU, E. T., S. L. PEREIRA, L. F. SILVEIRA, E. HÖFLING, AND A. WAJNTAL. 2005. Molecular phylogenetics and biogeography of Neotropical piping guans (Aves: Galliformes): Pipile Bonaparte, 1856 is synonym of Aburria Reichenbach, 1853. Molecular Phylogenetics & Evolution 35: 637-645.

PEREIRA, S.L., BAKER, A.J.& WAJNTAL, A. (2002). Combined nuclear and mitochondrial DNA sequences resolve generic relationships within the Cracidae (Galliformes, Aves). Systematic Biology 51(6): 946-958 http://individual.utoronto.ca/sergiolp/pdf/SB2002.pdf

PEREIRA, S.L. & BAKER, A.J. (2004). Vicariant speciation of curassows (Aves, Cracidae): a hypothesis based on mitochondrial DNA phylogeny. The Auk 121: 682-694. http://individual.utoronto.ca/sergiolp/pdf/Auk2004.pdf

Thomas Donegan, June 2007 (with helpful comments from Sergio Pereira and Dan Brooks and, on SACC linear order policy, from Van Remsen)

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

Comments from Stiles: "YES (essentially for both). The two subfamilies of cracids are solidly established; the logical thing to do with the chachalacas is to place them between the two as "subfamily incertae sedis" given the conflicting evidence regarding their placement. This would preserve the linear order of genera in B and call attention to the problem of where to place the chachalacas. For now, recognize both Mitu and Pauxi in the Cracinae, with a note that they may be congeneric."

Comments from Cadena: "YES on both, and I agree with Gary in that the best we can do with the chachalacas given conflicting evidence is to place them incertae sedis."

Comments from Remsen: "NO on A, however, as a matter of taste. Rather than have a third of the species in the family Incertae Sedis, I prefer to wait for data that places them one way or another. Three subfamilies would seem more logical than two plus Incertae Sedis. Also, I'm not convinced that the split between guans and currasows is all that deep. Clean, yes, but deep? [I need to look at branch lengths between other groups we rank at the subfamily level.]

"YES on B. This sequence best reflects the data published so far and adds information to our classification."

Comments from Robbins: "YES to both, although I put more faith in the three molecular data sets over the combined morphological/behavior/molecular data set in placement of Ortalis, i.e., I suspect that despite the historical inertia clouding our views, Ortalis is indeed aligned with the curassows."

Additional comments from Donegan: "Re Van Remsen's comment on the depth of the divisions in this family, Pereira et al. (2002) hypothesised the following periods (95% confidence interval) for major divisions in the Cracidae: Guans vs. Other Cracids - 26.9-40.6 million years ago (Early Oligocene) [extralimital Oreophasis: 26.6-36.1 mya (Early Oligocene)] Ortalis: 25.8-36.5 mya (Early Oligocene).

"Other generic-level divergences are postulated to have occurred in the Miocene or later.

"As the Guan / Curassow / Horned Guan / Chachalaca division took place at a similar period and so long ago, this could support the erection in due course of a new subfamily for Ortalis."

Comments from Zimmer: "YES on both. I agree with Van that three subfamilies seems to be the most logical ultimate course, but the conflicting data regarding placement of Ortalis precludes making such a move now. Better to place Ortalis as "incertae sedis" between the other two subfamilies until we have more definitive data."

Comments from Nores: "(A) NO. Considero que no debemos incluir subfamilias en sólo algunas de las familias en la SACC list. Si lo hacemos para algunas, lo tenríamos que hacer para todas, que sería como empezar de nuevo.

"(B.) YES. Me parece que la propuesta está bien fundamentada con varios trabajos meloculares coincidentes.
Nota. Quiero aquí resaltar el trabajo de Thomas Donegan en el SACC. A pesar de que no es miembro del Comité, el está permanentemente realizando interesantes propuestas y comentarios. Llama la atención lo fundamentadas de sus propuestas e incluso, como en este caso, con la posibildad de bajar de Internet los trabajos citados. Felicitaciones."

Comments from Stotz: " (A). NO. (B). The inability to place Ortalis clearly with respect to the other two groups makes it seem to me that creating subfamilies is premature at best in this case. In terms of rearranging the genera, I guess I don't see that it provides much meaningful information on relationships; our current order is consistent with the data on relationships among genera. Additionally placing Ortalis, which is essentially incertae sedis relative to the subfamilies in Cracidae, in that middle seems like an odd approach; typically you'd place such a taxon at the end."

Comments from Jaramillo: "YES (both) - In full agreement with Gary's logic [B]."

Comments from Schulenberg: "YES (although our current order pretty well matches what we think we need?)."

Comments from Pacheco: "(A) [No] Até que haja melhor refinamento quanto às relações de Ortalis com as subfamílias propostas. (B) [Yes] Considero a sequência linear aqui proposta é de fato mais informativa.