Wednesday, August 19, 2020

Not Hawks, Not Owls, and Maybe Not Even Parrots: the Parrot-like Mystery Birds of the Eocene

The rapid diversification of surviving mammal and bird groups in the wake of the Cretaceous–Paleogene (K–Pg) mass extinction has routinely been characterized as a pivotal episode in the origin of modern ecosystems. However, the very rapidity at which it occurred has made it hard for us to gain a fine-scale understanding of the evolutionary transitions that took place during this event. Probably as a result of this, determining how Paleogene mammals were related to modern mammals has been notoriously difficult

We've been slightly more fortunate when it comes to Paleogene birds—though isolated bones are frequently challenging to identify, more complete specimens can often be confidently linked to specific modern groups. Recognizable stem-swifts looked fairly similar to modern swifts, stem-penguins looked fairly similar to modern penguins, stem-rollers looked fairly similar to modern rollers, and so on.

However, there are nonetheless some well-preserved Paleogene bird fossils that have defied easy classification. Among these are a range of small, arboreal birds that have been recently likened to parrots... but read on.

One of these mystery bird groups is the halcyornithids. Fossils of these birds are fairly common in some Eocene deposits of Europe and North America. In the late 1990s and early 2000s, halcyornithids were more commonly known as "pseudasturids", because Halcyornis, represented only by an incomplete skull from the London Clay Formation of England, was not recognized as a member of this group until it was compared to more complete halcyornithid specimens by Gerald Mayr in 2007. Now that Halcyornis has been assigned to the group, Halcyornithidae (named in 1972) takes priority over Pseudasturidae (named in 1998).

Other named halcyornithids are more completely known: Pulchrapollia, also from the London Clay Formation, is known from a partial skeleton, whereas essentially complete skeletons are known for Pseudasturides and Serudaptus from the Messel Formation in Germany and the two species of Cyrilavis (C. olsoni and C. colburnorum) from the Green River Formation in the United States.

Holotype of Cyrilavis colburnorum, from Ksepka et al. (2011).

Halcyornithids were small (thrush-sized) birds with zygodactyl feet (having both the innermost and outermost toe on each foot pointing backwards), a feature found today in cuckoos, woodpeckers, and parrots, among others. The beaks of halcyornithids were quite short and robust, but don't exhibit clear specializations for taking specific foods, so they may have been generalists that fed on both fruits and invertebrates (Mayr, 1998). The structure of the brain cavity in Halcyornis suggests that it had fairly well-developed senses of sight, smell, and hearing, though it may not have been a particularly acrobatic flier (Walsh and Milner, 2011). This is potentially consistent with the short, rounded wings preserved in a specimen of Pseudasturides (Mayr, 1998).

Preserved wing feathering of Pseudasturides, from Mayr (1998).

Halcyornithids have long been known to science. In fact, Halcyornis was the first fossil bird ever to be given a scientific name, having been named in 1825! As might be expected for a fragmentary fossil that was named so long ago, the taxonomic affinities of Halcyornis have been in a state of flux for a long time, and it has been variously considered a coraciiform (thus closely related to kingfishers and rollers) or even a gull-like shorebird.

However, more complete halcyornithid specimens have made them only somewhat less vexing. Cyrilavis olsoni (the type species of Cyrilavis) was originally described in 1976 as a species of Primobucco. Houde and Olson (1989) recognized that it was unlikely to be a member of Primobucco (now considered a stem-roller), and thought it was more likely closely related to jacamars and puffbirds (which are in turn close relatives of woodpeckers and toucans). Mayr (1998) doubted that halcyornithids were particularly close to jacamars and puffbirds, but also found little evidence linking halcyornithids with any specific group of living birds. He later proposed in 2002 that halcyornithids were stem parrots, based on features of their hindlimbs and vertebrae.

Similarities have further been noted between halcyornithids and another group of small Eocene birds, the messelasturids, which are currently known from two genera: Messelastur from the Messel Formation and Tynskya from the Green River Formation. Messelasturids share with halcyornithids, among other features, large "brow ridges" above the eye sockets (also found in many raptorial birds) and the absence of an air sac opening in the humerus (upper arm bone). Their outermost toe on each foot was at least partially reversed, making them semi-zygodactylous if not fully zygodactylous. Unlike halcyornithids, messelasturids were additionally raptor-like in having a sharply hooked beak. Combined with their long, curved talons, it's likely that messelasturids were carnivores that hunted small animals.

The skull of Messelastur, from Mayr (2011).

Messelastur was described on the basis of two isolated skulls with attached neck vertebrae, and given its hooked bill, it was originally considered a type of hawk. Mayr (2005a) reported on a more complete specimen and included both messelasturids and halcyornithids in a phylogenetic analysis. His results supported his previous hypothesis that halcyornithids were stem parrots, but found owls as the closest living relatives of messelasturids. However, following study of another new Messelastur specimen, Mayr (2011) concluded that the features linking messelasturids with owls had been misinterpreted or were poorly substantiated. In an updated phylogenetic analysis, he found messelasturids and halcyornithids to be each other's closest relatives, and both of these groups as stem parrots.

Stem parrot affinities have also been hypothesized for Vastanavis from the Eocene Cambay Formation in India, based on similarities to Quercypsitta, an uncontested stem parrot from the Eocene of France (Mayr et al., 2010). Vastanavis is known from numerous disarticulated specimens, which collectively represent all major limb bones, but its skull is currently unknown. It had short, semi-zygodactyl feet with long claws.

Support for placing all these taxa on the parrot stem appeared to be building in 2012, when Dan Ksepka and Julia Clarke described Avolatavis from the Green River Formation. The only known specimen of Avolatavis comprises a partial skeleton including the tail, pelvis, and hindlimbs. Its feet, originally described as zygodactyl but later reinterpreted as semi-zygodactyl by Mayr et al. (2013), were stout and look well-suited for grasping. In their phylogenetic analysis, Ksepka and Clarke recovered halcyornithids, Messelastur, Vastanavis, and Avolatavis all as stem parrots, with halcyornithids and Messelastur being close relatives of one another, as previously suggested.

Holotype of Avolatavis, from Ksepka and Clarke (2012).

However, in the years following, other developments in avian phylogenetics have demanded further reevaluation of these fossils. For starters, molecular analyses have been painting an increasingly clear picture of how parrots are related to other extant birds, which had formerly been a highly controversial subject. These analyses consistently place parrots in a diverse group of mostly arboreal birds called Telluraves, in which the closest living relatives of parrots are passerines (together forming the clade Psittacopasserae). Psittacopasserans are in turn the closest living relatives of falcons (forming the clade Eufalconimorphae). This new understanding has provided a more solid framework for selecting which modern taxa need to be considered while assessing the phylogenetic position of parrot-like fossil birds. Although Ksepka and Clarke (2012) had rightly included both passerines and falcons in their analysis, their sampling of other telluravians had been limited.

Phylogenetic relationships among Telluraves, based on a consensus of recent molecular analyses. Some uncertainty remains regarding the position of Accipitriformes (hawks and kin), Strigiformes (owls), and Coliiformes (mousebirds), but this topology is the most widely recovered by recent studies.

Other advances came from the paleontological front, namely the recognition that a group of extinct zygodactylous birds, the appropriately-named zygodactylids, were likely stem passerines. Modern passerines do not have zygodactyl feet, but if they evolved from ancestors that did, it raises the possibility that many of the hindlimb characteristics used to place fossil taxa as stem parrots were in fact ancestral features of psittacopasserans, not specific to the parrot lineage.

Mayr (2015a) presented the results of several phylogenetic analyses, in which he included a broad range of extant telluravians, along with putative stem parrots and stem passerines. He consistently found that Psittacopes, another zygodactylous bird from the Messel that had long been considered an unequivocal stem parrot, was more likely to have been a stem passerine. Furthermore, placement along the parrot stem lineage was also not supported for halcyornithids, Messelastur, Vastanavis, or Avolatavis when the analyses were constrained to recover Psittacopasserae and Eufalconimorphae. Halcyornithids were instead found as a grade of stem psittacopasserans, Messelastur in an unresolved position at the base of Eufalconimorphae, and Vastanavis and Avolatavis outside of eufalconimorphs entirely. Mayr observed that these supposed stem parrots actually lack several features shared by parrots and stem passerines, notably a deep groove where the outermost toe attaches to the foot.

Mayr (2015a) did not try constraining the telluravian relationships outside of Eufalconimorphae to conform to recent molecular topologies, so I had a go at doing so myself using his dataset, and here are the results. The relationships within Eufalconimorphae are congruent with his findings. Of note is that the clade uniting Avolatavis, Eurofluvioviridavis, and Vastanavis is more closely related to eufalconimorphs than to other telluravians, despite being excluded from crown Eufalconimorphae. Additionally, Eocuculus (a mystery bird outside the scope of this blog post) is recovered as only a distant relative to all the telluravian taxa included.

Interestingly, Mayr's study found Vastanavis and Avolatavis as close relatives, consistent with previously proposed similarities between the two (Mayr et al., 2013). Joining them as well was Eurofluvioviridavis, a Messel bird that also had stout, semi-zygodactyl feet. As its name suggests, Eurofluvioviridavis was initially thought to have been closely related to Fluvioviridavis from the Green River Formation (Mayr, 2005b). However, whereas Fluvioviridavis was later reappraised as a strisorean, telluravian affinities have been considered more probable for Eurofluvioviridavis (Nesbitt et al., 2011), with Mayr (2015b) noting similarities to Avolatavis and Vastanavis in particular.

More recently, Ksepka et al. (2019) included an extensive phylogenetic analysis of both extant and fossil telluravians in their description of the stem passerine Eofringillirostrum. They corroborated several of Mayr's results, including Psittacopes as a stem passerine and halcyornithids as stem psittacopasserans, though messelasturids, Vastanavis, Avolatavis, and Eurofluvioviridavis were not included in their study. As I remarked in my blog post on Eofringillirostrum, I'd be interested in seeing their dataset further expanded with these taxa.

Phylogenetic results from Ksepka et al. (2019). Note that "Afroaves" should be labeled Australaves.

And that is the state of the art. Although the concept that there used to be a great diversity of stem parrots distributed across the Northern Hemisphere was compelling for a time, it currently appears more likely that these parrot-like birds occupied a range of disparate positions within Telluraves. Their fossils may thus offer the tantalizing potential of shedding light on the ancestral anatomy and lifestyle of this extremely diverse group. Telluravians today include raptorial predators, insectivores, piscivores, frugivores, generalist omnivores, and just about every other avian niche in between, but it has been suggested based on the interrelationships among living species that their last common ancestor was raptorial. Might this hypothesis find support in the raptor-like features that have been identified in the parrot-like mystery birds—the hooked bill in messelasturids, brow ridges in halcyornithids and messelasturids, and large foot claws in several of these species? Perhaps further research will tell.