A new Eocene species of presbyornithid (Aves, Anseriformes) from Murgon, Australia

Abstract

The taxa found in an Eocene deposit, near Murgon, Queensland, the only pre-Oligocene Paleogene site recording a terrestrial vertebrate fauna from Australia, are very significant for the insight they provide concerning the evolution of the Australian biota. Here we resolve the identity of fossils previously referred to the Graculavidae, waterbirds of then unresolved affinities. We taxonomically describe the first bird to be named from this fauna, Murgonornis archeri gen. et sp. nov., Presbyornithidae. Our findings reveal that presbyornithids were widespread globally in the earliest Eocene, and that this family had a history on Australia from at least 55 Ma until ca. 24 Ma, when they disappear from the fossil record.

Trevor H. Worthy [[email protected]], College of Science and Engineering, Flinders University, GPO Box 2100, Adelaide 5001, SA, Australia; Vanesa L. De Pietri [[email protected]], University of Canterbury, School of Earth and Environment, Private Bag 4800, Christchurch 8140, New Zealand; R. Paul Scofield [[email protected]], Canterbury Museum, 11 Rolleston Avenue, Christchurch 8013, New Zealand; Suzanne J. Hand [[email protected]], Earth and Sustainability Science Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia.

Keywords:

• Paleogene birds
• Presbyornithidae
• Australia
• Tingamarra Local Fauna
• Murgon

PRESBYORNITHIDS (Aves, Anseriformes) are long-legged birds known chiefly from the Northern Hemisphere. First described from the Eocene Green River deposits of North America (Wetmore 1926) as wading birds ancestral to (modern) avocets and stilts, it was many years before their anseriform affinities were recognized and accepted, as reviewed already (Ericson 1997, 2000, Mayr 2009, 2017, 2022, De Pietri et al. 2016, Zelenkov 2021). While the oldest member of the group is possibly Teviornis gobiensis Kurochkin, Dyke & Karhu, 2002 from the Maastrichtian of Mongolia (Kurochkin et al. 2002, De Pietri et al. 2016, Mayr 2022, but see Zelenkov 2021), the earliest unambiguous records of presbyornithids stem from the Paleocene (Fig. 1). Three species are known from the Paleocene and Eocene of North America, namely Presbyornis pervetus Wetmore, 1926, P. recurvirostra (Hardy, 1959), and the rare larger P. isoni Olson, 1994 (Ericson 2000, Mayr 2022). From the Paleocene and Eocene of Asia, Bumbalavis anatoides Zelenkov, 2021 and indeterminate species of Presbyornis are known from the Tsagan-Khushu locality, southern Mongolia, while the co-occurring Presbyornis mongoliensis Kurochkin & Dyke, 2010 was referred to the phoenicopterimorph taxon Juncitarsidae (Zelenkov 2021). No certain presbyornithids are known from Europe (Mayr 2022, but see Harrison & Walker 1976, 1979); reports from elsewhere in the Northern Hemisphere reveal undescribed taxa from Ellesmere Island in northern Canada and Algeria (Mayr 2022).

Figure 1. Map of the world showing the distribution of fossil presbyornithids.

In the Southern Hemisphere, the presbyornithid record is sparse (Tambussi & Degrange 2013). Telmabates antiquus Howard, 1955 from the lower Eocene of Argentina, South America was referred to Presbyornithidae by Ericson (2000). A smaller taxon, Telmabates howardae Cracraft, 1970, from the same Argentine deposit was regarded as family incertae sedis by Ericson (2000) due to the fragmentary material. An unnamed form was reported from the Eocene of Argentina by Tambussi & Noriega (1998).

Most recently, presbyornithids have been recognized from Australia with the referral of Wilaru tedfordi Boles, Finch, Hofheins, Vickers-Rich, Walters, & Rich, 2013, previously considered of charadriiform affinities, to the family (De Pietri et al. 2016) and the description of Wilaru prideauxi De Pietri, Scofield, Zelenkov, Boles & Worthy, 2020 (although described in 2016, the species name was not made available until De Pietri et al. (2020) republished the description as a correction with the article’s Zoobank Registration). The Australian taxa are the youngest known presbyornithids, with W. tedfordi being of late Oligocene–early Miocene age, and W. prideauxi slightly younger at ca 23.4–22 Ma. Species of Wilaru had shorter legs/tarsi than other presbyornithids and, notably, they were the first presbyornithids to exist in sympatry with anatids (Worthy 2009, De Pietri et al. 2016).

Australia was part of Gondwana until the Tasmanian Gateway first separated it from Antarctica (Bijl et al. 2013) between 50 and 49 Ma. Deep water existed by ca 35.5 Ma (Stickley et al. 2004), making terrestrial dispersal between Australia and Antarctica highly unlikely after this time (Beck & Ebach 2017). Since then, Australia has been an island continent. Fossil birds are virtually unknown in Australia before the late Oligocene, from which time onwards significant avifaunas are known from central Australian fluvio-lacustrine deposits and from numerous sites in the Riversleigh World Heritage Area in northwest Queensland (Vickers-Rich 1991, Archer et al. 2006, Boles 2006, 2017, Worthy & Nguyen 2020).

The only fossil terrestrial vertebrate fauna from the Cenozoic of Australia predating the latest Oligocene is that from Murgon in Queensland (Godthelp et al. 1992, 2001, Beck & Ebach 2017). Murgon’s Tingamarra Local Fauna, dated at about 55 Ma, is well known for its fossil mammals with three marsupialiforms (Thylacotinga bartholomaii Archer, Godthelp & Hand, 1993, Chulpasia jimthorselli Sigé, Archer, Crochet, Godthelp, Hand & Beck, 2009, Archaeonothos henkgodthelpi Beck, 2015), a bat Australonycteris clarkae Hand, Novacek, Godthelp & Archer, 1994, a ?condylarth eutherian Tingamarra porterorum Godthelp, Archer, Cifelli, Hand & Gilkeson, 1992, and an australidelphian marsupial Djarthia murgonensis Godthelp, Wroe & Archer, 1999. It also includes the oldest members of the mekosuchine crocodilian radiation in Australia, represented by two species of Kambara Willis, Molnar, Scanlon, 1993 (Willis et al. 1993, Salisbury & Willis 1996), Australia’s oldest snakes (Alamitophis tingamarra Scanlon, 2005 and Patagoniophis australiensis Scanlon, 2005), a trionychid turtle (Murgonemys braithwaitei White, 2001), and Australia’s oldest frog Lechriodus casca Tyler & Godthelp, 1993.

The Tingamarra Local Fauna is also of significant interest because it contains the oldest Cenozoic avifauna from Australia (Boles et al. 1994), although until now no taxa have been named. Of international significance, the first taxa noted were the world’s oldest passerine fossils, described by Boles (1995, 1997), providing fossil evidence for the Southern Hemisphere origin of passerines advanced on molecular grounds (e.g., Ericson et al. 2002, 2014). Several fossils, notably a coracoid, distal humerus and two distal tibiotarsi, were attributed by Boles (1999) to waterbirds and placed in the ‘rubbish bin taxon’ Graculavidae (a group traditionally thought to contain ‘transitional shorebirds’), although Boles simultaneously recognized them as not differing from presbyornithids, which at that time were classified in the Graculavidae. It was only subsequently that presbyornithids were recognized as anseriforms (e.g., as reviewed above). Most recently, Elzanowski & Boles (2012, 2015) have described two fossil quadrates attributed to a large anseriform, perhaps like Anhima, and to a coraciiform, respectively.

Ongoing excavations and processing of bulk matrix from the Murgon quarries has resulted in an accumulation of more than 40 specimens of fossil birds. In this contribution, we reassess the specimens Boles (1999) recognized as ‘graculavid’-like presbyornithids in the context of the relatively recent recognition of presbyornithids from Australia. Further, we identify several more specimens that can be associated with them and describe a single new presbyornithid taxon for these fossils.

Material and methods

Site description

The avian fossils described here were recovered from a mudstone interval of the Oakdale Sandstone outcropping on Tingamarra, a property near Boat Mountain, on the outskirts of Murgon, 270 km northwest of Brisbane, Queensland (Cook et al. 2013). The Tingamarra sediments consist of green, lacustrine, authigenic illite and smectite clays with carbonate stringers.

Nomenclature

We follow Dickinson & Remsen (2013) for modern taxa. Names for specific bone landmarks mainly follow Baumel & Witmer (1993), but sometimes terms are taken from Matsuoka & Hasegawa (2007) or Howard (1929) for the humerus, and from Elzanowski et al. (2012) for the coracoid. Terminology for soft tissues such as ligaments and muscles derive from Stegmann (1978), Baumel & Raikow (1993), and Matsuoka & Hasegawa (2007). Nomenclatural acts: To conform to the requirements of the International Code of Zoological Nomenclature and hence make available under that Code the new names contained herein when published online, this article and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The LSID of this publication is: urn:lsid:zoobank.org:pub:CF7BFB2B-4D4A-4575-B0D1-66FE34A9F5AE

Institutional abbreviations

AMNH, American Museum of Natural History, New York, USA; AR, specimens held at UNSW Sydney, New South Wales, Australia; QM F, fossil collections of the Queensland Museum, Brisbane, Queensland, Australia; SAMA, South Australian Museum, Adelaide, South Australia, Australia; USNM, National Museum of Natural History, Washington, DC, USA.

Anatomical abbreviations

When listing fossils, elements (singular or plural) are abbreviated as follows: hum, humerus; rad, radius; cmc, carpometacarpus; MII.1, manus digit II, phalanx 1; cor, coracoid; tib, tibiotarsus; tmt, tarsometatarsus. They also may be identified as L, left, or R, right, which may be prefixed by p, proximal, d, distal, or s, shaft to identify the part (pt) of the element, so, e.g., dL tib is distal left tibiotarsus.

Comparative specimens

Presbyornis pervetus: cor—USNM 618183, USNM 616561–67 (four sternal and three omal parts); hum—USNM 483163 (cast), USNM 616568, USNM 618204, USNM 618180; ulna—USNM 616569–74; cmc—USNM 618168, USNM 618226, USNM 618227; tib—USNM 483165 (cast), USNM 618192–96, USNM 618236; tmt—USNM 483166 (cast), USNM 618175–76, USNM 618177, USNM 618178, USNM 618213, USNM 618237. Wilaru tedfordi: AMNH 10723, dL hum; AMNH 10836, dR hum; AMNH 11409, dL hum; SAMA (AMNH 11441), dR hum; SAMA (AMNH 11451), dR hum; AMNH 11470, dR hum; AMNH 11476, dL hum; SAMA (AMNH 11452), dL hum; SAMA P.48925 (=AMNH 11442), holotype L hum; AMNH 10995, dL, pR ulna; AMNH 10777, dL rad; AMNH 10800 dL rad; AMNH 10813, dR rad; AMNH 11464, R rad; AMNH 11469, dL rad; AMNH 11471, dR rad; AMNH 11432, L cmc; AMNH 11474 + 10998, L cmc; AMNH 11414, cranial pt L cor; AMNH 11426, L cor; AMNH 11428, 2 pts R cor; AMNH 11433, R cor; AMNH 11458, L cor; AMNH 11473, cranial pt L cor; AMNH 11478, cranial pt R cor; SAMA P.58228, cranial pt R cor; AMNH 10885, dR tib; AMNH 11417, dR tib; AMNH 11423, dL tib; AMNH 11424, dR tib; AMNH 11427, dL tib; AMNH 11430, dL tib; AMNH 11455, dL tib. Telmabates antiquus: AMNH 3170, holotype, multiple elements including hum, ulna, rad, cmc, and distal tib; AMNH 3166, L cmc in two parts; AMNH 3169, multiple elements, including dR hum, pL cmc; AMNH 3171, multiple elements, including dL hum; AMNH 3173, dR hum; AMNH 3180, multiple elements, including dL tib, dR hum, phalanx proximalis digiti majoris; AMNH 3181, multiple elements including coracoids, hum, rad and distal tib; AMNH 3182, associated pL ulna and pR cmc; AMNH 3183, L cor; AMNH 8167, multiple elements, including dL hum. See Howard (1955) for full list of specimens. Telmabates howardae: AMNH 3189, dR tib.

Systematic palaeontology

Order ANSERIFORMES Wagler, 1831

Family PRESBYORNITHIDAE Wetmore, 1926

Remarks

The distal humerus is referred to Presbyornithidae based on the following features derived from those identified by Ericson (2000) and De Pietri et al. (2016): a deep and a distinct fossa brachialis with diagonally oriented ovoid scar for musculus brachialis; processus flexorius present but distally short; an almost obsolete sulcus scapulotricipitalis due to a poorly developed epicondylus dorsalis; large, deep impressio for the musculus pronator superficialis and attachment surface for ligamentum collaterale ventrale abut; attachment of ligamentum collaterale ventrale large, proximodistally elongate and tilted distally; ventral facies of processus flexorius with two equal-sized distinct pits for musculus pronator profundus cranially and for musculus extensor metacarpi ulnaris caudally; indistinct processus or tuberculum supracondylaris dorsalis. The Murgon humerus and the other referred bones are very similar to those of Presbyornis pervetus, Wilaru tedfordi and Telmabates antiquus. Of the 20 defining characteristics of the humerus and coracoid listed in the Diagnosis, these taxa only differ from the Murgon fossils in one or two characters, and otherwise differ in only minor degrees of expression; for example, relative shaft width (coracoid) or length of the processus flexorius on the humerus. Such similarities extend to the referred carpometacarpus and tibiotarsus, as characterized below.

Murgonornis gen. nov.

Type species

Murgonornis archeri gen. et sp. nov.

Diagnosis

As for the type species.

Etymology

The genus is named after Murgon, the locality from near which the fossils derive, gender masculine.

Zoobank registration

Zoobank genus ID: LSIDurn:lsid:zoobank.org:act:FCD62B8B-02DE-43C1-BDB6-5CD5F0246C5D

Murgonornis archeri gen. et sp. nov.

(Figs 2A, E–G, I, M, N; 3 A, G–I, M–O, Y, AA)

Figure 2. Presbyornithid humeri in A–D, cranial, E, caudal and F–H, distal views, and coracoids in I–L, O, ventral and M, N, P–R, dorsal views. A, E, F, Murgonornis archeri gen. et sp. nov., holotype QM F23735, dL humerus, from the Eocene Tingamarra Local Fauna; B, Wilaru tedfordi AMNH 11452, dL humerus; C, G, Presbyornis pervetus, USNM 618180, dL humerus; D, H, Telmabates antiquus AMNH 3170, dL humerus. I, M, Murgonornis archeri QM F23649, L coracoid (2 parts); N, O, M. archeri, QM F30291, R acrocoracoid; J, Wilaru tedfordi AMNH 11426, L coracoid, P, AMNH 11473, cranial part L coracoid; K, Q, Presbyornis pervetus USNM 618183, L coracoid; and L, R, Telmabates antiquus AMNH 3181, L coracoid. Abbreviations: cd, condylus dorsalis; ce, crista epimarginalis; cs, cotyla scapularis; cv, condylus ventralis; ed, epicondylus dorsalis; ev, epicondylus ventralis; fac, facies articularis clavicularis; fah, facies articularis humeralis; fb, fossa brachialis; fcu, pit for musculus flexor carpi ulnaris; fns, foramen nervi supracoracoidei; fo, fossa olecrani; gr, groove for insertion of ligamentum acrocoracoclaviculare superficiale; is, impressio musculi sternocoracoidei; le, labrum externum; liv, linea intermuscularis ventralis; mpp, scar for musculus pronator profundus; mps, scar for musculus pronator superficialis; pbcr, scar for palmar branch of musculus extensor carpi radialis; pf, processus flexorius; pp, processus procoracoideus; ss, sulcus supracoracoideus; sul, sulcus; tsd, tuberculum supracondylare dorsale; tsv, tuberculum supracondylare ventrale. Scale bars = 10 mm.

Figure 3. Presbyornithid tibiotarsi in A–E, G, cranial, F, H, J, lateral and I, K, L, medial views; proximal carpometacarpi in M, U, W, ventral, N, V, X, cranial, and O–R, caudal aspects; S, T, distal carpometacarpi in dorsocaudal aspect; Y, Z, distal ulnae in dorsal aspect; and AA, BB, distal radii in ventral view. Murgonornis archeri gen. et sp. nov.: QM F23736, A, I, dL tibiotarsus; QM F23737, G, H, dL tibiotarsus; M–O, QM F30376, proximal L carpometacarpus; S, QM F61038, distal R carpometacarpus; AA, QM F20871, distal left radius. Wilaru tedfordi: B, AMNH 11440, distal R tibiotarsus; P, U, AMNH 11432, L carpometacarpus. Presbyornis pervetus: C, J, K, USNM 618236, distal R tibiotarsus; Q (mirrored), V, W (mirrored), USNM 618227, right carpometacarpus. D, L, Telmabates howardae: AMNH 3189, distal R tibiotarsus, E (mirrored), F, Telmabates antiquus: AMNH 3180, distal L tibiotarsus; R (mirrored), X, AMNH 3169, proximal right carpometacarpus; T (mirrored), AMNH 3166, distal left carpometacarpus; Z (mirrored), AMNH 3170, distal left ulna; BB, AMNH 3170, distal left radius. Y, Indeterminate anseriform: QM F23346, dR ulna. Abbreviations: cf, fovea carpalis cranialis; cl, condylus lateralis; cm, condylus medialis; ctl, crista marking the proximal attachment for the transverse ligament; dtl, distal attachment for the transverse ligament; em, epicondylus medialis; et, elongate tuberculum; fcc, fovea carpalis caudalis; lcl, insertion scar for ligamentum collaterale lateralis; pe, processus extensorius; pp, processus pisiformis; ps, pons supratendineus; se, sulcus extensorius; sf, sulcus for musculus fibularis; sul, sulcus for the origin of musculus ulnometacarpalis dorsalis; tav, tuberculum aponeurosis ventralis; tc, tuberculum carpale; 1, origin for the branch of the tendon for musculus extensor indicis longus; 2&3, scar for the retinaculum enclosing the tendon for musculus extensor digitorum communis; 4, origin of the ligamentum digitorum metacarpale. Scale bars = 10 mm.

Diagnosis

A presbyornithid of similar size to Wilaru tedfordi characterized by the following combination of characters: Humerus: (1) the ligament attachment on the tuberculum supracondylare ventrale forms a facet whose proximal margin extends slightly proximal to the condylus dorsalis, is proximodistally twice as long as wide, and is separated from the ventral facies by a sharp crista along its whole length; (2) a large pit for the proximal head of musculus pronator superficialis abuts the facet on the tuberculum supracondylare ventrale and extends proximal to the facet; (3) the ventral facies of the processus flexorius has two distinct pits: the deep more cranial one is for musculus pronator profundus and is separated from that for musculus extensor metacarpi ulnaris (=m. entepicondylo-ulnaris) more caudally by a low ridge; (4) the tuberculum supracondylare dorsale has low dorsal prominence and is impressed by two scars facing dorsally, the larger and most proximal scar is interpreted as for the dorsal branch of musculus extensor carpi radialis; (5) the scar for the palmar branch of musculus extensor carpi radialis forms a pit slightly more proximally situated so its distal margin is level with the proximal margin of the pit for the dorsal branch, but it lies ventrally adjacent to it on the cranial facies; (6) there is a distinct epicondylus dorsalis that projects dorsally of the condylus dorsalis with a similar dorsal prominence to the tuberculum supracondylare dorsale; (7) the sulcus humerotricipitalis is undivided but is deeper ventrally; (8) the fossa olecrani is deep and excavates the base of the processus flexorius such that this process is ventrodorsally compressed and narrow at the level of the pits ventrally; (9) the distal margin of the fossa musculi brachialis is well defined and notably deepened at a level proximal to the tuberculum supracondylare ventrale and so is well proximal to the condylus dorsalis; (10) there is a deep, narrow sulcus between the condylus ventralis and condylus dorsalis cranially; (11) the two condyles have near equal cranial elevation; (12) the condylus ventralis in distal aspect is very rounded and appears hemi-spherically convex cranially, not somewhat flattened; (13) the processus flexorius is distally short, much less than the condylus ventralis. Coracoid: (14) is relatively short with a stout shaft; (15) the facies articularis clavicularis forms a smooth slightly medially convex uninterrupted curve that sternally forms a straight crest overhanging the sulcus supracoracoideus and is not hooked sternally at its ventral tip; (16) the sulcus supracoracoideus is non-pneumatic, deepened adjacent to the facies articularis humeralis, undercuts the facies articularis clavicularis, and the deepened part extends ventrally across two-thirds of the sulcus; (17) the cotyla scapularis is deep and round; (18) a small foramen nervi supracoracoidei passes through the processus procoracoideus close to the cotyla scapularis and adjacent to the shaft, and within the foramen there is an opening into the corpus of the bone; (19) the impressio musculi sternocoracoidei is shallow with lineae musculares traversing it at approximately 45° to the long axis of the bone; (20) ventrally, the surface near the sternal end has a distinct sulcus located medial to the linea intermuscularis ventralis that deepens sternally.

Wilaru tedfordi differs by: Humerus—the distal part of the ligamental insertion surface on the tuberculum supracondylare ventrale is more deeply excavated distally and especially ventrally in W. tedfordi, so that it faces distoventrally and has no defined margin separating it from the ventral surface, whereas it forms a distinct facet in M. archeri; the ventral side of the fossa olecrani excavates the base of the processus flexorius to a greater extent in W. tedfordi, resulting in the process being more ventrodorsally compressed. Coracoid—lack of a distinct sulcus (20) medial to the linea intermuscularis ventralis. Carpometacarpus—lack of a distinct fovea carpalis cranialis impacting the cranial margin of the trochlea carpalis (see below).

Presbyornis pervetus, of a similar age, known only from North America, differs notably by: humerus—with a more ventrally projecting epicondylus ventralis, a more distally projected processus flexorius, a proximodistally shorter facet on the tuberculum supracondylare ventrale, a less prominent epicondylus dorsalis (6) (see Fig. 2F cf. 2G), and a more lateromedially compressed ridge extending proximally from this tuberculum. Coracoid—has a slightly more slender shaft and lacks the ventral sulcus (20) medial to the linea intermuscularis ventralis. Carpometacarpus—having a more elongate distal synostosis of the metacarpals (see below).

Telmabates antiquus of a similar age, known only from Argentina, is a larger species, e.g., humerus distal width 17–18 mm, which differs further as follows: humerus—the epicondylus ventralis is less prominent, not extending ventrad of the tuberculum supracondylare ventrale and the epicondylus dorsalis is greatly reduced by comparison (6). Carpometacarpus—the ventral rim of the trochlea carpalis is more proximodistally elongate so that the length is greater than width, whereas in M. archeri, the width is greater than length. Coracoid—the shaft is narrower, less than the diameter of the cotyla scapularis, whereas the shaft is wider than the cotyla in M. archeri. Tibiotarsus—with a less prominent epicondylus medialis, a narrower sulcus extensorius at the proximal side of the pons supratendineus, and the distal margin of the pons is at right angles to the shaft axis rather than sloped proximomedially (see below).

Etymology

The species honours Michael (Mike) Archer for his outstanding contribution to vertebrate palaeontology in Australia. His energy and interests have driven the exploration of the fauna at Murgon and realized its crucial role in understanding evolution of vertebrates in Australia. Moreover, those efforts over several decades have driven understanding in reptile, bird, and mammal evolution through exploration of numerous faunas throughout Australia and elsewhere. In doing so, numerous students were introduced to palaeontology and many careers were forged under his influence.

Zoobank registration

Zoobank species ID: LSID urn:lsid:zoobank.org:act:DEC11ED4-D52B-4421-9FC9-7BBC95BBB28A

Holotype

QM F23735, dL humerus (Fig. 2A, E–G).

Paratypes

QM F23649, L cor, in two parts (Fig. 2I, M); QM F30291, R acrocoracoid (Fig. 2N, O); QM F20871, dL rad (Fig. 3AA), QM F30376, pL cmc (Fig. 3M–O); QM F61037, dR cmc; QM F61038, dR cmc (Fig. 3S); QM F23736, dL tib missing condylus medialis (Fig. 3G, H); QM F23737, dL tib missing condylus lateralis (Fig. 3A, I).

Referred specimens

QM F24447, pL manus phal. II.1; QM F20239, pR manus phal.II.2; QM F31298, pR manus phal.I.1. Tentatively referred is the slightly smaller QM F61039, R manus phal. I.1.

Type locality, unit and age

All fossils are from a mudstone interval in the Oakdale Sandstone exposed on the Tingamarra property of J. and M. Porter, Boat Mountain region, near Murgon, southeastern Queensland, Australia, 26°S 152°E; Tingamarra Local Fauna. Age: Potassium/argon dating of authigenic clays gave a minimal age of 54.6 ± 0.05 Ma for this site (Godthelp et al. 1992), i.e., earliest Eocene, which is congruent with biocorrelative evidence (Hand et al. 1994, Black et al. 2012, Beck & Ebach 2017).

Description and comparisons

Comparisons focus primarily on Wilaru tedfordi from the late Oligocene–early Miocene (ca 24–22 Ma) of Australia, the similar-sized Presbyornis pervetus from the Eocene of North America, and the slightly larger Telmabates antiquus from the Eocene of Argentina, which is the only named Eocene Southern Hemisphere presbyornithid.

Humerus

Measurements of the holotype dL humerus QM F23735 are distal width 13.4 mm, depth condylus dorsalis 7.3 mm. In addition to the features 1–12 of the Diagnosis, the humerus of Murgonornis archeri (Fig. 2A, E, F) has a small scar for the distal head of musculus pronator superficialis adjacent to the distal end of the tuberculum supracondylare ventrale; the sulcus scapulotricipitalis is very shallow; the origin of musculus extensor digitorum communis (Matsuoka & Hasegawa 2007, fig. 4) is immediately distal to the dorsal branch of musculus extensor metacarpi radialis on the dorsal surface of the tuberculum supracondylare dorsale and forms a slightly smaller pit; immediately distal to the pit for origin of musculus extensor digitorum communis is a shallow sulcus on the dorsal facies that extends from a deep pit in the epicondylus dorsalis that is mainly for the origin of musculus flexor carpi ulnaris (Matsuoka & Hasegawa 2007, fig. 4); this latter pit is bound by prominent distal and cranial margins and is weakly subdivided into two parts, with the more caudal part for the origin of musculus supinator shallower; proximal to this deep pit on the epicondylus dorsalis and caudal to the origin of musculus extensor digitorum communis is a distinct sulcus.

Humeri of Wilaru tedfordi (Fig. 2B) share all of these features with Murgonornis archeri QM F23735, differ only as described in the Diagnosis, and thus are the most similar among presbyornithids.

Humeri of Presbyornis pervetus are of similar size, e.g., USNM 618180 (Fig. 2C, G; see De Pietri et al. 2016), but differ from M. archeri QM F23735 by a more ventrally projecting epicondylus ventralis resulting in a concave, not straight, ventral profile; a more distally projected processus flexorius that results in less distal projection of the condylus ventralis beyond a line linking the distal margins of the condylus dorsalis and processus flexorius; a proximodistally shorter facet on the tuberculum supracondylare ventrale; a more lateromedially compressed ridge extending proximally from this tuberculum; and a less prominent epicondylus dorsalis.

Telmabates antiquus (Fig. 2D, H) is a larger species, e.g., humerus distal width 17–18 mm (cf. Table 1) and differs the most from Murgonornis archeri among compared presbyornithids. As noted in the Diagnosis, the humerus has a less prominent epicondylus ventralis that barely extends ventrad of the tuberculum supracondylare ventrale; the facet on the tuberculum supracondylare ventrale is flatter and parallel to the caudal shaft facies compared to slightly more proximally elevated in M. archeri; the pit for the proximal head of musculus pronator superficialis abutting the facet on the tuberculum supracondylare ventrale is smaller and does not extend proximal to the facet; and the epicondylus dorsalis has less projection dorsally of the condylus dorsalis.

Table 1. Measurements (mm) of humeri of Wilaru tedfordi compared to Murgonornis archeri.

Taxon	Catalogue number	Element	Distal width	Depth dorsal condyle
Wilaru tedfordi	AMNH 10723	dL hum	14.5	7.6
Wilaru tedfordi	AMNH 10836	dR hum	14	7.5
Wilaru tedfordi	AMNH 11409	dL Hum	13.5	7.8
Wilaru tedfordi	SAMA (AMNH 11441)	dR hum	14.6	8.5
Wilaru tedfordi	SAMA (AMNH 11451)	dR hum	14.4	8.2
Wilaru tedfordi	AMNH 11470	dR hum	13.4	7.4
Wilaru tedfordi	AMNH 11476	dL hum	14.6	8.2
Wilaru tedfordi	SAMA (AMNH 11452)	dL hum	15.1	8.3
Wilaru tedfordi	SAMA P.48925 = AMNH 11442 Holotype	L hum	15.1	8.8
Murgonornis archeri	QM F23735	dL hum	13.4	7.3

Radius

The paratype dL radius (Fig. 3AA; QM F20871) is referred to Murgonornis archeri because it reveals the following features that are essentially identical to bones referred to Wilaru tedfordi and those of Telmabates antiquus: (1) the ventral facies is dominated by a markedly prominent elongate tuberculum (see Howard 1955) of unknown function that extends from a somewhat proximodorsal position distoventrally to end adjacent to the tuberculum aponeurosis ventralis; (2) the tuberculum aponeurosis ventralis is marked by a large round flat insertion scar on the ventromedial corner for aponeurosis ventralis (= impressio retinaculum flexorium sensu Ericson (2000), see Baumel & Raikow (1993, fig. 5.5)); (3) the depressio ligamentosa is constrained to an area on the dorsal third of the distal width by the prominent elongate tuberculum (1); (4) the distoventral corner profile is rounded and the facies extending proximally from it is flat and medial to the tuberculum (1); (5) the facies articularis radiocarpalis is very slightly convex distally in ventral aspect; (6) the distodorsal corner of the radius, where it hosts the insertion of ligamentum radio-radiocarpale dorsale, forms a sharp angle that overhangs the ventral facies and projects lateral of the shaft; (7) in dorsal aspect, the sulcus tendinosus is a single broad shallow structure (anatids have two distinct sulci, but Anseranas and anhimids have a similar singular sulcus); (8) the ventrodistal area mesad of the sulcus tendinosus forms a rounded swollen area that is bound distomedially by the tuberculum aponeurosis ventralis; (9) on the compressed margo interosseus (caudalis) a well-marked ligament scar extends proximally, overlapping the proximal extent of the tuberculum (1) and is assumed to be for the ligamentum radioulnare interosseum.

The large tuberculum (1) is a unique feature of presbyornithids (see Howard 1955, Ericson 2000) among anseriforms. Anatids differ greatly in having two grooves to the sulcus tendinosus and a distinct central depressio ligamentosa.

Comparative width measurements of distal radii Wilaru tedfordi: AMNH 10777 6.8 mm, AMNH 10800 7.3 mm, AMNH 10813 6.7 mm, AMNH 11464 (total length 113.0 mm) 6.9 mm, AMNH 11469 6.7 mm, AMNH 11471 7.3 mm, vs Murgonornis archeri QM F20871 6.3 mm.

Carpometacarpus

Three carpometacarpi fragments are referred as paratypes to Murgonornis archeri because of their similarity to the compared presbyornithids. The proximal fragment QM F30376 (Fig. 3M–O) lacks the processus extensorius but preserves the trochlea carpalis including the complete distal extent of its rims and the processus pisiformis. It reveals the following features: (1) the rims of the trochlea carpalis have equal extent caudally and (2) distally; (3) the dorsal rim of trochlea carpalis forms an uninterrupted curve, lacking the notch characteristic of anatids; (4) the ventral rim, in ventral aspect, forms an even arched curve caudally that distally ends at a point of inflection level with the pisiform process and so defines a caudocranial width of the ventral trochlea carpalis that is wider than its proximodistal length (see Fig. 3M), but the rim continues distally beyond this inflexion point as per feature (1); (5) between the rims caudally is a deep fovea carpalis caudalis with a complex characteristic shape such that distally, beginning level with the end of the rims, a pair of foveae form a figure eight-shape that extends over three-quarters of the height of the trochlea carpalis and, extending farther proximally from this, a sulcus one-third as wide as the trochlea carpalis extends to a shallow round fovea whose cranial margin is at the mid-caudocranial point of the trochlea carpalis; (6) the processus pisiformis is prominent cranially; (7) a distinct fovea carpalis cranialis impacts the dorsal half of the trochlea carpalis.

Two distal carpometacarpi (QM F61037, QM F61038) are identical in preserved features; the more complete QM F61038 (Fig. 3S) is described here. It preserves the complete distal end including parts of the shaft of both metacarpi. Its complex morphology reveals the following features: (1) the articular facet for MIII.1 in dorsocaudal aspect (Fig. 3S) forms a distinct projection enclosing a deep notch of the same width as the projection, between it and the articulation for MII.1, and the caudal side of the facet for III.1 is distally flat and its proximal part is met by the strongly curved end of os metacarpale minus; (2) between the ossa metacarpalia minus et majus a sulcus is nearly bridged by an ossified retinaculum and marks the passage of the tendon for musculus interosseum palmaris (see Stegmann 1978, muscle 14); (3) caudally, on the os metacarpale majus are four tuberosities marking ligamental attachment scars (Fig. 3S), all arranged as per for Wilaru tedfordi; the first of these (scar 1) marks one side of the just-mentioned retinaculum (2); (4) the raised tuberculum of scar 1 is elongated distally and is the origin for the branch of the tendon for musculus extensor indicis longus that extends distally to insert on manus phalanx II.2 (Stegmann 1978, muscle 8); (5) scars 2 and 3 are a pair of well-marked scars on the caudal face of the os metacarpale majus (see Fig. 3S) that support the retinaculum enclosing the tendon for musculus extensor digitorum communis; the more ventral (scar 3) is relatively more proximal; (6) the fourth scar, located distal to scar 3, marks the origin of the ligamentum digitorum metacarpale (Stegmann 1978, fig. 4); (7) there is a shallow sulcus bound by scars 2, 3 and 4 that is about twice as wide as the sulcus tendineus leading into it. The width of the sulcus tendineus for the ligament of musculus extensor digitorum communis proximal to its distal expansion between scars 2 and 3 (Fig. 3S) is not preserved. The distal part only of scar 3 is preserved but is sufficient to reveal a steep elevation change on that half of the sulcus, thereby conforming with the primary sulcus tendineus being about half of the width of the sulcus between scars 2 and 3, as in Wilaru.

Most features in the specimens referred to Murgonornis archeri are very similar to those observed in Wilaru tedfordi (AMNH 11432, AMNH 11474 + 10998; Fig. 3P, U) and such similarity extends to the relative distal extent of the metacarpals, the short length of the distal synostosis, the shape of the facet for MIII.1, and the bulge ventrally on the distal extremity of the os metacarpale majus (= part of facies articularis digitalis major). The two major differences between specimens of M. archeri and Wilaru tedfordi are (1) the cranial margin of the trochlea carpalis in Wilaru tedfordi is straight and lacks the distinct fovea carpalis cranialis in its dorsal part (Fig. 3U); and (2) distally, the cranial side of the sulcus bound by scars 2, 3 and 4 is bound by a sharper crest. The autapomorphic feature in W. tedfordi of a rugose tip on the processus extensorius is non-assessable in M. archeri with the material at hand but is specifically variable within genera in anatids and relates to fighting behaviours.

Presbyornis pervetus, e.g., USNM 618227 (Fig. 3Q, V, W), shares features 1–9 for the proximal carpometacarpi but differs as follows: the caudal fossa is not as deepened and nor has the markedly figure-8 shape, although a similar sulcus extends proximally from it; the fossa supratrochlearis is deeper; and a distinct ridge separates the ventral facies of the processus extensorius from the fossa infratrochlearis.

Telmabates antiquus differs most obviously from Murgonornis archeri by the shape of the trochlea carpalis, which, as defined by its ventral rim, is more proximodistally elongate with a length that is greater than width, rather than a width greater than length. Specimens, e.g., AMNH 3182, 3169 (Fig. 3R, X), show a distinct fovea carpalis cranialis impacting the central-dorsal part of the trochlea carpalis (feature 7) as in Presbyornis pervetus and M. archeri, but unlike in Wilaru tedfordi, which lacks such a fovea. Telmabates antiquus is more similar to W. tedfordi than to P. pervetus in having a short distal symphysis and more projecting processus extensorius.

Distal wing phalanges

Four distal wing phalanges in the assemblage have a typical anseriform shape. Three are of appropriate size for Murgonornis archeri (assuming similar proportions to that of Tadorna tadornoides SAMA B39878, a female, to which other referred bones match well in size), so are tentatively referred to the new species. A manus phalanx II.1 referred to Te. antiquus, is typically anseriform in form; therefore, in the absence of any presbyornithid distal wing elements at hand, we used Tadorna tadornoides for comparisons.

The pL manus phalanx II.1 (QM F24447) differs from Ta. tadornoides only by the craniodorsal process being more squared in proximal aspect of the articular surface. The pR manus phalanx II.2 (QM F20239) is likewise very similar to the equivalent element in Ta. tadornoides. The main difference is the degree of ossification of the retinaculum, which spans a tendinal sulcus on the proximoventral cranial margin through which the tendon for musculus flexor digitorum profundus passes before inserting on the distal end of MII.2 (Stegmann 1978). In Ta. tadornoides, the ossified retinaculum is short but in QM F20239, the ossification is extensive, extending 4.5 mm along the sulcus. The proximal R manus phalanx I.1 (QM F31298), is consistent in size with the manus phalanges just described and differs very slightly from Ta. tadornoides. Its maximum proximal width is 5.8 mm. A complete R manus phalanx I.1 (QM F61039) has identical morphology to the preceding specimen but is rather smaller (maximum proximal width 4.8 mm, total length 18.3 mm) so, while likely to belong to the same genus as the other fossils, it possibly represents a second smaller species.

Coracoid

Two partial coracoids are referred as paratypes to Murgonornis archeri. The two specimens complement each other. QM F23649 (Fig. 2I, M) preserves the angulus medialis, the corpus, processus procoracoideus and most of the cotyla scapularis. Only the medial third of the sternal margin is preserved, with the lateral section including the entire processus lateralis lost. The entire facies articularis humeralis is broken off and lost. However, a separate fragment, found in the same bag of matrix (Boles 1999) preserves the ventral half of the processus acrocoracoideus including two-thirds of the facies articularis clavicularis and is clearly part of the same bone. Specimen QM F30291 (Fig. 2N, O) preserves the entire omal portion proximally of the cotyla scapularis.

The coracoid is characterized by the following additional features to those listed in the Diagnosis: (1) the processus procoracoideus projects as a robust short point medially, not cranially, and extends down the shaft at about 25% of the distance to the sternal facet; (2) the foramen nervi supracoracoidei is close (3.5 mm) to the cotyla scapularis; (3) the cotyla scapularis is broad (5.4 mm diameter) as well as deep and round; (4) the impressio musculi sternocoracoidea has two longer lineae musculares traversing it at about 45° to the shaft long axis from its centre to the lateral margin, with a shorter line emanating from the lateral margin between them; (5) the angulus medialis is acute, preceded by a short crista epimarginalis, or crista medialis, that sternally and dorsally bounds a sulcus passing over the medial margin; (6) the facies articularis interna is broad, where preserved in the medial half, and is bound omally by a raised crista labrum internum that also marks the edge of the impressio musculi sternocoracoidei; (7) the facies articularis sternalis ventralis has a distinct labrum externum that forms an elevated facet extending 9 mm from the angulus medialis; (8) the ventral face of the processus acrocoracoideus (facies apicalis), or opposite the facies clavicularis, has a single shallow groove for the insertion of ligamentum acrocoracoclaviculare superficiale in an oval rugose surface medial to the scar for impressio ligamenti acrocoracohumeralis; (9) ventrally, the linea intermuscularis ventralis passes from close to the lateral margin towards the sternal margin, but breakage means that while the base of the processus lateralis is present, its extent is unknown. However, mesad of this linea, a rounded ridge extends sternally towards the sternal margin and bounds a shallow sulcus that deepens sternally mesad to it.

The coracoids QM F23649 and QM F30291 share most of the above features with Wilaru tedfordi, e.g., AMNH 11426 (Fig. 2J, P). Specimen QM F23649 is slightly larger than the coracoids of W. tedfordi (Table 2). Otherwise, QM F23649 differs from W. tedfordi in having a deeper sulcus ventrally, near the sternal articulation, and medial to the linea intermuscularis ventralis.

Table 2. Measurements (mm) of coracoids of Wilaru tedfordi and Murgonornis archeri.

Taxon	Catalogue number	Length processus procoracoideus to sternal facet	Minimum shaft width	Length sternal side of cotyla scapularis to foramen
Wilaru tedfordi	AMNH 11414	–	4.4	1.1
Wilaru tedfordi	AMNH 11426	23.2	4.9	2.0
Wilaru tedfordi	AMNH 11428	–	4.3	1.0
Wilaru tedfordi	AMNH 11433	–	4.3	1.4
Wilaru tedfordi	AMNH 11458	23.3	4.8	3.0
Wilaru tedfordi	AMNH 11473	–	4.5	1.2
Wilaru tedfordi	AMNH 11478	–	5.1	1.7
Wilaru tedfordi	SAMA P.58228	–	4.2	1.2
Murgonornis archeri	QM F23649	28.7	5.3	3.0

Presbyornis pervetus USNM 618183 (Fig. 2K, Q) has all the same features (a prominent labrum internum, acute angulus medialis, deep circular cotyla scapularis, short processus procoracoideus, with a small foramen nervi supracoracoidei close to the cotyla, an obvious crista epimarginalis and few linea in the impressio musculi sternocoracoidei), but has a slightly more slender shaft and lacks the ventral sulcus medial to the linea intermuscularis ventralis.

Similarly, Telmabates antiquus (Fig. 2L, R) shares the same features as just noted for P. pervetus, and likewise differs from Murgonornis archeri with a relatively narrower shaft that has less diameter than the cotyla scapularis.

Tibiotarsus

Two distal left tibiotarsi are referred as paratypes to Murgonornis archeri. Fragment QM F23736 (Fig. 3G, H) is missing the condylus medialis; QM F23737 (Fig. 3A, I) is missing the condylus lateralis. Together they reveal the following morphological features: (1) the sulcus extensorius is marked by a sharp crista medially and becomes shallower towards the lateral shaft margin; (2) the proximomedial insertion scar for the transverse ligament (retinaculum extensorium tibiotarsi) is located proximal but close to the pons supratendineus and overhangs the sulcus extensorius; (3) the distolateral attachment scar for the transverse ligament is elongate and forms a prominent ridge that overlaps three-quarters of the proximodistal length of the pons supratendineus and terminates proximally in a sharp point level with the distal end of the medial scar for the transverse ligament; (4) the cranial tuberculum for the retinaculum musculus fibularis extends onto the lateral side of the proximal end of the crista for the transverse ligament; (5) the caudal/lateral tuberculum for the retinaculum musculus fibularis is prominent laterally, and the sulcus between the two tubercula faces mainly cranially in its proximal half but distally rotates onto the lateral facies; (6) the distal margin of the pons supratendineus is not perpendicular to the shaft, but rather is angled slightly proximally towards the medial side, the canalis extensorius is aligned between the condyles, and the length and width of the pons supratendineus are about equal; (7) there is a low tubercle just lateral to the distal opening of the canalis extensorius with a flattened facet facing distomedially that is for articulation with the craniolateral surface of the eminentia intercotylaris; (8) the epicondylus medialis is very prominent on the condylus medialis in cranial aspect, and in medial aspect is positioned cranial to the shaft; (9) the condylus medialis is entirely offset medially to the shaft as seen in cranial aspect, and is distally flattened with a distinct shallow notch; (10) the condylus lateralis in lateral aspect is flattened distally and the rim cranially forms a bulbous overhang of the lateral facies of the condyle; (11) a deep scar for the insertion of ligamentum collaterale laterale is located caudal to the cranial prominence of the condylus lateralis near level with the distal margin of the condyle (Fig. 3H, lcl). This feature appears to have phylogenetic significance as it appears to place presbyornithids outside of the clade comprising Anseranas and anatids. In anatids, this attachment lies much more proximal, level with the proximocranial extent of the condyle; in Anseranas, it is more distal than anatids but still less so than in Wilaru. All compared presbyornithids differ from M. archeri in that the distal margin of the pons supratendineus is aligned close to right angles to the shaft long axis (6).

Wilaru tedfordi is slightly larger than Murgonornis archeri (Table 3) and shares features 1–5, 7–11 with it (Fig. 3B); however, it differs by: a slightly narrower canalis extensorius, so the area mesad of the canalis is relatively broader; the distal scar for the transverse ligament forms a slightly less robust tubercle; the proximal margin of the condylus lateralis and the distal end of the caudal/lateral tuberculum for the retinaculum musculus fibularis are more widely separated; and the pons supratendineus is slightly longer.

Table 3. Measurements (mm) of tibiotarsi of Wilaru tedfordi and Murgonornis archeri.

Taxon	Catalogue number	Maximum distal width	Shaft width at mid-length of proximal scar of transverse ligament	Depth condylus medialis	Depth condylus lateralis
Wilaru tedfordi	AMNH 10885	9.36	6.1	9.2	–
Wilaru tedfordi	AMNH 11417	–	6.36	–	8.55
Wilaru tedfordi	AMNH 11423	10.45	6.40	–	8.27
Wilaru tedfordi	AMNH 11424	9.73	5.80	–	7.30
Wilaru tedfordi	AMNH 11427	9.90	5.85	9.46	7.62
Wilaru tedfordi	AMNH 11430	9.80	5.86	9.42	7.72
Wilaru tedfordi	AMNH 11455	9.80	5.65	9.69	7.48
Telmabates antiquus	AMNH 3180	12.1	–	11.3	–
Murgonornis archeri	QM F23737	–	5.20	ca 8.58	–
Murgonornis archeri	QM F23736	–	4.61		7.56

Presbyornis pervetus (Fig. 3C, J, K) shares features 1–5, 7–11 with M. archeri, but the distal/lateral attachment scar for the transverse ligament is smaller and does not extend so far proximally from the pons supratendineus, which is relatively longer.

Telmabates antiquus (Fig. 3D–F, L) is larger (Table 3) but shares most of the above features with M. archeri. It differs, in addition to (6), by a less prominent epicondylus medialis (8)—it is only visible in cranial aspect if the view has a medial component (Fig. 3D versus 3E); by a narrower sulcus extensorius where it enters the pons supratendineus resulting in its lateral margin being more widely separated from the scar for the laterodistal scar for the transverse ligament; and by a more elongate, albeit variably so, pons supratendineus.

Anseriformes family, genus et species indet.

(Fig. 3Y)

Remarks

The distal right ulna (QM F23346) is very similar to anseriforms in overall shape, and to presbyornithids such as Telmabates antiquus, in features as follows: (1) the tuberculum carpale is proximodistally elongate, length more than twice its cranial projection; (2) the distal profile of tuberculum carpale is at right angles to the shaft axis, not hooked distally; (3) the incisura tendinosa is shallow and short, and opens distally into the proximal end of a deep sulcus that marks the origin of musculus ulnometacarpalis dorsalis.

Measurements: maximum distal width 11.5 mm, proximodistal length of condylus dorsalis 9.1 mm, depth of condylus ventralis 7.7 mm; those for a small individual of Wilaru tedfordi AMNH 10995 are 8.7 mm, 8.0 mm, and 6.4 mm, respectively. QM F23346 is the size of Te. antiquus AMNH 3170 (Fig. 3). The close similarity to Te. antiquus suggests it may be from another presbyornithid/anseriform, but further material is required before an adequate constrained diagnosis is possible.

Discussion

In this work, we have described Murgonornis archeri gen. et sp. nov., attributing it to Presbyornithidae in Anseriformes. The species is the most common bird in the Eocene Tingamarra Local Fauna from Murgon with 12 bones referred to the taxon out of ∼50 avian specimens (authors’ data). Potentially another unregistered right manus phalanx I.1 that is smaller than but identical to one of the referred specimens is also attributable, given the large size range Ericson (2000) noted in the large samples of specimens assigned to Presbyornis pervetus. Larger anseriforms are evidenced by a partial quadrate (Elzanowski & Boles 2012). Indeterminate passeriforms are known from two bones (Boles 1995, 1997) and a coraciiform by one partial quadrate (Elzanowski & Boles 2015). Other than several pedal phalanges, some of which likely relate to Murgonornis archeri (e.g., QM F20872, a III.1; QM F30347, RII.1; see Boles 1999), vertebra and miscellaneous fragments, six potentially identifiable bird bones of at least one very small taxon remain to be assessed.

The otic process of a left quadrate (QM F23019) that was described and referred to as an anhimid-like anseriform by Elzanowski & Boles (2012) differs greatly from, and is much bigger than, that of Presbyornis pervetus. It is also much bigger than that of, for example, Tadorna tadornoides SAMA B39878, whose bones match well in size with those described herein as Murgonornis archeri. Therefore, based on morphology and size, this quadrate fragment is unlikely to belong to M. archeri and instead represents a larger, non-presbyornithid taxon. It is possible that the distal ulna described herein relates to this taxon.

It was suggested by Mayr (2009), based on the figures in Boles (1999), that the paratype distal humerus of Juncitarsus gracillimus Olson & Feduccia, 1980 (USNM 244330; see Olson & Feduccia 1980, fig. 28) was similar to QM F23735, described here as Murgonornis archeri. However, USNM 244330 differs from QM F23735 by the facet on the tuberculum supracondylare ventrale being more elongate, aligned more obliquely, and distinctly facing distodorsally, and the tuberculum supracondylare dorsale projecting distinctly cranially, as well as more dorsally.

The description of Murgonornis archeri reveals that presbyornithids were present in Australia during the early Eocene when it was conjoined with Antarctica, confirming their bi-hemispheric distribution by the early Paleogene across several continents—Eurasia, North and South America and Gondwana (Wetmore 1926, Howard 1955, Olson 1994, Ericson 2000, Zelenkov 2021) (Fig. 1). This, when combined with the presence of Wilaru tedfordi in the late Oligocene of Australia (De Pietri et al. 2016), reveals minimally a 30 Ma existence of the family in Australia. Nevertheless, lack of a consistent closer similarity of bones between M. archeri and one of the other compared presbyornithids precludes drawing conclusions on its relationships and hence origin. Except for the equally similar coracoids, bones of M. archeri differ variably from the other compared presbyornithids: the humerus is most similar to that of W. tedfordi; its carpometacarpus shares the presence of a fovea carpalis cranialis with P. pervetus and Te. antiquus but not with W. tedfordi, although trochlea carpalis shape is more like in the latter; and the tibiotarsus has a much stouter shaft than both W. tedfordi and P. pervetus and differs from all compared presbyornithids in a shorter pons supratendineus with a distal margin that slopes proximomedially to a greater extent. An obvious difference is seen between W. tedfordi and earlier presbyornithids in the relative length of the tarsometatarsus; the early Miocene Australian W. tedfordi had relatively shorter tarsi (see De Pietri et al. 2016); however, available fossils do not inform on the state for M. archeri, therefore, when this autapomorphic state for the later surviving presbyornithids was acquired is unknown.

It is not clear why presbyornithids disappeared in the Eocene elsewhere. To date, Australia is the only place that presbyornithids and anatids have been shown to have existed contemporaneously, with at least four anatid species coexisting in the same sites with W. tedfordi in the late Oligocene (Worthy 2009, De Pietri et al. 2016). Moreover, in the same deposits, there are two palaelodids (Palaelodus wilsoni Baird & Vickers-Rich, 1998, P. pledgei Baird & Vickers-Rich, 1998) that have leg proportions like those of W. tedfordi, and at least two phoenicopterids (Miller 1963, Vickers-Rich 1991) with long tarsi. These latter species likely occupied the deeper wading niche that presbyornithids did in the early Paleogene. Species of Presbyornis from the Eocene of North America were inferred to be gregarious filter feeders in lacustrine, sometimes saline, freshwater environments that either dabbled in bottom sediments and/or had a feeding mechanism like that of Stictonetta naevosa, which primarily filters food from within the waterbody just above the sediment (Olson & Feduccia 1980, Ericson 2000). However, Zelenkov & Stidham (2018) demonstrated that presbyornithids were functionally limited to filtering large objects and did so while wading, thus differentiating them from anatids that can filter much finer objects from water while swimming (Zelenkov 2020a). How all these taxa shared the aquatic niche space has not yet been examined, but presbyornithids continued to exist alongside anatids, phoenicopterids and palaelodids until at least the early Miocene (23–22 Ma) in Australia, when Wilaru prideauxi existed and is the youngest known member of the family (De Pietri et al. 2016, 2020). It is therefore somewhat mysterious as to why presbyornithids disappeared in the Eocene elsewhere, but the Australian situation suggests competition with anatids was not the reason.

The recognition and description of Murgonornis archeri from Australia as a presbyornithid further evidences the near global distribution of the group long before anatids appear in the fossil record. This better accords with recent phylogenetic analyses that have either recovered presbyornithids outside of Anseres, a clade comprising Anseranas and Anatidae but excluding anhimids (see Worthy et al. 2017); as the sister group to Anseriformes (see Tambussi et al. 2019); or alternatively as the sister group to Anseranatidae in parsimony analyses or sister to Anseriformes in Bayesian analyses (Field et al. 2020), rather than the earlier recovered sister group relationship with anatids (Ericson 1997, Livezey 1997). These scenarios are consistent with the much older fossil record of presbyornithids than that of anatids (De Pietri et al. 2016, Mayr 2022). Moreover, the presence of presbyornithids in Australia in the Eocene and in the late Oligocene–early Miocene makes it very likely that they had a continuous presence there during this period. Therefore, this requires that anatids evolved during a time when presbyornithids existed and were already a long established and widespread lineage. Given that presbyornithids are known from Asia, North and South America, and Australia, and that their absence from Antarctica (formed since its separation from Australia) and Africa is very likely due to a lack of explored fossil deposits, the possibility exists that anatids evolved in the presence of presbyornithids (contra Olson & Feduccia 1980). Australasia has a great over-representation of basal waterfowl lineages in its modern avifauna: for example, Anseranatidae, Dendrocygna, the cereopsines or sister group of anserines, Biziura, Oxyura, Malacorhynchus and Stictonetta. This observation, together with the knowledge that the late Oligocene–early Miocene (ca 24 Ma) true waterfowl (Anseres) faunas of Australia were diverse—an anseranatid and minimally four species in two genera of anatids (Worthy 2009, Worthy & Scanlon 2009)—shows that anatid evolution was long underway in Australia by the terminal Oligocene. The Australian continent separated from Antarctica about 49 Ma, after which shallow water existed on the Tasmanian Gateway until 35.5 Ma when the seaway deepened and the Antarctic Circumpolar Current was initiated (Stickley et al. 2004, Beck & Ebach 2017, Evangelinos et al. 2022). The resultant strong circum-Antarctic westerly winds led to global changes in climate and initiation of polar cooling.

Globally, the earliest appearance of undoubted anatids, albeit in low diversity, was in the late Eocene (MP20, ca 35 Ma)–early Oligocene of Europe and Asia, with species attributed to Romainvilliinae (Romainvillia stehlini Lebedinsky, 1927; Romainvillia kazakhstanensis Zelenkov, 2018; Saintandrea chenoides Mayr & De Pietri, 2013) (see Mayr 2008, 2009, 2022, Zelenkov 2018). These taxa are usually treated as the subfamily Romainvillinae and considered primitive anatids (e.g., Mayr & De Pietri 2013, Mayr 2022); however, Zelenkov (2018) suggested they should be recognized at the family level as they are more plesiomorphic than Dendrocygna and represent an intermediate form between presbyornithids and anatids. It should be noted, however, that Mayr & De Pietri (2013) had already observed that romainvillines were too derived to be ancestral to anatids, and Mayr & Smith (2001) had reported that they lived alongside anatids in the early Oligocene of Belgium. Nevertheless, these early anatids and other potential stem-anatids (e.g., Cousteauvia kustovia Zelenkov, 2020b) could be evidence for their evolution from presbyornithids in Europe or Asia (Zelenkov 2018, 2020a, 2020b). Diversity remained low in Europe through the late Oligocene–early Miocene, notably with three species of Mionetta Livezey & Martin, 1988 and a larger goose-like Cygnavus Lambrecht, 1931 (Livezey & Martin 1988, Worthy 2009, Mayr 2009, 2022, Zelenkov 2020a). The over-representation of basal waterfowl lineages in modern Australasian avifaunas and lack of such in the Northern Hemisphere suggests an austral origin for the group. The lack of Paleogene Southern Hemisphere fossils can be explained by the lack of fossil deposits containing terrestrial vertebrates; in sharp contrast to the Northern Hemisphere, there are none known in Australia between the 55 Ma Murgon deposit and the late Oligocene, and Paleogene avifaunas are unknown from Antarctica and exceedingly scarce in Africa and South America. However, if anatids evolved in the Australasian region of Gondwana, then the lack of evidence for them in the Tingamarra Local Fauna may limit this to sometime between ca 55 Ma and 35 Ma when they first undoubtedly appear in Europe. Moreover, if they evolved in the East Gondwana region of which Australia was then part, they would have done so in the presence of presbyornithids.

Acknowledgements

We thank many colleagues for their invaluable assistance at Murgon and in the UNSW lab, and in particular Mike Archer and Henk Godthelp. We thank Walter Boles and Jacqueline Nguyen for facilitating our re-examination of the Murgon bird fossils. T.H.W. is especially grateful to Helen James and Storrs Olson (Department of Birds) and Mark Florence (Department of Paleobiology), at the United States National Museum, Smithsonian Institution, Washington, DC, USA, for facilitating his visit and access to specimens in the USNM collections, and especially to Mark for taking images and cataloguing some of the key Presbyornis material. P.S. and V.D.P. thank Carl Mehling and Paul Sweet of AMNH for access to the material from Telmabates antiquus. We thank Nikita Zelenkov and an anonymous reviewer for suggestions that improved the text.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Murgon research by S.J.H. is supported by the Australian Research Council [DP180100792]. This study was supported by grants from the Marsden Fund Council from NZ Government funding [grant numbers: 16-CTM-01 and 21-UOC-040], managed by the Royal Society Te Apārangi.