Pleistocene raptors from cave deposits of South Australia, with a description of a new species of Dynatoaetus (Accipitridae: Aves): morphology, systematics and palaeoecological implications

Abstract

The Pleistocene fossil record of Australian raptors (Accipitridae: eagles, hawks and Old World vultures) is very poorly understood with only three distinct extinct species confirmed until now. The small Necrastur alacer remains unstudied; however recent research has established Cryptogyps lacertosus as a valid species that, along with the very large Dynatoaetus gaffae, has potential affinities with the Aegypiinae–Circaetinae clade. These, along with a single large living inland raptor, the Wedge-tailed Eagle Aquila audax, suggest that Australia had an impoverished diversity of large raptors compared to other similar continental regions. However, fossil material similar in size to A. audax from multiple Pleistocene cave deposits raises the possibility of further taxa and is here assessed. A partial skeleton from Green Waterhole/Fossil Cave in the Tantanoola District of South Australia is referred to Cryptogyps lacertosus on the basis of similar size to bones of this species and that the associated coracoid is identical to that found with a tarsometatarsus that was previously referred to this species, coming from Leaena’s Breath Cave on the Nullarbor Plain, Western Australia. This partial skeleton allowed additional material from the Old Collections, Wellington Caves, and Walli Caves in New South Wales to be referred to this species. A second species, more robust than A. audax, is identified from material from Victoria Fossil Cave at Naracoorte. Phylogenetic analyses determined this species to be the sister taxon of the much larger Dynatoaetus gaffae recently described from several Australian sites, and so we name it as Dynatoaetus pachyosteus sp. nov. Further fragmentary remains attributable to either A. audax, C. lacertosus or D. pachyosteus are recorded from sites in the Wellington Caves complex, NSW. This study increases the known diversity of raptors in Pleistocene Australia, consistent with the presence of the more diverse megafaunal prey that existed at the time.

Ellen K. Mather [[email protected]], College of Science and Engineering, Flinders University, GPO 2100, Adelaide, SA 5001, Australia; Michael S. Y. Lee [[email protected]], Earth Sciences Section, South Australian Museum, North Terrace, Adelaide 5000, Australia (College of Science and Engineering, Flinders University, GPO 2100, Adelaide, SA 5001, Australia); Diana A. Fusco [[email protected]], College of Science and Engineering, Flinders University, GPO 2100, Adelaide, SA 5001, Australia; John Hellstrom [[email protected]], Faculty of Science, School of Geography, Earth and Atmospheric Sciences, University of Melbourne, Grattan Street, Parkville, Melbourne, 3010, Australia; Trevor H. Worthy [[email protected]], College of Science and Engineering, Flinders University, GPO 2100, Adelaide, SA 5001, Australia.

Keywords:

• Pleistocene extinction
• Australia
• Aegypiinae
• Circaetinae
• avian predators
• avian scavengers
• taxonomy

MAINLAND Australia has 17 resident breeding species of accipitrid raptors (Accipitridae), comprising 12 genera and five subfamilies (Marchant & Higgins 1993, Debus 1998). Most genera found in Australia are not endemic, with the exception of Lophoictinia Kaup, 1847 and Hamirostra T. Brown, 1846. The Australian taxa are varied in their preferred habitats and diets, ranging from species in genera like Elanus Savigny, 1809 (open grassland and woodland, predators of small vertebrates and large invertebrates), to those in Haliaeetus Savigny, 1809 (coastal, fish), and Aquila Brisson, 1760 (open habitat, small to medium mammals) (Brown & Amadon 1968). Today, there are only two large (>2 kg) Australian eagles; the predominantly coastal White-bellied Sea Eagle Haliaeetus leucogaster (Gmelin, 1788), and the widely distributed Wedge-tailed Eagle Aquila audax (Latham, 1801).

The modern Australian avifauna notably lacks several subfamilies of large accipitrids, with no species present from Circaetinae (snake eagles), Harpiinae (harpy eagles and kin), Gypaetinae (Bearded Vulture and relatives) or Aegypiinae (Eurasian Griffon Vulture and relatives) (Nagy & Tökölyi 2014). The lack of harpiine and circaetine species in Australia is unusual given their occurrence in geographically close lands. The circaetines Spilornis cheela (Latham, 1790) (Indonesia) and Pithecophaga jefferyi Ogilvie-Grant, 1896 (Philippines), and the harpiines Harpyopsis novaeguineae Salvadori, 1875 and Macheiramphus alcinus papuanus (Mayr, 1940) (New Guinea) occur in rainforest habitats. New Guinea is part of the continent Sahul, and is only separated from Australia presently by the high sea levels of the Holocene; therefore, the lack of harpiines and circaetines in Australia can be partly attributed to the current limited amount of suitable rainforest habitat and climate change, as the extent of Australian rainforest has oscillated through the Quaternary (Donders et al. 2006, Sniderman et al. 2007, Sniderman & Haberle 2012, Sniderman et al. 2012).

The Pleistocene fossil record of Australian accipitrids is sparse, with most identified fossil specimens being highly fragmentary. Until recently, five extinct accipitrid species were recognized from Australian late Pliocene to Pleistocene deposits. These species were described by the naturalist and palaeontologist Charles Walter de Vis, who named Aquila brachialis (de Vis, 1889), Cryptogyps lacertosus (de Vis, 1905), Necrastur alacer de Vis, 1892, Asturaetus furcillatus de Vis, 1905 and Palaeolestes gorei de Vis, 1911. Palaeolestes was hesitantly assigned to the ‘Accipitres’, as the Accipitridae were known at the time, being based on a single phalanx that bore some similarity to those seen in living hawks and eagles (de Vis 1911).

The status of these species named by de Vis (1911) was later brought into question. Rich et al. (1982) determined that A. furcillatus was a synonym of the living Brown Falcon Falco berigora. Aquila brachialis has been suggested to be either a synonym of the extant A. audax (e.g., Condon 1975, Gaff 2002) or an indeterminate accipitrid (van Tets 1984). A reassessment by van Tets & Rich (1990) of the ‘accipitrid’ species named by de Vis found that only A. brachialis, ‘T.’ lacertosus and N. alacer were likely correctly identified as accipitrids; A. furcillatus was a Brown Falcon as already noted and P. gorei was likely non-avian. Necrastur alacer remains unstudied except for in an unpublished thesis by Gaff (2002), wherein it was considered to be distinct from living Australian accipitrids in terms of its small size (intermediate between Hamirostra melanosternon (Gould, 1841) and A. audax) and morphology.

Recent studies have confirmed that aegypiine vultures and large eagles in the Aegypiinae–Circaetinae clade were present in Australia during the Pleistocene. Cryptogyps (= Taphaetus) lacertosus (de Vis 1905) was about the same size of A. audax and is believed to have been a scavenger like most species in the subfamily (Mather et al. 2022). In contrast, the giant Dynatoaetus gaffae Mather, Lee, Camens and Worthy, 2023 was a very large predator, much like its relative, the modern-day P. jefferyi, the Philippine Eagle (Mather et al. 2023).

Here we describe these taxa and establish their relationships among accipitrids. Comparisons specifically include the fossil taxa C. lacertosus (a similar-sized species) and D. gaffae (much larger), two contemporary species broadly sympatric across southern Australia, and Hieraaetus (Harpagornis) moorei, the gigantic extinct eagle from New Zealand.

Australian Pleistocene accipitrid fossil sites

Australia has many Pleistocene fossil sites that are widely distributed across the country. Eight have produced fossil accipitrid material that cannot be assigned to any living Australian species, and which represent birds as large as or larger than Aquila audax. These sites are in Leaena’s Breath Cave, in the Nullarbor of WA; Victoria Fossil Cave, at Naracoorte, SA; Green Waterhole Cave (also known as Fossil Cave), at Tantanoola near Mt Gambier, SA; Warburton River and Cooper Creek in the Lake Eyre Basin, SA; Mairs Cave in South Australia, and the Walli Caves and Wellington Caves in NSW (Fig. 1). Some of this material was described in Mather et al. (2022, 2023), however, others remain undescribed, notably that from Victoria Fossil Cave and Green Waterhole, and some isolated fossils from Wellington Caves and Walli Caves.

Fig. 1. Map of Australia showing the locations of Leaena’s Breath Cave, Naracoorte Caves, Green Waterhole/Fossil Cave, Wellington Caves and Walli Caves.

Victoria Fossil Cave

Victoria Fossil Cave is the most significant cave for fossils in the Naracoorte Caves National Park and is key to the inclusion of the park in the Australian Fossil Mammal Sites (Riversleigh/Naracoorte) World Heritage Area, having extensive deposits that have provided fossils of many species of Pleistocene megafauna, as well as smaller vertebrates such as birds (van Tets & Smith 1974, Reed & Bourne 2000, Fraser & Wells 2006, Reed & Bourne 2009). The cave has multiple fossil sites within it, the most significant being Fossil Chamber, Grant Hall, and the Ossuaries (Reed 2003, 2006, Fraser & Wells 2006). Fossil Chamber was discovered in 1969 (Wells 1975) and played a crucial role in the successful World Heritage nomination in 1994. Like all caves at Naracoorte, Victoria Fossil Cave was formed in Tertiary limestone beds in the phreatic zone during the late Miocene or early Pliocene and has since been uplifted above the local water table (Sprigg 1952, Wells et al. 1984). Accumulation of fossil material did not begin until the middle Pleistocene, when erosion of the surrounding Naracoorte East Dune exposed openings into the caves (Wells et al. 1984), and it has been estimated that sites in the cave range between 500 ka and the present (Grün et al. 2001, Prideaux et al. 2007a, Macken et al. 2011, Macken & Reed 2013, Macken et al. 2013).

The diversity of extinct mammal fauna recorded from Victoria Fossil Cave has made it an extremely valuable site for assessing changes in environment, climate, and fauna through the Pleistocene (birds, reptiles and amphibians have yet to be described in any detail). However, very little has been documented on the accipitrid fossils from this site, with most existing records not mentioning accipitrids among their lists of birds (van Tets & Smith 1974, Reed & Bourne 2000, 2009), although Gaff (2002) discussed some specimens.

The material assessed in this study originates from the Fossil Chamber of Victoria Fossil Cave, which has produced fossil material dating 500–213 ka based on U/Th dating of flowstone caps and optical luminescence dating (Ayliffe et al. 1998, Arnold et al. 2022). Material excavated from this site had their stratigraphic data captured by recording the 3D location of the excavation unit in decimal feet with X, Y, and Z coordinates relative to a fixed datum near the base of the talus cone (see Reed 2003). For example, SAMA P28008 has coordinates of 60.5′–62.5′ [along the datum line], R1′–2′ [R = right of datum line], and −0.5′ to −1.0′ D/D [depth relative to datum].

Undescribed accipitrid material from Fossil Chamber of Victoria Fossil Cave includes a complete quadrate, humerus, ulna, carpometacarpus, a partial pelvis, and one complete femur that belongs to a hitherto undescribed species; other material was recently referred to D. gaffae (see Mather et al. 2023) and fossil material of Aquila audax including a tarsometatarsus (SAMA P59836) is known. Of the undescribed material, Gaff (2002) only mentioned the ulna.

Green Waterhole/Fossil Cave

Green Waterhole Cave (Fig. 2), also known by the name Fossil Cave and the Cave register numbers 5L81 and S123 of the Cave Exploration Group of South Australia (Inc.), Adelaide, is located in the Tantanoola district approximately 24 km northwest from Mt Gambier in SA. Primarily on account of the fossil deposits, it was added to the list of State Heritage sites of importance, State Heritage ID 26530, under the name ‘Green Waterhole – Tank Cave Fossil Complex’ (Designated place of palaeontological, speleological and geological significance) [https://maps.sa.gov.au/heritagesearch/HeritageItem.aspx?p_heritageno=28119]. Like the Naracoorte Caves, those in the Gambier region are formed within the Oligo-Miocene limestone shelf that is spread across southeast South Australia (Boutakoff 1963). Unlike the Naracoorte Caves, which are mostly dry, caves in the Gambier region are largely below the water table, with Green Waterhole Cave being no exception. Green Waterhole Cave is assumed to have first been formed during the late Miocene (Pledge 1980). The fossils occurred on the floor of the chamber at depths of 1–9 m.

Fig. 2. Map of the largely submerged Green Waterhole/Fossil Cave, redrawn from the original (Horne 1988), with the location of the cave in South Australia. Connecting points of the grid within the cave mark out the positions of the underwater anchor points for dive lines defining areas within which fossil bones were collected. Inward arrows around the collapse doline indicate a vertical drop. Black dot noted as Aslin Site 12/Site 07 marks the position of where most of the fossil material described in this study was found. Narrow parallel lines indicate air/water surface. Thick black lines mark the walls of the cave. Light grey outlines mark the presence of rocks and boulders. The well shaft affords entry from the surface to the underground lake.

The age of the fossils from Green Waterhole Cave has long been debated. Baird (1985) noted that recorded data on sea level transgressions indicated the cave entrance would have opened 125 ka at most, and that the fauna indicated that it could be no younger than 15 ka. This lower estimate was based off the findings of Horton (1984), indicating the Australian megafaunal extinction took place 26–15 ka, which has since been revised to 50–40 ka (Roberts et al. 2001, Gillespie et al. 2006, Saltré et al. 2016, Johnson et al. 2021). Fossils in the area termed Site 7, including the large accipitrid described herein and hundreds of other bird specimens, were deposited in sediment almost entirely composed of cave rafts—speleothems that form by crystallization on the surface of a water body, usually a pool. Their presence shows that a pool of water existed in the lower reaches of this chamber when the cave was largely drained of water during periods of lower water tables. Many other fossils accumulated on the subaerial surface of the now drowned slopes in this entrance chamber. It is likely that this pool was a primary attractant to the very numerous birds that accumulated in it (Baird 1985). As the fossils were found within the deposits of cave rafts, we consider that the birds and associated mammalian megafauna, including the accipitrid described herein, accumulated in the pool that had cave rafts forming on its surface. Such a pool could only exist when the local water table was more than 12 m lower than the 1980 water level, i.e., was below this part of the chamber. Prolonged lowered water tables existed between Marine Isotope Stages 2 to 5d when sea levels were lower: during MIS 2, sea levels were up to 120 m lower; during MIS 4–3 (71–29 ka), they were about 90–60 m lower; and during MIS 5a–5d (115–71 ka), they were higher but still more than 20 m below present (e.g., Miller et al. 2020). The earlier MIS 5e (130–115 ka) had sea levels similar to the present and so the chamber will have then been below the water level, as at present. Therefore, based on sea level curves, lowered water tables would have allowed a discrete pool to exist in the lower parts of the main entrance chamber between 115 and near 14 ka.

The fossil fauna from Green Waterhole Cave was first formally described by Pledge (1980), after multiple diving trips in the cave during the 1960s–1970s had yielded fossil bones and skeletons of extinct sthenurine and macropodine kangaroos. While the paper mostly focused on the kangaroo species present, Pledge noted the presence of many small bird bones. A formal survey of the cave was undertaken by the South Australian Underwater Speleological Society Inc. during 1987–1988. It was determined that the cave was approximately 70 m long overall, 30 m wide at the lowest chamber, and observed depths of over 15 m at certain points (Horne 1988; Newton 1988). There are minimally 23 bird species listed from Green Waterhole Cave, which includes accipitrids (Baird 1985, Baird et al. 1991, Reed & Bourne 2000). Currently, more than 958 lots (most passerines remain unidentified) are registered in SA Museum (see Supplemental Data 1 for species list).

The presence of bones of a large accipitrid has been noted from this site for over four decades (Rich & van Tets 1982, Baird 1985, 1991, Baird et al. 1991; Reed & Bourne 2000, Gaff 2002). The first accipitrid fossils were recovered from Green Waterhole Cave in 1979 and 1987 during dives initiated by Rod Wells of Flinders University, and included paired sets of carpometacarpi, ulnae and radii, a scapula, os metatarsale I, and a thoracic vertebra (Baird 1985, Newton 1988). Most were collected from ‘Site 7’ or nearby, see Fig. 2. An almost complete right coracoid, missing only the omal end, was later uncovered on an expedition led by T.H. Worthy and A. Camens with P. Horne, N. Skinner, D. Albano, M. Nielsen and I. Lewis as divers and/or surface support in 2006, from 2 m upslope of Aslin Site 12 tag, which is equivalent to Site 7. The radii, ulnae and carpometacarpi were observed by Gaff (2002) and interpreted as belonging to a large species of Perninae.

We propose that the partial skeleton from Green Waterhole/Fossil Cave and the undescribed smaller accipitrid material from Victoria Fossil Cave represent two accipitrid species, as comparable wing elements (ulna, carpometacarpus) occur in both sites. Apart from the material of a much larger eagle already referred to D. gaffae, and three bones identical to those of Aquila audax, there is minimally only one other species represented at Victoria Fossil Cave by these specimens and so we follow the parsimonious approach of referring the pelvic girdle elements and quadrate found in Victoria Fossil cave to the same taxon as that represented by the wing bones.

Leaena’s Breath Cave

Leaena’s (= Leana’s) Breath Cave is a 73 m long limestone cave located in the Nullarbor Plains of WA and is part of a collective of caves known as the Thylacoleo Caves (Prideaux et al. 2007b). The cave is minimally 4 million years old based on U-PB dating of speleothems (Woodhead et al. 2006). Leaena’s Breath Cave along with other caves in the Thylacoleo Caves assemblage has yielded fossils of remarkable quality, including complete skeletons (Prideaux et al. 2007b). The age of these fossils based on optical dating, U-Th dating and palaeomagnetic dating is thought to be 400–200 ka for specimens in the upper levels of the excavation, and 780–400 ka for specimens in the deepest part of the excavation (Prideaux et al. 2007b).

Undescribed accipitrid materials included in this study from this site are a sternal part of coracoid and the synsacrum of a pelvis; however, an associated tarsometatarsus was described by Mather et al. (2022). The coracoid was identified as an indeterminate accipitrid by Shute (2018).

Wellington Caves

The Wellington Caves are located in central-western NSW, 7 km south of the town Wellington. Large numbers of fossils have been collected from these caves since at least 1830, but provenance data and associated chronological and stratigraphical context is often poor (Dawson 1985). The caves contain multiple significant fossil deposits that range in age from Pliocene to Holocene (Fusco et al. 2023). Cathedral Cave yields the youngest fossils, from late Pleistocene to Holocene, and is the only location at Wellington Caves that has been subject to controlled stratigraphic excavations (Dawson & Augee 1997, Fusco et al. 2023). Fossils collected from Mitchell’s Cave, much of the Phosphate Mine, and Bone Cave (a discrete location within the Phosphate Mine) are older than those from Cathedral Cave, with faunal biocorrelation (e.g., Megirian et al. 2010) showing them to be early to late Pleistocene (Dawson 1985, 1995). The Big Sink and Koppa’s Pool (another discrete location in the Phosphate Mine) are considered to be Pliocene aged based on faunal biocorrelation (Dawson et al. 1999, Nipperess 2000).

The ‘Old Collections’ of Wellington Caves refers to fossil material held in the Australian Museum that was collected from the caves before August 1887 (Dawson 1985). This was prior to the opening of the Phosphate Mine and Big Sink in 1914 (Dawson et al. 1999). Dawson (1985) stated that fossils collected prior to 1884 are likely from Cathedral Cave or Mitchell’s Cave; however, the lack of associated provenance data for many of the fossils in the Old Collections means that their age cannot be determined beyond Pleistocene.

Walli Caves

Referred to as the Belubula Caves in earlier literature (see Frank 1974, Brocx et al. 2019), the Walli Caves are a cave complex of over 40 separate entrances/dolines 27 km northeast of the town of Cowra in New South Wales (Frank 1972, 1974). They are largely formed within the Palaeozoic Cliefden Caves limestone close to Licking Hole Creek.

Reported fossils from this site include species of Macropus Shaw, 1790, Vombatus Geoffrey, 1803 (as Phascolomys), Protemnodon Owen, 1873a, Thylacinus Temminck, 1824 and Thylacoleo carnifex Owen, 1859a (see Frank 1972, 1974). Due to a lack of material available for dating, the age of deposits within these caves are poorly understood (Frank 1974). The known accipitrid fossil material is restricted to a single distal ulna.

Materials and methods

Institutional abbreviations

AM, Australian Museum, Sydney, Australia. ANWC, Australian National Wildlife Collection, Canberra, Australia. FU, Palaeontology Collection Flinders University, Adelaide, Australia. FUR, Flinders University Vertebrate Collection, Adelaide, Australia. NMV, Museums Victoria, Melbourne, Australia. SAMA, South Australian Museum, Adelaide, Australia. WAM, Western Australian Museum, Perth, Australia. USNM, Smithsonian Museum of Natural History, Washington DC, USA. KU, University of Kansas Institute of Biodiversity, Lawrence, KS, USA. NHMUK, The Natural History Museum, Tring, UK. CM, Canterbury Museum, Christchurch, New Zealand. UM, University of Melbourne, Melbourne, Australia.

Anatomical abbreviations

DI.I, digit 1, phalanx 1. DII.I, digit II, phalanx 1. DII.2, digit II, phalanx 2. DIII.I, digit III, phalanx 1. DIII.2, digit III, phalanx 2. DIII.3, digit III, phalanx 3.

Anatomical nomenclature

The anatomical nomenclature advocated by Baumel & Witmer (1993) is followed for all bones. Taxonomic nomenclature follows Dickinson & Remsen (2013) and Gill et al. (2020) for composition of Accipitriformes, and Nagy & Tökölyi (2014) for subfamilial composition (excluding Milvinae).

Comparative material

Skeletons of a wide range of extant taxa were obtained on loan from museums and other institutions from across Australia and overseas and are listed in Mather et al. (2023), though the following corrections are made: all ANWC specimens should have a ‘B’ before their specimen number, and Hamirostra melanosternon specimen ANWC FALS-41 should be recorded as ANWC B18537.

The following specimens of fossil taxa were examined. Dynatoaetus gaffae: SAMA P59525 (partial skeleton including cranium, mandible fragments, two cervical vertebrae, three thoracic vertebrae, anterior end of the synsacrum, two caudal vertebrae, left and right sides furculum, right scapula, proximal right humerus, distal left ulna, proximal right radius, left carpometacarpus, left proximal manus phalanx major digit [or manus phalanx proximalis digiti majoris], two ribs, a tibiotarsus shaft, right tarsometatarsus, proximal right fibula, right metatarsal and left pedal phalanx I of digit I, left pedal phalanges I and II of digit II, right pedal phalanx I of digit IV, and an ungual phalanx); SAMA P 19157 (ungual phalanx); SAMA P17139 (ungual phalanx); SAMA P19158 (sternum); SAMA P14528 (distal humerus); SAMA P41514 (right femur); SAMA P28008 (distal right tarsometatarsus fragment); AM F.106562 (distal right tibiotarsus); SAMA P25218 (distal right tibiotarsus). Cryptogyps lacertosus: QM F5507 (distal right humerus); AM F.58092 (distal left and right humeri fragments); AM F.58093 (left tarsometatarsus); WAM 15.9.71 (proximal left tarsometatarsus). Hieraaetus moorei (Haast, 1872): casts of the types and originally referred bones, NMV P33029 (femur, CM AV 5104 pt, lectotype H. moorei Haast, 1872); NMV P33032 (tibiotarsus CM AV 5104 pt); NMV P33030 (tarsometatarsus CM AV 5104 pt); NMV P33031 (ungual phalanx CM AV 5104 pt); NMV P33027 (femur CM AV 5102 pt lectotype of H. assimilis Haast, 1874a); NMV P33028 (humerus, CM AV 5102 pt); NMV P33026 (ulna CM AV 5102 pt); and additional observations were made from figures in Holdaway (1991). Measurements were made with digital callipers, and rounded to the nearest 0.1 mm.

Phylogenetic methods

A total of 300 morphological characters were coded for the taxa listed below; for details of characters and states, see Mather et al. (2022, 2023). The new taxon from Victoria Fossil Cave was added to the matrix in Mather et al. (2023) and included in the analysis based on observations presented here, and the scorings for Cryptogyps were expanded in light of new material in this study (see Supplemental Data 2). Characters that were inapplicable on a specimen were coded using ‘–’, whereas missing data that was potentially scoreable (with more complete material) were coded using ‘?’.

Data from Burleigh et al. (2015) was used to provide molecular data for species used in this study, with the following genes being selected due to their good sampling across accipitriforms: cytochrome-b, cytochrome oxidase 1, NADH dehydrogenase 2, 12s RNA, RAG 1, and fibrinogen β introns 6 7 (see Mather et al. 2022, 2023 for full discussion and details of species used).

Phylogenetic comparisons were aimed primarily at determining the relationships of fossil specimens in the context of major living clades. A total of 47 species of Accipitridae, one species each of Pandionidae, Sagittariidae, Cathartidae, Threskiornithidae and Ciconiidae were sampled. The non-accipitrid species were selected due to their close relationship to Accipitridae (Pandionidae, Sagittariidae and Cathartidae) and physical similarities to large accipitrids (Threskiornithidae, Ciconiidae) as discussed in Mather et al. (2023).

Both parsimony and Bayesian analyses were used to analyse the data. The parsimony analyses of the morphological, molecular, and combined morphological-molecular matrices were analysed using PAUP 4.0b10, using heuristic searches (see Supplemental Data 3). Each search comprised of 1000 random addition replicates, and enabled TBR branch swapping, with NCHUCK set to 1000. The taxa Ciconia ciconia (Linnaeus, 1758), Threskiornis spinicollis (Jameson, 1835), Coragyps atratus (Bechstein, 1793) and Sagittarius serpentarius (J. F. Miller, 1779) were set as the outgroups. Once the heuristic searches had generated a set of most parsimonious trees (MPTs), a strict consensus tree was created from them. Support for clades was assessed using bootstrapping (200 replicates), and the majority-rule bootstrap consensus was set to a conlevel of 50 (only clades >50% shown).

For the Bayesian analyses (see Supplemental Data 4), MrBayes 3.2.7 was used via the platform CIPRES (Miller et al. 2010). For the molecular data, the optimal partitioning scheme and substitution models were obtained using PartitionFinder (Lanfear et al. 2012) using BIC and unlinked branch lengths. The analyses employed three partitions: morph, which included the morphological data; pfinder molec1, which included Cyt-b codons 1 and 2, CO1 codons 1 and 2, ND2 codons 1 and 2, 12s, Rag-1 codons 1, 2 and 3, and FGBint67; and pfinder molec2, which included Cyt-b codon 3, CO1 codon 3 and ND2 codon 3. The Morph partition had standard type data with correction for sampling only variable characters. The rates were set to gamma, with distribution approximated using four categories. The Molec1 partition had the GTR model, with substitution rates and frequencies having (separate) Dirichlet priors. The rates were set to Invgamma, with the gamma distribution approximated using four categories. The Molec2 partition also had a GTR gamma model, with all parameters separate to those in Molec1. The number of MCMC chains was set to 4 (incrementally heated at 0.1), the number of generations set to 50,000,000, the sample frequency set to 5000, with a burnin of 0.2 proving adequate. Each analysis was run four times (i.e., four runs each of four chains), from which a post-burnin majority-rule consensus tree was then derived. Ciconia ciconia was initially set as the sole outgroup taxon, due to limitations with MrBayes, but trees were later re-rooted on the Ciconia+Threskionis clade.

Mass estimation

The femora, humeri and coracoids of the fossil species were used for mass estimates. To achieve this, a range of measurements were taken from the bones, using a digital calliper for lengths and widths, or with a length of string for shaft circumference. These measurements were then used to predict the body mass with established formulae (Campbell & Tonni 1983, Campbell & Marcus 1992, Field et al. 2013).

U-Th dating

Calcite raft rubble was collected on 27–28 May 2006 from the submerged Site 7 in Green Waterhole Cave, placed into a plastic specimen jar and returned to the surface. Individual fragments were up to 20 mm in horizontal extent and all are approximately 1 mm thick. The calcite was clean, composed of sub-millimetre-scale clear calcite crystals, flat on the upper surface as expected for a calcite raft (Taylor et al. 2004), with no visible evidence of dissolution or overgrowth since falling from the pool surface and settling at its base.

The largest fragments were cleaned with a dental drill to reduce the potential for surface overgrowth or contamination. Five subsamples of approximately 0.25 cm² were taken from individual large flakes, giving an average sample mass of 60 mg. These were dissolved and their U and Th chemically extracted and analysed in 2007 following the procedure of Hellstrom (2003) using a Nu Instruments Plasma I inductively coupled plasma mass spectrometer at the School of Geography, Earth and Atmospheric Sciences, UM.

Results

U-Th dating of calcite rafts from Green Waterhole Cave

The raft fragments contained approximately 300 ng g^–1 of uranium and returned corrected uranium-thorium ages of between 59.1 and 61.1 thousand years before present (Table 1). It was not possible to constrain the likely initial thorium isotope ratio, hence an assumed initial ²³⁰Th/²³²Th activity ratio of 1.5 ± 1.5 was used, which covers > 95% of reported values for speleothems formed at temperate latitudes.

Table 1. U-series dating age estimates on five calcite raft samples from near Site 7, in Green Waterhole Cave.

Sample	U(ngg^−1)	[^230Th/^238U]^a	[^234U/^238U]^a	[^232Th/^238U]	[^230Th/^232Th]	Age(ka)^b	[^234U/^238U]i^c
Site2L81.1	313	0.5334(38)	1.1907(26)	0.03311(32)	16.1	59.1(4.7)	1.2253(41)
Site2L81.2	286	0.5375(64)	1.1867(56)	0.0354(13)	15.2	59.7(5.0)	1.2210(72)
Site2L81-3	311	0.5469(55)	1.1900(27)	0.03460(34)	15.8	61.1(4.9)	1.2258(44)
Site2L81-4	341	0.5446(23)	1.1895(22)	0.03589(23)	15.2	60.6(5.0)	1.2248(41)
Site2L81-5	314	0.5389(17)	1.1941(29)	0.03346(22)	16.1	59.6(4.7)	1.2296(45)

Analyses performed by John Hellstrom, University of Melbourne.

^aActivity ratios determined after Hellstrom (2003) and Drysdale et al (2012).

^bAge in kyr before present corrected for initial ²³⁰Th using eqn. 1 of Hellstrom (2006), the decay constants of Cheng et al. (2013) and [²³⁰Th/²³²Th]i of 1.5 ± 1.5. Dates reported as ka before present (BP), where the present is defined as the year 1950 CE.

^cInitial [²³⁴U/²³⁸U] calculated using corrected age. 2-s uncertainties in brackets are of the last two significant figures presented.

The uncertainties for the individual corrected ages of circa 5000 years predominantly reflect this uncertainty in estimation of the initial thorium correction, which is highly correlated between them. It is not possible to say whether the minor scatter between the corrected ages reflects variability in the actual initial Th isotope ratio (Hellstrom 2006), a minor impact of uranium mobility or post-depositional overgrowth, or genuine scatter in age of formation of the raft fragments.

The tight clustering of apparent ages at 60,000 years is remarkable and points to little dissolution or precipitation of calcite in this time, as either of these heterogeneous processes would produce a differential effect between individual raft fragments and lead to scatter of the apparent ages (Bajo et al. 2016). This close agreement indicates a relatively short interval of raft deposition, including deposition of the intercalated fossil material.

Systematic palaeontology

Class AVES Linnaeus, 1758

Order ACCIPITRIFORMES Vieillot, 1816

Family ACCIPITRIDAE Vigors, 1824

Cryptogyps lacertosus (de Vis, 1905)

(Figs 3–6)

Fig. 3. Coracoids and scapula referred to Cryptogyps lacertosus from Green Waterhole Cave, South Australia (A, B, D–F) and Leanna’s Breadth Cave, Nullarbor (C). SAMA P42487, right coracoid in A, ventral and B, dorsal view; C, right coracoid WAM 15.9.72 in dorsal view; right scapula SAMA P53845 in D, medial, E, lateral and F, cranial view. Abbreviations: Ac, acromion; CF, cranial fossa; CLA, crista lig. acrocoracoacromiali; CLS, coracobrachialis ligament attachment scar; CSc, corpus scapulae; CtS, cotyla scapularis; FAC, facies articularis clavicularis; FnS, foramen n. supracoracoidei; IS, impressio m. supracoracoidei; MD, margo dorsalis; MF, medial fossa; PPr, processus procoracoideus; TC, tuberculum coracoideum. Scale bars 10 mm.

Fig. 4. Large fossil accipitrid ulnae from Southern Australia. Mairs Cave, South Australia: A, Dynatoaetus gaffae ulna SAMA P59525 in ventral view; Cryptogyps lacertosus from Green Waterhole Cave, ulna SAMA P24323 in B, ventral, D, cranial and F, dorsal view; Dynatoaetus pachyosteus ulna SAMA P59029 in C, ventral, E, cranial and G, dorsal view. Abbreviations: CD, condylus dorsalis, CF, caudal fossa; CtV, cotyla ventralis; CV, condylus ventralis; IB, impressio brachialis; IR, incisura radialis; IT, incisura tendineus; ITC, incisura tuberculum carpale; O, olecranon; PLS, proximal ligamental scar; ScT, impressio m. scapulotricipitis; SI, sulcus intercondylaris; TC, tuberculum carpale; TLCV, tuberculum ligamentosa collateralis ventralis; VS, ventral scar. Scale bar 10 mm.

Fig. 5. Humeri (A–D), carpometacarpi (E–H) and radius (I, J) of eagle-sized accipitrids from Australia. Dynatoaetus pachyosteus. Holotype left humerus SAMA P41517 in A, caudal and B, cranial view, extant Aquila audax left humerus FUR 125 in C, cranial and D, caudal view, E, Dynatoaetus gaffae left carpometacarpus in ventral view, Cryptogyps lacertosus left carpometacarpus (Green Waterhole) in F, ventral and G, dorsal view; D. pachyosteus carpometacarpus from Victoria Fossil Cave in H, ventral view, C. lacertosus (Green Waterhole) radius in I, ventral and J, dorsal view. Abbreviations: CB, crista bicipitalis; CD, crista deltopectoralis; CH, cotyla humeralis; CV, condylus ventralis; DL, depressio ligamenti; DR, depressio radialis; DSH, dorsal sulcus humerotricipitalis; FADMaj, facies articularis digitalis major; FADMin, facies articularis digitalis minor; FAH, facies articularis humeralis; FAR, facies articularis radialis; FAU, facies articularis ulnaris; FB, fossa brachialis; FI, fossa infratrochlearis; FO, fossa olecrani; FPV, fossa pneumaticum ventralis; FS, fossa supratrochlearis; IC, incisura capitis; MPS, m. pectoralis scar; MSC, insertion of m. scapulohumeralis cranialis; OMM, os metacarpale minus; PA, processus alularis; PE, processus extensorius; PF, processus flexorius; PP, processus pisiformis; SLT, sulcus lig. transversus; SNC, sulcus nervi coracobrachialis; ST, sulcus tendineus; TAV, tuberculum aponeurosis ventralis; TBR, tuberculum bicipitalis radialis; TD, tuberculum dorsalis; TSD, tuberculum supracondylare dorsale; TSV, tuberculum supracondylare ventrale; VS, ventral sulcus. Scale bars 10 mm.

Fig. 6. Eagle-sized fossil accipitrid material from South Australia: Cryptogyps lacertosus, Green Waterhole Cave, thoracic vertebra SAMA P24329 in A, caudal, B, lateral, and C, cranial view; Dynatoaetus pachyosteus, right quadrate in D, medial, E, lateral and F, ventral view; C. lacertosus, Green Waterhole Cave, os metatarsale I in G, plantar and H, dorsal view. Abbreviations: CC, condylus caudalis; CL, condylus lateralis; CM, condylus medialis; CO, capitulum oticum; CP, condylus pterygoideus; CQ, cotyla quadratojugalis; CS, capitulum squamosum; FvC, fovea costalis; FV, foramen vertebrae; I, indentation; LM, lateral margin; PAT, processus articularis tarsometatarsalis; PO, processus orbitalis; Pr, projection; PV, processus ventralis; SR, sulcus ridges; ZCa, zygapophysis caudalis. Scale bar 10 mm.

1905, Taphaetus lacertosus de Vis, pl. 1, fig. 1.

1974, Taphaetus lacertosus de Vis; van Tets, p 58.

2022, Cryptogyps lacertosus (de Vis); Mather et al., p 8.

Diagnosis

Cryptogyps lacertosus is diagnosed by characters of the humerus and tarsometatarsus described in Mather et al. (2022). We add to those here to include the following characters: a coracoid where the width of the cotyla scapularis is less than a quarter of the shaft width, the processus coracoideus has little medial projection and little ventral curvature, and the shaft is broad relative to its length; a scapula with little expansion in width distally, and strong ventral curvature distally; an ulna with a markedly flattened dorsal facies along the shaft length, a non-pneumatic fossa present caudodorsal of the cotyla ventralis; a carpometacarpus where the distal end of the ventral trochlear rim is proximal of the processus pisiformes.

Referred material

SAMA P24329 fragmented thoracic vertebra; SAMA P53845 near complete right scapula, with only the distal end worn away; SAMA P42487 almost complete right coracoid with only the facies articularis clavicularis, processus acrocoracoideus and processus lateralis worn away; SAMA P24324 complete right ulna; SAMA P24323 complete left ulna; SAMA P24326 complete right radius; SAMA P24325 complete left radius; SAMA P24328 complete right carpometacarpus; SAMA P24327 complete left carpometacarpus; SAMA P53845 complete os metatarsale I. This material is referred to C. lacertosus first because the coracoid is identical to that found in Leanna’s Breadth Cave, which was in turn associated with a distinctive tarsometatarsus that was referred to this species by Mather et al. (2022). Second, these bones represent a similar sized bird to that represented by the lectotype distal humerus and other referred material for C. lacertosus (see Mather et al. 2022) and differ from Aquila audax and the other similar sized species described below. WAM 15.9.73 proximal left tarsometatarsus (see Mather et al. 2022). WAM 15.9.72 right coracoid missing omal end. AM F129566 distal right ulna. AM F54723 distal right ulna.

Locality, unit and age

SAMA P24329, SAMA P53845, SAMA P42487, SAMA P24324, SAMA P24323, SAMA P24326, SAMA P24325, SAMA P24328, SAMA P24327 derive from Site 7, Green Waterhole/Fossil Cave, 37°44′S; 140°31′E. Tantanoola district, South Australia, Australia. SAMA P53845 was recovered from the area encompassed by the grid C5–C6/N5–N6 (Fig. 2). Nearly all the vulture bones were found within a deposit of cave rafts with little terrigenous matter. Cave rafts form evaporatively on the surface of a pool in a cave so, when the fossils were deposited, they fell into a pool that was in the lower reaches of the chamber when this area of the cave was above the water table. The Uranium series dates obtained from calcite cave rafts (Table 1) reveal they formed in a narrow time interval 61.1–59.1 ka. This is consistent with deposition during the marine lowstand in Marine Isotope Stage 4, when global sea levels were ∼80 m lower than present ones (Miller et al. 2020) and the water table in the limestone housing the cave will have dropped accordingly. Rising sea levels commencing in the latest Pleistocene submerged the former pool and the surrounding slopes in and on which the fossils were deposited.

WAM 15.9.73 and WAM 15.9.72 were recovered from Leaena’s Breath Cave, Nullarbor, Western Australia, Australia, 31.4°S 128.1°E. WAM 15.9.73: Pit B1, Unit 3, depth 115–120 cm; WAM 15.9.72: Pit B, Unit 3, Quadrat 5, 100–105 cm depth, Early Pleistocene; collected by G. Prideaux in 2013, reported by E. Shute (2018).

AM F129566, no locality or stratigraphy data, however recorded as from Wellington Caves, New South Wales, Australia.

AMF 54723 was collected in 1966 or 1967 by R. M. Frank from Walli Caves, 27 km northeast of Cowra, in Wellington Valley, New South Wales, Australia.

Description

Measurements (mm)

Coracoid SAMA P42487, right: preserved omal-sternal length 73.0, preserved proximal height (cotyla scapularis to preserved processus acrocoracoideus) 33.8, facies articularis humeralis length 20.9, midshaft width 16.1, preserved sternal width 36.1. Radius, SAMA P24325, left: total length 196.6, proximal width 10.9, proximal depth (dorsal aspect) 10.1, shaft width 6.5, distal width 15.3. Radius, SAMA P24326, right: total length 195.2, proximal width 9.1, proximal depth 11.4, shaft width 6.5, distal width 15.3. Ulna, SAMA P24323, left: total length 208, proximal width 22.2, shaft width 9.5, distal width 20. Ulna, SAMA P24324, right: total length 208.8, proximal width 22.7, shaft width 9.6, distal width 19.9. Carpometacarpus, SAMA P24327, left: total length 108.4, proximal width 26.1, distal width excluding facies articularis digitalis minor 11.5, total distal width 17. Carpometacarpus, SAMA P24328, right: total length 108.7, proximal width 26.1, distal width excluding facies articularis digitalis minor 12.1, total distal width 16.5; Thoracic vertebra SAMA P24329: preserved height 33.9, preserved width 23.0, preserved length of centrum 21.7, cranial width of centrum 10.3, preserved cranial height of centrum 9.5, preserved caudal width of centrum 11.3, preserved caudal height centrum 12.6; Os metatarsale I SAMA P53845: length 26.1, distal width 12.5.

We argue that the fossil material represents a single individual on the following evidence: (1) a lack of duplicated elements; (2) that left and right sides are identical in all aspects; (3) the close proximity in which all fossil material was found; and (4) the size of all elements being within an expected range for a single skeleton.

Coracoid WAM 15.9.72: preserved length 52.9, midshaft width 15.4, preserved sternal width 32.9. This coracoid is referred to Cryptogyps lacertosus based on its discovery with the tarsometatarsus previously referred to C. lacertosus and that it represents a similar sized accipitrid; therefore, it is more parsimonious to consider these specimens represent one species of extinct vulture rather than two.

Coracoid

Two coracoids (Fig. 3A–C) are available: one from the Green Waterhole Cave individual and one from Leaena’s Breath Cave. These do not differ noticeably from each other. The Green Waterhole coracoid is well preserved, lacking only the tip of the processus lateralis, the facies articularis clavicularis, and the ventral section of the processus coracoideus. (1) The foramen n. supracoracoidei (Fig. 3A: FnS) is set against the corpus of the shaft rather than adjacent to the margin of the processus procoracoideus (Perninae; Gypaetinae; Aegypiinae; Circaetinae; most Haliaeetinae [Haliaeetus (Savigny, 1809), Milvus (Lacépède, 1799)]) and is positioned 7 mm distal to the cotyla scapularis. (2) The foramen has a medial pneumatic opening within it (Perninae; Gypaetinae; Aegypiinae; Circaetinae; few Haliaeetinae [Haliaeetus leucogaster]). (3) The sulcus m. supracoracoidei has a large pneumatic foramen dorsally within it that opens into the processus acrocoracoideus (Perninae; Gypaetinae; most Aegypiinae; Circaetinae; Aquilinae; Haliaeetinae; Buteoninae). (4) The facies articularis humeralis (Fig. 3B: FAH) is roughly twice as long as it is wide (Perninae; Gypaetinae; Aegypiinae; Circaetinae; Aquilinae; Haliaeetinae; Buteoninae). (5) The cotyla scapularis (Fig. 3B: CtS) is small, being less than a quarter of the width of the shaft (Perninae; Aegypiinae; Buteoninae), (6) and is shallow (Perninae; Aegypiinae; Circaetinae; Haliaeetinae; Buteoninae), (7) and ellipsoidal (6 by 5 mm) in shape (Perninae; Gypaetinae; Aegypiinae; Circaetinae; Aquilinae; Haliaeetinae; Buteoninae). (8) The omal margin of the processus procoracoideus (Fig. 3A: PPr) slopes sternomedially from the cotyla scapularis, (9) and has a short (8 mm) medial projection (Perninae; Gypaetinae; Aegypiinae; Haliaeetinae; Buteoninae), (10) with little to no curvature ventrally of the medial tip (Perninae; Gypaetinae; Aegypiinae; Circaetinae; Aquilinae; Haliaeetinae; Buteoninae). (11) The coracobrachialis ligament attachment scar (Fig. 3B: CLS) is positioned centrally in the shaft (Perninae; some Gypaetinae [Gypohierax Rüppell, 1836]; Circaetinae; most Aquilinae; few Haliaeetinae [H. leucogaster]; Buteoninae), (12) and is roughly triangular (Perninae; Gypaetinae; Aegypiinae; Circaetinae; Aquilinae; Haliaeetinae). (13) The impressio m. supracoracoidei (Fig. 3B: IS) is extremely shallow (Perninae; Gypaetinae; Aegypiinae; Circaetinae; Aquilinae; Haliaeetinae; Buteoninae). (14) The angulus medialis projects in a robust point that forms a 45° angle (Perninae; most Gypaetinae; most Aegypiinae [Gyps Savigny, 1809, Aegypius Savigny, 1809]; Aquilinae; Buteoninae). (15) The facies articularis sternalis dorsalis forms a dorsally prominent flange over the medial third of its extent, which is broad and rounded (most Perninae; most Gypaetinae; some Aegypiinae [Gyps]). (16) The shaft is broad and robust for its length (Perninae; some Gypaetinae [Gypohierax]; Aegypiinae).

Scapula

The right scapula (Fig. 3D–F). is near perfectly preserved, lacking only the caudal-most extremity. (1) The acromion (Fig. 3E: Ac) strongly projects cranially relative to the rest of the cranial end (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (2) The facies articularis clavicularis is laterally prominent, overhanging the lateral facies (Perninae, most Gypaetinae, most Aegypiinae, Circaetinae, Buteoninae). (3) The crista lig. acrocoracoacromiali has slight projection medially (some Aegypiinae [Necrosyrtes Gloger, 1841]). (4) A large, pneumatic fossa is present in the base of the acromion and visible in cranial view (Fig. 3F: CF) (few Perninae [Pernis Cuvier, 1816], most Gypaetinae [Gypohierax, Neophron Savigny, 1809], Aegypiinae, some Circaetinae [Spilornis G. R. Gray, 1840], Aquilinae, most Haliaeetinae, most Buteoninae). (5) The tuberculum coracoideum (Fig. 3D: TC) is flat, with very little cranial projection from the rest of the cranial end (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (6) A large pneumatic fossa is present in the medial face adjacent to the tuberculum coracoideum (Fig. 3D: MF) (Perninae, some Gypaetinae [Polyboroides Smith, 1829], some Circaetinae [Terathopius Lesson, 1830], some Aquilinae [Aquila chrysaetos (Linnaeus, 1758), Hieraaetus morphnoides (Gould, 1841), A. fasciata Vieillot, 1822], some Haliaeetinae [Haliaeetus leucogaster, Haliastur indus (Boddaert, 1783)]), (7) while there is no foramen on the lateral face (Perninae, some Gypaetinae [Polyboroides, Gypohierax], most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (8) The corpus scapulae (Fig. 3E: CSc) is elongate and narrow (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (9) and weakly expands in width distally (Perninae, Gypaetinae). (10) The margo dorsalis (Fig. 3D: MD) projects slightly dorsally relative to the collum (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (11) The corpus strongly curves ventrally towards the distal end (Gypaetinae, Aegypiinae, some Aquilinae [Spizaetus tyrannus (zu Wied-Neuwied, 1820)]).

Ulna

The left and right ulnae from Green Waterhole Cave are perfectly preserved (Fig. 4B, D, F). Additionally, there are two distal ulnae, one each from the Wellington and Walli Caves, that are assigned to the same species on the basis of morphological similarity (same degree of projection in tuberculum carpale, angle of the caudal margin of the condylus dorsalis). The following features can be observed. (1) The olecranon is set quite low, barely projecting proximal of the cotyla ventralis (Perninae, Gypaetinae, Circaetinae, few Aquilinae [Aquila fasciata], most Haliaeetinae). (2) The cotyla ventralis is large, about a third larger than the cotyla dorsalis (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (3) circular (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), and (4) deeply concave (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (5) The cotyla dorsalis is squarish in shape, with a short, distally projecting square-shaped processus cotylaris dorsalis (Perninae; Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (6) The cotylae are separated by a very prominent crista intercotylaris (some Perninae [Hamirostra, Pernis], Aquilinae, Buteoninae). (7) A shallow, distinct fossa is present caudodorsal of the cotyla ventralis (Gypaetinae, most Aegypiinae, Circaetinae, Buteoninae), (8) which is non-pneumatic (Gypaetinae, Circaetinae, Haliaeetinae, Buteoninae). (9) The impressio m. scapulotricipitis located caudal to the cotyla dorsalis is large and distinct (Perninae, Gypaetinae, Aegypiinae, Circaetinae, some Aquilinae [Aquila], Haliaeetinae, Buteoninae), and (10) shaped like an upside-down triangle (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (11) The incisura radialis is shallow and non-pneumatic (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (12) There are two ligamental scars present on the cranial face just distal of the incisura radialis, neither of which prominently protrudes cranially (Perninae, Gypaetinae, Aegypiinae, some Circaetinae, Aquilinae, Haliaeetinae, some Buteoninae [Ictinia Vieillot, 1816]). (13) The larger of the two is the dorsal scar (Fig. 4D: DS), which forms a short, concave, robust line orientated proximodistally, with the distal half curving ventrally (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (14) The ventral scar (Fig. 4D: VS) is a small circle about a quarter of the size of the dorsal tuberculum, is not concave, and is positioned ventrodistally adjacent to the dorsal scar (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (15) The tuberculum ligamentosa collateralis ventralis (Fig. 4E: TLCV) is barely prominent ventrally in cranial view, less so than the rim of the cotyla ventralis, and is proximodistally short, hardly overlapping the impressio brachialis (Perninae, Gypaetinae, some Aegypiinae), with the ligamental attachment scar on it extending proximally towards the olecranon. (16) The impressio brachialis (Fig. 4C: IB) is quite long (∼46.3 mm), and moderately deep for its size (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (17) When viewed in dorsal and ventral aspect, very little curvature is present in the shaft (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (18) The papillae remigales caudales are small and have a low profile in dorsal and ventral aspect (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (19) The tuberculum carpale (Fig. 4C: TC) prominently projects cranioventrally in dorsal and ventral view (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (20) The proximal dorsal margin of the tuberculum carpale merges evenly into the rest of the shaft (Perninae, Gypaetinae, Circaetinae, Aquilinae, Buteoninae). (21) The incisura tuberculum carpale (Fig. 4C: ITC) is quite deep and distinct, well separating the tuberculum carpale from the condylus ventralis ulnaris (Fig. 4B: CV) (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (22) The sulcus intercondylaris (Fig. 4B: SI) is shallow, forming a U-shape in ventral view (Perninae, some Aegypiinae, Aquilinae, Haliaeetinae). (23) The depressio radialis (Fig. 4C: DR) is shallow (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (24) The incisura tendineus (Fig. 4D: IT) is shallow, with the distal point of the caudal margin forming a projecting point that slightly overhangs the incisura (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (25) The dorsal margin of the dorsal condyle is oriented at an angle across the shaft (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Buteoninae). (26) The length of the dorsal condyle is roughly equal to its width as seen in dorsal aspect (Perninae, Gypaetinae, Aegypiinae, some Circaetinae [Terathopius], Buteoninae). (27) An unusual feature of the ulna is the extremely flattened dorsal facies along the entire length resulting in a distinct angularity of the dorsocaudal mid-shaft margin and of the dorsocranial shaft margin.

Radius

Both the left and right radii (Fig. 5I, J) from Green Waterhole Cave are complete and in near perfect condition, allowing for comparison of all features across taxa. (1) The cotyla humeralis (Fig. 5I: CH) has a large, semi-circular facet (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (2) which is shallow with little convexity (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (3) The tuberculum bicipitalis radialis (Fig. 5I: TBR) is quite prominent ventrally (some Perninae [Hamirostra, Lophoictinia], few Gypaetinae [Neophron], Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (4) with a shallow fossa medial to it (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (5) The facies articularis humeralis (Fig. 5I: FAU) is little projected laterally (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, some Buteoninae). (6) The ridges bounding the sulcus tendineus (Fig. 5J: ST) dorsally are flat (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (7) and the sulcus itself is shallow (Perninae, Gypaetinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (8) The facies articularis radiocarpalis (Fig. 5I: FAR) is flat in dorsal view (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (9) and orientated at a slight dorsoventral angle (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (10) The depressio ligamentosa (Fig. 5I: DL) is very shallow, and practically indistinguishable on the ventral face of the distal end (Perninae, Circaetinae, Aquilinae, most Haliaeetinae, Buteoninae). (11) The depressio ligamentosa is non-pneumatic (Perninae, Gypaetinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (12) The tuberculum aponeurosis ventralis (Fig. 5I: TAV) forms a roughly circular and prominent protrusion on the ventral face of the distal end (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae).

Carpometacarpus

A pair of complete left and right carpometacarpi (Fig. 5F, G) exist for the Green Waterhole Cave individual, which exhibits the following features: (1) The fossa infratrochlearis (Fig. 5F: FI) lacks pneumatization (most Perninae, Gypaetinae, few Aegypiinae [Necrosyrtes], Circaetinae, Aquilinae, Haliaeetinae, Buteoninae) and is shallow (Perninae, Gypaetinae, few Aegypiinae [Necrosyrtes], Circaetinae, Aquilinae, Haliaeetinae). (2) The fossa supratrochlearis (Fig. 5G: FS) lacks pneumatization and is shallow (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (3) A shallow proximodistally elongate sulcus is present on the ventral face between the processus pisiformis and processus extensorius (Fig. 5F: VS) (Perninae, most Gypaetinae, some Aegypiinae [Gyps, Aegypius, Torgos Kaup, 1828], Circaetinae, Aquilinae, most Haliaeetinae, Buteoninae). This sulcus extends from the cranial margin of the trochlea carpalis to the processus alularis. (4) The proximal margin of the processus extensorius has a shallow (Perninae, Gypaetinae, Aegypiinae, Aquilinae, Haliaeetinae), non-pneumatic fovea carpalis cranialis that extends to the base of the trochlea carpalis (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (5) The fovea carpalis caudalis is slightly deepened (Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae) and non-pneumatic (Perninae, Gypaetinae, few Aegypiinae [Trigonoceps Lesson, 1842], Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (6) The processus pisiformis (Fig. 5F: PP) moderately projects ventrally (Perninae, Gypaetinae, some Aegypiinae [Necrosyrtes], Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (7) The processus alularis (Fig. 5F: PA) projects slightly distally, creating a shallow notch between it and the shaft (few Perninae [Chondrohierax Lesson, 1843], most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (8) The processus extensorius (Fig. 5F, G: PE) projects mostly cranially with only slight proximal projection (Perninae, Gypaetinae, Aegypiinae, Circaetinae). (9) The sulcus tendineus (Fig. 5G: ST) is primarily located on the dorsal face, gradually curving onto the cranial face at mid-shaft length (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (10) The sulcus tendineus is narrow (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (11) The proximal end of the os metacarpale minus is non-pneumatic (most Perninae, Gypaetinae, some Aegypiinae [Trigonoceps, Sarcogyps Lesson, 1842] Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (12) The distal end of the ventral trochlear rim is positioned proximal to the base of the processus pisiformes and far proximal to the processus alularis (some Gypaetinae [Polyboroides]). (13) The facies articularis digitalis minor (Fig. 5F: FADMin) projects further distally than the facies articularis digitalis major (Fig. 5F: FADMaj) (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (14) The os metacarpale minus (Fig. 5G: OMM) is slightly arched caudally (Perninae, Gypaetinae, some Circaetinae [Spilornis], most Aquilinae, some Haliaeetinae [Milvus, Haliastur Selby, 1840]).

Thoracic vertebra

The thoracic vertebra (Fig. 6A–C) SAMA P’24329 from Green Waterhole Cave preserves the processus ventralis, both the cranial and caudal articular facets of the heterocoelous corpus, both zygapophyses caudales, one zygapophysis cranialis, and the foramen vertebrae. The shape of the facies articularis caudalis indicates it is the fourth thoracic vertebra (see summary for greater detail), and this determination was used when making comparisons. The description is as follows: (1) A large pneumatic foramen exists just proximal to the zygapophyses caudales (Fig. 6A, B: ZCa) (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (2) The foveae costales (Fig. 6B, C: FvC) are adjacent to the margin of the facies articularis cranialis of the corpus (some Perninae, Gypaetinae, Aegypiinae, Aquilinae). (3) The foveae costales are on lateral prominence of the corpus vertebrae in cranial view (most Perninae, Gypaetinae, some Buteoninae [Ictinia]). (4) A small, pneumatic foramina is present in the ventral half of the corpus (most Perninae, most Gypaetinae, some Aegypiinae [Necrosyrtes, Aegypius], Circaetinae, Aquilinae, Haliaeetinae, some Buteoninae [Ictinia]). (5) The shaft of the processus ventralis (Fig. 6C: PV) is robust, (Gypaetinae, Circaetinae, Aquilinae, Haliaeetinae) with its cranial least width roughly a third (3.4 mm) of the width of the facies articularis cranialis of the corpus (10.3 mm). (6) The processus ventralis ends in a bifid projection (most Gypaetinae, some Aegypiinae [Aegypius]) that would have been roughly equivalent in width to that of the facies articularis cranialis of the corpus, with the bifid tips strongly projecting laterally.

Os metatarsale I

The os metatarsale I (Fig. 6G, H) from the Green Waterhole Cave individual is unbroken and has excellent preservation of detail. (1) The os metatarsale I has a largely straight lateral margin (Fig. 6G: LM), with slight curvature towards the distal end (most Perninae, some Gypaetinae [Polyboroides], Aegypiinae). (2) The lateral margin of the distal end is prominently inflated, forming a near 160° angle two-thirds of the way down the shaft (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (3) The facet on the processus articularis tarsometatarsalis (Fig. 6H: PAT) is long, taking up more than half the ‘shaft’ length, but not quite extending to the proximal point of the distal inflation (some Perninae [Hamirostra, Lophoictinia], Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (4) A shallow indentation is present on the lateral margin of the processus articularis tarsometatarsalis (Fig. 6H: I) (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (5) The medial margin has a very weak protrusion (Fig. 6H: Pr) at the processus articularis tarsometatarsalis base (most Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Buteoninae). (6) The sulcus is bordered by low-projecting ridges (Fig. 6H: SR) at its distal termination (Perninae, Gypaetinae, Aegypiinae, some Circaetinae [Terathopius], Aquilinae, Buteoninae).

Remarks

The new material of Cryptogyps lacertosus differs from Aquila audax in the following characters (A. audax in brackets): the wing bones are notably more robust in C. lacertosus. Ulna: the dorsal facies is greatly flattened (rounded); the tuberculum ligamentosa collateralis ventralis is barely prominent and does not extend ventrally past the cotyla ventralis rim (ventrally prominent and extends past the cotyla rim); the fossa caudodorsal of the cotyla ventralis is inset into the cotyla facet (positioned adjacent to cotyla rim); the olecranon has little proximal projection to the cotyla ventralis (prominent projection); the dorsal margin of the dorsal condyle form a notable angle of approximately 70° relative to the shaft (very slight angle, 80°+). Carpometacarpus: the processus extensorius has little proximal projection (notable proximal upturn); the distal end of the ventral trochlear rim is proximal of the base of the processus pisiformis and well proximal of the processus alularis (adjacent with); the os metacarpale minus is arched towards the caudal end (largely flat). Scapula: the facies articularis clavicularis has extreme lateral prominence and distinctly overhangs the lateral facies (slight prominence); a large, deep pneumatic fossa is present in the medial face (absent); the corpus scapulae weakly expands in width distally (notably expands). Radius: the tuberculum bicipitalis radialis is largely flat and does not extend beyond the cotyla humeralis margin (peaked, prominently extends past margin); the tuberculum aponeurosis ventralis is less prominent ventrally (more prominent); the facies articularis ulnaris is dorsoventrally compressed (not compressed). Coracoid: the foramen n. supracoracoidei is set adjacent to the shaft corpus (set close to the medial margin of the processus procoracoideus); the foramen has a medial opening into the corpus (absent); the cotyla scapularis is less than a quarter the size of the facies articularis humeralis and is shallow (quarter of facies size, deep); the margin of the processus procoracoideus slopes sternomedially to the cotyla scapularis (consistently level with cotyla).

While some generic differences between C. lacertosus and D. gaffae have already been established (see Mather et al. 2023), the new C. lacertosus material reveals further differences from D. gaffae (character state in brackets) notably in the following ways. Ulna: the tuberculum carpale is strongly projecting (weakly projecting). Carpometacarpus: the ventral rim of the trochlea carpalis terminates well proximal of the processus pisiformes (adjacent to) Scapula: a pneumatic foramen is present in the base of the acromion in proximal view (absent), the distal margin of the facies articularis humeralis projects outwards from the shaft margin (continuous with shaft margin).

Dynatoaetus Mather, Lee, Camens and Worthy, 2023

Type species

Dynatoaetus gaffae Mather, Lee, Camens and Worthy, 2023.

Remarks

Referred to Dynatoaetus based on the following characters. Humerus: the fossa brachialis is notably deepened, the epicondylus ventralis is weakly projecting. Ulna: the tuberculum carpale has very little cranial projection. Femur: the muscle scar proximal to the fossa poplitea is circular in shape, the epicondylus lateralis is strongly projecting, the crista supracondylaris medialis is prominent, the sulcus between the condyles is very deep.

Dynatoaetus pachyosteus sp. nov.

(Figs 4–7)

Fig. 7. Fossil bones of species of Dynatoaetus, pelvis (A–D), femora (E–P). Dynatoaetus pachyosteus, pelvis SAMA P41516 in A, dorsal, B, ventral, C, right lateral and D, caudal view. Dynatoaetus gaffae right femur SAMA P41514 (E, G, I, K, M, O) compared to D. pachyosteus, SAMA P41513, Victoria Fossil Cave (SA) (F, H, J, L, N, P), in cranial (E, F), caudal (G, H), lateral (I, J), proximal cranial (K, L), distal caudal (M, N) views, and outlines of musculature scars on proximal lateral facies (O, P). Abbreviations: Ant, antitrochanter; AFR, anterior fossa renalis; CD, crista dorsolateralis; CSS, crista spinosa synsacri; CTF, crista trochanteris foramen; DIC, dorsal iliac crests; ECS, extremitas cranialis synsacra; EL, epicondylus lateralis; FLC, fovea lig. capitis; FT, fossa trochanteris; FP, fossa poplitea; IGL, impressio gastrocnemialis lateralis; IGM, impressio gastrocnemialis intermedia; ILCC, impressio lig. cruciati caudalis et cranialis; ILCL, impressio lig. collateralis lateralis; LIC, linea intermuscularis cranialis; PC, processus costales; TMGM, tuberculum muscularis gastrocnemialis medialis. Black scale bar 10 mm, black and grey scale bar 50 mm.

Diagnosis

A large accipitrid of similar size to Aquila audax and Cryptogyps lacertosus, with short, stout wing bones and very large and robust leg bones that exhibits the following unique combination of characters. Quadrate with (1) the processus orbitalis having a deepened sulcus on its medial facies. (2) There is no foramen basiorbitale nor a foramen rostromediale. (3) Condylus pterygoideus is prominent medially, projecting more so than the condylus medialis. Humerus with (4) a deep m. scapulohumeralis cranialis dorsally adjacent to the fossa pneumotricipitalis, (5) tuberculum supracondylaris dorsalis not prominent dorsally and only slightly so cranially, and (6) a markedly sigmoid curvature to the shaft. Pelvis with (7) a large, deep anterior fossa renalis. Femur with (8) the ligamental attachment scar proximal to the fossa poplitea, circular in shape and positioned centrally in the shaft.

Etymology

Combination of Ancient Greek πᾰχῠ́ς (‘pachys’, masculine), meaning ‘thick’, ‘large’, or ‘stout’, and the Ancient Greek ὀστέον (ostéon, neutral), meaning ‘bone’. The name references the size and robustness of the bones attributed to the species.

Holotype

SAMA P41517 (=FU 1141) left humerus missing only tuberculum ventrale.

Paratypes

SAMA P59030, complete right quadrate (62.75′–64.75′, R13.5′–15.5′, D/D −2.0′ to −2.5′); SAMA P59029, complete left ulna (–7′ to −3′, R5.5′–7.5′, D/D −1.72′ to −1.97′); SAMA P41515, right carpometacarpus missing distal end and most of os metacarpale minus (57′–60.5′, R12′–13′, D/D −1.75′ to −2.75′); SAMA P41516, partial pelvis preserving synsacrum and most of right lateral side (57.5′–62.5′, R13′–14.5′, D/D −0.5′ to −1.0′); SAMA P41513, complete right femur (56.5′–58.5′, R1′–2′, D/D −0.5′ to −1.0′). Five specimens, MNI = 1. All can be attributed to the upper part of Unit 7 as defined by Reed (2003, Fig. 6.3O). All differ from similar-sized A. audax and C. lacertosus, where comparable, and are substantially smaller than D. gaffae, which is known from the same deposit; therefore, all are referred to D. pachyosteus.

Type locality, unit and age

Excavation co-ordinates (decimal feet) 64.5′–67′, R9′–10′, D/D −0.5′ to −1.0′, Main Fossil Chamber, Victoria Fossil Cave, Naracoorte, South Australia. Early–Middle Pleistocene, between 300–150 ka (Arnold et al. 2022).

Zoobank registration

Zoobank LSID of publication: urn:lsid:zoobank.org:pub:16EEAC4F-71AF-4D64-9128-175D5E815491; Zoobank LSID of new species: urn:lsid:zoobank.org:act:3A539E7E-FDC0-4B2B-A12A-3D51DA8D2537.

Description

Measurements (mm)

Holotype: total length 178.3, proximal width [tuberculum dorsale—crista bicipitalis] 37.1, width fossa pneumotricipitalis 11.7, length crista deltopectoralis from tuberculum dorsale 59.2, least shaft width 18.7, maximum distal width 33.3, condylus dorsalis width 9.3, condylus dorsalis depth 19.4, condylus ventralis width 14.8, condylus ventralis depth 10.7, tuberculum supracondylaris ventralis to processus flexorius depth 15.0.

Victoria Fossil Cave, Quadrate SAMA P59030: height 21.3, mediolateral width across condyles 15.5. Pelvis SAMA P41516: length synsacrum 104.7. Femur SAMA P41513: total length 131.1, proximal width 32.5, preserved proximal depth 23.8, shaft width 18.9, least shaft circumference 59, distal width 32.3 (without lateral projection), distal width 35.8 (with lateral projection), width condylus medialis 17.1, height condylus medialis 19.2, depth condylus medialis 25.5, width condylus lateralis 13.1, height condylus lateralis 19.0, depth condylus lateralis 31.3. Humerus SAMA P41517; total length 178.3, proximal width [tuberculum dorsale—crista bicipitalis] 37.1, width fossa pneumotricipitalis 11.7, length crista deltopectoralis from tuberculum dorsale 59.2, least shaft width 18.7, maximum distal width 33.3, condylus dorsalis width 9.3, condylus dorsalis depth 19.4, condylus ventralis width 14.8, condylus ventralis depth 10.7, tuberculum supracondylaris ventralis to processus flexorius depth 15.0. Ulna SAMA P59029: total length 195.4, proximal width 21.4, shaft width 9.7, distal width 15.8. Carpometacarpus SAM P41515: proximal width 26, shaft width 7.6.

Humerus

The holotype is a complete right humerus (Fig. 5A, B) lacking only the tuberculum ventrale. It is approximately comparable in length to that of a female Aquila audax but is much stouter and more sigmoid. The humerus is quite robust for its length. (1) The tuberculum dorsale (Fig. 5B: TD) is a small facet directed proximally, slightly prominent dorsoproximally, and proximodistally aligned/level with the sulcus lig. transversus (Fig. 5B: SLT) in cranial view (Perninae, most Gypaetinae, Aegypiinae, some Circaetinae [Terathopius], Haliaeetinae, Buteoninae). (2) The sulcus lig. transversus is deep but limited to the ventral side of the caput (Perninae, Haliaeetinae). (3) The intumescentia humeri is inflated into a slight mound (most Perninae, some Haliaeetinae [Haliaeetus albicilla (Linnaeus, 1758), H. leucocephalus (Linnaeus, 1766)]). (4) The margin of the fossa pneumotricipitalis ventralis (Fig. 5A: FPV) forms a broad semi-circle (Gypaetinae, Aegypiinae, some Haliaeetinae [Haliaeetus]). (5) The insertion of the m. scapulohumeralis cranialis (Fig. 5A: MSC) is deep (some Gypaetinae [Polyboroides], Buteoninae). (6) The attachment scar for the m. scapulohumeralis caudalis is not prominent and forms a broad scar on the ventral rim of the crista bicipitalis (Gypaetinae, Aegypiinae, some Aquilinae [Spizaetus Vieillot, 1816], some Haliaeetinae [Milvus migrans (Boddaert, 1783), Haliastur]). (7) The incisura capitis (Fig. 5A: IC) is shallow relative to the caput (tuberculum ventrale is broken so depth relative to it not assessable) (Perninae, Aegypiinae, most Aquilinae, some Haliaeetinae [Haliaeetus, Milvus milvus (Linnaeus, 1758)]). (8) The insertion scar at the base dorsally of the incisura capitis is round and very faint (Perninae, Gypaetinae, Aegypiinae, some Circaetinae [Spilornis], Aquilinae, Haliaeetinae, Buteoninae). (9) The angulus deltopectoralis of the crista deltopectoralis (Fig. 5B: CD) is roughly level with the distal margin of the crista bicipitalis (Fig. 5A, B: CB) (Aegypiinae, Circaetinae, Aquilinae, some Haliaeetinae [Haliaeetus, Milvus milvus]). (10) The dorsal margin of the section of the crista deltopectoralis proximal to the angulus is flat (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, most Aquilinae, Haliaeetinae, Buteoninae). (11) The angulus deltopectoralis forms a rounded angle (∼100°) in ventral view (some Perninae [Hamirostra, Lophoictinia], Gypaetinae, Aegypiinae, Circaetinae, most Aquilinae, Haliaeetinae). (12) The distal termination of the crista is positioned on the cranial facies of the shaft displaced from the dorsal margin, allowing the shaft margin to be visible alongside it (Perninae, Gypaetinae, some Aegypiinae [Gyps, Aegypius], some Circaetinae [Spilornis], Aquilinae), (13) and merges with the shaft without change in angle in ventral view (Perninae, Gypaetinae, Aegypiinae, some Circaetinae [Spilornis], few Buteoninae [Buteo rufofuscus (J. R. Forster, 1798)]). (14) In cranial view, the crista prominently projects dorsally from the shaft, distinctly further than the tuberculum dorsale (Aegypiinae [Gyps and Necrosyrtes slightly dorsal, Aegypius distinctly dorsal]). (15) A sulcus is present on the proximal caudal face of the crista deltopectoralis (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (16) In cranial view, the insertion scar for the m. pectoralis (Fig. 5B: MPS) is robust and extends proximally to level with the distal base of the crista bicipitalis (some Perninae [Hamirostra, Chondrohierax], some Gypaetinae [Polyboroides], Circaetinae, Aquilinae, Buteoninae). (17) The crista bicipitalis forms a distinct distoventrally convex flange distally (some Perninae [Hamirostra, Lophoictinia], Gypaetinae, Aegypiinae, some Circaetinae [Spilornis], most Aquilinae, Haliaeetinae, Buteoninae). (18) A faint sulcus nervus coracobrachialis (Fig. 5B: SNC) is present extending proximodorsally from the distal margin of the crista bicipitalis (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (19) The shaft is markedly sigmoid in its curvature (Aquilinae [most similar to Spizaetus], few Buteoninae [Buteo nitidus (Latham, 1790)]). (20) The scar for the m. latissimus dorsi forms a distinct line and angularity caudally on the shaft (Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, some Haliaeetinae [Haliaeetus]). (21) The processus flexorius (Fig. 5A: PF) does not protrude caudally in ventral view (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (22) In cranial view, the processus flexorius and condylus ventralis (Fig. 5B: CV) are roughly equal in distal extent (Perninae, Circaetinae, Aquilinae). (23) Distally, the processus flexorius is rounded (some Perninae [Lophoictinia], Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, some Haliaeetinae [Milvus migrans], Buteoninae). (24) The insertion scar for the pronator superficialis is large and deep and close to the ventrocranial margin of the tuberculum supracondylare ventralis (some Perninae [Hamirostra], Circaetinae, Aquilinae, most Buteoninae). (25) The tuberculum supracondylare dorsale (Fig. 5B: TSD) lacks dorsal projection but is slightly prominent cranially (Gypaetinae, Aegypiinae, some Circaetinae [Spilornis], some Aquilinae [Spizaetus]). (26) The fossa olecrani (Fig. 5A: FO) is shallow (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (27) The fossa m. brachialis (Fig. 5B: FB) is deep (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae), (28) and non-pneumatic (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (29) The dorsal margin of this fossa is broadly separated from the ventral shaft margin by a distance over a third of fossa width (Gypaetinae, some Circaetinae [Spilornis], Haliaeetinae). (30) The palmar scar for the m. extensor carpi radialis is a shallow oval/line, while the dorsal scar is a large, shallow circle (Perninae, Gypaetinae, Aegypiinae, Haliaeetinae, Buteoninae). (31) There are two shallow insertion scars for the m. flexor carpi ulnaris (most Perninae, Aquilinae). (32) The tuberculum supracondylare ventrale (Fig. 5B: TSV) is aligned proximodorsally across the shaft (most Perninae, Gypaetinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (33) The proximoventral margin of the tuberculum supracondylare ventrale is sharply prominent cranially (Perninae, Aquilinae, Haliaeetinae, Buteoninae). (34) The dorsal sulcus for the m. humerotricipitalis is broad, spanning roughly half the dorsoventral shaft width beside it (some Perninae [Chondrohierax, Pernis], some Gypaetinae [Polyboroides, Gypohierax], Circaetinae, Aquilinae, some Haliaeetinae [Haliaeetus], Buteoninae) and is twice as wide as its ventral counterpart.

Ulna

One ulna (Fig. 4C, E, G) is attributed to Dynatoaetus pachyosteus, which is well preserved and complete. (1) The olecranon (Fig. 4E: O) is quite low, barely projecting proximal of the cotyla ventralis (Perninae, Gypaetinae, Circaetinae, few Aquilinae [Aquila fasciata], most Haliaeetinae). (2) The cotyla ventralis is quite large, area about twice the size of the cotyla dorsalis (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), (3) circular (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), and (4) moderately concave. (5) The cotyla dorsalis is square-like in shape, with a short, distally projecting well rounded processus cotylaris dorsalis enclosing a slight notch ventrally, this notch being absent in Cryptogyps (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (6) The cotylae are separated by a moderately prominent crista intercotylaris (few Perninae [Pernis, Hamirostra], Aegypiinae, Aquilinae, Buteoninae). (7) A very shallow, faint fossa is present caudodorsal of the cotyla ventralis and does not impact on the rim of the cotyla, unlike in Cryptogyps where the homologous fossa is placed more ventral and clearly interrupts the rim of the fossa (Fig. 4D: CF) (Gypaetinae, most Aegypiinae, Circaetinae, Buteoninae), (8) which is non-pneumatic (Gypaetinae, Circaetinae, Haliaeetinae, Buteoninae). (9) The impressio m. scapulotricipitis (Fig. 4F: ScT) located caudal to the cotyla dorsalis is small and distinct (most Perninae, some Aquilinae [Hieraaetus, Spizaetus], some Buteoninae [Buteo Lacépède, 1799]), and (10) triangular (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (11) The incisura radialis (Fig. 4E: IR) is shallow and non-pneumatic, and much narrower than in Cryptogyps (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (12) There are two ligamental scars present on the cranial face just distal of the incisura radialis, neither of which prominently protrudes cranially (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (13) The larger of the two is the dorsal scar (Fig. 4D: DS) (Perninae, Gypaetinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), which in the fossil forms a short, concave, robust line orientated, with the proximal and distal ends notably curving ventrally. (14) The ventral scar (Fig. 4D: VS) is a small circle about a quarter of the size of the dorsal tuberculum, is not concave (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae), and is positioned distally adjacent to the dorsal scar in the fossil. (15) The tuberculum ligamentosa collateralis ventralis (Fig. 4E: TLCV) is elongate proximodistally, more so than Cryptogyps, and overlaps the impressio brachialis, and is more prominent ventrally than that of Cryptogyps in cranial view, with equivalent ventral prominence to that of the ventral margin of the cotyla ventralis (few Gypaetinae [Polyboroides]) and the ligamental attachment scar on it extends caudally on the ventral face (Fig. 4B: PLS) and extending proximally towards the olecranon. (16) The impressio brachialis (Fig. 4C: IB) is quite long and shallow (Perninae, Gypaetinae, Aegypiinae, most Circaetinae, most Aquilinae, Haliaeetinae, Buteoninae). (17) When viewed in dorsal and ventral aspect, very little curvature is present in the shaft, but is more so than in Cryptogyps (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (18) The papillae remigales caudales are small and have a low profile in dorsal and ventral aspect (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (19) The tuberculum carpale (Fig. 4C: TC) has very little cranioventral projection (some Perninae, most Gypaetinae, Circaetinae, Aquilinae, some Haliaeetinae [Haliaeetus], Buteoninae), to the point of being almost flattened in the fossil. (20) The proximal dorsal margin of the tuberculum carpale smoothly grades into the rest of the shaft (Perninae, Gypaetinae, Circaetinae, Aquilinae, Buteoninae). (21) The incisura tuberculum carpale (Fig. 4C: ITC) is barely evident (some Perninae, most Gypaetinae, Circaetinae, Aquilinae, some Haliaeetinae [Haliaeetus], Buteoninae) on account of the very small tuberculum carpale. (22) The sulcus intercondylaris (Fig. 4B: SI) is very shallow, forming a U-shape in ventral view but does not notch the distal end (Perninae, most Gypaetinae, most Aegypiinae, few Aquilinae [Hieraaetus moorei]). (23) The depressio radialis (Fig. 4C: DR) is shallow (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (24) The incisura tendineus (Fig. 4D: IT) is shallow (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (25) The dorsal margin of the dorsal condyle (Fig. 4C: CD) is oriented at an angle across the shaft (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Buteoninae). (26) The length of the dorsal condyle is roughly equal to its width in dorsal aspect (Perninae, Gypaetinae, Aegypiinae, some Circaetinae [Terathopius], Buteoninae). (27) The dorsal shaft facies lacks the overt flattening present in the ulnae shaft of Cryptogyps.

Carpometacarpus

A near complete right carpometacarpus (Fig. 5H) missing only the distal end and the os metacarpale minus is known from Victoria Fossil Cave and is attributed to Dynatoaetus pachyosteus. (1) The fossa infratrochlearis (Fig. 5F: FI) lacks pneumatization (most Perninae, Gypaetinae, few Aegypiinae [Necrosyrtes], Circaetinae, Aquilinae, Haliaeetinae, Buteoninae) and is shallow (Perninae, Gypaetinae, few Aegypiinae [Necrosyrtes], Circaetinae, Aquilinae, Haliaeetinae). (2) The fossa supratrochlearis (Fig. 5G: FS) lacks pneumatization and is shallow (Perninae, Gypaetinae, most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (3) A shallow proximodistally elongate sulcus is present in the ventral face between the processus pisiformis and processus extensorius (Fig. 5F: VS) (Perninae, most Gypaetinae, some Aegypiinae [Gyps, Aegypius, Torgos], Circaetinae, Aquilinae, most Haliaeetinae, Buteoninae). This sulcus extends from the cranial margin of the trochlea carpalis to the processus alularis. (4) The proximal margin of the processus extensorius has a shallow fovea carpalis cranialis (Perninae, Gypaetinae, Aegypiinae, Aquilinae, Haliaeetinae), that is non-pneumatic (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (5) The fovea carpalis caudalis is slightly deepened (Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae) and non-pneumatic (Perninae, Gypaetinae, few Aegypiinae [Trigonoceps], Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (6) The processus pisiformis (Fig. 5F: PP) moderately projects ventrally (Perninae, Gypaetinae, some Aegypiinae [Necrosyrtes], Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (7) The dorsal facies between the processus alularis (Fig. 5F: PA) and the shaft has a shallow notch (few Perninae [Chondrohierax], most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (8) The processus extensorius (Fig. 5F, G: PE) projects mostly cranially with only slight proximal projection (Perninae, Gypaetinae, Aegypiinae, Circaetinae). (9) The sulcus tendineus (Fig. 5G: ST) is primarily located on the dorsal face, gradually curving onto the cranial face at the half-shaft length. (10) The sulcus tendineus is narrow (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (11) The proximal end of the os metacarpale minus is non-pneumatic (most Perninae, Gypaetinae, some Aegypiinae [Trigonoceps, Sarcogyps] Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (12) The distal end of the ventral trochlear rim extends distally to the processus pisiformes and essential level with the processus alularis, and thus is much more extensive distally than Cryptogyps (most Accipitridae).

Quadrate

The right quadrate (SAMA P59030) from Victoria Fossil Cave is almost complete, revealing all details of the processus oticus, cotylae and most of the processus orbitalis (Fig. 6D–F). It exhibits multiple differentiating features from A. audax and so is assumed to belong to Dynatoaetus pachyosteus. (1) The capitulum oticum (Fig. 6D: CO) is twice as large as the capitulum squamosum (Fig. 6E: CS) (some Perninae [Pernis], Gypaetinae, most Aegypiinae, Circaetinae, some Aquilinae [Hieraaetus], Buteoninae). (2) There is no caudal foramen under the capitulum (most Gypaetinae, few Aegypiinae [Sarcogyps], Circaetinae, some Aquilinae [Hieraaetus, Spizaetus], Haliaeetinae, Buteoninae). (3) There is no foramen pneumaticum caudomediale (most Perninae, Gypaetinae, Aegypiinae, Circaetinae, most Aquilinae, most Haliaeetinae, Buteoninae). (4) There is no foramen pneumaticum basiorbitale (most Aegypiinae, few Haliaeetinae (Haliaeetus leucocephalus]). (5) There is no foramen rostromediale (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, few Haliaeetinae [Haliaeetus albicilla, Haliastur sphenurus (Vieillot, 1818), Milvus migrans], Buteoninae). (6) The condylus lateralis (Fig. 6F: CL) is roughly twice the size of the condylus medialis (Fig. 6F: CM (some Perninae [Chondrohierax], most Gypaetinae, Aegypiinae, Circaetinae, some Aquilinae [Aquila fasciata, Spizaetus], most Haliaeetinae). (7) The processus orbitalis has a deep sulcus on its medial face (shallow variant in some Aegypiinae [Gyps, Aegypius], Haliaeetinae). (8) The cotyla quadratojugalis (Fig. 6E: CQ) takes up two thirds of the lateral face and is deep and fully enclosed (Perninae, most Gypaetinae, Aegypiinae, Circaetinae, some Haliaeetinae [Haliastur, Milvus migrans]). (9) The condylus pterygoideus (Fig. 6D: CP) is very prominent medially (extends rostrally and medially past the condylus medialis) (most Gypaetinae [Neophron, Polyboroides]) (10) and is well separated dorsally from the condylus medialis (some Perninae [Hamirostra, Lophoictinia], some Gypaetinae [Neophron], most Aegypiinae, some Circaetinae [Terathopius]).

Pelvis

The partial pelvis SAMA P41516 from the type locality, Victoria Fossil Cave (Fig. 7A–D). SAMA P41516 has most of the right lateral side, excluding the caudal-most region, preserved and is short and robust compared to other Australian accipitrids of similar size so is referred to Dynatoaetus. (1) The crista iliaca dorsalis (Fig. 7A: DIC) are not separated above the 5th and 6th synsacral vert (this is 15–20 mm in front of the acetabular region) (some Perninae [Hamirostra], Gypaetinae, Aegypiinae, Circaetinae, some Aquilinae [Spizaetus], few Haliaeetinae [Haliaeetus albicilla], Buteoninae), but damage precludes assessing separation more anteriorly. (2) It lacks foramina intertransversaria (Perninae, most Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, most Haliaeetinae, Buteoninae). (3) The crista spinosa synsacri (Fig. 7D: CSS) is flat, barely protruding from the dorsal surface (Perninae, most Gypaetinae, Aegypiinae, most Aquilinae, Haliaeetinae, Buteoninae). (4) The antitrochanter (Fig. 7A, B, D: Ant) projects further laterally than the crista dorsolateralis ilii (Fig. 7D: CD) (Gypaetinae, Aegypiinae, Circaetinae, some Aquilinae [Aquila], few Haliaeetinae [Haliaeetus leucogaster]). (5) The fossa iliocaudalis is very shallow (Perninae, some Gypaetinae [Polyboroides, Gypaetus Storr, 1784], most Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (6) There is no processus ventralis on the extremitas cranialis synsacra (Fig. 7C: ECS) (Perninae, Gypaetinae, some Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (7) The anterior fossa renalis (Fig. 7B: AFR) is extremely large and deep (few Gypaetinae [Gypohierax], most Aegypiinae, Circaetinae, Aquilinae, few Haliaeetinae [Haliaeetus leucocephalus, H. albicilla]). (8) The sutura iliosynsacralis forms an angled line, with the dorsal half parallel to the dorsoventral axis and the lower half angled roughly 45° relative to the axis (Perninae). (9) The processus costales (Fig. 7B: PC) immediately cranial to the anterior fossa renalis are robust and fused laterally to brace the ilium (Perninae, most Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, some Buteoninae [Buteo]).

Femur

The femur SAMA P41513 is nearly perfectly preserved (Fig. 7F, H, J, L, N, P). It is large and robust, yet considerable smaller than that of Dynatoaetus gaffae, found in the same deposits and furthermore differs in morphological details so it is referred to Dynatoaetus pachyosteus. It exhibits the following features. The femur is extremely large and robust compared to all living large Australian accipitrids (see Table 2), and in Australasia only specimens of Dynatoaetus gaffae and Haast’s eagle Hieraaetus moorei outsize them. (1) The fovea lig. capitis (Fig. 7H: FLC) is deep (few Gypaetinae [Polyboroides], Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae) and (2) large relative to the caput (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (3) The fossa trochanteris (Fig. 7E: FT) is very shallow (Perninae, most Gypaetinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (4) The crista trochanteris has one very large pneumatic foramen penetrating it medially (Fig. 7E, F: CTF) (Perninae, Gypaetinae, Aegypiinae, Circaetinae, most Aquilinae, Haliaeetinae, Buteoninae). (5) The depression distad to the facies articularis antitrochanterica on the caudal face is very shallow (Perninae, most Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, most Buteoninae). (6) The linea intermuscularis cranialis (Fig. 7E: LIC) is positioned at mid-lateromedial width (most Gypaetinae, Aegypiinae, Circaetinae). (7) The proximal point of the epicondylus lateralis (Fig. 7F: EL) is strongly projecting laterally (Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (8) The crista supracondylaris medialis is extremely prominent, as formed by the tuberculum muscularis gastrocnemialis medialis (Fig. 7H: TMGM). (9) The impressio gastrocnemialis lateralis immediately adjacent to the prominent epicondylus lateralis (Fig. 7F, L: IGL) is shallow and large (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Haliaeetinae, Buteoninae). (10) The condylus lateralis and medialis are separated by a deep notch caudally (Perninae, Aegypiinae, Aquilinae, Haliaeetinae, Buteoninae). (11) The impressio gastrocnemialis intermedia (Fig. 7G, H: IGM) is circular in shape (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Haliaeetinae, Buteoninae), (12) and is positioned centrally in the shaft (most Perninae, most Gypaetinae, most Aegypiinae, Circaetinae, most Aquilinae). (13) The fossa poplitea (Fig. 7G: FP) is deep (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae). (14) The impressio lig. cruciati caudalis et cranialis (Fig. 7G: ILCC) form two distinct shallow sulci (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (15) The pars cranialis is relatively twice as wide as the pars caudalis (Perninae, Gypaetinae, Aegypiinae, Circaetinae, Aquilinae, Haliaeetinae, Buteoninae). (16) The pars caudalis is slightly deeper than the pars cranialis (Circaetinae, Aquilinae). (17) The impressio lig. collaterale laterale (Fig. 7F: ILCL) spans roughly two-thirds the caudo-cranial depth of the condylus lateralis (Perninae, Aegypiinae, Buteoninae).

Table 2. Comparative measurements of the humerus, femora and tibiotarsus in Dynatoaetus pachyosteus to D. gaffae, Aquila audax, Hieraaetus moorei, Haliaeetus leucogaster, Aegypius monachus, Gyps coprotheres and Necrosyrtes monachus.

Species	Humerus length	Humerus proximal width	Humerus least shaft width	Humerus distal width	Femur length	Femur proximal width	Femur least shaft width	Femur distal width	Tibiotarsus distal width	Tibiotarsus least shaft width
Dynatoaetus pachyosteus	178.3	37.1	15.7	33.3	131.1	32.5	16.2	32.3	–	–
Dynatoaetus gaffae	–	54.5	–	44.8	145.5	38.3	18.6	35.2	27.6	14.6
Aquila audax	179.7, (10), 167.6–191.8, 9.0	34.4, (10), 32.7–36.5, 1.7	11.8, (10), 10.6–13.4, 0.9	29.5, (10), 28.1–31.4, 1.2	112.4, (10), 103.3–117.7, 4.4	24.9, (10), 23.1–27.7, 1.6	11.3, (10), 10.1–12.3, 0.8	23.3, (10), 21.5–25.6, 1.5	19.2, (10), 17.8–20.8, 1.2	10.1, (10), 9.4–11.1, 0.6
Hieraaetus moorei	232.4, (20), 208–259.9, 16.2	52.7, (18), 42.3–58.3, 4.5	17.9, (21), 15.6–21.0, 1.7	41.3, (21/20), 37.1–47.3, 3.0	162.8, (17), 140.3–176.0, 10.5	42.2, (17), 36.0–47.5, 3.8	18.3, (18), 15.3–19.9, 1.5	45.4, (18), 36.2–52.5, 5.1	34.1, (15), 27.3–39.8, 4.0	14.2, (17), 11.8–16.6, 1.6
Haliaeetus leucogaster	189.4, (10), 180.4–197.3, 6.4	35.5, (10), 32.9–37.8, 1.6	13.1, (10), 11.9–14, 0.6	30.6, (10), 28.6–32.7, 1.4	102.2, (10), 96.9–106.3, 3.4	24.3, (10), 22.4–25.8, 1.2	11.9, (10), 10.8–12.8, 0.6	21.6, (10), 20.1–23.1, 1.2	18.1, (10), 16.3–19.4, 1.2	9.9, (10), 8.8–11.4, 0.7
Aegypius monachus	212.7	37.4	14.3	36.2	112.6	28.3	11.6	23.8	18.7	9.6
Gyps coprotheres	260.1	45.5	18.0	35.8	134.2	32.1	15.2	30.4	23.1	12.3
Necrosyrtes monachus	154.5	28.4	10.7	26.5	79.1	18.4	8.6	16.6	13.9	7.5

For Aquila audax, Haliaeetus leucogaster and Hieraaetus moorei, the data represents the average, number of specimens, range, and standard deviation of measurements. Specimens used: Aquila audax FUR 085, FUR 125, SAMA B46162, SAMA B46992, SAMA B31109, SAMA B46161, SAMA B47814, SAMA B32850, SAMA B32849, SAMA B49025; Haliaeetus leucogaster SAMA B49459, SAMA B58700, SAMA B46632, SAMA B56381, SAMA B59108, SAMA B56082, SAMA B48927, SAMA B49688, SAMA B48773, SAMA B37344; Hieraaetus moorei data taken from Worthy and Holdaway (2002); Aegypius monachus R553; Gyps coprotheres ANWC 22724; Necrosyrtes monachus USNM 620646. *Breakage prevented from measuring from the same position as other specimens.

Remarks

Dynatoaetus pachyosteus can be distinguished from D. gaffae (state in brackets) in the following features. Humerus: the tuberculum supracondylare ventrale is rounded in ventral view (peaked) and extends to the insertion of the m. pronator profundus (ends proximal to the insertion). The dorsal margin of the fossa brachialis is broadly separated from the ventral shaft margin by a distance over a third of fossa width (narrow separation, less than a quarter of shaft width). The dorsal scar for the m. extensor carpi radialis is a broad circle that spans up to half the epicondylus ventralis length (spans a third of length). Ulna: the ulnae are very similar between species, with the major differing factor being their size. Carpometacarpus: the fovea carpalis caudalis is slightly deepened (shallow). The facies caudal to the processus pisiformes is flattened (prominent muscle scar present). The processus extensorius is slightly upturned towards the proximal tip (mostly horizontal). Pelvis: the VFC pelvis is markedly smaller in size than that of D. gaffae. Femur: in cranial view, the base of the facies articularis antitrochanterica forms a slightly cranially protruding ridge (flattened line). The m. obturatorius lateralis scar is directly adjacent to the m. ischiofemoralis extensorius scar (separated by gap). The impressio gastrocnemialis intermedia prominently protrudes caudally from the caudal surface (flat, does not protrude). The base of the epicondylus lateralis forms a distinct angle with the crista tibiofibularis (continuous with crista). The crista supracondylaris medialis is extremely prominent, as formed by the tuberculum muscularis gastrocnemialis medialis (slight prominence, mostly flattened). The base of the fossa poplitea is slightly pneumatic (not pneumatic).

Dynatoaetus pachyosteus can be distinguished from Cryptogyps lacertosus (lectotype or Green Waterhole specimen state is in brackets) by the following features. Humerus: the processus flexorius and condylus ventralis have roughly equal distal extent (processus flexorius much shorter); tuberculum supracondylare dorsale essentially lacking and not prominent of epicondylus dorsalis (large, prominent dorsally); interior margin of the tuberculum supracondylare ventrale is angled across the shaft (parallel to shaft) and facet for ligament insertion on it is broadly ovoid (more elongate); the dorsal sulcus of the m. humerotricipitalis is broad and takes up at least half the shaft width (dorsal sulcus narrow, half the width of the ventral part of the sulcus); fossa m. brachialis moderately deep (substantially deeper). Ulna: the cotyla ventralis is shallow in ventral view (deep, quite concave); the dorsal facies of the shaft is rounded (flattened); the tuberculum ligamenti collateralis ventralis extends distally alongside the impressio brachialis and has a slight but noticeable ventral prominence (positioned proximal to impressio brachialis and completely flat); the fossa on the caudal margin of the cotyla ventralis is positioned in line with the crista intercotylaris and is shallow (ventrally adjacent to crista and deep); the olecranon is less robust (more robust); a muscle insertion is present on the ventral facies distal to the incisura radialis (all insertions on cranial facies); the caudo-dorsal facies is rounded (facies greatly flattened compared to the rest of shaft); the tuberculum carpale has very little cranial projection (prominent cranial projection); the angle of the ventral condylus dorsalis margin cuts across the shaft (angle more closely aligned to the shaft); the distal margin of the condylus dorsalis is level to that of the condylus ventralis (extends distal to the condylus ventralis). Carpometacarpus: the ventral rim of the trochlea carpalis terminates distal to the processus pisiformis and level with the processus alularis (terminates far proximal to processus pisiformis).

Dynatoaetus pachyosteus can be distinguished from Hieraaetus moorei (whose state is in brackets) by the following features: the sizes of the limb bones are significantly smaller than those of H. moorei. Quadrate: there is no foramen basiorbitale (present), the processus orbitalis has a deep sulcus on the medial side (no sulcus), the processus orbitalis has a strongly proximo-medial angle (angled more medially); the condylus pterygoideus strongly projects medially, with almost equal medial extent to the medial condyle (moderate projection with roughly half the extent of medial condyle). Humerus: the proximal margin of the insertion for the m. pectoralis is equivalent to the distal margin of the crista bicipitalis (ends well distal of the crista bicipitalis); the distal end of the crista deltopectoralis is continuous with the shaft (at an angle with the shaft); the dorsal insertion of the sulcus lig. transversus is separated from the rest by a prominent ridge (continuous); the dorsal fossa m. humerotricipitalis is deep (shallow); the facies between the tuberculum supracondylare dorsale and the epicondylus dorsalis is weakly convex (prominently convex); the distance between the ventral margin of the fossa brachialis is narrow, between a quarter to a fifth of the shaft width (extremely narrow less than a fifth of shaft width), the interior margin of the tuberculum supracondylare ventrale is oriented at a high angle across the shaft (parallel to shaft). Ulna: the depressio radialis is distinctly deepened (shallow, indistinct). Carpometacarpus: the os metacarpale minus is arched (flattened), the processus extensorius is oriented at a 140° angle (90° angle). Pelvis: the section of facies directly proximal to the antitrochanter has a deepened fossa present (no fossa in H. moorei), the antitrochanter has greater lateral projection than the crista dorsolateralis (less lateral extent in H. moorei), the processus costales of the vertebra acetabularis merge to form a narrow bridge that connects to the lateral margins (broad bridge in H. moorei). Femur: the tuberculum muscularis gastrocnemialis medialis is prominent (flattened in H. moorei); the proximal margin of the fossa poplitea is distinct from the rest of the cranial facies (continuous, less distinct in H. moorei); the muscular attachment proximal to the fossa poplitea is circular in shape and positioned central in the shaft (elongate oval, slightly offset laterally in H. moorei).

Accipitridae indet.

Referred material

AM F129563, AM F129564, ungual phalanges; AM F152515, a partial carpometacarpus. WAM 15.9.622, a pelvis synsacrum.

Locality, unit and age

Wellington Caves, New South Wales, Australia: AM F129563, AM F129564, AM F152515, no locality or stratigraphy data recorded, Pleistocene unconstrained. Leaena’s Breath Cave, Western Australia, Australia: WAM 15.9.622, found on surface of cave floor, Pleistocene unconstrained.

Description

Two large, robust ungual phalanges with deep, distinct foramina: AM F129563 TL (preserved) = 40.1 mm, height proximal articular surface = 15.5 mm (some breakage), width proximal articular surface = 14.3 mm; AM F129564 TL (preserved) = 36.6 mm, height proximal articular surface = 15.5 mm, width proximal articular surface = 12.3 mm. A shaft + distal part of a carpometacarpus, with surrounding bone and cave soil matrix: distal width = 22.5 mm. WAM 15.9.622 is a large synsacrum found on the surface of the cave floor, with a preserved length of 85.7 mm and preserved width of 30.5 mm.

Remarks

Multiple species of large accipitrids are known from Wellington Caves, including A. audax (distal ulna CCW ID 4879 [field number]), C. lacertosus (distal humeri AM F58092, tarsometatarsus AM F58093), and D. gaffae (distal tibiotarsus AM F106562). The ungual phalanges are from the Old Collections, making their precise site of origin unknown. The phalanges could be D. pachyosteus or D. gaffae based on their size. The carpometacarpus is from a sedimentary layer in Cathedral Cave (layer 13) that has a mean modelled age of 64 ka (Fusco et al. 2023) and could be either A. audax, C. lacertosus or D. pachyosteus based on size. WAM 15.9.622 was labelled as A. audax but could also potentially belong to C. lacertosus, or D. pachyosteus.

Aquila audax

Referred material

SAM P41565, left coracoid; SAM P41518, right femur; SAMA P59836, tarsometatarsus.

Locality, unit and age

Excavation co-ordinates (decimal feet) 67′–68′, R3.5′–16.5′, D/D −1.0′ to −1.5′ (P59836); 65′, R7′, D/D 0.0′/–0.5′ (P41518); excavation data unknown (P41565); Main Fossil Chamber, Victoria Fossil Cave, Naracoorte, South Australia. Early–Middle Pleistocene, 300–150 ka (Arnold et al. 2022).

Remarks

This fossil material is identical to that of living A. audax, based on comparisons to FUR 125 and 085. This material demonstrates that the species coexisted with D. gaffae and D. pachyosteus in South Australia.

Body mass estimation

Using the mass algorithms of Field et al. (2013), Campbell & Marcus (1992), and Campbell & Tonni (1983), body mass was predicted for different fossil bones of Cryptogyps lacertosus, Dynatoaetus gaffae and D. pachyosteus (Table 3). The predicted body mass for D. pachyosteus varied considerably by element. The length of the humerus gave the very low value of 2.9 kg (Table 3), within the size range of a male Aquila audax (see Marchant & Higgins 1993). This seems unlikely based on the other mass estimates and the overall robustness of the bones. Worthy & Holdaway (2002) commented on how humerus length consistently produced low mass predictions compared to that of living raptors. In contrast, the least shaft circumference (LSC) gave a predicted mass of 9 kg, which would be more consistent with a particularly large, robust eagle. The LSC of the femur generated an even greater predicted mass of around 13 kg (Table 3), but it has been noted that predictions derived from the femur of birds sometimes overestimate actual mass (Handley et al. 2016). The coracoid of C. lacertosus generated a predicted mass of 11–12 kg based on the length of the humeral articular facet and shaft width, though total length gave the result of 3 kg. Dynatoaetus gaffae had consistently high results, ranging from 12 kg (tibiotarsus LSC, Table 3) to more than 19 kg (femur LSC, Table 3), though this is partly because there were no complete humeri or coracoids to get total length measurements from.

Table 3. Mass estimates of the tibiotarsus, femur, humerus, and coracoid from Dynatoaetus pachyosteus sp. nov., D. gaffae and Cryptogyps lacertosus. Usage of log10 and ln varied between formulae, see cited sources for details.

Specimen	Element	Measurement (mm)	Regression input (log10 or ln)	Predicted body mass (log10 or ln)	Predicted body mass (g)	Formula
Dynatoaetus gaffae AM F.106562	Distal tibiotarsus	45 (LSC)	1.6	4.1	12,116.8	Campbell and Marcus 1992 tib LSC
		45 (LSC)	1.6	4.1	12,280.0	Campbell and Tonni 1983 tib LSC
Dynatoaetus pachyosteus SAMA P.41513	Femur	55 (LSC)	4.0	9.5	13,461.6	Field et al. 2013 fem LSC
		55 (LSC)	1.7	4.1	13,521.2	Campbell and Marcus 1992 fem lsc
Dynatoaetus gaffae SAMA P.41514	Femur	64 (LSC)	4.2	9.9	19,366.9	Field et al. 2013 fem LSC
		64 (LSC)	1.8	4.3	19,485.0	Campbell and Marcus 1992 fem lsc
Dynatoaetus pachyosteus SAMA P.41517	Humerus	52 (LSC)	3.9	9.1	9023.8	Field et al. 2013 hum LSC
		178.3 (length)	5.2	7.9	2863.4	Field et al. 2013 hum L
Cryptogyps lacertosus SAMA P.42487	Coracoid	20.9 (HAF length)	3	9.4	12,295.5	Field et al. 2013 cor HAF length
		16.1 (SW)	2.8	9.3	11,247.7	Field et al. 2013 cor SW
		73	4.3	8	3037.5	Field et al. 2013 cor L (from lateral process)

Predicted weight is measured in grams (g). Abbreviations: BM, body mass; cor, coracoid; fem, femur; HAF, humeral articular facet; hum, humerus; L, length; LSC, least shaft circumference; SW, shaft width; tib, tibiotarsus.

The large discrepancies between the different mass estimates can be explained by the fact that the algorithms were created using data collected from a wide range of bird species. This is particularly evident in the length measurements, where shorter bone lengths would be generally associated with lower body mass, not taking into account relative stoutness expected to be greater in short-winged flapping taxa compared to gliding taxa. Most likely, the true mass ranges for the fossil species would fall somewhere between the proposed estimates. Based on similarly sized species, D. gaffae could have reached up to 12 kg in mass, C. lacertosus would probably have a similar range to A. audax of 3–6 kg, while D. pachyosteus would have been at least several kilograms lighter than D. gaffae.

Phylogenetic analyses

Analysis 1: parsimony, morphology + DNA, ordered characters

Using the combined molecular and morphological data, six MPTs were found, each with a tree length of 1857 steps, and the strict consensus tree is in Fig. 8. Monophyly of the Accipitridae was strongly supported (bootstrap support 86%), and only five nodes had support less than 50%.

Fig. 8. Strict consensus tree from parsimony analysis of molecular and morphological data for accipitrid raptors (and outgroups). The six most-parsimonious trees had tree length = 1857, CI = 0.2143, HI = 0.7857, RI = 0.5767. Bootstrap values are given at each node.

Dynatoaetus pachyosteus grouped with D. gaffae and aegypiines (these three taxa forming a polytomy), while Cryptogyps lacertosus was strongly supported as being basal to all extant Aegypiinae (83%). The Aegypiinae–Circaetinae clade as a whole had weak support (45.2%) with moderate support for a sister relationship to the Aquilinae–Harpiinae–Accipitrinae–Buteoninae–Haliaeetinae clade (64%).

The living Aegypiinae plus C. lacertosus form a clade sister to that of D. gaffae + D. pachyosteus supported by three unambiguous (optimization-independent) characters: character 143 state 0 (CI 0.29) (humerus, processus flexorius does not project distal to the condylus ventralis), 154 state 0 (0.14) (humerus, interior margin of tuberculum supracondylare ventrale oriented parallel relative to shaft margin), and 168 state 1 (distal ulna, tuberculum carpale has prominent protrusion). However, all three characters have high homoplasy.

The Dynatoaetus clade is supported by two unambiguous characters: character 120 state 1 (proximal humerus, the intumescentia humeri has a flattened or slightly concave surface) and character 205 state 1 (proximal femur, the proximal projection of the crista trochanteris from the facies articularis antitrochanterica is low). Both characters have high homoplasy.

Analysis 2: Bayesian inference, morphology + DNA, ordered

The Bayesian analysis produced a tree with very similar topology to that in the parsimony analysis using the same data, with support values for nodes very strong in all but a few cases (Fig. 9).

Fig. 9. Majority-rule consensus tree for Bayesian analysis of combined molecular and morphological data (ordered, unlinked branches) for accipitrid raptors (and outgroups). Posterior probabilities (expressed as percentages) shown at nodes. Fossil taxa are coloured red.

The Circaetinae–Aegypiinae had weak support as a monophyletic clade (posterior probability = 0.69). The Circaetinae clade had strong support (PP = 0.89) as did the Aegypiinae clade (PP = 1). All the clades within Aegypiinae had PP >0.9. Cryptogyps lacertosus had weak support for being basal to all extant vultures PP = 0.69. Dynatoaetus pachyosteus formed a clade with D. gaffae that had weak support (PP = 0.71) and was basal to Cryptogyps and extant Aegypiinae.

Conclusions

Both phylogenetic analyses resolved Dynatoaetus pachyosteus as part of a clade with D. gaffae, with the only difference being the level of support. Both analyses also recovered Cryptogyps as basal to extant Aegypiinae, and the Dynatoaetus clade as basal to the Cryptogyps–Aegypiinae clade. Parsimony morphology + DNA analyses provided weaker conclusions (in terms of both topological comparisons to accepted relationships, and support values), while the corresponding Bayesian inference analysis provided stronger results.

Discussion

The surveyed fossils could be separated into two taxa based on differences in size and morphology (see Fig. 10 for artistic reconstruction). The first, described as Dynatoaetus pachyosteus, was present only in deposits from Victoria Fossil Cave (Naracoorte, South Australia). This species had a similar wingspan to Aquila audax but was more robust (Table 2, 3), and phylogenetically resolved as a member of the Circaetinae–Aegypiinae clade and a close relative of D. gaffae and aegypiines. The second taxon represented is Cryptogyps lacertosus, and is more abundant and widespread, now being recorded from the Lake Eyre Basin, the Wellington Caves and Walli Caves (Wellington, New South Wales), Green Waterhole Cave (Tantanoola, South Australia), and Leaena’s Breath Cave (Nullarbor Plain, Western Australia).

Fig. 10. A reconstruction of a flock of Cryptogyps lacertosus (left side) and several individuals of Dynatoaetus pachyosteus (right side) feeding on a carcass of Diprotodon optatum in the Late Pleistocene Naracoorte landscape. Artwork by John Barrie.

Phylogenetic relationships

The diversity and phylogenetic relationships of Australian fossil accipitrids have been little studied, likely because of their incompleteness. Here we have described relatively fragmentary fossils and derived phylogenetic hypotheses that were constrained by molecular data for living species. Dynatoaetus pachyosteus was consistently resolved as closely related to the extinct D. gaffae, then to Cryptogyps lacertosus, and aegypiines in all phylogenetic analyses.

The Dynatoaetus genus does not have any specific autapomorphies, though it is notable in the robustness of the limb bones. For Cryptogyps, a distinct autapomorphy is the greatly flattened dorsal facies of the ulna shaft.

The phylogenetic proximity of D. pachyosteus to D. gaffae strongly supports our contention they are congeneric. This could indicate that prior to the Pleistocene mass extinction, Australia had an endemic diversity of an Aegypiinae–Circaetinae lineage, similar to the extant endemic pernine kites Hamirostra melanosternon and Lophoictinia isura. It is surprising that D. pachyosteus and D. gaffae align more closely to aegypiine vultures than circaetines, as their pelvic limb morphology is more akin to active-hunting eagles rather than scavengers. The two subfamilies are thought to have diverged 20–16 Mya, with the initial diversification of modern Circaetinae occurring 18–14 Mya compared to 10–8 Mya for Aegypiinae (see Nagy & Tökölyi 2014, Mindell et al. 2018). It may be that the ancestors of the Dynatoaetus clade diverged from the aegypiine stem lineage before this point, and so retained more predatory features. Alternatively, D. gaffae and D. pachyosteus may be circaetines that have some superficial similarities to aegypiines that other extant circaetines lack.

Diversity of Pleistocene Australian Accipitridae

Dynatoaetus pachyosteus increases the diversity of known Pleistocene Accipitridae in Australia. Added to Cryptogyps lacertosus, D. gaffae and Aquila audax, this brings the number of large inland Pleistocene accipitrid species up to at least four. The fossil taxon Necrastur alacer is rather smaller, nor does it have a specific age assigned to its locality (Worthy & Nguyen 2020) and could date to either the Pliocene or the Pleistocene. Assuming all living Australian accipitrids were present during the Pleistocene, this brings the total diversity for that time up to at least 20 species and six to seven subfamilies.

More significantly, the presence of three additional large accipitrids during the Pleistocene in inland Australia solves the mystery of the scarcity of large raptors on the continent. In contemporary Australia, the only large inland accipitrid is A. audax, which both hunts and scavenges. The largest females of this species can potentially weigh just over 5 kg, though the average is 4.2, while males can reach up to 4 kg but on average weigh 3.1 (Marchant & Higgins 1993). Having only one large predatory accipitrid present in a continent the size of Australia is highly unusual, more especially so considering that until the late Pleistocene there was a diverse range of mammals that would have provided suitable food resources. In Africa, for example, there are now at least 10 predatory eagles of a similar size to or larger than A. audax, and 12 species of accipitrid vultures from both subfamilies (Dickinson & Remsen 2013). Africa is substantially larger than Australia, however, so a better comparison might be the accipitrid diversity present in the similarly sized region of North America south of the arctic and sub-arctic zones. There are two species of large eagles commonly found in this region today: the Bald Eagle Haliaeetus leucocephalus and the Golden Eagle Aquila chrysaetos across the United States and Canada, with the typically Eurasian White-tailed Eagle Haliaeetus albicilla and Steller’s Sea Eagle Haliaeetus pelagicus (Pallas, 1811) occasionally sighted in northern regions such as Alaska (Clark & Wheeler 1983). An additional two large species, the Harpy Eagle Harpia harpyja (Linnaeus, 1758) and the Solitary Eagle Buteogallus solitarius (von Tschudi, 1845) are also present in North America but are restricted to the tropics in Mexico. However, North America is also inhabited by five species of scavenging New World cathartid vultures: the Turkey Vulture Cathartes aura (Linnaeus, 1758), the Black Vulture Coragyps atratus, and the California Condor Gymnogyps californianus (Shaw, 1797) in the United States, and the Lesser Yellow-headed Vulture Cathartes burrovianus Cassin, 1845 and King Vulture Sarcoramphus papa (Linnaeus, 1758) from Mexico. The fossil record also reveals that the North American accipitrid fauna was much more diverse up until the Late Pleistocene, with an additional four large eagles, one New World vulture, two Old World vultures, and two teratorns known to have coexisted with the extant species at sites such as Rancho la Brea (Jefferson 1991). This is an extremely diverse assemblage of large predatory and scavenging birds compared to what is known from Pleistocene Australia. Therefore, the presence of three additional large Australian accipitrids, especially ones that appear to have been capable of hunting large prey or had an obligate scavenging ecology, brings the family’s diversity closer to what might be expected from a continent of Australia’s size.

Palaeoecology

The morphology of Dynatoaetus pachyosteus is notably robust, with short robust wing bones and relatively larger robust pelvis and leg bones. The pelvis is especially deep and robust in the anterior ilial region, capable of housing large muscles, and the femora are very large, up to 14% longer and 30% wider in the shaft on average than for Aquila audax (see Table 2). Having such a relatively robust pelvis and leg bones, combined with a tarsometatarsal morphology suggesting the presence of robust talons, is convergent on the morphology of Hieraaetus moorei and indicates that this bird was capable of predating on large prey. The wing bones are short but robust, suggesting a predominantly flapping flight behaviour again similar to that envisaged for H. moorei (Holdaway 1991, Worthy & Holdaway 2002).

That the legs were so large and robust compared to the wings strongly indicates that D. pachyosteus was an active predator which likely targeted mammalian prey. To estimate the size range of the potential prey of D. pachyosteus, observations of other large accipitrids known to attack large animals can be compared. Hieraaetus moorei is an example of an accipitrid specialized in hunting large prey. It is estimated to have had a wingspan of 2–3 m and a mass of 10–15 kg, making it the largest accipitrid known to have ever existed (Holdaway 1991, Brathwaite 1992, Worthy & Holdaway 2002), although the giant Caribbean Gigantohierax suarezi Arredondo & Arredondo, 2002 rivals it in size (Suarez 2020). Despite its great size, the wings of H. moorei were comparatively short, which is thought to be an adaptation to flying through the forested landscape of Holocene New Zealand (Brathwaite 1992). Fossil evidence indicates that it regularly preyed upon species of the moa genera Dinornis Owen, 1843, Emeus Reichenbach, 1852, Euryapteryx Haast, 1874b and Pachyornis Lydekker, 1891, with fossil skeletons of individuals estimated to be up to 200 kg in weight bearing the marks of attacks (Worthy & Holdaway 2002, Bunce et al. 2005). Among living raptors, predation upon animals much larger than the hunter is not unheard of. Stephanoaetus coronatus is a living eagle native to Africa, documented as weighing up to 4.7 kg in females (Ferguson-Lees & Christie 2001). It has a notably powerful build compared to other eagles of a similar size, and has been documented hunting small antelopes, including the 4.5–5.9 kg suni (Nesotragus moschatus von Dueben, 1846) (Lawson 1986), and has been recorded killing a Harvey’s Duiker (Cephalophus harveyi Thomas, 1894) that was roughly two-thirds the size of an adult (see Brown & Amadon 1968). The average adult Harvey’s Duiker weighs 15 kg, so this juvenile was likely at least twice the weight of the attacking eagle. The diet of A. audax is today typically dominated by (introduced) rabbits, but it has also been documented attacking juvenile and small or weakened adult eastern grey kangaroos (Macropus giganteus Shaw, 1790), either in pairs or groups (Marchant & Higgins 1993, Fuentes & Olsen 2015). In these attacks, the eagles targeted the kangaroo’s head in repeated strikes until it either escaped or collapsed from injury and exhaustion (Fuentes & Olsen 2015). The average adult A. audax weighs approximately 3.5 kg, excluding size differences between males and females (Marchant & Higgins 1993), while the average adult M. giganteus weighs 31 kg in females and 56 kg in males (Pearse 1981).

Based on its large, powerful build and the comparative observations above, it is entirely plausible for D. pachyosteus to have hunted the juveniles and weakened adults of extinct species of megafauna such as in the giant flightless bird Genyornis newtoni Stirling and Zietz, 1896, or species of giant kangaroos such as those in Protemnodon, Sthenurus Owen, 1873a and Procoptodon Owen, 1873b. These animals may have been slower and less agile compared to the smaller kangaroos and ratites that survived the Pleistocene extinction and might have been easier for the heavily built D. pachyosteus to pursue and attack. This is broadly similar to the likely ecology of D. gaffae, but there may have been some niche separation of preferred prey and/or habitat between these species. Besides hunting, D. pachyosteus would have probably supplemented its diet through scavenging from carcasses like most living eagles do today, though it would have been competing with the specialist scavenger species like C. lacertosus.

The new phylogenetic interpretation of C. lacertosus shows that it was the sister taxon to all living aegypiine vultures, rather than being sister to the Gyps genus (Mather et al. 2022). Notably, support for the relationship increased in comparison to the Mather et al. 2022 phylogeny. This shows that the referral of the Green Waterhole/Fossil Cave partial skeleton to C. lacertosus, despite only one overlapping element with previous material, enhanced the support for aegypiine affinity of this taxon. However, unlike most aegypiine vultures, it appears to have lacked the same degree of pneumatism in the ulna and carpometacarpus present in most of the larger species. As greater pneumaticity in these bones is an adaptation for long distance soaring flight (Hertel 1946), this suggests that C. lacertosus was less suited for such behaviour.

Implications and interpretations

One aspect often noted about Pleistocene Australia is the paucity of large, terrestrial, mammalian predators; as of writing, the only species thought to have been capable of killing large prey animals is Thylacoleo carnifex, with evidence suggesting they were social animals that may have hunted in groups (see Arman & Prideaux 2016). While species of Thylacinus Temminck, 1824 and Sarcophilus Cuvier, 1837 were also present on mainland Australia at this time, it is thought that they would only have been capable of killing smaller prey (Wroe et al. 2007, Rovinsky et al. 2020). The giant varanid Varanus priscus (Owen, 1859b) has been suggested to have hunted large mammals in a similar manner to the Komodo dragon V. komodoensis Ouwens, 1912 (Fry et al. 2009), though others theorize that it mostly subsisted from scavenging (Wroe 2002). The low diversity of mammalian carnivores has led to suggestions that the Pleistocene carnivore guild of Australia was dominated by the large reptiles; the snake Wonambi naracoortensis Smith, 1976, the crocodilians Quinkana fortirostrum Molnar, 1981, Gunggamarandu maunala Ristevski et al., 2021 and Paludirex gracilis (Willis and Molnar, 1997), the giant varanid lizard Varanus priscus and the still living V. komodoensis (see Hocknull et al. 2009) inhabited Australia at the same time as the extant Crocodylus porosus Schneider, 1801 and C. johnsoni Krefft, 1873. However, most of these species are strongly associated with water, which would limit their range to riverine, coastal and wetland environments. In addition to this, the theory of reptilian predators playing a significant role in Australian ecosystems during the Pleistocene was disputed by Wroe (2002). The role of avian predators during this time period has rarely been discussed.

The presence of vultures (C. lacertosus) and multiple large eagles (D. gaffae, D. pachyosteus, A. audax) during the Pleistocene has very significant implications for our understanding of the dynamics of the Australian fauna at the time. Vultures play significant roles in scavenging guilds where they are present, being vital in the consumption of carcasses, potentially reducing the spread of harmful pathogens (Ogada et al. 2012a, 2012b), aiding in the location of food by other predators (Kane & Kendall 2017), and facilitating energy flow through food webs (Wilson & Wolkovich 2011). Cryptogyps lacertosus most likely performed these same roles. The fact that three species of large eagles—D. gaffae, D. pachyosteus and A. audax—coexisted with each other further suggests that some degree of specialization into separate niches must have been present among these species. Both species of Dynatoaetus were active hunters, as evidenced by their large and robust leg bones unlike any vulture, and it is quite likely that their presence restricted A. audax to a narrower niche than the one it occupies in the present.

Interactions between fossil species

All known extinct Pleistocene accipitrids seem to have had overlapping ranges. Cryptogyps lacertosus is known from Leaena’s Breath Cave in the Nullarbor, WA, the Warburton River, SA, Green Waterhole in Tantanoola, SA and the Wellington Caves, NSW. Dynatoaetus gaffae is known from Mairs Cave in SA, Victoria Fossil Cave in Naracoorte, SA, Coopers Creek in SA, and Wellington Caves in NSW. Dynatoaetus pachyosteus is currently only known from Victoria Fossil Cave in Naracoorte. This overlap is not surprising; Aquila audax juveniles are known to disperse up to 800 km from their parents’ home range (Debus 1998), and it is quite likely that these extinct accipitrids would have also been wide-ranging dispersers. The presence of three (four including A. audax) large accipitrids coexisting in the same environments indicates that there was likely some form of niche partitioning, either in the form of preferred diet, preferred habitat, or both. Dynatoaetus pachyosteus and D. gaffae were active hunters rather than obligate scavengers, but they likely also played a role as scavengers alongside C. lacertosus. Most accipitrids will scavenge carrion, and it has been documented that some species of vultures often use eagles as a means of finding carcasses (Kane et al. 2014). Along with A. audax, D. pachyosteus and D. gaffae may have been used as a visual indicator for the presence of a carcass based on observations of their flight behaviour by scavenging species.

Role in the Australian Pleistocene

During the late Pleistocene, Cryptogyps lacertosus would have been one of the primary scavenger species on the continent until its extinction. The scavenging guild of the time would also have included the other large raptors, and facultative mammalian scavengers such as species of Thylacinus and Sarcophilus, which subsequently went extinct on the mainland roughly three thousand years ago (White et al. 2018). Larger mammalian predators like Thylacoleo carnifex, and large reptiles like the giant monitor lizard Varanus priscus possibly also played a role before their extinction in the late Pleistocene. In the present day, most scavenging across Australia is carried out primarily by avian species such as Aquila audax and the raven Corvus coronoides Vigors & Horsfield, 1827, and by the more recently introduced mammals such as wild dogs, feral cats and foxes (Read & Wilson 2004, Forsyth et al. 2014).

While the environmental impacts of the extinction of Australian vultures is difficult to determine over 50–40 ka later, it is possible to make inferences based on observations of current vulture declines across the globe. In India, the drastic decline in vulture populations means that carcasses are more frequently scavenged by feral dogs, resulting in their population booming and an increase in the transmission of diseases such as rabies (Markandya et al. 2008). This phenomenon is also observable in other places where vultures have declined or been extirpated, with facultative scavenging mammals arriving in greater numbers to feed on carrion and potentially increasing transmission of pathogens (Ogada et al. 2012b). The ecological release of less specialized avian scavengers can also occur following vulture decline, as the removal of apex scavengers has been documented to result in the less efficient ‘mesoscavengers’ increasing in their abundance (O’Bryan et al. 2019). In the case of Australia, this is likely what has allowed A. audax to become such a widespread generalist across the continent. In some cases, it seems that facultative scavengers are unable to replace vultures, leaving the carrion consumption role primarily to decomposers instead (Hill et al. 2018). The loss of these species from the Australian ecosystem in the late Pleistocene would have therefore had significant effects on the structure of the terrestrial Australian ecosystems.

Acknowledgements

We thank Phillipa Horton, Maya Penck and Mary-Anne Binnie (SAMA), Jacqueline Nguyen and Matthew McCurry (AM), Judith White and Joanne Cooper (NHMUK), Chris Milensky (USNM), Mark Robbins (KU), Tim Ziegler and Karen Roberts (NMV), Mikael Siversson (WAM), and Leo Joseph and Alex Drew (ANWC) for access to collections and specimens. Elizabeth Scharsachs and family provided accommodation in Tring to EM. We also thank Rod Wells (FU) for copies of original maps and locality information. Peter Horne, Dave Albano, Neville Skinner, Mark Neilsen, Ian Lewis, Dave Fielder, Andrea Gordon, and Katrin and Gerret Springer assisted with Green Waterhole Cave expeditions. Jacqueline Nguyen (AM) transported Wellington Caves material in 2020. The South Australian Department of Environment and Heritage facilitated permits for Green Waterhole Cave excavations in 2006 (C25141-1), 2007 (C25141-2) and 2008 (C25151-2).

Disclosure statement

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

Additional information

Funding

The Birdlife Australia Raptor Group funded the artistic reconstruction seen in Fig. 10 via an annual grant (2022). Two anonymous reviewers and the Alcheringa Editorial Board contributed helpful comments and feedback during the review process.