Proposal (703) to South American Classification Committee
Elevate Steatornithidae and Nyctibiidae to rank of Order
Note: A nearly identical version of this proposal has now been approved unanimously by the North American Classification Committee.
Synopsis: To maintain the monophyly of our current Caprimulgiformes and Apodiformes, this would elevate two families to the rank of Order: Steatornithiformes and Nyctibiiformes.
Background: Our current classification treats the Caprimulgiformes as containing three families: Caprimulgidae (nightjars), Nyctibiidae (potoos), and Steatornithidae (oilbird). Our Apodiformes contains two families: Apodidae (swifts) and Trochilidae (hummingbirds). These two orders have long been regarded as closely related. Traditional classifications also place the Old World Podargidae (frogmouths) and Aegothelidae (owlet-nightjars) in the Caprimulgiformes. Recent genetic data (e.g., Ericson et al. 2006, Hackett et al. 2008, Prum et al. 2015) are concordant in finding that the latter is actually sister to Apodidae + Trochilidae, and also that these three families are embedded in the Caprimulgiformes, thus making traditional Caprimulgiformes paraphyletic with respect to Apodiformes.
Here is the relevant portion of the tree from Hackett et al. (2008):
And here is the relevant portion of the tree from Prum et al. (2015):
Cracraft (2013) in Dickinson & Remsen (2013) [despite my objections] maintained the monophyly of Caprimulgiformes by elimination of Apodiformes as an order and inclusion of Trochilidae and Apodidae as families of the Caprimulgiformes. If this proposal is voted down, then Cracraft’s solution is the simplest alternative option.
However, an expanded Caprimulgiformes would include several lineages that are as old or older than many other taxa ranked traditionally as orders; it would also be spectacularly heterogeneous in terms of morphology – think of the profound differences, for example, between a potoo and a hummingbird.
Here is a broader view of the Prum et al. (2015) time-calibrated tree, with geological time periods along the bottom; nodes are calibration points, which are enumerated in the Supplementary material. The resolution here is lousy; so if anyone needs a pdf, just let me know:
With all appropriate caveats concerning the uncertainty of the underlying data, let’s use this figure as a gauge of relative lineage ages. If you draw an imaginary vertical line through the tree in the very early Eocene at roughly 54 mya, the following lineages are predicted to have been evolving separately at that point (with taxa currently ranked as families by SACC marked in red):
5. traditional Apodiformes (Trochilidae + Apodidae + Hemiprocnidae/inae)
7. Cuculiformes + Otidiformes
12. Phoenicopteriformes + Podicipediformes
22. our current Pelecaniformes minus Threskiornithidae
Thus, the lineages currently called Families in Caprimulgiformes are as old or older than most lineages we label as Orders.
If you zoom out to the full view of the tree in this figure, the following lineages also intersect the line through the early Eocene:
23. all ratites plus tinamous
32. Upupiformes + Bucerotiformes
Thus, the signal is even stronger when one looks at the entire figure – lineages as old as ca. 54 mya are consistently ranked in our classification as Orders or even multiple Orders. Of the 5 exceptions, 4 are in traditional Caprimulgiformes. The fifth is the Threskiornithidae (for which I will do a follow-up proposal to NACC).
I emphasize that I recognize that the Prum et al. tree represents preliminary analyses of new data, and that modifications are inevitable. Nonetheless, note that the topology and chronology are generally consistent with other data, fossil (see Mayr tree below) and genetic – in other words, this is not a radical overhaul of what we know about relationships and how we portray them in hierarchical classification. Using Prum et al. (2015), however, at least represents an objective approach to higher classification that differs from the current data-free approach that is maintained by historical momentum.
Below is the figure from Mayr’s (2014) paper that maps the oldest fossils for crown group birds. (I know the resolution isn’t good – let me know if you need a pdf):
The topology differs from that of Prum et al., but the lineage ages, reconstructed on the basis of fossil data, are similar, namely all of the caprimulgiform lineages are ancient, all projected to be evolving separately since the Paleocene or early Eocene, i.e. as old or older as most taxa we rank as orders.
So, I propose the following higher-level classification of the group labeled as Strisores by Mayr and Prum et al. (and based on the topology in Prum et al. 2015); brackets indicate extralimital taxa for which we do not have to endorse the ranks explicitly:
[Order Podargiformes (extralimital)
[Order Aegotheliformes (extralimital)
[Family Hemiprocnidae (extralimital)]
For those of you accustomed to thinking of the old Caprimulgiformes as consisting of several similar family-level taxa of night birds, consider that the phenotypic differences among these groups is masked somewhat by a degree of convergent evolution on cryptic coloration. Remove that, and these birds differ dramatically from one another. The echolocating Oilbird is the only nocturnal frugivore in Aves and really bears no morphological resemblance to any other bird. Likewise, the potoos bear little resemblance to any other birds and have bill and eyelid morphology found in no other group. The owlet-frogmouths are just bizarre birds that don’t seem to resemble anything else. Swifts and hummingbirds likewise are unique groups in birds, and once you take away parallel extreme adaptations for flight in terms of reduced feet and elongated wings, they share little in terms of plumage and morphology – one could even make an argument based on lineage age that they should also be treated as separate orders (in fact Pam Rasmussen has a NACC proposal pending to treat them as orders). The morphological distinctiveness of each of these groups is certainly related to the enormous amount of time since they shared common ancestors.
I recommend a YES vote on the proposal. A NO would necessarily generate a proposal (by someone else) to treat them all in the same order Caprimulgiformes (or perhaps some hybrid classification such as including Aegothelidae and Trochilidae in Apodiformes, and potoos and oilbirds in same order, each separate from Caprimulgiformes).
CRACRAFT, J. 2013. Avian higher-level relationships and classification: nonpasseriforms. Pp. xxi-xliii in The Howard and Moore Complete Checklist of the Birds of the World, 4th Edition, Vol. 1. Non-passerines (E. C. Dickinson & J. V. Remsen, Jr., eds.). Aves Press, Eastbourne, U.K.
DICKINSON, E. C., AND J. V. REMSEN, JR. (eds.). 2013. The Howard and Moore complete checklist of the birds of the World. Vol. 1. Non-passerines. Aves Press, Eastbourne, U.K., 461 pp.
MAYR, G. 2014. The origins of crown group birds: molecules and fossils. Palaeontology 57: 231–242.
PRUM, R. O., J. S. BERV, A. DORNBURG, D. J. FIELD, J. P. TOWNSEND, E. M. LEMMON, AND A. R. LEMMON. 2015. A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing. Nature 526: 569-573.
Van Remsen, February 2016
Comments from Claramunt: “A tentative NO. This is a difficult decision. I’m inclined to choose the alternative of including Apodidae and Trochilidae into Caprimulgiformes. Here are my reasons:
“1) Except for the Aegothelidae-Apodidae-Trochilidae clade, relationships among the other families are not well resolved. Hackett et al. (2008) recovered a Nyctibiidae/Steatornithidae clade branching first, but with no strong support. Osteological data suggested Steatornithidae alone branching first (Mayr 2010, J. Zool. Syst. Evol. Res. 48:126-137). Finally Prum et al. (2015) show Caprimulgidae branching first and the highest support (P =1) for all nodes in the main analysis (figure 1), but a species-tree method resulted in a basal polytomy (figure S3), so I don’t think that figure 1 is the final word on caprimulgiform interrelationship. In any case the point is that intermediate solutions of 2 or 3 orders are not warranted, instead wee need to lump everything or split everything, as in the current proposal.
“2) The proposal would result in six orders where we had two before, and would create a nearly complete redundancy between the Order and Family categories. Moreover, most of the families already contain relatively few species in a single genus, and one, a single species. Therefore, from pure (anonymous) taxonomic considerations, merging Apodidae and Trochilidae into Caprimulgiformes seems a more conservative, less radical, solution, and would result in a more balanced classification. So, are there other considerations that make the conservative solution undesirable?
“3) I don’t see a signal of Caprimulgiformes sensu lato being unusually old. First of all, the age of Caprimulgiformes in Prum et al. is reasonable but the age of many other basal nodes in the tree are underestimated, the result of a maximum age constraint at the base of the tree that is too young. My own analyses (Claramunt & Cracraft 2015: http://advances.sciencemag.org/content/1/11/e1501005) show that Anseriformes and Galliformes (stem age: 72 Ma) are considerably older than Caprimulgiformes (67 Ma). The later is indeed the oldest order within Neoaves, but it is followed closely by Opisthocomiformes (66 Ma, just 1 million year younger and with no chance of being split apart), Gruiformes (65 Ma), Charadriiformes (65 Ma), and a long list of gradually younger orders. At the level of families, the oldest nightbird family (in our tree) is Nyctibiidae (61 Ma), followed by Steatornithidae (60 Ma), which are younger than other four families of Neoaves in the SACC region, Columbidae (64 Ma), Opisthocomidae (66Ma), Cariamidae (63 Ma), Falconidae (62 Ma), and younger than other five families outside the SACC region (Gaviidae, Musophagidae, Otididae, Coliidae, Leptosomidae); all these older clades are both families and orders, so it could be argued that Nyctibiidae and Steatornithidae also deserves to be orders. Old families that are not orders at the same time are in the Galloanseres (Anhimidae 69 Ma, Anatidae 68 Ma, Megapodiidae 58 Ma). Within Neoaves, old families that are not orders are younger than the oldest nightbird families, the oldest being Strigidae and Tytonidae at 56Ma, followed by Fregatidae, Threskiornithidae and Meropidae at 54 Ma. If the 5 million years difference enough to consider the nightbird families too old, given calibration uncertainties and confidence intervals? Bottom-line, I see some signal of nightbirds being old but I don't see them as unusually old, even if comparisons are restricted to Neoaves.
“4) The inclusion of swifts and hummingbirds into Caprimulgiformes would add a tremendous diversity in number of species but in terms of broad ecophenotypic types, it would add just two: diurnal aerial specialists (swifts, Caprimulgiformes already have nocturnal aerial specialists), and the unique hummingbird nectarivore type. With these additions, Caprimulgiformes join a group of megadiverse orders like Gruiformes, Charadriiformes, and Passeriformes. Yet, an expanded Caprimulgiformes (including swifts and hummingbirds) remain cohesive in overall body proportions and internal anatomical details. All are medium to small sized birds, have big heads, big eyes, long wings, short feet, and, except for the nectarivore hummingbirds, big mouths. And cohesion is not just because of superficial similarities; similarities in the internal anatomy revealed the affinities between Apodiformes and Caprimulgiformes, and paraphyly of the later in the firs place (Mayr 2002 J Ornithol 143:82–97). Of course each subclade has its own characteristics, some of them unique among birds, but that is already reflected in the classification at the level of Family.
“5) I admit that if swifts would not exist, inclusion of hummingbirds and nightbirds in a single order would be more difficult to endorse. But swifts do exist and somewhat fill the gap between nightbirds and hummingbirds (or at least provide a “stepping stone”). Actually, together they tell a very interesting story of a clade of diurnal aerial specialists (swifts and hummingbirds) originating from a group of nocturnal birds, and there are very interesting fossils documenting this transition (Mayr 2002, Ksepka et al. 2013 Proc. Royal Soc. B 280).
“Therefore, so far I don't see compelling reasons for not adopting the more conservative solution of including all families into an expanded Caprimulgiformes, as in the Howard & Moore list.
Comments from Stiles: “YES. One important point glossed over by Santiago is that the idea of a broad Caprimulgiformes, implying that the Apodiformes evolved from nocturnal ancestors, is that such an argument ignores the reorganization of the visual system of nocturnal vs. diurnal birds. Evolution from nightbirds requires the loss and reacquisition of visual pigments as well as major morphological adjustments of the retina that seem to me to be highly improbable. Hence, I favor continued recognition of Apodiformes at the ordinal level, as well as recognition of Steatornithiformes and Nyctibiiformes, each of which have long been evolving independently since the Eocene. As I see it, this does not conflict with either the fossil evidence of Mayr or the Prum et al. calibrations (within the error bars for same). The “redundancy” of family and ordinal names in Steatornithiformes and Nyctibiiformes simply recognizes that speciation has been limited in these groups, at least since the Eocene; especially for Steatornithiformes, the extremely specialized diet and reproductive biology seems a likely reason for this.”
Comments from Pacheco: “YES. The two formulations (Van and Santiago) seem acceptable from the data, but I prefer to an arrangement in order level more directly adjusted to the phenotype (when possible) than a more comprehensive arrangement. I admit, however, that the strongest argument in my vote is this: ‘all of the caprimulgiform lineages are ancient (...) as old or older as most taxa we rank as orders’.”
Comments from Jaramillo: “YES. This is difficult, both suggested arrangements have problems as I see them. Santiago has detailed many of the issues. But in the end, putting all of these birds together in a single order I find more troubling.”
Comments from Areta: “YES. This is largely a matter of taste on how to better accommodate deep branches in the tree of life nomenclaturally. Based on morphological and ecological differences, it makes more sense to me to separate the bizarre Steatornithidae and Nyctibiidae (as well as Podargidae and Aegothelidae) in different orders than to lump them with Hemiprocnidae, Apodidae and Trochilidae, the latter of which (despite their notable differences) form a more coherent grouping. There is always the possibility of using the term Strisores to refer to all these families conforming a diverse assortment of birds.”
Comments from Robbins: “YES. There are a lot of assumptions, from problematic node support to even more questionable dating of these lineages, that should cause one to pause. However, if taken at face value, it seems reasonable to elevate Steatornithidae and Nyctibiidae to ordinal level. Nevertheless, I have no strong opinions on this and could easily be convinced to adopt Cracraft’s convention.”
Comments from Zimmer: “YES. The idea of treating Tufted Coquette, Great Potoo, Oilbird, Fork-tailed Palm-Swift and Long-trained Nightjar in the same Order, is extremely unpalatable to me. I understand the arguments for it, but I just can’t get there. I’d feel a lot better about elevating a few families to ordinal level, even given some of the uncertainties of the underlying data with respect to questions of lineage dating and node support.”