Diatryma geiselensis Fischer, 1978

Diatryma geiselensis Fischer, 1978

Differential diagnosis. The gastornithiform species from the Geisel valley differs from:

• Gastornis parisiensis Hébert, 1855 in: mandibular rami dorsoventrally deeper and processus coronoideus better developed; rudiments of furcula absent; trochlea metatarsi II reaching farther distally (other differences concerning the morphology of the scapulocoracoid and the tarsometatarsus are mentioned in the differential diagnosis of the taxon Diatryma ).

• Gastornis russelli Martin, 1992 in: being much larger; trochlea metatarsi II reaching farther distally (other differences concerning the morphology of the tarsometatarsus are mentioned in the differential diagnosis of the taxon Diatryma ).

• Gastornis laurenti Mourer-Chauviré and Bourdon, 2020 in: upper beak proportionally longer; mandibular rami dorsoventrally deeper and processus coronoideus better developed; trochlea metatarsi II reaching farther distally (other differences concerning the morphology of the tarsometatarsus are mentioned in the differential diagnosis of the taxon Diatryma ).

• Diatryma sarasini Schaub, 1929 in: trochlea metatarsi II proportionally larger and reaching farther distally; trochlea metatarsi III not asymmetrical (in D. sarasini the lateral rim of this trochlea reaches farther distally than the medial one).

• Diatryma gigantea Cope, 1876 in: mandible dorsoventrally deeper; furcula absent or more strongly co-ossified with scapulocoracoid (see description); body of coracoid pierced by a large foramen; tarsometatarsus stouter; trochlea metatarsi II proportionally larger.

Remarks. Meaningful comparisons with Diatryma regens , of which only a single pedal phalanx is known ( Marsh, 1894), are not possible. Zhongyuanus xichuanensis is based on a distal tibiotarsus

( Hou, 1980). This element exhibits a similar morphology in all gastornithiforms and shows some damage owing to compression and breakage in all

D. geiselensis specimens, so that only limited comparisons with Z. xichuanensis are possible.

Referred specimens. Fischer (1962, 1978) made only few attempts to assign to particular individuals the various isolated bones he described. Hellmund

(2013) presented a clearer overview of the association of these remains and estimated that they represent nine individuals (the new skull described in the present study was not included). Because some specimens lack exact stratigraphic data,

there remains a possibility that they stem from individuals with such data, and we determine a mini-

mum number of eight Diatryma geiselensis individuals in the collection of GMH (including the new skull).

• GMH Dia 2 (holotype scapulocoracoid); there are no exact stratigraphic data, but Fischer (1962, 1978) noted that the specimen is likely to be from site XIII, so that it possibly belongs to the same individual as GMH XIII-40-1954.

• GMH Dia 3 (right humerus lacking distal end); there are no exact stratigraphic data, but Fischer (1962) noted that the specimen may be from site XIII, so that it possibly belongs to the same individual as GMH XIII-40-1954.

• GMH Dia 8 (left femur); Hellmund (2013) considered the exact provenance of this specimen to be uncertain, but Fischer (1962) indicated that it stems from the “Unterkohle” of Neumark West and may belong to the same individual as GMH XIII-40-1954.

• GMH Dia 12 (left tarsometatarsus); no exact stratigraphic data exist, but the specimen may stem from one of the individuals found in the “Unterkohle” of Neumark West.

• GMH XIII-40-1954 (right femur, right tibiotarsus); from Neumark West (“Unterkohle”, early Lutetian, MP 11), found in 1954.

• GMH XIII-108-1954 (rib); from Neumark West (“Unterkohle”, early Lutetian, MP 11), found in

PALAEO- ELECTRONICA.ORG

1954; whether the specimen belongs to the same individual as GMH XIII-40-1954 cannot be determined.

• GMH XIV-258-1954 (proximal end of left tibiotarsus); from Neumark West (“Unterkohle”, early Lutetian, MP 11), found in 1954.

• GMH XIV-2476-1955 (proximal and distal ends of right femur); from Neumark West (“Unterkohle”, early Lutetian, MP 11), found in 1955.

• GMH XIV-4730-1956 (partial mandible), GMH XIV-658-1956 (partial pelvis), GMH XIV-2042- 1956 (left femur), GMH XIV-4628-1956 (right femur), GMH XIV-2091-1956 (left tibiotarsus), GMH XIV-4005-1956 (pedal phalanx), from Neumark West (“Unterkohle”, early Lutetian, MP 11), found in 1956.

• GMH XVIII-1178-1958 (skull); from Neumark Süd (“untere Mittelkohle”, middle Lutetian, MP 12), found in 1958.

• GMH IL-38-1969 (pedal phalanges); from Geiselröhlitz (“untere Mittelkohle”, middle Lutetian, MP 12), found in 1969.

• GMH XXXV-481-1963 (right femur, left tibiotarsus) and GMH XXXV-202-1963 (right tibiotarsus); from Neumark Süd (“obere Mittelkohle”, late Lutetian, MP 13), found in 1963; contrary to Hellmund (2013), we consider it likely that both tibiotarsi are from the same individual.

• GMH XLI-200-1968 (partial upper beak, mandible, atlas and three further vertebrae, rib, partial scapulocoracoid, partial pelvis, right femur and partial left femur, right tibiotarsus and distal portion of left tibiotarsus, right tarsometatarsus, six pedal phalanges); from Geiselröhlitz (“obere Mittelkohle”, late Lutetian, MP 13), found in 1968.

Remarks. Fischer (1967) reported a putative partial gastornithiform sternum from the Geisel valley

(GMH XIV-3979c-1956), but this specimen was misidentified and actually is a right ilium of the perissodactyl mammal Lophiodon . Incidentally, the specimen was published in a study, in which meta-

tarsals of Lophiodon were introduced as a new avian taxon (“Saurornis matthewsi”); this latter identification was corrected by Fischer (1987),

whereas the sternum was still assigned to the Gastornithiformes by Hellmund (2013). A putative gastornithiform humerus (GMH XIV-3979a-1956),

which was listed by Hellmund (2013), is the right radius of Lophiodon . A putative scapulocoracoid

(GMH XIV-3979b-1956), also listed by Hellmund

(2013), is likewise misidentified even though its true identity is elusive.

Notes on the comparative osteology of Diatryma geiselensis . Fischer (1978) described a partial upper beak (rostrum) of D. geiselensis , which belongs to the most complete individual GMH XLI-200-1968 ( Figure 1 View FIGURE 1 C-F). Together with two partial mandibles, this specimen until now constituted the only skull remains of D. geiselensis .

Here we report a nearly complete but strongly flattened skull from the collection of GMH, which was stored in two parts (neurocranium and upper beak) that have now been reassembled ( Figure 1 View FIGURE 1 A-B). The specimen (GMH XVIII-1178-1958) was mistaken for crocodilian remains ‒ as per the index card of the fossil ‒ and is from a smaller individual than the previously known D. geiselensis beak and mandibles. The fossil is only the second complete skull of a gastornithiform known so far, with the other being that of the D. gigantea skeleton described by Matthew and Granger (1917). Unfortunately, GMH XVIII-1178-1958 is strongly flattened and the bone surfaces of the specimen are shattered throughout, so that subtle osteological details cannot be recognized. The maximum length of GMH XVIII-1178-1958, from the caudal end of the left processus paroccipitalis to the tip of the beak, is 347 mm.

The culmen (dorsal ridge) of the upper beak of GMH XVIII-1178-1958 is less dorsally vaulted than in Diatryma gigantea and the tomia (cutting edges) are more strongly sigmoidally curved ( Figure 2A, C View FIGURE 2 ). The beak of this specimen is also less strongly dorsally vaulted than in the partial beak GMH XLI-

PALAEO- ELECTRONICA.ORG

200-1968 ( Figure 1G View FIGURE 1 ), which was described by Fischer (1978). The rostrum of the latter specimen (GMH XLI-200-1968) is mediolaterally wider than in D. gigantea ( Figure 1F, H View FIGURE 1 ); owing to the flattening of GMH XVIII-1178-1958, the width of its beak cannot be reliably assessed. The bone surface of the D. geiselensis beak exhibits numerous neurovascular foramina and vascular grooves. Mourer-Chauviré and Bourdon (2020) already commented on the shape of the nostril of D. geiselensis , which is preserved on the right side of specimen GMH XLI-200-1968, but was not figured by Fischer (1978). The nostrils of GMH XVIII-1178-1958 conform to those of GMH XLI-200-1968. They are of similar size to the nostrils of D. gigantea and open rostrally into shallow furrows at mid-height on the rostrum level with the quadratojugal bar ( Figure 1D View FIGURE 1 ); these furrows were also noted for G. laurenti and D. gigantea by Bourdon et al. (2016). The damaged surface of the upper beak of GMH XLI-200-1968 shows a “spongy” interior.

As in Gastornis laurenti and Diatryma gigantea (see Bourdon et al., 2016), there are a distinct nasofrontal hinge and lateral projections at the base of the upper beak. GMH XVIII-1178-1958 lacks the bones of the palate and the jugal bars are likewise not preserved. The orbits are not clearly visible in the specimen, but must have been situated rostral to the large processus postorbitalis ( Figure 2A View FIGURE 2 ). As in D. gigantea , the neurocranium is strikingly short relative to the entire length of the skull. Andors (1988: 135) noted a peculiar rostral restriction of the orbit of D. gigantea , which he considered to possibly be the result of a hyperdevelopment of the fossae temporales. The fossae temporales of GMH XVIII-1178-1958 are likewise very deep. In D. gigantea , a massive bony bridge connects the processus zygomaticus with the processus postorbitalis and forms a secondary temporal opening. This orbitozygomatic junction (sensu Elzanowski and Mayr, 2018) is not visible in GMH XVIII-1178-1958 and appears to be broken. The condylus occipitalis is very large and caudally protruding, as it also is in D. gigantea .

Two gastornithiform mandibles, which are from specimens GMH XIV-4730-1956 and GMH XLI-200-1968, are known from the Geisel valley ( Figure 3 View FIGURE 3 A-E). In both, the rami are dorsoventrally deeper than in Gastornis laurenti , G. parisiensis ( Figure 3 View FIGURE 3 F-G), and Diatryma gigantea ( Figure 2C View FIGURE 2 ). In both specimens from the Geisel valley, the symphysis mandibulae is wider than in G. laurenti , G. parisiensis , and D. gigantea ; whether this is an artefact of preservation is difficult to ascertain. The processus coronoideus is more prominent than in G. parisiensis and G. laurenti (see Mourer-Chauviré and Bourdon, 2020: figure 1) and narrower than in D. gigantea . As noted by Fischer (1962, 1978), the mandibular surface is distinctly pitted by neurovascular foramina and covered with vascular grooves. In both specimens, the rostral tip of the mandible is more strongly raised dorsally than in G. parisiensis .

The holotype of Diatryma geiselensis is a scapulocoracoid ( Figure 4 View FIGURE 4 M-N), which lacks exact collection data but is likely to stem from the “Unterkohle” of Neumark West ( Fischer, 1978) and therefore possibly belongs to specimen GMH XIII-40-1954. We identified a second, previously unrecognized partial scapulocoracoid in the collection of GMH, which is from the individual GMH XLI-200- 1968 and was stored amongst unsorted material ( Figure 4 View FIGURE 4 K-L). The specimen is distinctly larger than the scapulocoracoid of the holotype but otherwise shows an identical morphology. Fischer (1962, 1978) noted the presence of a foramen nervi supracoracoidei in D. geiselensis and its absence in D. gigantea ; furthermore, unlike in D. geiselensis , there is an elongated process extending from the cranial end of the scapula on the medial side of the scapulocoracoid of D. gigantea ( Figure 4O‒Q View FIGURE 4 ). This process was considered to be the acromion of the scapula by Matthew and Granger (1917). At the time of Fischer’s studies, no correctly identified scapulocoracoid of Gastornis was known. Martin (1992) showed that the shoul- der girdle bones identified by Lemoine (1881) do not belong to Gastornis and are not even of avian origin. Martin (1992) identified a previously undescribed scapulocoracoid of G. parisiensis from Mont de Berru, and photographs of this compound bone, which has a distinctive “boot-shaped” outline, are here for the first time shown ( Figure 4 View FIGURE 4 A-F). Unlike in D. geiselensis and D. gigantea , the scapula and coracoid of G. parisiensis are angled relative to each other, whereas both bones are situated in the same axis and plane in the Diatryma species. The coracoid of G. parisiensis exhibits a small foramen nervi supracoracoidei, and there is a narrow, rod-shaped process that is likely to represent the shaft of the furcula ( Figure 4 View FIGURE 4 A-B). Based on comparisons with the scapulocoracoid of G. parisiensis , the stouter process in the same area of the D. gigantea coracoid (which was longer at the time of the discovery of the specimen but has since lost its tip; compare Figures 4O and 4P View FIGURE 4 ) probably also represents the omal portion of a co-ossified furcula. Because there is a deep notch in the coracoid of D. gigantea ( Figure 4O View FIGURE 4 ), it is unclear whether the large foramen piercing the scapulocoracoid of D. geiselensis is indeed a foramen nervi supracoracoidei or whether it represents an opening that formed by fusion of the remnants of the furcular shaft with the coracoid (which would explain the absence of a furcula in D. geiselensis ). The facies articularis humeralis of D. geiselensis is less deeply incised into the body of the scapulocoracoid than in D. gigantea . As in G. parisiensis , a slit-like gap separates the ventral portions of the facies articulares humerales of the coracoid and scapula in the holotype coracoid of D. geiselensis . In both, D. geiselensis and D. gigantea , there is a marked longitudinal ridge along the midline of the ventral surface of the corpus of the coracoid.

PALAEO- ELECTRONICA.ORG

Even though fossils from the Geisel valley are often deformed by diagenetic flattening, this does not severely affect the two Diatryma geiselensis tarsometatarsi, which are largely three-dimensionally preserved, with only moderate dorsoplantar compression. These two tarsometatarsi are stouter than those of the Gastornis species, of which a complete tarsometatarsus is known ( G. parisiensis , G. russelli , and G. laurenti ; Figure 5 View FIGURE 5 ). The elongation indices (ratio length of bone to mediolateral width of proximal end) of the two tarsometatarsi from the Geisel valley are ~2.21 (GMH XLI-200- 1968) and ~2.78 (GMH Dia 12). For G. parisiensis , this index is 2.96 (MNHN L3092) and 3.32 (MNHN L3093), whereas it is 2.79 in D. gigantea (based on a complete tarsometatarsus figured by Andors, 1988: pl. 17) and 2.52 in a Diatryma -like tarsometatarsus from the London Clay of the Isle of Sheppey (see discussion). The ratio of the proximal (mediolateral) width of the bone to the width of the midsection of the shaft is 1.58‒1.77 in G. parisiensis (MNHN L3092 and MNHN L3093) and 1.97‒2.46 in D. gigantea (the larger value is based on the tarsometatarsus figured by Andors, 1988: pl. 17, the smaller value is from USNM 15118 [based on a photo provided by T. Worthy]). For D. geiselensis , this ratio is difficult to estimate, because part of the proximal end of the tarsometatarsus is broken in GMH Dia 12, whereas the shaft is damaged in GMH XLI-200-1968; for the latter specimen the ratio is ~2.24. The tarsometatarsus of D. geiselensis differs from the distal tarsometatarsus of the D. sarasini holotype in that the trochlea metatarsi II reaches farther distally and the trochlea metatarsi III has a symmetric shape (in G. sarasini , the lateral rim projects farther distally than the medial one). The two largely complete tarsometatarsi of D. geiselensis differ in size and some morphological features ( Figure 5I, K View FIGURE 5 ). Whereas the foramen vasculare distale is well

PALAEO- ELECTRONICA.ORG delimited in GMH Dia 12, it appears to be distally open in GMH XLI-200-1968, and the trochlea metatarsi II is proportionally larger in the smaller specimen GMH Dia 12 than in the larger GMH XLI-200-1968. In both specimens the trochlea metatarsi III is proportionally larger than in the Gastornis tarsometatarsi, and the trochlea metatarsi IV is mediolaterally wider. In GMH XLI-200-1968, the trochlea metatarsi IV reaches less far distally than the trochlea metatarsi II (the condition in GMH Dia 12 cannot be established, because the distal portion of the trochlea metatarsi IV is broken).