The Metaves Hypothesis

“Perhaps the one thing that Metaves have most in common is having been characterized by taxonomists as problematic.” — Fain and Houde, 2004.

In 2004, Fain and Houde published their Parallel Radiations paper, which proposed that most of the troublesome taxa (except bustards, cuckoos, cranes, and turacos) were part of a single basal clade in Neoaves, Metaves. Metaves totals to 13 orders containing over 900 species, almost 10% of all living bird species. Fain and Houde called the rest of Neoaves, comprising about 85% of living birds, Coronaves. They argue that Metaves and Coronaves both diversified into some of the same types of forms (the parallel radiations of the title).

The Fain and Houde analysis was based on a single gene, the 7th intron of β-fibrinogen. The overall picture they found is quite similar to arrangement presented here, although it differs in detail. In their analysis, Neoaves consisted of two parts, Metaves and Coronaves. Coronaves included a shorebird clade (Charadriae), and a waterbird clade roughly equivalent to the Pelecanae. The rest of Neoaves remained unresolved.

Subsequent analyses by Ericson et al. (2006a) and Hackett et al. (2008) included other genes in the analysis. This results in a more reasonable arrangement of many taxa. They still divide Neoaves into Metaves and Coronaves. At that level, the only real disagreement is whether hoatzins belong in Metaves (Fain and Houde; Ericson et al.) or Coronaves (Hackett et al.). Even with extra genes included, the Metaves/Coronaves split is still based on β-fibrinogen. However, Hackett et al. also include the 5th intron of β-fibrinogen, and this may be responsible for placing the hoatzin outside Metaves.

There had been hints before that some of the Metaves were related (e.g., grebes and flamingos), and that some bird families might be wrongly placed (e.g., tropicbirds), but proposal that they formed a single clade came as a surprise. Ericson et al. (2006a) followed this up with a more detailed analysis that probably does a decent job of portraying the higher taxonomy of much of Neoaves. Hackett et al. (2008) also recovered Metaves, except for the hoatzin, which ended up in Coronaves at the base of the waterbird clade (Pelecanae).

At present, we have reasonable confidence that owlet-nightjars, hummingbirds, and swifts group together, and all have been placed in Apodiformes. The frogmouths, oilbirds, nightjars, and potoos are believed to be their closest relatives, although genetic support for this is surprisingly weak. If correct, the larger group can be called Strisores. Grebes and flamingos also seem related, as do kagu and sunbittern, but their relationships to other, but neither group has strong affinities with anything else. The pigeons and doves are also problematic. Perhaps the sandgrouse are related, maybe the mesites, but again the divisions between them are deep and they aren't closely related to anything else. With only one gene pulling these disparate groups together, Metaves is controversial.

In particular, studies based on mitochondrial DNA do not support the Metaves hypothesis (e.g., Gibb et al. 2007; Slack et al. 2007; Brown et al. 2008; Morgan-Richards et al. 2008; Pratt et al. 2009; Pacheco et al., 2011). Metaves depends on the interpretation of one rather messy gene fragment (the seventh intron of the β-fibrinogen gene). Although there's not anything obviously wrong, it cannot be used in consistent fashion across such a broad swath of species. As it also lacks support from other genes, this leaves open the possibility that the Metaves hypothesis is just plain wrong. That said, most currently available alternate topologies do not give any better results for the taxa in Metaves, and major features seem to change with each subsequent analysis.

For example, the deep mitochondrial study by Morgan-Richards et al. (2008) groups tropicbirds with accipiters and kagu with woodpeckers and passerines while Gibb et al. (2007) attribute their own grouping of owls with parrots to long-branch attraction. Ericson et al. (2006a, supplement) also present a tree that excludes the β-fibrinogen gene, which places the tropicbirds as sister to the turacos and kagu and sunbittern sister to the cuckoos, while Brown et al. (2008) put the tropicbirds next to the penguins. It is also notable that the oilbird, potoos, frogmouths, and nightjars do not group together when the β-fibrinogen gene is ignored.

Although these mitochondrial studies raise issues concerning the Metaves hypothesis, they have not reached the point where they can produce a comprehensive phylogenetic tree. Indeed, the limited taxon sampling means that branches may move substantially as data is added. Just compare the recent effort by Pratt et al. (2009) with Morgan-Richards et al. (2008) or Gibb et al. (2007). Even Pacheco et al. (2011) punts on many bird taxa.

Another recent alternate is Cracraft et al. (2004). At present, this is the only modern alternative that is both sensible and truly comprehensive.

There are also alternative forms of the Metaves hypothesis. One is given by Ericson et al. (2006a) in the supplementary material. In that tree, the Hoaztin is basal in Neoaves. The next branch includes the rest of the Columbimorphae. After that, the next division is between the Strisores and Coronaves. Hackett et al.'s (2008) version is a bit different from either of the Ericson versions, with the absence of the hoatzin being the major difference. They also mention the possibility that Metaves may not be a clade, although it shows as a clade in their maximum likelihood estimates.

The taxonomy presented here presumes that Metaves is a natural grouping, containing birds that are more closely related to each other than to the rest of Neoaves. It relies on Fain-Houde (2004), Ericson et al. (2006a) and especially Hackett (2008) for the organization of the non-passerines. I have inserted several higher level groups to organize the tree. See the next page for more on Coronaves.

Plausible alternatives would group the Columbiformes and Mirandornithes with the Charadriimorphae (e.g., Brown et al. 2008; Pacheco et al., 2011, figure 1A) and put the Strisores in Passerae (Ericson et al., 2006, ESM-6; Wang et al., 2011).

The Structure of Metaves

Metaves breaks into two parts. The first contains what looks like a rather heterogeneous collection of families: sandgrouse, pigeons, tropicbirds, mesites, flamingos, grebes. The other group contains kagu and sunbittern in a basal position, followed by the nightjars and allies, hummingbirds, and swifts.

Following Huxley (1867), the nightjars and relatives (potoos, oilbird, frogmouths) and the Apodiform grouping of owlet-nightjars, swifts, treeswifts, and hummingbirds are collectively referred to as the Strisores. The position of the owlet-nightjars is surprising as they had previously been thought closer to the main nightjar group. The exact relation of the other Strisores remains uncertain (e.g., Barrowclough et al., 2006; Mariaux and Braun, 1996; Mayr, 2002). In fact, they usually do not form a clade in DNA studies! (The oilbirds are particularly problematic.) Hackett et al. (2008) and Ericson et al. (2006a) are two of the few where they are a clade.

Given the lack of solid information, and uncertainty about the relationships of these birds, I've put the Apodiformes (with owlet-nightjars) in one order, and separated the nightjars, frogmouths, potoos, and oilbird in their own orders Caprimulgiformes, Podargiformes, Nyctibiiformes, and Steatornithiformes. It may be possible to consolidate some of these orders in the future, or perhaps one will move to an entirely different place on the tree.

In comparison, in the Sibley-Monroe list, the families in Metaves are scattered, but many are recognized as problematic. The Sibley-Monroe list has the Podicipediformes (grebes) and Phoenicopteriformes (flamingos) as isolated orders with unclear affinities. The Phaethontidae (tropicbirds) are in the Pelecaniformes (Brown et al. (2008) group them with the hawks). The Rhynochetidae (kagu), Eurypygidae (sunbittern), and Mesitornithidae (mesites) are considered Gruiformes. The Pteroclidae (sandgrouse) are considered Charadriiformes and the Columbiformes (pigeons and doves) are an isolated order.

The Strisores fare better. Gill (1995) had already noted the Caprimulgiformes may be related to the Apodiformes.