The remains of a unique fossil bony fish were discovered in late Eocene (39.1–36.7 Ma: NZ Kaiatan stage) mudstone at Burnside near Dunedin, New Zealand in the 1930s and subsequently named and described by Frederick Chapman. He interpreted the type specimen as being a large-scaled relative of the modern Thyrsites of the Gempylidae (Scombroidei: Trichiuroidea), known to be swift, large oceanic predators. However, Chapman is unlikely to have seen all of the fossil, and did not discuss the skull and caudal skeleton. Additional material now allows these to be included in the expanded description herein, including key morphologic features of the fish such as the presence of premaxillary fangs. This study describes the rather complex history of the specimen and re-examines this significant fossil fish in the University of Otago collections, giving a more complete understanding of Eothyrsites morphology, paleoecology and relationships. In summary, we suspect Eothyrsites represents an ancestral form of gempylid, closely related to the gemfish group, an important Southern Hemisphere macrofossil record from the Eocene seas around Zealandia.

Introduction

English born and later Melbourne-based Frederick Chapman (13 February 1864–10 December 1943) was the inaugural Australian Commonwealth Palaeontologist, known especially for his work on Foraminifera, and his books on Australasian fossils. In 1902 Chapman was appointed by the Victorian government to the newly created position of palaeontologist to the National Museum, Melbourne. While at the museum in 1914 Chapman was sent fish remains from New Zealand, his research later published in 1918 (as NZGS Paleontological Bulletin 7). Chapman’s descriptions covered 43 species of fish, 37 of which are chondrichthyian. His teleosts included: scales of an elopid identified as Thrissopater, from the Cretaceous of North Canterbury; Scombroclupea, a supposed clupeid from the Weka Pass area, Canterbury (presumably Oligocene); ‘Diplomystus’ from the Cretaceous of Marlborough; teeth of a sea-bream Sargus, from the Oligocene of Canterbury; and distinctive pharyngeal toothplates from two species of the wrasse Labrodon, found in mid-Cenozoic sediments in Canterbury (Chapman 1918).

From 1932 to 1943 Chapman was awarded for his efforts, being made an Honorary Fellow of the Royal Society of New Zealand (Benson 1944).

In 1933 Chapman was visited in Melbourne by William Noel Benson, University of Otago geologist, who provided Chapman the opportunity to study two fossil fish brought over from Otago collections (see note by Chapman 1934). The first fish was from Abbotsford, Dunedin, an impressive Early Eocene (Heretaungan NZ stage) skull that Chapman assigned to the now abandoned Cretaceous genus ‘Portheus’; Errol White (BMNH) suggested to Benson in 1945 that this specimen may be closer to the scombrid genus Scombrinus sp. (as noted in King et al. 2009 after Rust 2000).

The second fossil fish was also discovered near Dunedin, a partial skeleton from Burnside.

After first writing a popular newspaper article based on the specimen, which he named Eothyrsites or ‘dawn barracoota’ (see Chapman 1934), Chapman (1935) formally described this specimen based on three mudstone slabs containing some skull material, pectoral fins and trunk. He interpreted the specimen as a relative of Thyrsites atun (the snake mackerel or snoek, sometimes called ‘New Zealand barracouta’) of the family Gempylidae (Nakamura and Parin 1993; Froese and Pauly 2022). This and the other 10 species of modern gemfishes found in New Zealand waters are swift, large oceanic predators, often characterised by elongate bodies and fang-like teeth (Stewart 2015).

Soon Chapman was working on other fossils, publishing on Australian cidaroid echinoderms in 1934. He had retired in 1935 when the article on the two New Zealand fossil fish was published by the Royal Society. In 1936 he returned to the National Museum of Australia as honorary palaeontologist. According to his assistant, micropaleontologist Irene Crespin (1979), he was well liked: ‘outstanding among Chapman’s qualities were his quiet gentlemanly manner, his patience and his willingness to pass on his experience and guidance to all-comers’. Chapman was a prolific writer, publishing five books and some 500 scientific papers, some in collaboration, on geology, palaeontology and zoology. He died at Kew, Australia in 1943.

In the 1990s further material was located in the Otago University collection which appears to belong to Eothyrsites – additional fragments which contain more of the skull and the caudal fin. These have undergone examination and preparation (Rust 2000). It seems unlikely these pieces were ever examined by Chapman.

However, a series of letters between Benson and several paleontologists/geologists concerning a second fossil fish specimen from Burnside, dated between 1955 and 1956, are held in archives of the Hocken Library, Dunedin. Duplicates of this correspondence and a single additional letter from A. W. B. Powell (conchologist at Auckland Museum) to Benson about the fish are held by Auckland Museum.

Briefly summarising these letters: Benson contacted Powell asking if he would delay in studying the ‘new’ fossil fish material from Burnside that had been received by Professor B. J. Marples for the Otago University Museum, then packed in a box and sent to Auckland Museum (Benson 1955). Powell was overseas at the time and Auckland Museum director Gilbert Archey replied to Benson that the material had not been located (Archey 1956). Upon his return Powell wrote that the specimen had not been unpacked and would be held until further instruction was received from Benson (Powell 1956). Benson also wrote to fossil fish specialist Errol White at the BMNH, who had been an earlier acquaintance (Benson 1956), asking for assistance in revising Chapman’s description, and mentioned the material (still in Auckland) requiring further preparation. White (1956) later replied declining to study the fish himself but suggested that Benson contact Belgian paleoichthyologist Edgard Casier. However Benson was busy completing his geological map of the Dunedin area and passed away not long after, in 1957. At this stage we cannot say if Casier was ever contacted about the fossil, or when the material was returned, probably by Powell, to the University of Otago collection.

In any case, Benson felt strongly that the original description of this unique New Zealand fossil genus and species warranted further examination and expansion. Even today the fossil record of marine teleost material from the New Zealand Cenozoic consists mostly of isolated bones or incomplete specimens (see review in King et al. 2009), so this relatively complete example is especially noteworthy.

Materials and methods

According to Chapman (1935) the type series consisted of three syntypes – associated pieces of mudstone matrix containing semi-articulated fish remains; although the skeleton is crushed, elements are generally well preserved with only minimal distortion. His original description included OU 6329 a, a piece with vertebrae and scales (Chapman’s Figs 2 and 5), together with OU 6329 b, a small piece with numerous scales (a detached part of the above); OU 6332, a piece containing pectoral fin (Chapman’s Fig. 4); and OU 6854 d (once labelled 5982), a piece containing mandible (Chapman’s Fig. 3).

In addition to these we note pieces OU 6854 a, b, c plus three smaller fragmentary pieces, as yet unnumbered. This is thought to be the material sent away and discussed by Benson in his letters (1955, 1956), but all originating from the same locality and horizon as Chapman’s specimen, the Burnside Mudstone, a unit from which macrofossils are rare.

It seems likely the material was collected at two different times, possibly decades apart. Chapman described part of the material in 1935. One of Benson’s letters to Powell (in 1955) stated: ‘ … in the Otago University museum a recently received fossil fish, from the Upper Eocene Burnside mudstone, which I fancied might be either Eothyrisites holosquamatus, Chapman or something related to it’. Benson clearly believed the ‘recently received’ material was a separate specimen.

The possibility exists that this assemblage constitutes parts of two separate individuals, however it seems more likely that they represent counterparts of a single skeleton, and they are treated as such herein. The remains all exhibit similar preservation, are all parts of a fish skeleton of consistent size, and there are no duplicated elements. Many pieces bear equivalent characteristic scales (hence Chapman’s species name holosquamatus), strongly suggesting all the fragments are from the same original specimen, separated in space and time during recovery and subsequent storage.

Taphonomic processes post mortem but before burial could explain aspects of the preservation of Eothyrsites, in particular the partial disintegration of the skeleton. In quiet oceanic depositional settings some fish carcasses have been shown to refloat, or partially refloat, in the days after death due to gases developing in the abdominal cavity. Often the head and tail may droop or may even detach before the carcass later re-sinks to the seafloor as a jumbled heap (Schäfer 1972). Hence we find in this specimen, the skeleton of Eothyrsites with overlapping elements, semi-articulated sections of trunk, a skull with the lower jaw detached from the neurocranium and upper jaw, and partially isolated fins and scales.

The fossil material is housed in the paleontology collection in the Department of Geology, University of Otago, Dunedin, New Zealand (OU). NMNZ refers to material held at the Museum of New Zealand Te Papa Tongarewa; BMNH refers to the British Museum of Natural History, now Natural History Museum (NHMUK).

Some of the additional pieces recovered required additional matrix removal before examination, to allow investigations of their skeletal structures in as much detail as possible. They were prepared using thin needles and studied using a stereomicroscope. Measurements were taken using callipers to the nearest 0.1 mm. Comparative information was derived mainly from the literature, although images of modern osteological material of Thyrsites atun (NMNZ P.002334, 002659) and Rexea solandri (NMNZ P.002339) were provided by A. Stewart from the national fish collection at Te Papa, Wellington.

Locality

The initial fossil material was collected, presumably by quarry workers during 1929–33, at the Burnside Marl Pit, formerly the quarry of the Milburn Lime and Cement Co. Ltd., also known as the Burnside Cementworks, Dunedin, New Zealand: 45.8961°S, 170.4423°E ().

Figure 1. Map of the fossil locality at Burnside, Dunedin, South Island, New Zealand. Map produced with S. Read (OU Geology).

The marl or Burnside Mudstone (Benson 1968) is late Eocene in age (39.1–36.7 Ma: NZ Kaiatan stage, equivalent to Bartonian –Priabonian age). Details of the locality are recorded in the New Zealand Fossil Record File database (www.fred.org.nz), with the registration number I44/f0176; the map reference is NZMS 260 I44 115 757.

Geological setting and paleoenvironment

The regional distribution and lithology of the Burnside Mudstone has been discussed by Benson (1968) and Bishop and Turnbull (1996). The formation reaches up to 45 metres thick at its type locality from which the fossil fish was recovered (), being a fine-grained, massive mudstone, largely light grey in colour, soft semi-plastic, calcareous (up to 20% carbonate), with only occasional macrofossils. These include thin-shelled bivalves, a sea-pen (Hamilton 1958), and rare penguin bones (OU collections). Microfossil analyses suggest continuous deposition of the sediment probably occurred in quiet deeper waters (perhaps 200 m depth) of the outer shelf (see note by Hornibrook in Hamilton 1958). This is consistent with the preservational taphonomy of the fish skeleton discussed above; in addition there is little evidence of physical disturbance or scavenging of the remains before burial in the muddy sediment.

Figure 2. Stratigraphic column of locality. Boundaries are based on McMillan and Wilson (1997).

Systematic paleontology

Infraclass teleostei mueller 1845

Series Pelagiaria Betancur-R et al. 2017.

Order Scombriformes Betancur-R et al. 2017.

Suborder Scombroidei (sensu Carpenter et al. 1995)

Trichiuroidea (sensu Nakamura & Parin 1993)

Family Gempylidae Gill, 1862.

Genus Eothyrsites Chapman 1935.

Type and only species: Eothyrsites holosquamatus Chapman 1935.

Diagnosis

Danilchenko (1960, 1967) listed some key morphological features of fossil members of the family Gempylidae applicable here, notably the presence of an elongate, laterally compressed body, large head with lower jaw protruding and jaws bearing anterior large canine teeth (fangs). Gill cover smooth, without spines. Vertebrae count typically 30–55, with ribs on parapophyses. Anterior dorsal fin long, composed of comparatively thin hard rays, united with the short second (soft) dorsal fin. Anal fin opposite the second dorsal. Accessory finlets usually present. Pelvic fins below or slightly behind the pectorals, sometimes reduced to a pair of spines. Caudal fin moderately developed, deeply forked; caudal peduncle slender. However, advances in the study of molecular and morphologic (particularly skull) characters suggest the Gempylidae is actually paraphyletic and can better be termed a grade within the Trichiuroidea (summarised in Beckett et al. 2018, see Discussion section).

Type material: Consists of seven labelled syntypes (from 10 associated pieces of mudstone matrix containing semi-articulated fish remains from what we consider a single individual) currently catalogued as OU 6329, OU 6332 a, OU 6332 b; plus OU 6854 a, b, c and d (see ).

Figure 3. Eothyrsites holosquamatus Chapman, composite photograph of syntypes and associated material (OU 6329 a, b, OU 6332, OU 6854 a, b, c, d). Key to image: A, partial skull, OU 6854 c; B, piece containing lower jaw and opercular elements, part of tooth-bearing mandible labelled by Chapman, OU 6854 d; C, reverse side of OU 6854 d with preopercular labelled; D, close-up of maxillary teeth, OU 6854 a; E, abdominal region including vertebrae, rays, scales and small anal? fin, OU 6329 a; F, pectoral fin, OU 6854 a; G, partial pectoral fin labelled by Chapman, OU 6332; H, partial? fin, reverse side of OU 6332; I, caudal endoskeleton, OU 6854 b; J, trunk portion with scales labelled by Chapman, OU 6329 b. Note B, C, E, G, H and J were examined by Chapman (1935). All scale bars 20 mm unless indicated otherwise.

The original three syntypes described by Chapman (1935) were OU 6329 a + b, OU 6332 and OU 6854 d. Note that the OU 6854 d was at some stage marked 5982, and catalogued with reference to the scales.

Description of type specimen

Overall body shape: From the available material we suspect the type of Eothyrsites holosquamatus originally had a total body length in the order of 700 mm. The body was probably somewhat elongate, laterally compressed and slender, as for gemfish (Roberts and Stewart 1997), but perhaps not to the degree seen in most other modern gempylids (e.g. Matsubara and Iwai 1952; Nakamura and Parin 1993; Stewart 2015).

Skull: Mudstone slab OU 6854 c (A, ) contains much of the cranium and upper jaw, while the lower jaw is detached and fragmented, exposed along with opercular elements on piece OU 6854 d (B). We estimate the large skull of Eothyrsites was elongate and narrow, measuring approximately 190 mm in length anteroposteriorly (the posterior portion of the cranium, beyond the lateral commissure at the posterior margin of the orbit, including exoccipitals and basioccipital, is missing in the specimen), and 85 mm across transversely in the orbital region. This is however slightly larger and broader than the modern skull material of an adult Thyrsites atun (NMNZ P.002334, 002659) and Rexea solandri (NMNZ P.002339), as well as the early gempylid crania described in Beckett et al. (2018). Although incomplete, an estimated skull length of approximately 30% body length is similar to modern gemfish such as Rexea spp. (e.g. Roberts and Stewart 1997).

Figure 4. Eothyrsites holosquamatus Chapman, photograph of skull OU 6854 c, ventral view.

When viewing the skull as exposed from the underside (), a long and narrow (approximately 100 × 20 mm) parasphenoid can be seen centrally. This compares well with those seen in modern skulls of Thyrsites and Rexea, and an English fossil specimen of a gempylid NHMUK PV OR 41318, shown in Figs 7 and 8 in Beckett et al. (2018). The parasphenoid, which meets the basisphenoid posteriorly, extends anteriorly to the posterior margin of the vomer and meets this near the lateral ethmoids. The anterior tip of the vomer is missing and unfortunately it is not possible to determine if teeth were present, an important morphologic character (see Beckett et al. 2018). The skull roof (comprised largely of the paired frontal bones), the occipital crest and other features of the dorsal surface of the Eothyrsites skull are obscured by matrix. Towards the posterior of the cranium it is difficult to determine the sutures and shape of elements in the otic and occipital regions, including the pterotic and prootic, due to poor preservation.

Figure 5. Eothyrsites holosquamatus Chapman, line drawing of skull OU 6854 c, ventral view. Abbreviations: af, anterior premaxillary fang; boc, basioccipital; eth, ethmoid; fr, frontal; mx, maxilla; os, orbitosphenoid; pas, parasphenoid; pl, palatine; pmx, premaxilla; pro, prootic; pto, pterotic; spo, sphenotic; su, supraorbital, vo, vomer.

Eothyrsites probably had relatively large eyes. An estimated orbital length (horizontal across sclerotic) is 50 mm, which would constitute about 7% total body length of 700 mm, consistent within the range of modern gemfish such as Rexea (Roberts and Stewart 1997); or 26% skull length, slightly larger than those of modern Thyrsites atun (Stewart 2015), but similar to other fossil gempylids (Danilchenko 1960). The orbits were just posterior of the midlength of the head and behind the mouth, with the dorsal orbital margin probably coincident with the dorsal profile of the skull as for modern gemfish. Part of the curved circumorbital series (jugal + supraorbital?) or possibly a partial, slightly crushed sclerotic ring can be seen in the right orbital region of the skull (). While Carnvale (2006) noted a report of the presence of ossified sclerotics in the gempylids Gempylus and Nesiarchus, he also pointed out that these ossifications are absent from other gempylid genera and the trichiurids. Sclerotic bones are apparently absent in the extinct genus Hemithyrsites (Danilchenko 1960).

Jaw elements: The mouth of Eothyrsites was large, with the upper jaws strengthened by the premaxilla being fixed and hence non-protrusible, typical of gempylids and other members of the Scombroidei. In the key study of scombroid phylogeny, Johnson (1986, p. 21) also noted: ‘an inner row of two or three enlarged, fang-like teeth near the premaxillary symphysis characterises the outgroup taxa, Sphyraena and the gempylids. Absence of these specialised teeth is a synapomorphy of scombrids and billfishes’. Although many of the teeth of Eothyrsites are difficult to see in the fossil specimen as the skull is crushed, the presence of at least one slightly curved fang-like tooth, 11 mm long, at the anterior of the upper jaw ( and ), suggests Eothyrsites had at least one pair of symphysial teeth. We follow Bemis et al. (2019) here in use of the term ‘fang’. There is no obvious barb on the Eothyrsites premaxillary fang, such as those seen in the cutlassfish Trichiurus (Bemis et al. 2019), and it appears to be more conical than compressed in section. This compares well with a 12 mm symphysial fang seen in a modern Thyrsites specimen (NMNZ P.002334) and a 15 mm fang in Rexea solandri (NMNZ P.002339).

In Eothyrsites each premaxilla also supported a number (at least nine) of smaller inwardly curved, conical (but not lanceolate) lateral teeth. A modern Thyrsitesspecimen (NMNZ P.002334) premaxilla bears 20 smaller (4–5 mm length) conical teeth; Rexea solandri (NMNZ P.002339) has a similar number of slightly larger (up to 7 mm) teeth. The lateral premaxillary teeth of Trichiurus lepturusand the snake mackerel Gempylus serpens are reduced in number but larger than those of Eothyrsites (see Bemis et al. 2019). Another gempylid, the oilfish Ruvettus, has similar teeth but Ruvettus (and Lepidocybium) don’t have the enlarged fangs at the anterior (A. Stewart pers. comm.).

There was probably a row of slightly smaller teeth in the lower jaw of Eothyrsites. Chapman noted and figured five ‘mandibular’ teeth in series on OU 6854 d, each 4 mm long, laterally compressed yet somewhat more slender than those of the premaxilla and slightly inwardly curved (B). We suspect these actually represent dentary teeth, as immediately adjacent to the oral fragment of the dentary this piece of the specimen also contains part of the posterior lower jaw – including at least a broad triangular 80 × 50 mm portion of the posterior end of the articular and angular. This exhibits a notable facet for the articular condyle of the quadrate and resembles in shape that seen in Thyrsites atun (NMNH P.002659). Unfortunately the anterior portion of the dentary is not obvious in the specimen, so we are unable to assess whether Eothyrsites also exhibited enlarged dentary fangs (cf. Promethicthyes) or had a protruding lower jaw (as for Thyrsites and Rexea).

Opercular series: A number of partly overlapping broad, flat elements that presumably include the opercular bones are present in the specimen on OU 6854 d (C). In fact this piece also contains part of several incomplete suspensorium and lower jaw elements (possibly the quadrate as well as the angulo-articular) compressed together under the gill cover. Chapman noted the preopercular, which is probably in fact the lower rounded portion of the interopercular, which occur paired with the subopercular and flattened, each side measuring approximately 60 × 30 mm. Typical for the family, the gill cover elements are smooth, without any spines or significant ornament (Danilchenko 1960). Part of several branchiostegal rays may just be visible on the edge of the specimen.

Postcranial skeleton and fins

Vertebral column: At least eight articulated amphicoelus trunk centra are present in the specimen piece OU 6329 (E, ), of length 20 mm anteroposteriorly and with a height of 18 mm. Toward the anterior and posterior of the body some are damaged and there appear to be vertebrae missing. Portions of at least three centra of similar size are present on the smaller unnumbered fragments. We would expect a total vertebral count of at least 30 as in modern gempylids (Potthoff et al. 1986; Nakamura and Parin 1993), with gemfish such as Rexeaspecies typically having 34 (Roberts and Stewart 1997). Chapman noted the abdominal vertebral centra of Eothyrsites as being of similar dimensions to those in Thyrsites, but more robust.

Figure 6. Photograph of part of abdominal region of Eothyrsites holosquamatusChapman, syntype OU 6329, lateral view.

Ribs are not well preserved, those present are moderately long and slender (∼4 mm wide) and angled caudally (seen in OU 6329).

Squamation: The body of Eothyrsites holosquamatus appears to have been covered in reasonably large (on average 4 mm but up to 7 mm diameter), robust cycloid scales. One piece of the specimen, OU 6329 b, seems to reveal a 50 × 50 mm portion of trunk covered by overlapping scales (J). These ovoid-shape scales have smooth margins, appear to lack prominent radii, and due to preservation the circular growth rings are only just visible. Chapman (1935) noted that modern Thyrsites atun only has similar scales along its lateral line, the majority of its scales being very much reduced in size.

Pectoral fins: Eothyrsites had relatively large and elongate pectoral fins, which are still articulated in the specimen. According to Chapman these are narrow, slender, and only slightly curved, 70 mm in length, 21 mm at widest, as in the left pectoral figured as OU 6332. However, in addition, fragment OU 6854 a bears a second (right) fin, somewhat more complete distally (F). It is comprised of at least 10 rays, the longest being approximately 140 mm in length and reaching some 40 mm in width. The modern gemfish Rexea typically have 14 pectoral fin rays (Roberts and Stewart 1997). With the pectoral fins at 20% total body length, this is slightly longer than the pectoral fins of modern Thyrsites atun, and in this respect resembles those seen in the longfin escolar Scombrolabrax (Nakamura and Parin 2001). Unfortunately the pieces of Eothyrsites containing pectoral fin are broken away in the region of the pectoral girdle, although a small part (40 × 30 mm) of the coracoid or postcleithrum may be present on OU 6332.

Dorsal fin: Gempylids typically exhibit a notable spinous dorsal fin. We are uncertain if this is preserved in the specimen. Note that Johnson (1986) considered the absence of supraneural bones as a synapomorphy of the Gempylidae (including Trichiuridae), Scombridae and billfishes (Istiophoridae and Xiphiidae). Gempylidae are characterised by having a spinous dorsal fin develop prior to the soft dorsal fin, seen in young larvae of the group (Johnson 1986). The dorsal-fin spines of the other gempylid genera, except Thyrsitops, are recorded as bearing serrations at least on their posterolateral keels (Carnvale 2006).

We note the presence of a small fin comprising ∼10 rays measuring up to 40 mm in length, seen in piece OU 6329 (). This may represent the anal or pelvic fin, or alternatively, it could represent a dorsal or anal ‘finlet’.

Caudal skeleton: Eothyrsites has a rather small forked caudal fin, the preserved endoskeleton (OU 6854 b) measuring approximately 80 mm from tip to tip of the lobes, and is comprised of articulated elements that become broader distally and once supported the principal rays (I). These distal rays and the procurrent rays are not present in the specimen. Despite being difficult to define due to preservation in Eothyrsites, three vertebrae once supported the fin rays. The preural and ural centra (urostyle = PU + U1) appear fused to the second ural centrum (U2) to form the terminal caudal compound centrum (). Three epurals articulate to the upper side of this, while the ventral margin supports at least three hypurals (H1–3). The specimen also suggests two slightly broader central hypurals (H4 + H5) that once supported the principal rays from their distal margin. The parhypural is not obvious and was perhaps preserved on the counterpart of the specimen. Overall a similar caudal arrangement to the specimen is seen (although not exclusively) in modern fishes of the Gempylidae. These exhibit a caudal skeleton where the overlying procurrent fin rays are usually supported by the neural and haemal elements of the three terminal centra (Potthoff et al. 1986).

Figure 7. Restoration of caudal skeleton of Eothyrsites holosquamatus Chapman, based on syntype OU 6854 b, lateral view.

Discussion

Some anatomical features of Eothyrsites appear similar to the extant barracuda (family Sphyraenidae), suggesting a relationship within the broader suborder Scombroidei. However, upon further study, the suspicion that Eothyrsites may have actually belonged to the Sphyraenidae (Rust 2000; King et al. 2009) is inaccurate. For example, the caudal skeleton does not suggest inclusion in Sphyraenidae. The modern barracuda Sphyraena has a caudal skeleton with fused hypurals, urostyle and uroneurals (Hollister 1937). Eothyrsites () does not exhibit this degree of caudal fusion into a reduced number of solid hypural plates, and in this respect also differs from the higher Scombridae and billfish (Istiophoridae and Xiphiidae).

In addition, the anterior and posterior large fang-like premaxillary symphysial teeth of the barracuda Sphyraena are somewhat compressed, bilaterally symmetrical and typically exhibit a coloured enamell on the cutting edges, or at a notable commissural barb found near the posterior apical portion of the tooth (Ballen 2020). This is in contrast with the smooth margins on the teeth of Eothyrsites. Also, besides having large fang-like but asymmetrical dentary symphysial teeth, the smaller dentary teeth of Sphyraena are notably lanceolate in profile and bear fine serrations (Ballen 2020), features not seen in Eothyrsites, although the teeth of the lower jaw are poorly preserved in our specimen.

We believe Chapman was correct in assigning the specimen to the Gempylidae within the Trichiuroidea. Obviously, there is difficulty in identifying some morphological characters in our specimen of Eothyrsites due to preservation, but it generally appears consistent with the family level diagnosis of Danilchenko (see diagnosis section). Danilchenko (1960, 1967), Monsch and Bannikov (2011) and Bannikov (2017) also discuss several extinct genera of gempylids from the Oligocene of Europe, including Thyrsitocephalus (vom Rath 1859), which Chapman (1934, 1935) compared to the Dunedin fossil. Thyrsitocephalus alpinus from Canton Glarus, Switzerland, was a much smaller fish (length 16 cm), the head being a fifth of the body length with rather small orbits and 12 recurved teeth (none mentioned as being enlarged as fangs) in each jaw. It had a large anterior dorsal fin with more than 20 spines, small paired fins and many finlets. Like Eothyrsites it had an elongate body (54 vertebrae) and a forked caudal fin, but scales are not known (vom Rath 1859). Another of these fossil gempylids, Hemithyrsites, is also a smaller fish but regarded as coeval with Promethicthyes(see discussion in Monsch and Bannikov 2011), although most of the fossil specimens are small/sub-adult/juvenile? Its vertebrae are typically twice as long as high, hence more elongate than those of Eothyrsites. Hemithyrsites has 17 caudal vertebrae, unfused hypurals, and a complete covering of cycloid scales (Danilchenko 1960, 1967; Monsch and Bannikov 2011).

The prominent cycloid scales are a distinctive character of the Burnside fossil. Chapman himself noted the contrast in scales present between Eothyrsites and modern Thyrsites, those in the latter being very much reduced (apart from those in the lateral line region). He expressed some of his thoughts on the loss of large scales as an example of a ‘degenerative process’ in Chapman (1934). However, the bodies of adult specimens of modern gemfish, Promethicthyes prometheusand Rexea solandri in particular, are also entirely covered by small cycloid scales (Nakamura and Parin 1993; Roberts and Stewart 1997; Stewart 2015).

The fishes commonly named gemfishes are characterised (Roberts and Stewart 1997) by a slightly elongated body, a moderate number of vertebrae, the presence of supernumerary finlets on the caudal peduncle, a reduction of pelvic fins and the presence of small fangs on the lower jaw. The gemfish are also united in having two pairs of teeth on the vomer (Beckett et al. 2018). Certainly the skull and jaws (with anterior fangs and smaller posterior teeth) of Eothyrsitescompare well with modern gemfishes. Like Thyrsites, large adult specimens of modern Rexea reach up to 1–1.5 m in length in New Zealand waters (Nakamura and Parin 1993; Stewart 2015). All have forked caudal fins.

Cranial characters for gempylids are taxonomically important but often difficult to distinguish in fossil specimens where the crania are often crushed (Beckett et al. 2018). Use of technology such as x-ray computed tomography is now aiding studies of early gempylid skull specimens from the Eocene London Clay for example (Beckett et al. 2018). Beckett et al. (2018) also provide several detailed phylogenetic analyses of the Gempylidae (which they designate as a ‘grade’) and the closely related Trichiuridae, within the grouping Trichiuroidea. They do note a fossil skull from the London Clay, NHMUK PV OR 41318, as being similar to Rexea in exhibiting a narrowing of the frontals above the orbit and the pterotic ending before the posterior margin of the neurocranium. Unfortunately a lack of readily identifiable cranial characters in our specimen due to compression and thus far being only exposed from the underside, makes comparison and placement of Eothyrsites in their phylogenetic framework somewhat difficult at this stage. Further preparation and study of the skull in the future using novel techniques such as employed by Beckett et al. (2018) should aid understanding of the species relationships.

In conclusion, Chapman’s establishment of Eothyrsites holosquamatus as a unique genus and species appears justified, however we suspect Eothyrsitesrepresents an ancestral form of gempylid, perhaps more closely related to gemfish such as Promethicthyes or Rexea than Thyrsites.

Acknowledgements

Many years of fossil research, collection and curation at the University of Otago by Ewan Fordyce in particular have provided inspiration and information about the ancient life of the New Zealand region. The authors gratefully acknowledge Ewan Fordyce, Daphne Lee (OU) and Andrew Stewart (Te Papa) for their encouragement and assistance in this project. The reviewers are thanked for their constructive comments. Andrew Stewart also kindly reviewed an early stage of the manuscript. Stephen Read (OU) assisted with production of the locality map. Staff at the Hocken Library and Auckland Museum provided access to personal correspondence of key figures in Zealandian natural history. We also acknowledge the broad contributions Frederick Chapman and William Noel Benson made to Australasian paleontology.

Disclosure statement

No potential conflict of interest was reported by the authors.