Proposal (781) to South American Classification Committee

Change the generic classification of the Trochilinae (part 2)

The genetic tree of the Trochilidae by McGuire et al. (2014) showed that the generic taxonomy of the hummingbirds was incongruent with phylogeny at many points, and that the tribe Trochilini, popularly known as the Emeralds, epitomized this conflict. The bewildering array of problems to be resolved was summarized briefly by Stiles et al. (2071a), and a resolution of these in a new generic taxonomy was presented by Stiles et al. (2017b). Based upon the genetic tree, we divided the Trochilini into four groups. In part 1 of this proposal, we treated Groups A, B and C; this proposal deals with the largest and difficult group of genera, Group D. We divided this group into seven subgroups, within each of which we recognize from one to four genera. The nomenclatural issues are more complicated in this group, and for more detail and explanation of our resolution of these, see Stiles et al. (2017a and 2017b). As in part 1, we work through the groups in the order in which they appear in the figure, reproduced below.

Subgroup D1 includes only two genera, Leucippus and Phaeochroa, separated on long branches. Recent classifications of Leucippus have included several other species, but the phylogeny places all of these in different subgroups, leaving fallax, its type species, alone in what becomes a monotypic genus. Both occur in dry habitats, Phaeochroa mainly on the Pacific slope of Middle America and northern Caribbean Colombia, L. fallax in dry to desertic habitats in extreme northern Colombia and Venezuela. Phaeochroa is also monotypic in most classifications, with cuvierii as its type, although some would split roberti of the Caribbean slope of Middle America as a separate species. Both fallax and cuvierii are rather dull-colored but they differ in pattern: fallax is uniform buff below, whereas cuvierii is mostly green below, speckled or scaled with buffy-whitish; cuvierii is much larger, and the outer primaries are flattened and thickened much like those of Campylopterus in group B; indeed, Schuchmann include cuvierii in Campylopterus for this reason. However, this course is refuted by the phylogeny: the “sabre” wings of each were derived independently. We advocate recognizing both of these as monotypic genera.

Subgroup D2 includes two well-separated clades. The first includes two species formerly included in Leucippus in recent taxonomies, taczanowskii and baeri. The phylogeny precludes their inclusion in Leucippus, but the generic name Thaumasius is applicable, its type species being taczanowskii. These two species share a rather dull, brownish plumage, differing in pattern and size, but both occupy relatively dry habitats of the Pacific slope of extreme southern Ecuador and northern Peru. The second clade includes three species on long branches, currently segregated in three monotypic genera: Taphrospilus hypostictus, Eupetomena macroura and Aphantochroa cirrochloris. In plumage, hypostictus and cirrochloris are dull in coloration although differing in pattern; macroura is very different in plumage, dark blue with a longer, forked tail. However, cirrochloris and macroura are sister species sharing a similar pattern of distribution in lowland southeastern South America as well as “saber” wings resembling those of Campylopterus in the males, whereas hypostictus occupies mainly subtropical elevations along eastern Andean slopes from southern Colombia to northern Bolivia, and lacks the modified primaries of the other two. This species has been included in Leucippus by some authors, but the phylogeny precludes this treatment. Here, three options are available: a) lump all three in a single genus, for which Eupetomena takes priority; b) lump Aphantochroa and Eupetomena because of their sister status, shared distribution and modified primaries in males, while maintaining a monospecific Taphrospilus or c) maintain three monospecific genera. We consider a) the worst option because it gives no information regarding relationships and

would subsume considerable genetic differences; option c), while preserving stability, also ignores relationships among these species. We therefore prefer option b), which is most informative in this respect as well as in distribution and morphology, although its two species are widely divergent in plumage color. However, such “color clashes” also occur in several other subgroups in group D, as detailed below.

Subgroup D3 first includes a clear outlier with no close relatives, “Leucippus” chlorocercus, (it could even be considered a subgroup by itself), which therefore requires its separation in a monotypic genus Talaphorus, which was originally described for it. Its distribution, along the upper reaches of large Amazonian rivers, is also unique. Its inclusion in Leucippus in the past was due to its dull colors and conservative plumage evolution having masked its genetic distinctiveness. Next in this subgroup is a distinct clade including only the genus Trochilus with one or two species. Unique in morphology and distribution, Trochilus clearly merits generic rank.

Next in subgroup D3 are two well-separated clades formerly included in Amazilia (but such inclusion is refuted by the phylogeny; see below). The first clade comprises three Mexican species: violiceps, wagneri and viridicauda, for which the generic name Leucolia is applicable. We note here that we had accepted viridicauda as its type following the recommendation of Elliot, but this was incorrect because it was described after Leucolia was named; we have submitted a manuscript (Stiles et all, submitted) substituting violiceps as the type species to correct this error. The final clade in D3 includes ten species in two compact clusters separated by a very short branch, such that they all should be considered as a single genus Saucerottia. The four species of the first cluster are Middle American, the six of the second cluster occur from southern Middle America into northern South America; all share certain morphological features and glittering green over the chest or the entire underparts. In the past, Saucerottia had often been considered a subgenus within Amazilia, but generic status is supported by the phylogeny.

Subgroup D4 includes only three species: rutila, yucatanensis and tzacatl; because the first is the type species of Amazilia, these three constitute the necessarily much-restricted genus Amazilia. The genetic distinctness of this subgroup was not recognized heretofore due to numerous convergences in plumage with species of several other subgroups.

Subgroup D5 includes two successive outliers on relatively long branches, then two much more closely related clusters of three and five species. The first outlier is the species amazilia, often considered the type species of the genus Amazilia in the past, but the genetic tree does not support its inclusion therein. We have proposed that the earliest generic name for this species is Amazilis (see Stiles et al. (2017a) regarding the nomenclatural complexities involved). This species appears to be an old isolate at the southwestern extreme of the distribution of what we dubbed the “amazilian complex”, and it also has a unique plumage pattern. The second outlier, on a slightly shorter branch, is “Amazilia” franciae, which we also consider to constitute a separate genus, for which the name Uranomitra is applicable. It is the only member of the complex with a montane distribution and marked sexual dichromatism, as well as several morphological differences from others in this complex.

The third group of species in subgroup D5 provides the most extreme mismatch between the genetic data and plumage features. The first cluster includes three species currently placed in three genera: Chrysuronia oenone, Leucippus goudoti and Amazilia versicolor; the second, four species in three genera: Leucippus coeruleogularis, Hylocharis grayi, Amazilia brevirostris (usually including chionopectus) and A. leucogaster. The white-bellied species versicolor, brevirostris and leucogaster were included in Agyrtria by Schuchmann (1999), but Agyrtrina is correct. Two options exist here: a) give each cluster a separate generic name; or b) combine both clusters in a single genus. For the first option, the generic name Chrysuronia has priority; for the second, Eucephala (the original genus name of grayi). For the second option, Chrysuronia takes priority over Eucephala. We favor the second option because of the short branch connecting the two clusters, and the enlarged genus Chrysuronia is scarcely more heterogeneous in plumages than either cluster produced by the first; in addition, option b combines in the same genus members of two genera that must disappear in the interest of priority, Agyrtrina and Lepidopyga. The two most divergent species in male plumage, “Lepidopyga” goudoti and coeruleogularis, are not even sisters in the phylogeny, and the “Agyrtrina” species also appear in both clusters. The females of all of these species are more or less “white-bellied”, as are the males in the monomorphic species.

In subgroup D6, the species albicollis is an outlier on a long branch, and we favor continuing to recognize its distinctness by maintaining it in the monospecific genus Leucochloris; its plumage is also unique. The second cluster in this subgroup, on a fairly long branch, includes two species: “Amazilia” lactea and fimbriata, which share a similar plumage pattern, differing merely in colors. The phylogeny precludes their inclusion in Amazilia, and for them, the generic name Chionomesa is available and applicable. Both species are widely distributed in lowland cis-Andean South America, fimbriata more northern, lactea more southeastern in ranges. We favor placing both in Chionomesa. The third and similarly distinct cluster in subgroup D6 includes the species sapphirina and chrysura, which we consider should constitute the restricted genus Hylocharis, of which sapphirina is the type species. These share a unique plumage feature and both occur in southeastern South America, although sapphirina has a broad but disjunct range in South America north of the Amazonian watershed.

The fourth cluster in subgroup D6 includes two species previously placed in either Leucippus or Amazilia: chionopectus and viridicauda. These species are less distinct genetically from the preceding cluster, such that they could be included in Hylocharis, but they are widely discordant in distribution and ecology, being found at middle and upper elevations in the Andes of Peru and Bolivia, such that we consider them to best represent a distinct genus. Their distinctiveness had previously been suggested by Peters, but neither he nor we found a generic name applicable to them. We therefore proposed a new genus Elliotia for them. Unfortunately, this name was found to be preoccupied, and we have submitted a manuscript substituting another name. We therefore suggest to SACC members that they evaluate the evidence for generic status of these two species under the generic name yet to be published.

Subgroup D7 includes two genetically distinct clusters. The first contains the species “Amazilia” amabilis, decora and rosenbergi. For these, the generic name Polyerata is applicable, with amabilis as its type species. The circumscription of this genus by some recent authors included several other species that the phylogeny placed in other subgroups, but Polyerata as here restricted is clearly valid, and we advocate its recognition. The second subgroup includes five species arranged in a stepwise cascade with short branches separating them: “Amazilia” candida, “Hylocharis” eliciae, “Hylocharis” cyanus, Juliamyia julie and Chlorestes notatus. Inclusion of any of these species in Amazilia or Hylocharis is not supported by the phylogeny. Because all of the branch lengths are short, we consider that any further subdivision of this group would be arbitrary and could require the resurrection of at least one generic name and the probable erection of one new genus, we prefer considering all of these species congeners; the generic name Chlorestes takes priority. Thus, the generic name Juliamyia is placed in synonymy: in fact, the two most closely related species are notatus and julie, which differ considerably in male plumages but share one unique feature, their strongly rounded tails. Once again, in this group as a whole, female plumages are much more similar than those of the males, and that of the one monomorphic species, candida, also fits the situation in the enlarged Chrysuronia above.

Finally, we leave two Middle American species unclassified (incertae sedis) because we lack genetic information for them and are reluctant to place them on the basis of plumage characters that have been repeatedly shown to exhibit homoplasy: “Amazilia” luciae and “Amazilia” boucardi.

We now present the results of this generic rearrangement for valuation by the SACC in the following series of proposals.

1. A. Consider subgroup D1as a single genus, for which Leucippus takes priority.

B. Recognize Leucippus and Phaeochroa as distinct monospecific genera. Given the long branches and morphological distinctiveness, we strongly favor this alternative.

2. A. Maintain baeri and taczanowskii in Leucippus.

B. Recognize the genus Thaumasius for these two species. We strongly favor this option, because option A would produce a polyphyletic Leucippus.

3. A. Continue to recognize three monospecific genera Aphantochroa, Eupetomena and Taphrospilus.

B. Lump Aphantochroa into Eupetomena while maintaining a monospecific Taphrospilus.

C. Lump all three into a single genus Eupetomena. We favor option B as being most concordant regarding relationships, morphology and distributions, although cirrochloris and macroura differ strongly in plumage; we consider option C as the worst alternative.

4. A. Recognize the monospecific genus Talaphorus for chlorocercus. Given its great genetic distinctness, there really is no other sensible option here.

5. . [extralimital --- advisory only] A. Recognize the genus Leucolia for the extralimital species viridifrons, violiceps, and wagneri. Again, there is no real alternative: they cannot remain in Amazilia, no other generic name previously applied for them accords with the genetic tree and they share a characteristic distribution.

6. A. Split the genus Saucerottia into two genera, one Middle American and the other found from southern Middle America into South America.

B. Maintain a single genus Saucerottia including both groups above. We strongly favor this option given the close relationships and morphological congruence of these two groups.

7. A. Restrict the genus Amazilia to the species in subgroup D4.

B. Continue to recognize a broader Amazilia, although its limits would be difficult to define and would subsume too much genetic divergence. We strongly favor option A.

8. A. Recognize the genus Amazilis as a monospecific genus for the species amazilia based upon its distinctness genetically and in plumage and distribution.

B. Include more of subgroup D5 in Amazilis. We favor option A.

9. A. Recognize the monospecific genus Uranomitra for the species franciae, given that it is so distinct genetically, morphologically and in its highland distribution from the following species cluster.

B. Include franciae and the following group in Amazilis. This option would produce a very heterogeneous group subsuming a great deal of genetic, morphological and distributional diversity. We favor option A.

10. A. Divide the remaining ten species of group D5 into two genera, Chrysuronia and Eucephala.

B. Include all eight species in the genus Chrysuronia. Although decidedly heterogeneous in male plumages, the combined group is little more so than each of them separately, the groups are closely related and this option would include members of two genera that must be sunk due to phylogeny but are not sisters in the phylogeny. We favor option B.

11. A. Continue to separate albicollis in a monospecific genus recognizing its genetic distinctiveness and unique plumage. We favor this option, because option B would subsume much genetic and distributional divergence.

B. Combine Leucochloris with one or more clusters of subgroup D6.

12. A. Recognize Chionomesa for the species fimbriata and lactea. We favor option A, because the other subgroups are approximately equal in genetic distinctiveness and differ greatly in distribution.

B. Combine these with one or more clusters in subgroup B6.

13. A. Recognize a restricted Hylocharis for the species sapphirina and chrysura. All other species previously included in this genus fall out in different parts of the phylogeny, and the genus as here restricted shows a unique plumage feature and a largely congruent distribution.

B. Include at least the following group in Hylocharis reflecting genetic similarity. We favor option A, for the reasons given more fully below.

14. A. Recognize a new genus for the species chionogaster and viridicauda (here called Elliotia, but this is preoccupied; its name to be supplied in a manuscript submitted).

B. Include these species in Hylocharis. We favor option A, because option B would produce a morphologically, ecologically and biogeographically incoherent grouping.

15. A. Recognize a restricted genus Polyerata for the species amabilis, extralimital decora, and rosenbergi. This proposal is novel only in its restriction; the inclusion of several other species in some classifications is precluded by the genetic data, although we also do not include two species for lack of genetic data (see below). There is no really feasible alternative here.

16. A. Divide the remaining cluster in subgroup D7 into two to four genera.

B. Consider the five species in this cluster congeneric under the name Chlorestes. We favor this option because these species occur in a stepwise cascade with very short branches between them, such that any such subdivision would be arbitrary and would require at least one new generic name.

Gary Stiles, March 2018

__________________________________________________________

Comments from Areta: “I vote YES on the following: 1B, 2A, 3B, 4A, 5A, 6B, 7A, 8A, 9A, 10B, 11A, 12A, 13A, 14A, 15A, 16B

Comments from Zimmer:

“1B. YES. The branch lengths in the tree and the morphological/ecological distinctions between fallax and cuvierii would argue against treating them together in a restricted Leucippus. That Phaeochroa did not cluster with Campylopterus or, particularly, with Aphantochroa (with which there are numerous behavioral as well as morphological parallels) is one of many surprises (to me) revealed by the phylogeny.

“2B. YES. Any other treatment would leave us with a polyphyletic Leucippus.

“3B. YES. This one pains me! It’s difficult to reconcile the plumage distinctions between the colorful, long/deeply fork-tailed macroura and the drab, almost featureless cirrochloris, enough to consider them in the same genus. However, they are clearly sister species in the phylogeny; they share a similar geographic distribution (which the phylogeny reveals to be more important than plumage differences or similarities in several cases); both have the “sabrewing” morphology found in Campylopterus; and both are persistently vocal, aggressive birds that assume prominent perches and routinely bully other hummingbirds at feeders. That’s enough to nudge me over the line in favor of lumping Aphantochroa into Eupetomena, which has priority. My second choice would be to maintain all three species in monotypic genera, but, as Gary points out, that configuration would not be informative regarding the sister relationship of macroura and cirrochloris. I have no appetite for lumping Taphrospilus in with the other two species.

“4A. YES. This is the only option given the genetic distance and branching patterns reflected in the tree. On morphology and its restriction to scrubby habitats, I would have guessed chlorocercus to be closer to Thaumasius (baeri and taczanowskii). But, as is the case with L. fallax, biogeography and narrow habitat specialization prove to be more important clues to genetic relatedness than plumage.

“5A. YES. This is the only option given the relationships shown in the phylogeny, and makes sense biogeographically. I assume that the three species we are talking about recognizing in Leucolia are violiceps, wagneri and viridifrons (from which wagneri is a split), NOT viridicauda or viridigaster (neither of which have Mexican/Middle American distributions) as cited in different parts of the Proposal. The tree correctly includes viridifrons.

“6B. YES. Hairsplitting in my opinion, to attempt to separate these 10 species into different genera.

“7A. YES. I think this greatly restricted version of Amazilia makes more sense genetically and biogeographically than the expansive alternative.

“8A. YES. I strongly favor this option. Lumping all of subgroup D5 into Amazilis would make for a very heterogeneous, difficult to diagnose genus.

“9A. YES. Genetics and distribution favor this approach, in my opinion.

“10B .YES. Both options leave us with groupings that are genetically cohesive, but difficult to diagnose on plumage, and I don’t see the genetic distance between the two clusters, nor the slightly greater morphological heterogeneity introduced by lumping the two clusters together as obstacles to treating them all in an expanded Chrysuronia.

“11A. YES. Leucochloris albicollis is distinct genetically, and in plumage, and, has a geographic distribution restricted to the greater Atlantic Forest & temperate forest biome of southeastern SA. It is also noteworthy for its far-carrying, distinctive, and incessantly delivered songs. I can’t see lumping it in with other members of the subgroup D6. The wording in this sub-proposal suggests a preference for Option B, but I don’t think that is what Gary intended.

“12A. YES. This (fimbriata and lactea) form a genetically and morphologically cohesive pair, and any other arrangement within this subgroup would contain too much heterogeneity for my tastes.

“13A. YES. One of many surprises (for me) revealed by the phylogeny, was the polyphyletic nature of Hylocharis. The bill morphology (with the laterally expanded base) and color pattern (bright red with dark tip) shared by all of the species in the genus (as currently constituted) made for a pretty distinctive grouping. However, Cynanthus latirostris, Hylocharis/Basilinna leucotis and H./B. xantusii have similar bills, and those species are nowhere close to sapphirina, cyanus, chrysura, grayi/humboldtii or eliciae in the phylogeny. I would also note, that, in addition to being united by some plumage features, H. sapphirina and H. chrysura differ from eliciae and cyanus in their vocal behavior. It has always been my impression that eliciae and (perhaps to a lesser extent) cyanus males form exploded leks to sing (and the vocalizations are somewhat similar as well), whereas I have never observed any obvious lekking behavior in sapphirina or chrysura. At any rate, I would strongly oppose lumping any other members of subgroup D6 into a reconstituted Hylocharis with sapphirina and chrysura.

“14A. YES. Including these two species within a pared-down Hylocharis (to the exclusion of several morphologically similar species formerly included in that genus) would be unpalatable to me, even if acceptable on genetic distances alone.

“15A. YES. For reasons spelled out in the Proposal.

“16A. YES. I would favor recognizing two genera here: one with the plumage outlier and sexually monomorphic candida, and an expanded Chlorestes containing notatus, julie, eliciae, and cyanus. “A.” candida appears to be more different genetically from the rest of the group, it is the only sexually monomorphic species in the bunch, the only one in which the male is relatively dull in plumage, and, the only one of the bunch whose distribution is entirely within Middle/Central America. Treating them all in a single genus is certainly defensible on genetic grounds, but, given a choice, I would prefer a more morphologically cohesive grouping.”

Comments by Claramunt:

“3A. Tempting to try to get rid of one monospecific genus but divergence is so great between Eupetomena and Aphantochroa.

“5. No. The obvious alternative is to include these two species into Saucerottia, given that the similar looking cyanocephala is in that clade

“8.B, 9.B. Separating amazilia and franciae into monotypic genera would not prevent the remaining clade to be heterogeneous (both amazilia and franciae were included in the same genus, Amazilia, suggesting that they are not very distinctive.

The risk with the following proposed changes is that some are based on flimsy phylogenetic evidence (low node support).

“’I’m on the fence regarding the Hylocharis clade. I think it could be perfectly fine to have a diverse Hylocharis, from fimbriata to viridicauda in which color variation follows Gloger’s rule. Just for lack of a strong conviction, I’ll go with the recommendation:12.A, 13A, 14A

15A. Recognize a restricted genus Polyerata. On the edge in this one. They could be included in Chlorestes instead of their own genus. But since a sister relationship is not well-supported, the use of Polyerata may be a solution for now.