Reappraisal of early Miocene rails (Aves, Rallidae) from central France: diversity and character evolution

Amended diagnosis

Humerus with (1) small tuberculum dorsale (Fig. 1A,D); (2) elongated and dorsally concave crista deltopectoralis (Fig. 1B); (3) ventrally protruding epicondylus ventralis (Fig. 1H); (4) processus flexorius short, not surpassing distally the condylus ventralis (Fig. 1B,H); (5) fossa m. brachialis elongated and ventrally situated (Fig. 1B); (6) concave tuberculum supracondylare dorsale (Fig. 1H); tarsometatarsus with (7) deep fossa parahypotarsalis medialis (Fig. 1EE); (8) oval foramen vasculare distale (Fig. 1X); (9) trochlea metatarsi II not reaching distally past proximal margin of trochlea metatarsi III (Fig. 1EE); (10) trochlea metatarsi II without marked furrow dorsally (Fig. 1Y). Agrees with the early Oligocene rallid Belgirallus in (3), (4), (6), (8)–(10) and differs from it in (5).

Remarks

Apart from Palaeoaramides christyi, Cracraft (1973) recognized two additional species of Palaeoaramides from the Miocene of France, namely P. beaumontii (Milne-Edwards, 1869) from the middle Miocene (MN 6) of Sansan and P. minutus Cracraft, 1973 from the middle Miocene of La Grive-Saint-Alban (MN 7–8). The former, as Pararallus beaumontii (Milne-Edwards, 1869), was moved to the genus Palaeoaramides by Cracraft (1973) because its tibiotarsus did not differ from that of P. christyi. No tarsometatarsus is known from ‘Palaeoaramides’ beaumontii, and the humerus of an indeterminate species of Palaeoaramides, here for the first time described, is very different from that of the Sansan rail, differing mainly in the shape and orientation of the processus flexorius and in the proximodistally elongated tuberculum dorsale. Therefore, we do not consider Cracraft's generic referral to be correct. The small rail P. minutus, known from a tarsometatarsus, was only tentatively placed in Palaeoaramides, and was later referred to Aves incertae sedis by Mlíkovský (2002), who doubted it was a rail. Tertiariaporphyrula lungi Kuročkin and Ganea, 1972 from the late Miocene (MN 9–10) of Moldova was included in Palaeoaramides by Olson (1977) because Kuročkin and Ganea had erected the genus ‘Tertiariaporphyrula’ for Rallus beaumontii as well, which, as already detailed, was synonymized with Palaeoaramides by Cracraft (1973). The holotype of ‘Palaeoaramides’ lungi is a tibiotarsus and, unfortunately, this element does not display any distinctive features that would, beyond doubt, justify such a referral. ‘Palaeoaramides’ tugarinovi Kuročkin, 1980 from the middle Pliocene of Mongolia was described from a humerus and therefore was compared to the humerus of ‘Palaeoaramides’ beaumontii, which, as noted, does not correspond with the type material of Palaeoaramides Lambrecht, 1933.

Lectotype

MNHN Av. 2868, complete right tarsometatarsus, chosen by Cracraft (1973).

Locality and age

Saint-Gérand-le-Puy area, perhaps Montaigu, Allier, France; late Oligocene–early Miocene. For more details see De Pietri et al. (2011a).

Measurements

See Milne-Edwards (1869–1871), p. 148, and Cracraft (1973), p. 28.

Remarks

The material originally referred to Palaeoaramides christyi by Milne-Edwards (1869–1871) consists of a right tarsometatarsus (MNHN Av. 2868; Fig. 1W,X,AA,EE) and right tibiotarsus (MNHN Av. 2867; Fig. 1P,R). Milne-Edwards implied that these elements may have once been associated and thus belonged to the same individual. However, this is not explicitly stated ‘… parmis lesquels c'est trouvait une patte presque entière d'un Ralle de grande taille’, and as a result, Cracraft (1973) designated the tarsometatarsus as the lectotype of the species. We note that associated material in the localities of the Allier Basin is not uncommon and that it is indeed a possibility these bones belonged to the same individual, which would also be supported by the measurements provided by Milne-Edwards (1869–1871, p. 148).

Rallus eximius Milne-Edwards, 1869, a slightly smaller species than P. christyi, was synonymized with P. christyi by Lambrecht (1933), with the justification that the size difference for tarsometatarsi was within the range expected for a single species. This view was retained by Cracraft (1973), but R. eximius was maintained as a distinct species in Palaeoaramides by Brodkorb (1967). However, as noted by Milne-Edwards (1869–1871), there are morphological differences that are not congruent with this hypothesis (see below).

Description and comparisons

The tibiotarsi of rails show little variation among most taxa. As in all Rallidae except Gallinula chloropus, Tribonyx ventralis and species of Fulica Linnaeus, 1758, an ossified bridge over the sulcus for musculus fibularis brevis is present (Fig. 1R–T); its absence in one of the referred Palaeoaramides specimens (NMB Sau. 4984) is an artefact of preservation (see below). The lateral tuberositas retinaculum extensoris (for the transverse ligament) on the cranial surface adjacent to the sulcus extensorius (Fig. 1R) is elongate and ridge-like as in Aramides saracura – in Rallus aquaticus, for instance, it is short and subtriangular in shape. The tubercle lateral to the distal margin of the pons supratendineus (Fig. 1T) (for the abutment of the eminentia intercotylaris of the tarsometatarsus) is not as well developed as in species of Aramides Pucheran, 1845.

The condylus lateralis is craniocaudally very round in lateral aspect as in most examined rails. In Fulica atra and Gallinula chloropus, the cranial rim is flattened. As in many rails (e.g. Rallus aquaticus), the proximal margin of this condyle is strongly notched medially in cranial view. Contrary to the condition in most examined rails, however, the epicondylus medialis (Fig. 1R) is not prominent, which is also the case in Aramides saracura, Sarothrura pulchra and Himantornis haematopus.

Unlike the majority of extant rails, the hypotarsus of Palaeoaramides christyi bears no closed canals and displays a very short and straight crista medialis hypotarsi (Fig. 1AA). A similar morphology is nevertheless present in some of the examined rails, such as in Crex crex (Fig. 1DD), Hypotaenidia philippensis, some specimens of Himantornis haematopus and a fossil rail from the middle Miocene of Sansan. In all other rails, except Sarothrura pulchra (in which both canals are closed, see later; Fig. 2DD), an ossified ligament closes the sulcus for the tendon of m. flexor digitorum longus (fdl) and that for the tendon of m. flexor perforatus digiti II (fpd) remains open (Fig. 1CC). The hypotarsus is otherwise very uniform within rails (see, however, discussion concerning Paraortygometra Lambrecht, 1933 and Sarothrura pulchra), and whether this sulcus is open or closed appears to be the major difference among taxa. The fossa parahypotarsalis medialis is well marked in Palaeoaramides; such is also the case in species of Aramides, Himantornis haematopus, Laterallus leucopyrrhus and Rallina fasciata but not in the other examined rails.

The distal end of the tarsometatarsus of P. christyi is most similar in proportions and configuration of the trochleae to that of small rails such as Sarothrura pulchra and Laterallus leucopyrrhus, being mediolaterally compressed and with the trochlea metatarsi II turned far plantarly (Fig. 1X, compare with 1Z). It differs from all recent rails in that the articular surface of trochlea metatarsi II does not bear a marked furrow (contra Fig. 1Z), a condition also present in the early Oligocene taxon Belgirallus. The plantar projection on trochlea metatarsi II is not rounded but elongated (Fig. 1EE). This condition is present in species of Aramides, Sarothrura pulchra and Laterallus leucopyrrhus, as well as in the coeval Paraortygometra porzanoides. The distal end further differs from that of many extant rails in that the trochlea metatarsi II is distinctly proximally situated, with its distal rim not reaching the proximal base of trochlea metatarsi III in medial view (Fig. 1EE). Among the examined modern taxa, this condition is present in Sarothrura pulchra, Laterallus leucopyrrhus, Porzana porzana, Rallus aquaticus and Hypotaenidia philippensis; this feature is also present in Belgirallus (Mayr and Smith 2001; figure 6). As in the latter and unlike in most recent rails, the foramen vasculare distale (Fig. 1X) is not distinctly round but elongated, which had already been noticed by Cracraft (1973). In Belgirallus it is nevertheless more distally positioned.

Holotype

MNHN Av. 2865, right tarsometatarsus (see Cracraft 1973 for measurements).

Locality and age

Saint-Gérand-le-Puy area, perhaps Montaigu, Allier, France; late Oligocene–early Miocene. For more details refer to De Pietri et al. (2011a).

Amended diagnosis

Differs from the only slightly smaller and coeval Palaeoaramides christyi in (1) medial margin of base of hypotarsus more elongated plantarly (Fig. 1BB); (2) crista lateralis hypotarsi more elongated proximodistally (Fig. 1FF); (3) crista lateralis hypotarsi lacking sharp angle distally (Fig. 1FF, compare with 1EE); (4) distal end of tarsometatarsus overall more mediolaterally compressed, with trochleae metatarsorum II and IV not flaring as medially or laterally, respectively (Fig. 1Y).

Remarks

Although ‘Rallus’ eximius was synonymized with Palaeoaramides christyi by Lambrecht (1933), this classification was not retained by Brodkorb (1967), who assigned ‘R.’ eximius to Palaeoaramides, but noted the possibility it may be the female of P. christyi. The morphological differences between the type tarsometatarsus of Palaeoaramides christyi and that of ‘Rallus’ eximius, some of which were already reported by Milne-Edwards in his original description (1869–1871), suggest that Palaeoaramides eximius may indeed be a different species than P. christyi. Whether the differences we note here are a result of individual variation remains to be evaluated. Until more material is found, we recommend classifying Palaeoaramides eximius as a distinct species.

Referred specimens

Left humerus (NMB MA 3151); right coracoid (NMB Pa. 10489), left distal tibiotarsus (NMB Sau. 4984); right distal tarsometatarsus (NMB MA 7348).

Locality and age

Humerus: Montaigu, Allier, France; early Miocene, MN 2. Coracoid: Paulhiac, Allier, France; early Miocene, MN 1.

Measurements

Humerus – maximum length: 53.1; proximal width: 10.8; distal width: 7.9; minimum width of shaft: 3.5. Coracoid – length as preserved: 15.2; maximum length of facies articularis humeralis: 4.1; length from omal border of facies articularis humeralis to medial margin of cotyla scapularis: 4.9; width of shaft at level of foramen nervi supracoracoidei: 2.8. Tibiotarsus – distal width: 7.8; minimum width of shaft: 3.7. Tarsometatarsus – distal width: 7.8; minimum width of shaft: 3.1.

Remarks

It is not possible to refer the above-mentioned material to either Palaeoaramides christyi or P. eximius, as these species have similar-sized leg bones, and these elements could belong to either one. This is nevertheless the only size range known for species of Palaeoaramides from the early Miocene of France, and therefore, inclusion within this genus is justified.

The coracoid and humerus have not been described for Palaeoaramides sensu stricto, with the diagnosis referring to characters of the humerus in Cracraft (1973) being actually based on the humerus referred to the middle Miocene ‘Palaeoaramides’ beaumontii from the French locality of Sansan. Attribution of P. beaumontii to Palaeoaramides, which was based on a tibiotarsus, is likely to be erroneous, as the associated humerus is clearly different from those known from Saint-Gérand-le-Puy (personal observation).

Description and comparisons

The coracoid (Fig. 1M) is about the size of that of Aramides saracura (Fig. 1N). The foramen nervi supracoracoidei is large and rather sternally situated (Fig. 1M). A large foramen nervi supracoracoidei is also present in Himantornis haematopus (see Olson 1973; figure 2) and to some degree in the early Oligocene rail taxon Belgirallus (Mayr and Smith 2001; figure 6). For the most part, the poor preservation of the specimen does not allow for thorough comparisons. The crista procoracoidei (terminology after Livezey 1998; Fig. 1N) is incompletely preserved with its medial margin missing, suggesting that the corpus coracoidei was mediolaterally wider and unlike the coracoid of some rails in which the processus procoracoideus is distinctly set off from the medial margin of the corpus coracoidei (e.g. Gallinula chloropus, species of Porphyrio Brisson, 1760, and Fulica; Fig. 1O). The medial and lateral margins of the shaft appear to have been parallel, as observed by Mayr and Smith (2001) for Belgirallus, but this is also the case in species of Aramides (Fig. 1N) and some other rails. Overall proportions suggest this coracoid was elongate and slender with mediolaterally compressed omal and sternal ends (as in Rallus aquaticus, species of Aramides; Fig. 1N), whereas other rails (e.g. Fulica/Gallinula, Crex crex, Porzana porzana; Fig. 1O) display a stouter coracoid with widened omal and sternal extremities (see also Olson 1973; figure 2). As in species of Aramides but also Belgirallus, the lateral margin of the cotyla scapularis (Fig. 1M) does not project past the lateral outline of the corpus coracoidei.

The humerus (Fig. 1A,B) is approximately the size of the humerus of Aramides cajanea (Fig. 1C), but slightly more slender. The crista bicipitalis is proximodistally very short; its ventral margin is markedly round and ventrally protruding (Fig. 1B), being similar to Rallus aquaticus, and unlike the rather straight (e.g. Fulica atra and Gallinula chloropus) or less ventrally protruding crista bicipitalis (e.g. extant small rails, species of Aramides and Porphyrio; Fig. 1C) present in all other examined rails. The crista deltopectoralis (Fig. 1B) is only weakly developed cranially and lacks the subrectangular shape present in Crex crex, Porzana porzana and, to a lesser degree, representatives of Porphyrio. The crista deltopectoralis is otherwise well developed in the majority of the examined modern rails, a notable exception being species of Aramides. Dorsally, however, this crista lacks the platform-like structure (offset dorsally from the corpus) present in most rails (Fig. 1F, compare with 1G) and, unlike most examined recent taxa, is therefore markedly concave dorsally. This feature is likely to be related to the poorly developed tuberculum dorsale (Fig. 1D). This tubercle is better developed and proximodistally more elongated as a crista m. supracoracoidei in all other examined rails (e.g. Fig. 1E). A nearly identical tubercle is nevertheless present in a late Oligocene/early Miocene rail from the Mainz Basin in Germany, Rhenanorallus rhenanus Mayr, 2011 (figure 2). The fossa pneumotricipitalis (Fig. 1D) is shallower than in species of Aramides, resembling that of Sarothrura pulchra and Rallus aquaticus. The caput humeri (Fig. 1B) is dorsoventrally compressed and proximally prominent.

At the distal end, the processus flexorius is only weakly developed and does not project as far distally compared with that of extant rails (Fig. 1B, compare with 1C). The epicondylus ventralis (Fig. 1H) is more ventrally situated than the ventrodistal rim of the processus flexorius. Thus, the distal end matches that of representatives of Belgirallus (see Mayr and Smith 2001; figure 6). A similar condition is also present in species of Aramides, albeit the processus flexorius is more elongated in this taxon (Fig. 1C). A ventrally protruding processus flexorius is present in many rails with the exception of Porzana porzana, Crex crex (Fig. 1J), Fulica atra, Gallinula chloropus and species of Porphyrio, and also in the middle Miocene rail from Sansan, ‘Palaeoaramides’ beaumontii. In these taxa, the epicondylus ventralis and distal margin of the processus flexorius are at the same level ventrally (Fig. 1J). The tuberculum supracondylare ventrale is very prominent cranially (Fig. 1H). The distal end further resembles that of both species of Belgirallus and to a lesser degree Aramides saracura in the presence of a concave tuberculum supracondylare dorsale (Fig. 1H). The fossa m. brachialis (Fig. 1B) is medially positioned in Palaeoaramides and species of Aramides, whereas it appears to be wider and therefore more dorsally expanded in Belgirallus.

Despite the noted similarities with the extant genus Aramides (Fig. 1C), the humerus of Palaeoaramides is more slender in comparison and lacks the pronounced ventral curvature of the distal end, in which the epicondylus medialis reaches farther ventrally than, or to the level of, the crista bicipitalis (compare Fig. 1B with 1C). At the proximal end, the caput humeri is mediolaterally compressed compared with Aramides, the crista bicipitalis is proximodistally shorter, the crista deltopectoralis is markedly dorsally concave and proximodistally longer, and the tuberculum dorsale is distinctly smaller.

Type species

Baselrallus intermedius De Pietri and Mayr sp. nov.

Etymology

After the Swiss city of Basel, where the referred material is deposited (NMB).

Diagnosis

As for species.

Holotype

NMB MA. 2514, left humerus.

Etymology

Refers to the intermediate morphology between Palaeoaramides and that of other rails.

Diagnosis

Differs from Palaeoaramides in humerus with (1) proportionally shorter shaft; (2) crista deltopectoralis shorter, better developed and dorsocranially oriented (Fig. 1K); (3) tuberculum dorsale proximodistally elongated (Fig. 1E); (4) fossa m. brachialis extending further dorsally (Fig. 1I); (5) tuberculum supracondylare dorsale convex (Fig. 1I); (6) tuberculum supracondylare ventrale smaller and markedly convex; and (7) processus flexorius proximodistally wider (Fig. 1I, compare with 1H).

Type locality and age

Montaigu, Allier, France; early Miocene (MN 2).

Referred specimens

Left tibiotarsus (NMB SG 234004) with proximal end missing; perhaps Montaigu, Allier, France; early Miocene (MN 2).

Measurements

Humerus – maximum length: 39.7; proximal width: 8.5; distal width: 5.9; minimum width of shaft: 2.6. Tibiotarsus – distal width: 5.9; minimum width of shaft: 3.

Remarks

The tibiotarsus from NMB was considered by Cracraft (1973) to belong in Palaeoaramides christyi despite its much smaller size compared with the measurements of the other material attributed to this species (see table 7, p. 28, in Cracraft 1973). Some differences in morphology to the type tibiotarsus of P. christyi also do not support this referral. For this reason, we tentatively refer it to the same species as the humerus NMB MA 2514 based on size.

Description and comparisons

The humerus of Baselrallus intermedius gen. et sp. nov. (Fig. 1K,L) resembles that of Palaeoaramides in several features: it displays a short and ventrally protruding crista bicipitalis, a dorsally markedly concave crista deltopectoralis, a weakly distally developed processus flexorius and an epicondylus ventralis that is more ventrally situated than the distal margin of the processus flexorius. Contrary to the condition in Palaeoaramides, the crista deltopectoralis is slightly better developed, it is dorsocranially instead of cranially oriented (Fig. 1K) and is also proportionally shorter. The shaft of the humerus of Baselrallus intermedius is slightly shorter than that of Palaeoaramides. The tuberculum dorsale is better developed proximodistally (Fig. 1E) and resembles that of Heliornis fulica (Fig. 2G). At the distal end, the fossa m. brachialis (Fig. 1I) extends further dorsally than that of Palaeoaramides, and the tuberculum supracondylare ventrale is less prominent and markedly convex, as is also the case in Paraortygometra porzanoides (see below). Also, the epicondylus ventralis (Fig. 1I) is more proximally situated than in Palaeoaramides. The ventral margin of the processus flexorius is proportionally wider distoproximally than that of P. christyi, and its distal rim reaches to the level of the condylus ventralis (Fig. 1I).

Tibiotarsus NMB SG 234004 (Fig. 1Q,S,V) differs from the same element of Palaeoaramides in that the condylus medialis, in distal view, displays a straighter orientation cranially (Fig. 1V), as opposed to it flaring medially as in Palaeoaramides (Fig. 1U). Also, the lateral attachment for the transverse ligamenta is proximodistally more elongated than that of P. christyi (Fig. 1R).

Lectotype

MNHN Av. 2871, right tarsometatarsus (designated by Cracraft 1973). Although a humerus and femur were also attributed to Paraortygometra porzanoides by Milne-Edwards (1869), emphasis was put on the tarsometatarsus, which Mlíkovský (2002) considered to be the holotype of the species. Because Milne-Edwards's description is indeed based on three different bones, however, we follow Cracraft (1973).

Newly referred material (all from NMB)

Left coracoid lacking sternal end (NMB Sau. 4985); right proximal humerus (NMB MA 1108); right humerus lacking proximal end (NMB MA 1502); left humerus (NMB MA 3154); right humerus (NMB MA 9786); left humerus (NMB Ph. 2770); right humerus lacking proximal end (NMB SG 5710); left humerus lacking proximal end (NMB SG 7718); left humerus (NMB SG 7719); right humerus lacking proximal end (NMB SG 7721); left humerus (NMB SG 234005); right ulna (NMB MA 3165); left ulna (NMB SG 234006); left proximal ulna (NMB SG 234007); right ulna (NMB SG 234008); left femur (NMB SG 7709); right femur (NMB SG 19371); right femur (NMB SG 23705); right distal tarsometatarsus (NMB Ph. 3082); right tarsometatarsus (NMB SG 234009).

Locality and age

The newly referred material comes from the ‘Saint-Gérand-le-Puy’ locality of Montaigu, Allier, France; early Miocene, MN 2; with the exception of NMB Sau. 4985, which is from the locality of Saulcet, Allier, France; early Miocene, MN 1. The material (humeri, femora, tarsometatarsi) listed by Cracraft (1973) perhaps comes from Montaigu as well, but this is not certain.

Amended diagnosis

Small rail, humerus differs from Palaeoaramides in (1) caput humeri and proximal caudal margin of bone together with tuberculum dorsale joined to adjacent area distal to this margin by raised surface (‘platform-like’ surface; Fig. 2H); (2) tuberculum dorsale with proximodistally elongated crista m. supracoracoidei (Fig. 2A); (3) crista deltopectoralis more prominent cranially, with wide and pronounced convex dorsal margin (Fig. 2B,H); (4) caput humeri ventrodorsally wider and less protruding proximally (Fig. 2B); (5) tuberculum supracondylare dorsale poorly developed (Fig. 2B); (6) tuberculum supracondylare ventrale (Fig. 2B) less prominent and markedly round cranially. Differs from the late Oligocene–early Miocene Rhenanorallus rhenanus Mayr, 2011 in (2) and (7) processus flexorius (Fig. 2B) smaller and less ventrodistally protruding. Further differs from Palaeoaramides in (8) tarsometatarsus stouter (Fig. 2X); (9) hypotarsus with two closed canals for the tendons of the muscles flexor digitorum longus (fdl) and flexor perforatus digiti II (fpd, Fig. 2CC); (10) fossa parahypotarsalis medialis shallow.

Remarks

The newly referred material is attributed to Paraortygometra porzanoides because this is the only small rail known from localities of the Saint-Gérand-le-Puy area, and because the humerus, femur and tarsometatarsus match the type material, as described by Milne-Edwards (1867) and Cracraft (1973).

Several rails from the late early to the late Miocene of Europe and Asia have been referred to Paraortygometra by previous authors (e.g. Cracraft 1973; Cheneval et al. 1991; Mlíkovský 2002). Microrallus feifari Švec, 1983 from the early Miocene (MN 4) of Dolnice, Czech Republic, was synonymized with Paraortygometra porzanoides by Mlíkovský (2002), but no reason for this action was provided. Although similar in size and overall morphology, the illustrations of the distal humerus provided by Švec (1983) indicate that the processus flexorius at least had a very different shape compared with that of Paraortygometra. Cheneval et al. (1991) referred material from the early Miocene (MN 4a) of Thailand to Paraortygometra porzanoides although the trochlea metatarsi II extends farther distally in the tarsometatarsus of the Thai fossil than in that of P. porzanoides. Cracraft (1973) further tentatively referred several rallid remains from the middle Miocene of La Grive-Saint-Alban (MN 7–8), France, to P. porzanoides. While we have not examined this material, the difference in age alone renders it very unlikely for these rails to have belonged to the same species, which was also noted by Cracraft (1973). Fragmentary material from the late Miocene (MN 12) of Spain was also referred to P. porzanoides by Cheneval and Adrover (1993); we note, however, that aside from the temporal gap between these records of P. porzanoides, the tuberculum carpale of the ulna in the Spanish specimen exhibits a very different shape compared with that of the species from Saint-Gérand-le-Puy (Fig. 2N).

Description and comparisons

Although fragmentary, the previously unknown coracoid (Fig. 2J,K) is overall similar to the corresponding bone of some small rails such as Rallina fasciata and Sarothrura pulchra (Fig. 2L). The facies articularis humeralis is elongated in lateral direction, bearing a salient dorsal projection (Fig. 2K). As in Sarothrura pulchra and some other small rails, the processus acrocoracoideus (Fig. 2K,L) does not project greatly in the omal direction (compare with Fig. 1O). The ventral facet of the facies articularis clavicularis is particularly well developed in some rails (e.g. Zapornia flavirostra, Porzana porzana and Rallus aquaticus), but unfortunately, this condition cannot be determined for Paraortygometra porzanoides as the area is damaged. The foramen nervi supracoracoidei is small (Fig. 2J).

The sternal end of the coracoid is missing, and therefore, several features cannot be assessed. Only the omal rim of the impressio m. sternocoracoidei is preserved (Fig. 2K), which suggests it was considerably deep, as the bone becomes much thinner at this point. Similar to what can be observed in Sarothrura pulchra, there is a dorsally pronounced ridge marking its medial margin in the omal direction (Fig. 2K), albeit not as marked as in the latter. The humerus of Paraortygometra porzanoides (Fig. 2A,B) is elongated and slender compared with that of Sarothrura pulchra (Fig. 2D,E) and is overall most similar to that of Laterallus leucopyrrhus and Rallus aquaticus in proportions. The crista deltopectoralis is more strongly developed cranially than in Palaeoaramides and, unlike in this taxon, is dorsally convex as in all examined recent rails (Fig. 2H). The caput humeri (Fig. 2B) does not project as proximally and is not as dorsoventrally compressed as in Palaeoaramides, a morphology present in most rails. The caudal aspect of the proximal humerus is characterized by a proximodistally elongated tuberculum dorsale as a crista m. supracoracoidei (Fig. 2A), albeit it is not as elongated as that of Sarothrura pulchra (Fig. 2E); this feature is present in all examined recent rails. In Rallina fasciata and Laterallus leucopyrrhus, this tubercle is more prominent dorsally, proportionally shorter and very much set off from the crista deltopectoralis. Paraortygometra porzanoides differs from Rallus aquaticus in that the proximal and distal margins of the tuberculum dorsale are aligned at the same level (Fig. 2A). Contrary to Cracraft (1973), the fossa pneumotricipitalis is not deep and is indeed comparable to that of Palaeoaramides.

At the distal end, the fossa m. brachialis is ventrally well marked, becoming shallower dorsally (Fig. 2B). The scar is, however, not as deep as that of some taxa such as Crex crex, Zapornia flavirostra, Fulica atra and Gallinula chloropus. It closely resembles that of Baselrallus intermedius in that it is not so medially situated. As in Palaeoaramides, the epicondylus ventralis is more ventrally situated than the distal tip of the processus flexorius (Fig. 2B, see also figure 3 in Mayr 2010), and the latter does not project as distally as that of some rails (e.g. Rallina fasciata). As in most rails, Paraortygometra displays a distal ventral curvature of the humerus, being very similar to Sarothrura pulchra but displaying a proportionally longer shaft. The opposite condition (i.e. in which the crista bicipitalis projects distinctly further ventrally than the processus flexorius and therefore the distal end appears ‘straighter’) is found in Porzana porzana, Crex crex (Fig. 2C), Rallus aquaticus, Gallinula chloropus and Fulica atra. The tuberculum supracondylare dorsale (Fig. 2B) is poorly marked and resembles the one present in Sarothrura pulchra and Laterallus leucopyrrhus. The previously unknown ulna is proportionally short compared with the length of the humerus and thus very similar in proportions to that of Rallus aquaticus and Zapornia flavirostra. Proportionally longer ulnae were found in all other examined small rails. As in Sarothrura pulchra, the cotyla dorsalis (Fig. 2P,Q) is almost as wide as it is deep, and its dorsal margin is uniform throughout, in contrast to the truncated condition present in all other examined rails (Fig. 2R). The tuberculum ligamenti collateralis ventralis (Fig. 2N) is very well marked. Paraortygometra porzanoides cannot be distinguished from other rails based on additional features of the proximal end, and the distal end is likewise very uniform within Rallidae. Contrary to Sarothrura and most examined rails, however, the dorsally situated tubercle that borders the incisura tendinosa at the distal end (Fig. 2M) is very much pronounced; a similar condition was found only in Gallinula chloropus. The femur (Fig. 2S,T) closely resembles the corresponding bone of some small rails, such as Sarothrura pulchra and Laterallus leucopyrrhus. Paraortygometra differs from Gallinula chloropus and Fulica atra in that the impressio ansae m. iliofibularis (Fig. 2S) is situated much more proximally in these taxa, while being almost adjacent to the proximal portion of the trochlea fibularis in all other examined rails. It further differs from most other examined rails in that the fossa poplitea (Fig. 2T,U) is rather shallow – a condition also present in Sarothrura pulchra and Rallus aquaticus. It is also distinguished from Porzana porzana in that the distal end is not as mediolaterally compressed and from Crex crex in that the trochanter femoris is not as well developed. A very well-developed tubercle proximal to the fossa poplitea (Fig. 2S), situated slightly medially, is present in Paraortygometra porzanoides – such a feature was not found to be as developed in any of the examined rallids. A peculiar feature of P. porzanoides is the well-marked and rounded depression on the caudal surface of the condylus medialis (fovea tendineus m. tibialis, Fig. 2S), which was found to be absent in all examined Rallidae but present in the Messel rail (Messelornithidae Hesse, 1988) Itardiornis hessae Mourer-Chauviré, 1995 (Mourer-Chauviré 1995; figure 22, plate II). A very shallow and irregularly shaped scar is present in some species of Rallus Linnaeus, 1758. The tarsometatarsus of Paraortygometra porzanoides resembles the relatively shorter tarsometatarsus of Crex crex and Porzana porzana in proportions, being also stouter than in other examined rails. The morphology of the hypotarsus (Fig. 2CC) is nevertheless nearly identical to that of Sarothrura pulchra (Fig. 2DD), in which there are two closed canals for the tendons of the muscles flexor digitorum longus and flexor perforatus digiti II. As noted by Olson (1977), this morphology is also present in the extinct flightless rail Nesotrochis steganinos Olson, 1974, from the Quaternary of the Caribbean island of Hispaniola. In N. steganinos, however, the hypotarsus is plantarly more elongated (exceeding the dorsoplantar width of the cotyla medialis) and mediolaterally narrower than in Paraortygometra porzanoides and Sarothrura. The tarsometatarsus of N. steganinos does not resemble that of P. porzanoides in any other feature. Other rails, fossil and extant, also display two enclosed tendinal canals in the hypotarsus: Australlus disneyi (Boles 2005) from the Oligo-Miocene of Riversleigh, Australia (Boles 2005; Worthy and Boles 2011), and for example the extant gallinule Tribonyx ventralis (Fig. 2EE), albeit the overall morphology of the hypotarsus is different in these taxa.

As in Sarothrura pulchra (Fig. 2AA), the fossa hypotarsalis medialis is shallow (also in Heliornis fulica) and the impressio ligamenti collateralis lateralis is well marked.

The distal end of the tarsometatarsus is very similar to that of Palaeoaramides in that the trochleae are mediolaterally compressed, with trochlea metatarsi II strongly plantarly oriented and not flaring as medially as in all recent rails. As in Palaeoaramides, trochlea metatarsi II is distinctly proximally situated (but also in Sarothrura pulchra and a few other rails), and unlike in all recent rails, a marked furrow on its articular surface is absent. Contrary to Palaeoaramides, however, the foramen vasculare distale (Fig. 2V) is round rather than slit-like.