Proposal (522) to South American Classification Committee
An alternative classification of nightjar species in the New World
Background: This proposal is based on the proposals #465 and #501 by Mark Robbins and the committee member comments, which were not in agreement with inclusion of Uropsalis, Eleothreptus, Nyctidromus, Nyctiprogne, Lurocalis, and some Caprimulgus spp. in a large genus Hydropsalis.
Because there have been a number of doubts and questions regarding the way we should allocate the New World nightjars to genera on the basis of the Han et al. paper, several members recommended breaking the proposal into several parts to be voted on separately. The problem centers upon Han et al.’s “NW 3” clade – a large and very heterogeneous grouping that many are not content with subsuming into a single genus Hydropsalis. By contrast, there seems to be little objection to the generic arrangements in the other two clades: “NW 2” for the nighthawks (all species in Chordeiles) and “NW 1” to include Antrostomus (for rufus and carolinensis) and Nyctiphrynus for the other species in our area. Clearly we have to arrive at some kind of consensus, because the status quo, with species in Caprimulgus scattered all over the map, is clearly wrong – hopefully the stepwise approach will help. We arrange the subproposals “from the bottom up” in the sequence given by Han et al.
A. Lump all of NW 3 into Hydropsalis. Presumably most of the committee will vote NO, as this was the sticking point for passage of proposal 501.
B. Split Lurocalis from the remainder of the clade. This one should not be a problem, as the deepest split in NW 3 is precisely this, and several committee members have noted the morphological and behavioral distinctness of Lurocalis (besides, if we don’t do this split, there is little justification for any others given the topology!) So, presumably, this should be YES.
C. Recognize Nyctiprogne as a genus (including veilliardi). The alternative would be to lump it, but into what? Behaviorally, it seems rather more like Lurocalis, but the topology as it stands would group it as sister to the remainder of a broad Hydropsalis. Given the long branch-length to this taxon, we recommend a YES.
D1. Split Nyctidromus from the rest of the Hydropsalis group. This is a fairly well-supported clade with three species: albicollis, anthonyi and nigrescens.
D2. Split Nyctidromus, and further split nigrescens from the other two species as the genus Nyctipolus. The branch length to nigrescens in this clade is quite long, and this could be a justification for so doing; it was one of the suggested moves in proposal 501. We have no strong recommendation on this one.
E1. Split parvulus and whitelyi from the remainder of Hydropsalis in the genus Setopagis. However, the topology indicates that parvulus is sister to (whitelyi + the remainder of Hydropsalis), but with rather poor bootstrap support and very short branches for these separations. The alternatives here would be:
E2. Split parvulus and whitelyi in two monotypic genera; Setopagis for the former, a new genus (since there appears to be no name uniquely available for this species) for the latter. If evidence from other areas (song, morphology etc.) suggests a close relationship, a NO would be appropriate; if they are really quite different birds, YES (this is what the topology suggests: branch lengths to both are quite long). A further point here is the third species in Cory and Hellmayr’s now polyphyletic Nyctipolus, hirundinaceus; this isn’t in the Han et al. study but is recognized by Peters and by us; where does it fit in?
Here, one could adopt E1 as at least a temporary option, in view of the rather poor support for any split, or E2 (with the necessity of describing a new genus for whitelyi). We have no strong recommendation.
F1. Split the clade Eleothreptus +”C.” longirostris from the remainder of the Hydropsalis group as the genus Eleothreptus. These form a distinct clade with good support. However, males of Eleothreptus (sensu stricto) are morphologically bizarre, whereas longirostris looks like a mainstream “Caprimulgus” (albeit with more conspicuous white areas on the wing and tail than most). Given that some committee members favor keeping Eleothreptus (sensu stricto) separate, the alternative would be:
F2. Split Eleothreptus from Hydropsalis, and further split longirostris from Eleothreptus, in the genus Systellura Ridgway 1912. The type species of Systellura in Ridgway’s description was ruficervix, which was lumped into longirostris by Peters (and to our knowledge, this has been followed by all authors since). The only problem that could arise would be someone re-splitting ruficervix from longirostris, and showing that they are not closely related (i.e., sister species), in which case a new genus would have to be described for longirostris. However, given the status quo with ruficervix as a race of longirostris, there is no problem with the use of Systellura. It depends upon how much weight is given the more bizarre male morphology of Eleothreptus (sensu stricto). We tentatively suggest YES for F1, NO for F2.
G1. Split Uropsalis from the remaining species of the now-reduced Hydropsalis group. Again, the two species of Uropsalis do stand apart morphologically and behaviorally and form a clade; however, the remaining four species (cayennensis, maculicaudus, torquata and climacocerca) form a sister clade and are also more or less long-tailed, with flight displays but are lowland birds, whereas Uropsalis is Andean. A NO vote would place both groups in Hydropsalis, a YES would separate Uropsalis. We would suggest a YES.
H1. Lump cayennensis, maculicaudus, torquata and climacocerca in Hydropsalis; these are all more or less long-tailed, slender birds by comparison with most other nightjars, and form a compact, fairly well-supported lowland clade with short branch lengths. The problem is how to split this group, if one wants to.
H2. Split cayennensis and maculicaudus from Hydropsalis in Antiurus Ridgway 1912. However, this doesn’t fit the topology, because maculicaudus is sister to (cayennensis + Hydropsalis) with fairly high support, thus making Antiurus paraphyletic. The type species of Antiurus is maculicaudus, so one option would be:
H3. Separate maculicaudus in Antiurus, keep cayennensis in Hydropsalis. A further option:
H4. If one wishes to further split cayennensis from Hydropsalis, the name Thermochalcis Richmond 1915 is available for this species (Richmond argued that the name Stenopsis Cassin, 1851, also with type cayennensis, was preoccupied). These latter options depend upon how far one is prepared to go with monotypic genera, and effectively how one prefers to weight morphological vs. molecular evidence.
Hopefully these suggestions will help to define the problem more clearly in order to reach a consensus on this admittedly difficult group!
The nagging problem is what to do with “Caprimulgus” hirundinaceus – anyone have any information or ideas on this one?
Manuel Nores and Gary Stiles, May 2012
Comments from Robbins: “I would recommend that the committee give the Hydropsalis clade some context by looking at the entire phylogeny that is presented in Han et al. 2010. By examining the Old World Caprimulgus clade I underscore two taxa (Macrodipteryx) that have evolved extremes in plumage morphology that rival those “aberrant” taxa embedded within Hydropsalis (Eleothreptus, Uropsalis). Macrodipteryx is deeply embedded within Caprimulgus and it would take the recognition of a minimum of four genera to maintain that genus. As Nores y Stiles outlined, based on the Han et al. data, to maintain the sacred segmentata/lyra and anomalus in separate genera would require the recognition of a minimum of eight genera! What should be apparent from examining the entire molecular-based phylogeny is that within these monophyletic clades the evolution of presumed secondary sexual characters is not unusual (not surprisingly, the same can be said for another character suite, i.e., vocalizations). Having pondered caprimulgid relationships for > 25 years based solely on morphology and vocalizations, I was particularly surprised with the molecular data and I quickly realized that morphology and vocalizations (except for mostly sister relationships) often lead to wrong conclusions about relationships. Thus, after getting over the shock of the genetic results I have embraced those data and now marvel at the divergent evolution of plumage and voice within each of these monophyletic clades. I don’t need a unique name attached to every novelty (whether it is plumage morphology and/or vocalizations) to appreciate affinities within this fascinating group.
“One final comment: I am now privy to an enlarged molecular-based data set that provides additional node support in several parts of the caprimulgid tree (M. Braun, N White, pers. comm.). Based on those unpublished data I no longer support proposal 501 that I submitted, i.e., I support including nigrescens, anthonyi, and albicollis in Hydropsalis.
Comments from Remsen:
A. NO. This would be perhaps the most morphologically and ecologically heterogeneous genus in W Hemisphere.
D. I go for D2, i.e. Nyctipolus for nigrescens (unless someone can point out why this would make sense in Nyctidromus.
E. I go for E1 and live with potential paraphyly of Setopagis than for naming new genus until additional data demand the latter.
F. I go for F2, i.e. recognizing both Eleothreptus and Systellura unless someone can convince me that female Eleothreptus share features with longirostris. Cory (1918) had them in the same genus, Thermochalcis, so they must share some features.
G1 – YES.
H. I’ll go with H1, i.e. remaining 4 species in Hydropsalis.
Comments from Zimmer: “Yuck, what a mess! After contemplating all of the generic splitting covered by this stepwise approach, I was almost ready to throw in with Mark, whose comments regarding the heterogeneity contained within Old World Caprimulgus are pretty spot-on. However, I cannot bring myself to lump all of NW3 into Hydropsalis. For me, this is less about the extreme sexually selected characters found in males of Eleothreptus/Macropsalis, and more about the ecological extremes within a single genus that would include members so divergent as Lurocalis versus virtually everything else, as well as the extreme vocal differences between those species with far-carrying, throbbing vocalizations on the one hand, or those with relatively weak, clicking vocalizations on the other. Although perhaps just as consistent with the genetic data of Han et al. 2010 as the 8-genus approach, and certainly a lot easier to deal with, I don’t find it particularly informative, especially if we are setting Nyctiphrynus and several former Caprimulgus apart in Antrostomus.
“So, with that in mind, here goes:
G. YES. But aren’t we forgetting forcipata? I don’t see it mentioned anywhere in this, and the proposal speaks of only 2 species of Uropsalis (Andean segmentata and lyra). I would think that forcipata has to go with the other two, even though its distribution is so disjunct. Like segmentata and lyra, forcipata does seem to be more of a montane, or at least foothill bird, and tail morphology, vertical roosting behavior and flight displays all place it with the other two species.
H. H1 (remaining four species in Hydropsalis.)
I. “Caprimulgus” hirundinaceus – This one is a puzzle, and I don’t have any real answers. Vocally, I would put it closer to Blackish Nightjar (“Caprimulgus” nigrescens), which it also resembles in its habit of often roosting on rocky, open ground. Its most common vocalization is a short, whistled “PEEER” repeated monotonously at short intervals. But it also gives some sputtering “put putput” notes and an abbreviated purr, both of which are reminiscent of things that nigrescens does.”
Comments from Stiles:
D- YES to D2, at least tentatively; in view of Kevin’s comments, I’d also place hirundinacea in Nyctipolus, pending genetic data.
E- YES to E1, at least until these branches become better resolved
F- YES to F2, but with the following suggestion. Although both “Eleothreptus” are odd, it looks to me as though they are so in very different ways; candicans appears much more “normal” in overall form, differing mainly in the large amount of white in the males (at least in the HBW plate), and I fail to see how it could really be considered so like the truly bizarre anomalus as to be considered congeneric with it. Its female does look more like longirostris (again, from eyeballing the HBW plate). I see little sense in three monotypic genera in this well-resolved clade, but I’d go for splitting of anomalus in a monotypic Eleothreptus and including candicans in Systellura with longirostris s (males of which also have rather more white in wings and tail than most “Caprimulgus”), at least until genetic data shows that it is indeed closer to anomalus (I note that Han et al. didn’t have candicans in their phylogeny).
G- YES, probably including forcipata, which I don’t have experience with.
H- YES to H1: there has been sufficient splitting, and all branch lengths are short, so place all four in Hydropsalis.”
Comments from Pacheco:
D. YES to D2. Traditionally (e.g. Cory & Hellmayr 1918) the Genus Nyctipolus contained N. nigrescens and N. hirundinaceus. I agree with Zimmer, by field experience, that these two species are related.
E. YES to E1, for now.
F. YES to F2
G. YES, but subordinating forcipata to Uropsalis, at least.
H. YES to H1
Comments from Pérez-Emán: “This is a very difficult proposal, because the options seem to be extreme: one genus against up to twelve potential genera! Plumage and vocalization differences, the potential for convergence evolution, as well as their ecological differences and the lack of support for many of the nodes in the molecular phylogenetic hypothesis, place decisions on the particular weight we put on each character. However, as suggesting one genus seems to force too much variability into one taxon, we need to decide on how many genera are appropriate. If we are using the Han et al. (2010) study to understand phylogenetic relationships among these taxa, then we face the problem of too many weakly supported nodes (in relation to the New World 3 clade) and the lack of natural history information for some of the species. For me, this scenario calls for a conservative strategy of accepting many genera until new data suggest a different course of action.
D: D2. Mark mentioned evidence indicating strong support to include nigrescens in a clade with albicollis and anthonyi. However, this node would suggest close relationship but not necessarily species within the same genus. Decision here could go either way.
E: I go for E2 as molecular evidence doesn’t suggest close relationships between parvulus and whitelyi. One could argue that it is better to lump both species into the same genus to avoid naming a new genus. However, molecular evidence is against it, the only available whitelyi vocalization (by Brian O’Shea) is different from the ones I have heard from parvulus and there is a lack of basic and comparable natural history information. Consequently, to be consistent in my evaluation of each aspect of this proposal, I suggest E2 is the most appropriate solution for the moment.
G: YES. Not only Uropsalis members are very different morphologically and behaviorally but also there is a lack of support to consider them close relatives to any other member of the group.
Comments from Jaramillo: “
A. NO – Creates an undiagnosable genus,
difficult to define, use, or defend!
B. YES – This is reasonable.
C. YES – recognize Nyctiprogne.