Narrow-beaked trogons from the early Eocene London Clay of Walton-on-the-Naze (Essex, UK)

We describe multiple partial skeletons of a new trogon species from the early Eocene London Clay of Walton-on-the-Naze (Essex, UK), which are among the oldest fossils of the Trogoniformes. Eotrogon stenorhynchus, gen. et sp. nov. has a much narrower and more gracile beak than extant trogons, which denotes different ecological attributes of the fossil species. Eotrogon stenorhynchus already had the heterodactyl foot characterising extant trogons, even though the trochlea for the second toe is smaller than in crown group Trogoniformes. Differences to extant trogons in the wing and pectoral girdle elements suggest that E. stenorhynchus was less adapted to short-term hovering, which may indicate different foraging techniques. We also report a partial tarsometatarsus from the early Miocene of France that is likely to belong to Paratrogon gallicus, a species previously only known from humeri. P. gallicus is the earliest modern-type trogon, and we show that the newly identified tarsometatarsus does not support the proposed referral of this species to the African taxon Apaloderma. We identify skeletal features that suggest a sister group relationship between Apaloderma and all other crown group Trogoniformes, but the exact affinities of Paratrogon remain poorly resolved. (http://zoobank.org/urn:lsid:zoobank.org:pub:73B64B84-11C2-4D50-8540-099CF86B6CA1).


Introduction
Trogons (Trogoniformes) are the only group of birds with a heterodactyl foot, in which the second toe is permanently reversed. These colourful birds are mainly insectivorous or frugivorous; they breed in tree holes and are able to dig their nesting cavities into rotten wood with their powerful beaks. The extant species occur in tropical or subtropical latitudes of Africa, Asia and the Americas, and trogons are particularly diversified in the New World, where most species and genera occur (Collar 2001). According to current taxonomies, such as the IOC World Bird List, the New World species are classified into the taxa Pharomachrus, Euptilotis, Priotelus, and Trogon, the three African species belong to Apaloderma, and the Asian species are assigned to Apalharpactes and Harpactes.
Sequence-based phylogenies supported a sister group relationship between the Trogoniformes and a clade, which includes bucerotiform, coraciiform, and piciform birds (Prum et al. 2015;Kuhl et al. 2021). The clade including the Trogoniformes, Bucerotiformes, Coraciiformes, and Piciformes was termed Eucavitaves (Yuri et al. 2013). However, molecular phylogenies yielded conflicting results concerning the interrelationships of the extant trogoniform species. Analyses of mitochondrial and ribosomal sequence data showed the African taxon Apaloderma to be outside a clade formed by the Asian and New World trogons (Espinosa de los Monteros 1998). By contrast, an analysis of combined nuclear and mitochondrial sequence data supported a nonmonophyly of New World trogons, which resulted as successive sister taxa of a clade formed by the Old World taxa Apaloderma, Apalharpactes, and Harpactes (Moyle 2005). Analyses of individual nuclear sequence data yielded inconsistent results, but an analysis of combined nuclear and mitochondrial data recovered a sister group relationship between Apaloderma and a clade including the Asian and New World taxa (Johansson and Ericson 2005). This relationship also resulted from an analysis of ultraconserved element loci (Oliveros et al. 2020).
Trogons have a scant Paleogene fossil record. The hitherto oldest member of the group is Septentrogon madseni from the early Eocene (54.5 million years ago [Ma]) Fur Formation in Denmark (Kristoffersen 2002a). The holotype of this species is an isolated neurocranium, but Kristoffersen (2002b) described further putative trogoniform remains from the Fur Formation, which have yet to be formally published.
Foshanornis songi from the early Eocene of China was likened to the Trogoniformes by Zhao et al. (2015), but the authors also noted differences from trogons in some skeletal features, such as a proportionally shorter crista deltopectoralis of the humerus. A distal tibiotarsus from the early Eocene (53.6-52.8 Ma) Nanjemoy Formation in Virginia (USA) was tentatively assigned to the Trogoniformes by Mayr (2016a). Weidig (2003) briefly described a postcranial skeleton of a possible trogoniform from the middle Eocene Green River Formation (Wyoming, USA), but the specimen still awaits its formal publication.
The oldest well-represented trogoniform specimens are from the latest early or earliest middle Eocene (~ 48 Ma) of Messel in Germany. This site yielded two articulated skeletons of Masillatrogon pumilio, a species that exhibits heterodactyl feet and was smaller than all extant trogons (Mayr 2005(Mayr , 2009. Primotrogon wintersteini from the early Oligocene (32-34 Ma) of the Lubéron region in France is also represented by two skeletons (Mayr 1999(Mayr , 2001. Fragmentary remains of Primotrogon-like trogoniforms were reported from the early Oligocene of Germany (Mayr 2005) and Belgium (Mayr and Smith 2013). A partial skeleton of an unnamed trogoniform exists from the early Oligocene of Switzerland (Olson 1976).
The early Miocene Paratrogon gallicus is the oldest fossil record of a modern-type trogon (and, in fact, the only pre-Pleistocene one). This species was described on the basis of two isolated humeri from the Saint-Gérand-le-Puy area in France (Milne-Edwards 1867-71;Mlíkovský 2002; according to latter author, one of these humeri appears to be lost). Mlíkovský (2002) also mentioned an undescribed record of P. gallicus from the early Miocene of the Czech Republic.
Here we report multiple partial skeletons of trogoniform birds from the early Eocene London Clay of Walton-on-the-Naze (Essex, UK), which stem from the collection of the late Michael Daniels (Fig. 1). The existence of a trogoniform tarsometatarsus in the Daniels collection was already mentioned by Mayr (1999: 432), but having been in a private collection until recently, the specimens remained undescribed, even though the tarsometatarsus and a partial skeleton were figured by Mayr (2022). In addition, we describe the first tarsometatarsus of P. gallicus and comment on the origin and interrelationships of crown group Trogoniformes.
Whereas the trogoniform fossils from the London Clay are the oldest fossils of Pan-Trogoniformes (the clade including the stem and crown groups), the specimens of Paratrogon are the earliest records of modern-type Trogoniformes. Therefore, the fossils span critical dates in the evolution of trogons and provide new information on the evolutionary history of these birds.

Material and methods
The fossils are deposited in the Royal Belgian Institute of Natural Sciences, Brussels, Belgium (IRSNB); the Geo-

Diagnosis
Differs from Masillatrogon, Primotrogon, and all crown group Trogoniformes in a mediolaterally narrower rostral portion of the upper beak. Furthermore differs from Masillatrogon in a longer acromion of the scapula (Fig. 2a, b) and a proportionally somewhat longer carpometacarpus (ratio ulna length: carpometacarpus length 1.9 versus 2.0-2.1 in M. pumilio). Distinguished from Primotrogon in that the coracoid has a longer processus procoracoideus and a hook-shaped processus acrocoracoideus. Differs from Foshanornis in that rostrum of upper beak is proportionally shorter and dorsoventrally deeper, humerus with longer crista deltopectoralis. Comparisons with Septentrogon are not possible owing to a lack of overlap in the fossil material (see comments below).

Taxonomic remarks
The only unambiguously identified fossil of Septentrogon madseni from the Danish Fur Formation is the holotype neurocranium, which belongs to a larger species than the fossil from the London Clay (see species diagnosis). The S. madseni holotype cannot be directly compared with the fossils from the London Clay, so we are unable to exclude the possibility that the fossils from the London Clay belong to the taxon Septentrogon. However, there is currently no basis for a referral to the latter taxon other than the similar age and geographical proximity of the respective fossil localities. A referral to Septentrogon would also be poorly substantiated, because there are undescribed fossils of other trogoniformlike birds in the Daniels collection (which appear to be more distantly related to the new taxon and crown group Trogoniformes), and it is currently not possible to unambiguously differentiate any of these from Septentrogon.

Diagnosis
As for genus. The new species is smaller than Septentrogon madseni; the limb bones correspond to those of Primotrogon wintersteini in their length (Table 1) and the neurocranium of P. wintersteini measures about 21 mm and is therefore distinctly shorter than that of S. madseni, which has a neurocranium length of 25 mm (Kristoffersen 2002a); a heterodactyl tarsometatarsus from the Fur Formation, which may belong to S. madseni (Fig. 2n;Kristoffersen 2002b), has a length of 15.0 mm (versus 11.3 mm in the new species). The new species is larger than Masillatrogon pumilio (Table 1).

Remarks
The specimens show some differences in size,  (Fig. 3a). Meanwhile, the fragile structure has fallen off the sediment, and even though its caudal section is now damaged, the overall proportions can be well assessed. Most notably, the beak of Eotrogon stenorhynchus, gen. et sp. nov. is much narrower than that of extant trogons and has a more delicate appearance. The beaks of Masillatrogon (Fig. 3h) and Primotrogon are likewise distinguished from the London Clay fossil and more similar to those of extant trogons (Fig. 3f, g). The tip of the upper beak of the new species is less ventrally deflected than in crown group Trogoniformes. The nostrils are large and ovate, and proportionally larger than in extant trogons. Furthermore, unlike in crown group Trogoniformes, there is no ossified nasal septum (septum nasi osseum). The holotype includes a nearly complete quadrate ( Fig. 3i-l), which differs from that of crown group Trogoniformes  in that the capitulum oticum does not form a medial projection, the incisura intercapitularis is more pronounced, and the condylus ventralis is less ventrally protruding and rostrocaudally wider. The holotype also comprises both pterygoids (Fig. 3q-t). As in extant trogons (Fig. 3u, v), but unlike in most other small land birds, the bone exhibits a distinct articulation facet for a basipterygoid process, which is proportionally smaller than in crown group Trogoniformes.
In NMS.Z.2021.40.87 the rostral section of the mandible is preserved (Fig. 3y). As it is, the mandible appears to be much narrower mediolaterally than that of extant trogons (Fig. 3z), but the two rami were glued together by the collector, which may have distorted the original shape of the mandible. The tip of the pars symphysialis is missing.
The os basiurohyale of the hyoid apparatus is rod-like and of similar shape to that of extant trogons (Fig. 3w, x); the caudal portion of the urohyale is broken.
Only a few vertebrae are preserved in the specimens. The axis ( Fig. 3aa-cc) has a proportionally shorter processus articularis caudalis than the axis of extant Trogoniformes ( Fig. 3dd-ff). As in extant trogons, the dens is very long.
The coracoid ( Fig. 4h-k) has a longer processus procoracoideus than the coracoid of Primotrogon and crown group Trogoniformes. Unlike in the latter, the tip of the processus procoracoideus is ventrally inflected. The cotyla scapularis is deeply concave, whereas it is shallow in Primotrogon and extant trogons (the transition from a cuplike cotyla scapularis to a flat facies articularis scapularis occurred multiple times independently within neornithine birds; Mayr 2021). The extremitas sternalis of the coracoid of NMS.Z.2021.40.89 shows an incipient medial projection of the shaft, which is very pronounced in extant trogons. The processus lateralis is longer and narrower in the sterno-omal direction than it is in crown group Trogoniformes.
The scapula (Fig. 4f) has a relatively straight blade and a long acromion. In concordance with the cup-like cotyla scapularis, the bone exhibits a well-developed tuberculum coracoideum.
The furcula (Fig. 4b, c) is narrowly U-shaped. As in extant Trogoniformes, the extremitas sternalis bears a welldeveloped apophysis furculae. The extremitas omalis also corresponds to that of extant trogons in its shape and is only slightly widened.
The cranial portion of the sternum is preserved in the holotype, whereas NMS.Z.2021.40.89 includes a fragment of the lateral margin of the bone. As in extant Trogoniformes, the spina externa is well developed and slightly bifurcated (Fig. 4a). The lateral margin of the sternum exhibits four processus costales.
No complete humerus is preserved in the specimens, but a reconstruction of the bone based on the overlapping portions from the left and right humeri of the holotype (Fig. 5c) indicates similar overall proportions to the humerus of ◂ crown group Trogoniformes. However, the proximal end is proximodistally narrower and there are several differences in detail to that of extant trogons. As in Masillatrogon, the caput humeri is more caudally deflected than in extant Trogoniformes. There is a small embossment in the distal portion of the incisura capitis. The crista deltopectoralis is long and of subrectangular shape, with a straight dorsal margin. Also, as in Masillatrogon, the tuberculum dorsale is notably smaller than in crown group Trogoniformes (Fig. 5i-k) and the early Miocene Paratrogon (Fig. 5g, h), and the crista bicipitalis is not as prominent and markedly convex as it is in extant trogons. On the distal end of the bone, the fossa musculi brachialis is more extensive and less deeply recessed than in crown group Trogoniformes. The cranial surface of the condylus ventralis bears an elongate depression.
The ulna (Fig. 6a, b) exceeds the humerus distinctly in length. The condylus ventralis forms a pointed and ventrally directed projection, which is a distinctive derived characteristic of trogoniform birds (Fig. 6b, c). The cotyla dorsalis has a somewhat less convex dorsal margin than that of extant trogons; as in the latter, it reaches farther distally than the cotyla ventralis. The cotyla ventralis is proportionally smaller than in crown group Trogoniformes and unlike in the latter it does not extend onto the olecranon.
The carpometacarpus (Fig. 6d, e) corresponds well to that of extant trogons (Fig. 6f, g) in its overall proportions. As in the latter, the spatium intermetacarpale is very wide, even though it is somewhat narrower than in crown group Trogoniformes (in Masillatrogon the spatium appears to be even narrower). Unlike in extant trogons, a small tuberculum intermetacarpale is present (Fig. 6e). The distal projection   of the os metacarpale majus is less developed than in extant trogons (Fig. 6f).
Unlike in crown group Trogoniformes, the caudal margin of the phalanx proximalis digiti majoris is convex (Fig. 6h,  i), whereas it is concave in the proximal section of the bone in extant trogons, which results in a cleaver-like shape of the phalanx (Fig. 6j). NMS.Z.2021.40.84 includes an os carpi radiale, which is proximodistally narrower than the corresponding ossicle of extant trogons (Fig. 6k, l).
A fragment of the pelvis is preser ved in NMS.Z.2021.40.86, but does not allow meaningful comparisons with the pelvis of extant trogons.
Both tibiotarsi are present in NMS.Z.2021.40.86 (Fig. 7a, b), and although the bones are not complete, their complementary preservation and overlapping sections allow a reconstruction of the original length of the tibiotarsus (Fig. 2k), which would have been shorter than the humerus. The crista cnemialis cranialis is more proximally projected than it is in crown group Trogoniformes, in which it does not protrude beyond the proximal articular facets of the tibiotarsus (Fig. 7c). In Masillatrogon the crista cnemialis cranialis is equally prominent (Fig. 2j). The distal end of the tibiotarsus corresponds well with a distal tibiotarsus from the early Eocene Nanjemoy Formation in Virginia, USA (Fig. 7e), which was tentatively assigned to the Trogoniformes by Mayr (2016a). As in the latter specimen, there is a prominence of subtriangular shape lateral to the pons supratendineus. The narrow condyli are widely separated. The sulcus extensorius is situated in the medial portion of the tibiotarsus; by contrast, this sulcus is centrally located in extant trogons (Fig. 7c).
The tarsometatarsus (Fig. 7f-i) resembles that of Apaloderma (Fig. 7m-p) in its proportions, whereas the tarsometatarsi of Harpactes (Fig. 7q-t) and Pharomachrus (Fig. 7y-bb) are stouter, and that of Trogon (Fig. 7u-x) has wider proximal and distal ends. With the trochlea metatarsi II being strongly plantarly deflected, the bone shows the characteristic morphology associated with a heterodactyl foot. The hypotarsus forms a prominent crista medialis (Fig. 7h); unfortunately, its lateral portion is damaged, so that the configuration of the hypotarsal canals/sulci cannot be discerned (Apaloderma differs from other extant trogons in that there is a sulcus rather than a canal for the tendon of musculus flexor hallucis longus; Fig. 7o, Mayr 2016b). The fossa infracotylaris dorsalis is distinct (Fig. 7f), whereas this fossa is very shallow or entirely absent in crown group Trogoniformes. The foramina vascularia proximalia are situated more closely together than they are in extant trogons; in Trogon the lateral foramen vasculare proximale is absent (Fig. 7u). The tuberositas musculi tibialis cranialis is more centrally situated than in crown group Trogoniformes, in which it is located near the medial margin of the shaft. As in extant trogons, there is no crista medianoplantaris and the small and shallow fossa metatarsi I is situated far proximally at the beginning of the distal third of the tarsometatarsus. The foramen vasculare distale is large as it is in Apaloderma, whereas the foramen is smaller in Asian and New World trogons (the foramen is particularly small and proximally situated in Trogon). The trochlea metatarsi II is strongly plantarly deflected, but unlike in extant trogons it does not form a medial flange. In distal view, the trochlea metatarsi II also does not show the edge-like medial projection seen in extant Trogoniformes (Fig. 7i, p). The trochlea metatarsi III is somewhat wider than in extant trogons. The trochlea metatarsi IV is slightly worn but appears to have been less cylindrical than in extant trogons.

Taxonomic remarks
Paratrogon gallicus was so far known from two isolated humeri; a photograph of the only currently traceable specimen is shown here for the first time (Fig. 5g, h). We identify a partial tarsometatarsus of a similar-sized trogoniform species from the geographical area where the previous Paratrogon fossils were found. Even though direct comparisons of the material are not possible owing to a lack of overlap in the bones, it is the most parsimonious assumption that the new partial tarsometatarsus belongs to P. gallicus or a closely related species.

Description and comparisons
The specimen is the first tarsometatarsus referred to P. gallicus. The shaft of the bone is slender as it is in Apaloderma (the tarsometatarsi of Harpactes and Pharomachrus have a stouter shaft). The distal end is mediolaterally narrower than in Trogon and Pharomachrus. As in extant Asian and New World trogons, but unlike in the African taxon Apaloderma, the foramen vasculare distale is very small. Unlike in Eotrogon stenorhynchus, the trochlea metatarsi II forms a pronounced medial flange, which corresponds to that of crown group Trogoniformes in its size. Compared to extant trogons, the distal margin of the trochlea metatarsi IV is oriented at a more acute angle relative to the longitudinal axis of the tarsometatarsus (in extant trogons, it runs perpendicular to the tarsometatarsus axis; Fig. 7); the lateral rim of the plantar articular surface of this trochlea is less prominent than in extant trogons but appears to be somewhat abraded in the fossil.

Evolutionary significance of the fossils from the London Clay
An assignment of the tarsometatarsus NMS.Z.2021.40.85 to the Trogoniformes is well supported by the plantarly deflected trochlea metatarsi II of the specimen, which indicates a heterodactyl foot and is a diagnostic apomorphy of the Trogoniformes. The tarsometatarsus is not preserved in the holotype of Eotrogon stenorhynchus, gen. et sp. nov., but the other postcranial elements closely resemble the corresponding bones of crown group Trogoniformes and various characteristic shared derived features allow an unambiguous identification, including a humerus with a pronounced fossa musculi brachialis, an ulna with a condylus ventralis forming a pointed projection, and a carpometacarpus with a wide spatium intermetacarpale.
The specimens from Walton-on-the-Naze represent the earliest fossil Trogoniformes, even thoughdepending on their unknown exact stratigraphic provenancethey may only be slightly older than the holotype of Septentrogon madseni (according to Kristoffersen 2002a, the S. madseni holotype has an age of 54.5 Ma, whereas the sediments of the Walton Member of the London Clay Formation date from 54.6-55 Ma ;Jolley 1996;King 2016;Collinson et al. 2016).
Eotrogon stenorhynchus is clearly outside a clade formed by Primotrogon wintersteini and crown group Trogoniformes (Fig. 8). Unlike in P. wintersteini and extant trogons, the coracoid of E. stenorhynchus has a long processus procoracoideus and a hooked processus acrocoracoideus; a similar coracoid morphology occurs in early Paleogene bucerotiform and coraciiform birds and is therefore likely to be plesiomorphic for Eucavitaves. Unlike in Primotrogon and crown group Trogoniformes, the cotyla scapularis of the coracoid is deeply concave. Furthermore, the tuberculum dorsale of the humerus is smaller in E. stenorhynchus than in P. wintersteini and crown group Trogoniformes.
From an ecomorphological point of view, it is of particular interest that E. stenorhynchus has a much narrower beak than Masillatrogon, Primotrogon, and crown group Trogoniformes. Unlike in extant trogons, the beak of E. stenorhynchus also lacks an ossified nasal septum, which in the extant species adds additional strength to the beak. The disparate shape of the beak suggests that E. stenorhynchus had a different diet or foraging technique than its extant relatives; as detailed below, this assumption is consistent with differences in the postcranial bones.
Cavity nesting is likely to be a plesiomorphic trait of Eucavitaves, but the extant representatives of the clade employ a great variety of substrates and not all are able to excavate cavities themselves. Trogons are known for the ability to dig nesting sites into rotten wood with their wide and robust beaks (Collar 2001). The more gracile beak of Eotrogon possibly indicates that the fossil taxon did not show this specialised behaviour to the same degree as extant trogons (if so, the specializations of crown group Trogoniformes may have evolved due to increased competition for secure breeding spaces in natural tree holes during the Cenozoic). The disparate beak shapes probably also account for other differences between Eotrogon and more advanced Trogoniformes, especially those concerning the morphologies of the quadratum and axis vertebra.
In their postcranial anatomy, early Eocene stem group Trogoniformes are more similar to the crown group taxa, but there are also some notable differences. On the one hand, E. stenorhynchus and Masillatrogon pumilio are distinctly smaller than all species of crown group Trogoniformes, with a very small size likely being plesiomorphic for trogons. Stem group representatives of the Bucerotiformes, which are the taxon branching next in the phylogeny of Eucavitaves, are equally small (Mayr 2000;Mayr et al. 2020), so that a small size may be plesiomorphic for Eucavitaves as a whole. Morphologically, early Eocene stem group Trogoniformes likewise resemble coeval stem group Bucerotiformes in, for example, the shape of the humerus and carpometacarpus and the short tarsometatarsus, and these similarities are probably also plesiomorphic for Eucavitaves.
E. stenorhynchus also shows some differences to extant Trogoniformes in the morphology of the pectoral girdle and wing bones. For example, the tuberculum dorsale of the humerus is smaller in the fossil species, and the coracoid has a longer processus procoracoideus and a more deeply excavated cotyla scapularis. We consider it likely that these differences are functionally correlated and due to a more strongly developed supracoracoideus muscle in extant trogons. This muscle inserts on the tuberculum dorsale of the humerus and elevates the wing. Its tendon passes through the triosseal canal of the shoulder joint, which is bounded by the procoracoid process of the coracoid. In trogons, flattening of the cotyla scapularis may also be related to increased capability for flapping flight, even though across neornithine birds as a whole there appears to exist no strong correlation between the morphology of the coracoscapular articulation and a particular flight technique (Mayr 2021).
Extant trogons forage by sallying flights from perches and employ a foraging technique termed sally-gleaning; trogons are also capable of short-term hovering in order to pluck fruits or flowers (Ávila et al. 1996;Collar 2001;Gonsioroski et al. 2021). The differences in the wing and pectoral girdle skeletons of early Paleogene and extant Trogoniformes suggest that the fossil species were less adapted to shortterm hovering and that these capabilities evolved in the trogoniform stem lineage. How exactly this relates to possible differences in feeding behaviour remains elusive, but we hypothesise that Paleogene stem group Trogoniformes may have been less sedentary than their extant relatives, which The short tarsometatarsi of early Paleogene stem group Trogoniformes indicate perching habits, and these birds certainly lived in forested environments. Among extant trogons, the larger species tend to be more frugivorous than the smaller ones, which are predominantly insectivorous (Collar 2001). Early Paleogene trogons are distinctly smaller than their extant relatives, which suggests that these birds were also insectivorous. That a predominantly insectivorous diet is plesiomorphic for crown group Trogoniformes is also indicated by the fact that the species of the African taxon Apaloderma, which is the sister taxon of other crown group Trogoniformes (see next section), are mainly insectivorous and do not eat fruit (Collar 2001).
Judging from the plantarly deflected trochlea metatarsi II as well as the position of the second toe in the articulated skeletons from Messel, E. stenorhynchus and Masillatrogon pumilio had heterodactyl feet. However, the trochlea metatarsi II of E. stenorhynchus does not form a marked medial flange, which guides the flexor tendon of the reversed second toe in crown group Trogoniformes (Fig. 7i, p). Possibly, this more pronounced heterodactyl foot of crown group Trogoniformes is related to their sedentary behaviour, even though it may also be due to other behavioural characteristics of trogons, such as the ability to excavate nesting cavities in tree trunks (which requires clinging to vertical surfaces).
The fossil record shows trogons to be an ancient group of birds. Already by the early Eocene, stem group Trogoniformes may have had a wide distribution across the Northern Hemisphere. The crown group representatives also underwent substantial range expansions from their centre of origin (which is likely to have been in Africa or Eurasia). Crown group Trogoniformes were unable to colonise Madagascar and the region east of Wallace's Line, with these areas having been separated by unsurmountable seaways from Africa and Asia, respectively, during most of the Cenozoic.

Phylogenetic interrelationships and origin of crown group Trogoniformes
As detailed in the introduction, sequence-based analyses yielded conflicting results concerning the interrelationships of crown group Trogoniformes. However, some studies  Oliveros et al. (2020); divergence dates of the fossil taxa are hypothetical. The numbered nodes are characterised by: 1 beak mediolaterally wide and of subtriangular shape, with width at base being more than half of beak length; 2 coracoid with shallow facies articularis scapularis, rounded processus acrocoracoideus, and short processus procoracoideus, humerus with large tuberculum dorsale; 3 caudal width of beak subequal to length. [Colour online] supported a sister group relationship between Apaloderma and all other extant trogons. From a morphological point of view, this is also suggested by the fact that, unlike in Asian and New World trogons, the hypotarsus of Apaloderma does not exhibit a closed canal for the tendon of musculus flexor hallucis longus ( Fig. 7o and Mayr 2016b), which is likely to represent the plesiomorphic hypotarsal condition. In further contrast to Asian and New World trogons, the tuberositas musculi tibialis cranialis of Apaloderma does not form a distinct tubercle and is situated more centrally (Fig. 7). Finally, and also unlike in Asian and New World trogons, the tarsometatarsus of Apaloderma exhibits a large foramen vasculare distale, which, by outgroup comparisons with Paleogene stem group Trogoniformes, probably represents a plesiomorphic feature.
Calibrated molecular data suggest an origin of the trogoniform crown group either in the Oligocene, 26.7 Ma (Moyle 2005), or at the Oligocene/Miocene boundary, some 23 Ma (Oliveros et al. 2020). These dates correspond well to the age of Paratrogon gallicus from the early Miocene of France, which is the sole pre-Pleistocene fossil record of a modern-type trogoniform. Therefore, the exact phylogenetic affinities of this fossil species (which was not used as a calibration point in the aforementioned studies) are of potential significance for dating the timing of the origin of crown group Trogoniformes.
Paratrogon gallicus was assigned to the extant taxon Apaloderma by Mlíkovský (2002), who noted that the "entepicondylar prominence" (= epicondylus ventralis; Fig. 5g, i, j) is larger and more distinct in the fossil species than in all extant trogons except for the species of Apaloderma. Humeri of Apaloderma were not available to us, but we do not see major differences in the humerus of Paratrogon and those of Harpactes and Trogon (Fig. 5g-k). Irrespective thereof, the morphology of the newly referred tarsometatarsus does not support an assignment of Paratrogon to the taxon Apaloderma, and the small foramen vasculare distale may suggest closer affinities to the clade including Asian and New World trogons. However, the distal margin of the trochlea metatarsi IV of Paratrogon is oriented at a more acute angle relative to the longitudinal axis of the tarsometatarsus than in all extant Trogoniformes, and with regard to the orientation of the distal margin of the trochlea metatarsi IV, Paratrogon corresponds to Eotrogon stenorhynchus (Fig. 7f, j). If a more obliquely oriented distal margin of the trochlea metatarsi IV is plesiomorphic for trogons, Paratrogon would be outside crown group Trogoniformes.
The exact provenance of the Paratrogon gallicus lectotype, the humerus, is not known, but most fossils from the Saint-Gérand-le-Puy are from the stratigraphic level MN2a (Mlíkovský 2002) and therefore have an age of about 21 million years (e.g., Mennecart et al. 2016). The new tarsometatarsus is from the locality Saulcet, the fossiliferous sediments of which are somewhat older and date from 21.5-23 Ma (De Pietri et al. 2011;Mennecart et al. 2016). Even though the currently known fossil material is not sufficient for an unambiguous placement of Paratrogon, it documents the presence of essentially modern-type Trogoniformes in the earliest Miocene and provides at least an approximate minimum age estimate for the origin of crown group Trogoniformes.
We note finally that extant trogons exhibit strikingly different tarsometatarsus morphologies, which appear not to have been commented upon by previous authors (Fig. 7m-z). Apart from disparate proportions, the bone differs in the size and numbers of the vascular foramina on the proximal and distal ends, with the foramen vasculare distale being particularly large in Apaloderma, and with Trogon lacking a lateral foramen vasculare proximale. The foramina vascularia proximalia serve for the passage of the arteriae metatarsales plantares, whereas the foramen vasculare distale transmits the arteria metatarsalis dorsalis (Midtgård 1982). The reduced size or complete loss of some of these foramina suggests that there are fewer, or smaller, arteries that supply the plantar surface of the foot, but the functional reasons for this pattern are unknown. According to Midtgård (1982: 561), metatarsal arteries could be reduced owing to an elongation of the tarsometatarsus, but in the case of trogons a reduction of the plantar arteries may be due to a limited use of the feet, with extant Trogoniformes most of the time sitting motionless on their perches. However, because this habit is common to all extant trogon species, further research is needed to identify particular anatomical or physiological attributes that led to a reduction of the plantar arteries.
Funding Open Access funding enabled and organized by Projekt DEAL.
Data availability All data supporting the findings of this study are available within the paper. This article has no additional data.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.