Two new Palaeocene osteoglossomorphs from Canada, with a reassessment of the relationships of the genus †Joffrichthys, and analysis of diversity from articulated versus microfossil material

Two new Palaeocene osteoglossomorphs from Canada, with a reassessment of the relationships of the... Abstract A single block containing five articulated osteoglossomorphs was recovered from the Paskapoo Formation of southern Alberta, during development of a residential community in Calgary. Two of the specimens represent a new species of †Joffrichthys, and the other three represent a new genus and species of osteoglossomorph. The discovery of a new species of †Joffrichthys led us to re-examine the type species and to recode many of the characters that have been used in phylogenetic analyses. In particular, we interpret the caudal skeleton of †Joffrichthys to have 16 branched principal rays, not 15, which indicates this genus does not belong in Osteoglossiformes, and removes it from the osteoglossid/heterotidine affiliations previously reported. We assessed the relationships of the two new taxa using a modified data matrix including new outgroups and corrected data, with and without the inclusion of †Ostariostoma. Our results show that †Joffrichthys is a basal member of the superorder, and not a member of the Heterotidinae, but the other new taxon is left incertae sedis in the superorder. We also provide data on the early history of osteoglossomorphs in North America provided by isolated elements from Cretaceous and Palaeocene microfossil sites that complement and supplement that provided by articulated specimens. Cretaceous, Heterotidinae, Hiodontiformes, Joffrichthys, Ostariostoma wilseyi, Osteoglossidae, Osteoglossomorpha INTRODUCTION Osteoglossomorpha is a basal teleostean lineage that may be the most primitive living lineage of Teleostei (e.g. see review in Arratia, 1997). Osteoglossomorph fishes are known from deposits as early as the Early Cretaceous and are still extant today. They have a greater diversity of lineages in the fossil record than they do in the modern fauna, although the modern fauna includes a much larger number of species, particularly within Mormyridae (freshwater elephantfishes). Despite many years of study, the phylogenetic relationships of the superorder are still not well resolved. The oldest fossil osteoglossomorphs are from Early Cretaceous deposits of China (Wilson & Murray, 2008; Murray, You & Peng, 2010). In North America, the oldest known is †Chandlerichthys strickeri Grande, 1986, from the mid-Cretaceous deposits of Alaska. This fish, preserved as a carbon film with few clear details, was allied with the Osteoglossomorpha by Grande (1986) and seems to fit this group well in the details that he could discern. Slightly younger are the Campanian †Cretophareodus alberticus Li, 1996, from the late Campanian Belly River Group of southern Alberta, and †Wilsonichthys aridinsulensis Murray et al., 2016, and a second, as yet undescribed, osteoglossomorph recently recovered from the late Maastrichtian Scollard Formation of central Alberta (Murray et al., 2016). All of these are deep to moderately deep-bodied forms with relatively large dorsal and anal fins, from freshwater deposits. Disarticulated fish elements from the Late Cretaceous vertebrate microfossil localities have also been referred to the Osteoglossomorpha based on comparison with recent and fossil members of the group (Brinkman & Neuman, 2002; Neuman & Brinkman, 2005; Brinkman et al., 2013; Brinkman, Newbrey & Neuman, 2014). These provide additional data on the diversity and distribution of the group, as well as relative abundance within assemblages. Palaeocene osteoglossomorphs in North America are less common than Cretaceous forms. †Ostariostoma wilseyi Schaeffer, 1949, from deposits of Late Cretaceous or early Palaeocene age of the Livingston Formation of Montana, was tentatively identified as a hiodontid; however, isolated elements referred to this taxon bring this identity into question, instead suggesting gonorhychiform relationships (Brinkman, Neuman & Divay, 2017). The only confirmed Palaeocene North American osteoglossomorph species are †Joffrichthys symmetropterus Li & Wilson, 1996a, from the Paskapoo Formation of Alberta, and †J. triangulpterus Newbrey & Bozek, 2000, from the Sentinel Butte Formation of North Dakota. Additionally, an indeterminate species of †Joffrichthys was reported from the Smoky Tower locality in north-central Alberta (Li & Wilson, 1996a). Although a hiodontid was reported by Wilson (1980) on the basis of isolated elements (scales, parasphenoids, maxillae and a partial tail), these remains, along with those Wilson (1980) assigned to Osteoglossidae, were all later referred to †Joffrichthys (Li & Wilson, 1996a). †Joffrichthys has been placed within the subfamily Heterotidinae (Osteoglossidae) since it was first described. This relationship implies a very long history for the subfamily and family, and, based on sister-group relationships, a very long history for many other extant osteoglossomorph lineages. We here describe two new osteoglossomorph taxa from the Paskapoo Formation of Alberta, and assess the relationships of them and the genus †Joffrichthys, based on a reinterpretation of the type species. With the additional data on the morphology of early osteoglossomorphs provided by these specimens, we also evaluate the vertebrate microfossil remains from the Late Cretaceous previously referred to the Osteoglossomorpha and review the data they provide on the diversity and distribution of the group within North America during the Late Cretaceous. Geology The Palaeocene beds of west-central Alberta consist of fluviolacustrine sediments deposited by rivers flowing in an easterly direction (Hoffman & Stockey, 2000). Four lithological units of Palaeocene age in Alberta are currently recognized. These are the upper (Palaeocene) part of the Scollard Formation, upper (Palaeocene) part of the Willow Creek Formation, and the exclusively Palaeocene Paskapoo and Porcupine Hills formations. The Scollard Formation and Willow Creek Formation both contain the Cretaceous/Palaeogene (K/Pg) boundary, which divides the units into Cretaceous and Palaeocene parts (Sweet & Braman, 1992). Although the Paskapoo Formation has produced many articulated fishes, articulated specimens of teleosts have not been recovered from the Palaeocene part of the Scollard and Willow Creek formations. Vertebrate microfossil localities containing isolated fish elements are present, but these have not yet been thoroughly studied. The Paskapoo Formation is widespread in western Alberta, extending from Calgary in the south to the Hinton area north-west of Edmonton. The formation was deposited by low energy fluvial systems under humid conditions, with associated overbank environments including ponds and oxbow lakes (Hoffman & Stockey, 2000). The rivers depositing Paskapoo sediments trended slightly south of east (Carrigy, 1971). Many of the fish from the Paskapoo Formation have been recovered from outcrops (Wilson, 1980). A particularly significant locality is the Joffre Bridge road cut, which preserved a mass death layer containing hundreds of fishes. Actinopterygian fishes recovered from the mass death layer include an amiid, osmerid (†Speirsaenigma lindoei), percopsid (†Massamorichthys wilsoni) and the osteoglossomorph †J. symmetropterus (Wilson & Williams, 1991; Li & Wilson, 1996a; Murray, 1996). The Porcupine Hills Formation was deposited in seasonally dry conditions by rivers that trended slightly east of north. It differs from the Paskapoo Formation in its mineralogical composition and the presence of caliche (Carrigy, 1971). Strata of the Porcupine Hills Formation disconformably overlie those of the early Palaeocene levels of the Scollard Formation in the Calgary region and the Willow Creek Formation farther south (Jerzykiewicz, 1997; Lerbekmo & Sweet, 2000). Both the Porcupine Hills Formation and the Paskapoo Formation are present in Calgary, with the Paskapoo overlying the Porcupine Hills Formation. Three mass death assemblages of gar fish have been recovered from Calgary and area, although it is uncertain whether these are from the Paskapoo Formation or the Porcupine Hills Formation. One of these, TMP 2013.009.0001, a specimen in the Royal Tyrrell Museum, is exceptional in that it preserves the remains of 24 fully articulated three-dimensionally preserved fish. However, articulated teleosts have not previously been recovered from the Palaeocene beds in Calgary or from the Porcupine Hills Formation further south. The material described here came from Calgary and was found during the development of a subdivision on the north edge of the city. The large block contains the remains of five osteoglossomorph fishes (Fig. 1). The block was recovered by Mr Edgar Nernberg of Calgary during a routine excavation of a basement with a backhoe in mid-March 2015. Two weeks later he contacted Dr Darla Zelenitsky at the University of Calgary about the discovery; she then reported it to Dr Francois Therrien at the Royal Tyrrell Museum where the specimen is now curated. Zelenitsky visited the site at the time of the report, although the location where the osteoglossomorph block was found had been fully built over with a home basement. Small blocks and rubble of sandstone similar to that of the osteoglossomorph block were present immediately around the constructed basement but contained no visible fossils. Since the block was recovered from a relatively high elevation in the city, it is interpreted as having come from the upper portion of the Paskapoo Formation. The age of the Paskapoo Formation in Calgary has been estimated by magnetostratigraphic (Lerbekmo & Sweet, 2000, 2007), radiometric (Lerbekmo et al., 2008) and biostratigraphic studies (Krause, 1978; Fox, 1990; Demchuk & Hills, 1991). For the latter, both palynology and mammals indicate that the base of the Paskapoo Formation in the Calgary area is middle Palaeocene and the upper beds are late Palaeocene in age. Thus, the block is assumed to be from the late Palaeocene. Figure 1. View largeDownload slide Photograph of the whole block from the Paskapoo Formation showing the relative positions of the five specimens. †Joffrichthys tanyourus sp. nov., holotype TMP 2015.011.0003 (A) and paratype TMP 2015.011.0002 (B); †Lopadichthys colwellae gen. et sp. nov., holotype TMP 2015.011.0001 (C), paratype TMP 2015.011.0004 (D) and paratype TMP 2015.011.0005 (E). Fish are oriented with the dorsal surface towards the top of the page (B, C), with the dorsal surface towards the bottom of the page (A), with the dorsal surface to the left (D) and with the dorsal surface to the right (E). Scale bar is in centimetres. Figure 1. View largeDownload slide Photograph of the whole block from the Paskapoo Formation showing the relative positions of the five specimens. †Joffrichthys tanyourus sp. nov., holotype TMP 2015.011.0003 (A) and paratype TMP 2015.011.0002 (B); †Lopadichthys colwellae gen. et sp. nov., holotype TMP 2015.011.0001 (C), paratype TMP 2015.011.0004 (D) and paratype TMP 2015.011.0005 (E). Fish are oriented with the dorsal surface towards the top of the page (B, C), with the dorsal surface towards the bottom of the page (A), with the dorsal surface to the left (D) and with the dorsal surface to the right (E). Scale bar is in centimetres. MATERIAL AND METHODS The new osteoglossomorph material described here is preserved on a bedding plane of a single large block held in the collections of the Royal Tyrrell Museum of Palaeontology (TMP), Alberta, Canada. Five fishes are present; three are complete and two are partial specimens, one of which preserves the anteroventral half of the body, and the other preserves the posterior two-thirds of the body. The block containing the five fishes (Fig. 1) with the individual fish specimens designated by the numbers TMP 2015.011.0001, TMP 2015.011.0002, TMP 2015.011.003, TMP 2015.011.0004 and TMP 2015.011.0005. Comparative fossil and extant material examined is from the TMP, the Royal Ontario Museum (ROM), Toronto, Ontario; University of Alberta Museum of Zoology (UAMZ), and University of Alberta Laboratory for Vertebrate Palaeontology (UALVP) Edmonton, Alberta; University of California Museum of Paleontology (UCMP) Berkeley, California; and the National Museum of Tanzania (latex peels of specimens collected by the Wembere Manonga palaeontological expedition, with catalogue numbers prefixed WM). A dagger symbol (†) is used to indicate fossil taxa. The new Paskapoo osteoglossomorph fossil material was prepared by Allan Lindoe, UALVP, using a microscribe and needles. To highlight the relief, specimens were coated with ammonium chloride prior to taking photographs. Silicone peels were made of the specimens and used to confirm some anatomical characters after descriptions were made based on the original fossil material. Comparative material examined †Joffrichthys symmetropterus UALVP 23705 (holotype), UALVP 31545, 37128; †J. triangulpterus UALVP 51921 (cast of Field Museum specimen PF12171a,b as †J. symmetropterus); †Joffrichthys sp. UALVP 34770; †Singida jacksonoides Greenwood & Patterson, 1967, WM 241/96, WM 298/96, WM 314/96, WM 315/96, WM 536/96; †Chauliopareion mahengeense Murray & Wilson, 2005, WM 378/96, WM 492/96; †W. aridinsulensis TMP 2012.020.1493 (holotype), TMP 2012.020.1498; †O. wilseyi Shaeffer, 1949, UALVP 52610 (cast of the holotype, Princeton University Geological Museum specimen PU14728); Amia calva Linnaeus, 1766, UAMZ 1260 and three unnumbered skulls. Phylogenetic analysis The phylogenetic analysis is based on a data set for Osteoglossomorpha that has been used in a number of publications, with a few modifications. The original data matrices are from Li, Wilson & Grande (1997) and Hilton (2003). The characters were assessed for overlap and clarity and combined in a single matrix by Wilson & Murray (2008); the character list we use here is from that publication. Additional taxa were added by Wilson & Murray (2008), as well as Murray et al. (2010; †Shuleichthys brachypteryx from the Cretaceous of China) and Murray et al. (2016; †W. aridinsulensis from the Maastrichtian of Alberta). Caudal fin ray counts for †J. symmetropterus and †C. mahengeense were reassessed and modified after examination of the holotype and referred material for the former and latex peels of the latter (see Discussion). We also modified some of the data for Elops, making changes based on Ridewood (1904), Taverne (1974) and Schultze & Arratia (1989); many of these changes were providing a code for which there was previously a question mark in the data matrix. This corrected data matrix was used for our first analysis, to compare the position of the new taxa based on the previously used data matrix. A number of named fossil taxa have not been included in our, nor a number of other (see Wilson & Murray, 2008), phylogenetic analyses. In particular, African genera such as †Ridewoodichthys Taverne, 2009 (based on disarticulated jaw bones and a caudal skeleton), †Paradercetis Casier, 1965 (based on a partial skull), and †Chanopsis Casier, 1961, have been excluded, but all were placed in the more derived family Osteoglossidae by Taverne (1975, 1976a, 2009, 2016a) and so should not affect the relationships of more basal members of the superorder examined here. We also excluded †Kipalaichthys Casier, 1965, which was considered an osteoglossomorph of uncertain affinities by Taverne (1976a, b) and is not well known for the skull. Several Asian taxa have been excluded; these were not readily available to us but future plans of several researchers include detailed reanalyses of these taxa. Three European genera from Eocene marine deposits of Italy (Taverne, 1998) named for articulated fossil material are also not included in our analyses. Taverne (1998) considered one to belong to the family Osteoglossidae (†Thrissopterus Heckel, 1856), and reported another (†Monopteros Volta, 1796) as lacking the caudal skeleton and both that taxon and Foreyichthys Taverne, 1979 as lacking many skull elements; this makes it difficult to code either for the phylogenetic analysis. In the second analysis, we assessed the taxa used in the previous data matrix and replaced the outgroup taxa, so that we could reduce the amount of missing data. We removed the ichthyodectiform †Cladocyclus, and replaced it with the extant, non-teleost A. calva, which we coded based on the text and figures of Grande & Bemis (1998) and personal examination of skeletal material. With the inclusion of Amia, we changed character 68 (number of epurals) state 0 from ‘two or three’ to ‘two or more’, and we modified character 71 (number of hypurals) state 0 from ‘seven’ to ‘seven or more’ to accommodate the more numerous epurals and hypurals of Amia. We also removed the data for Clupeoidei that were taken from Li et al. (1997), and instead coded for the extinct clupeomorph order †Ellimmichthyiformes based on †Armigatus dalmaticus Murray et al., 2017 and †A. namourensis Forey et al., 2003, with additional data from a three-dimensional skull of †Diplomystus (from Forey, 2004). We also included Clupeiformes, the extant clupeomorph order, based on Dorosoma (data taken from Grande, 1985). †Eohiodon is considered here to be a valid taxon following Murray et al. (2010), and not subsumed into Hiodon as supported by Hilton & Grande (2008). We combined the data for the three mormyrids into a single polymorphic taxon and did the same for the notopterids, following Murray et al. (2010), and also excluded †Ostariostoma from one of the analyses as discussed below. A single new character was also added to the matrix. This character, the expansion of the parapophysis on the first centrum, was noted by Forey & Hilton (2010). Based on their work, character 88 is formulated as: parapophysis on first centrum: 0) not expanded or hypertrophied; 1) expanded or hypertrophied to reach under the occiput. This feature is discussed further in the Phylogenetic Analysis section. The final character state matrix for this modified data set is given in Appendix 1, with a list of the characters and states in Appendix 2. The original and new data matrices were manipulated in Mesquite v. 3.10 build 765 (Maddison & Maddison, 2016) and trees were visualized in the same software. Data were analysed in PAUP v. 4.0a152 (Swofford, 2002). We used parsimony analysis with an heuristic search, using simple stepwise addition and TBR branch swapping, with all characters unordered and unweighted for all analyses. Anatomical abbreviations aa, anguloarticular; ang, angular; art, articular; boc, basioccipital; brst, branchiostegal rays; ch, anterior ceratohyal; c, centrum; cl, cleithrum; cor, coracoid; den, dentary; ds, dermosphenotic; ect, ectopterygoid; end, endopterygoid; ep, epural; epi, epioccipital; es, extrascapular; exo, exoccipital; fr, frontal; hh, hypohyal; hy, hypural; hyo, hyomandibula; ic, intercalar; io, infraorbital; iop, interopercle; l, left; le, lateral ethmoid; m, mesethmoid; met, metapterygoid; mx, maxilla; n, neural arch; na, nasal; naap, neural arch articulation pit; nsp, neural spine; op, opercle; os, orbitosphenoid; pa, parietal; par, parapophysis; pd, predorsal bone; ph, parhypural; pmx, premaxilla; pop, preopercle; ps, parasphenoid; pto, pterotic; pts, pterosphenoid; ptt, post-temporal; pu, preural centrum; q, quadrate; r, right; ra, retroarticular; rap, rib articulation pit; scl, supracleithrum; soc, supraoccipital; sop, subopercle; sym, symplectic; tb, toothed bone of pharyngeal arches; un, uroneural; u, ural centrum. SYSTEMATIC PALAEONTOLOGY Teleostei Müller, 1845 Osteoglossomorpha Greenwood, Rosen, Weitzman, & Myers, 1966 incertae sedis †Joffrichthys Li & Wilson, 1996a Included species: †Joffrichthys symmetropterus Li & Wilson, 1996a; †J. triangulpterus Newbrey & Bozek, 2000. Emended diagnosis: Deep-bodied osteoglossomorph fishes with six hypurals, no epurals, one distinct uroneural, six pelvic fin rays, kidney-shaped opercle ornamented laterally with striations radiating from area of facet for articulation with hyomandibula, and large dorsal and anal fins positioned posteriorly. Differs from Osteoglossidae and Notopteridae by having 18 principal rays (16 branched) in the caudal fin; from Notopteridae by lacking a long anal fin confluent with the caudal fin, and from Mormyridae by lacking the rounded snout, elongate jaw bones and other specializations of those families; from Hiodontiformes by lacking the posteriorly recurved spine on the opercle; and from basal members of the superorder by having the supraorbital sensory canal ending in the frontal not the parietal (which is similar to the more derived Osteoglossidae), and having the dorsal and ventral arms of the post-temporal equal in length (this last feature is probable, but not clearly visible in the type species). †Joffrichthys tanyourus sp. nov. Holotype: TMP 2015.011.0003, a complete fish preserved in right lateral view (Fig. 2). Figure 2. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 2. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Paratype: TMP 2015.011.0002, the anteroventral portion of a fish preserving the head and ventral body, missing the dorsal and caudal regions, preserved in right lateral view (Fig. 3). Figure 3. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. Figure 3. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. Type locality and age: Paskapoo Formation, Calgary, Alberta, Canada; late Palaeocene in age. Etymology: The specific epithet is for the Greek tany meaning long and oura meaning tail. The two previously described species are named for the shape of their dorsal and anal fins, but we name this third species for having a longer caudal peduncle than the others. Diagnosis: Differs from †J. symmetropterus and †J. triangulpterus by having 26 abdominal vertebrae (compared to 22 in †J. symmetropterus and 23–25 in †J. triangulpterus), falcate dorsal and anal fin (compared to rounded fin margins in †J. symmetropterus and triangular fins in †J. triangulpterus), 11 predorsal (supraneural) bones (compared to 16 in †J. triangulpterus and 18 in †J. symmetropterus). Further differs from †J. triangulpterus by having the pelvic fin positioned near to the centre point between the origins of the pectoral and anal fins rather than closer to the anal fin and having about four centra between the insertion of the first anal and dorsal fin pterygiophores, rather than 8–9 centra. Further differs from †J. symmetropterus by having the caudal peduncle longer than deep instead of deeper than long (unknown in †J. triangulpterus). Description General body form This is a moderately deep-bodied fish, with a terete shape (Figs 2, 3). Both specimens are preserved in lateral view, indicating this species was probably laterally compressed in life. The complete specimen (holotype) is 205 mm in total length, and the paratype (incomplete specimen) would have been slightly smaller. The anal and dorsal fins are positioned posteriorly, with the dorsal fin inserting more anteriorly than the anal fin. Both fins are falcate, with the anterior rays being the longest, then the fin steps down to a shorter depth posterior to the midpoint. The caudal peduncle is longer than it is deep. The head is triangular in lateral view, much deeper posteriorly than at the jaws, and is about one quarter of the standard length (SL) in the holotype. The following description applies to the holotype, with the paratype agreeing in details where they can be seen unless otherwise noted. Counts and measurements are given in Table 1, along with those for specimens of the two other species. Table 1. Counts and measurements (in mm) for the two specimens of †Joffrichthys tanyourus sp. nov., J. triangulpterus (taken from Newbrey & Bozak, 2000) and J. symmetropterus †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 Measurements are in millimetres. ? indicates some uncertainty in the data reported here. *These counts and measurements were made by the authors and some data differ from those reported by Li & Wilson (1996a). †Counts and measurements for this specimen were made by adding numbers from the part and counterpart which preserve different parts of the fish. View Large Table 1. Counts and measurements (in mm) for the two specimens of †Joffrichthys tanyourus sp. nov., J. triangulpterus (taken from Newbrey & Bozak, 2000) and J. symmetropterus †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 Measurements are in millimetres. ? indicates some uncertainty in the data reported here. *These counts and measurements were made by the authors and some data differ from those reported by Li & Wilson (1996a). †Counts and measurements for this specimen were made by adding numbers from the part and counterpart which preserve different parts of the fish. View Large Neurocranium and skull roof The head is preserved in both TMP 2015.011.0003 (holotype; Fig. 4) and specimen TMP 2015.011.0002 (Fig. 5). A mesethmoid is present but no details can be determined. The nasal is a broad, flat, bone, with a notch and expansion anteriorly (Fig. 4); it is about twice as long as it is wide. The left and right nasal bones are completely separated by the mesethmoid. There appears to be a canal in the nasal extending from the posteromedial corner to the midpoint of the lateral edge. The frontal is roughly trapezoidal, widening gradually posteriorly; there is no distinct anterior expansion as found in †Phareodus and †Brychaetus (anterior supraorbital shelf of Hilton, 2003: fig. 11). The sensory canal on the frontal is enclosed in bone. The frontals are more than twice as long (antero-posterior length) as they are broad (medial-lateral width). The parietals are about one-third as long as the frontals. They are roughly rectangular in shape and do not bear a sensory canal. The supraoccipital crest is high and prolonged posteriorly. There is a large, roughly square, extrascapular bone positioned over the area between the parietal and the pterotic (Fig. 5). The pterotic is a long bone, reaching from the posterior portion of the frontal to the posterior end of the hyomandibular head. It bears an enclosed canal and a dorsal flange in its midpoint (Fig. 5). Figure 4. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 4. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 5. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. Figure 5. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. The supraoccipital is large, and angles dorsally above the level of the rest of the skull. Below this area in both specimens are bony remains that we interpret as the basioccipital and epioccipitals, with an additional unidentified bone in the paratype (Fig. 5), that could be a portion of the post-temporal. Ventrally, the parasphenoid bears large caniniform teeth along all of its visible length (Figs 4, 5). Posteriorly, the bone angles dorsally (Fig. 5). Based on the position of the visible part of the parasphenoid, which is ventral compared to the basioccipital condyle and vertebral column, either the posterior portion of the parasphenoid or the anterior portion of the basioccipital, or both, must extend significantly dorsoventrally in order for the two bones to meet one another. Jaws, suspensorium and branchial arches The terminal jaws are robust and large, bearing large caniniform teeth (Figs 4, 5). The articulation between the mandible and quadrate is positioned under the middle of the orbit. The anterior end of the premaxilla is enlarged into a low, rounded, ascending process that extends for about half the length of the bone. The premaxilla bears multiple rows of teeth with the teeth in the outside row being the largest (best seen in TMP 2015.011.0003, holotype, but not visible in the figures). There are six premaxillary teeth in the external row of teeth in the paratype (Fig. 5), and four teeth with several tooth sockets preserved in the holotype (TMP 2015.011.0003; Fig. 4). The maxilla appears deep in the paratype, but quite narrow anteriorly in the holotype; we interpret this as differences in the orientation of the preserved elements and incomplete preservation of the maxilla in the holotype. The anterior head of the maxilla narrows to articulate with the premaxilla, and the posterior end of the bone is gently rounded. A single row of 11 (TMP 2015.011.0002, paratype) or 12 (TMP 2015.011.0003 holotype) teeth is preserved on the maxilla. There are no supramaxillae. The dentary is relatively elongate and slender compared with †J. symmetropterus and has a relatively smaller symphysis. The ventral edge is slightly curved. An enclosed mandibular sensory canal that opens through four or more pores is located near the ventral edge of the bone. Similarly to the premaxilla, multiple rows of teeth are present on the dentary with the outer row of teeth being the largest (best seen in the holotype). The retroarticular is clearly a separate ossification, and is confined to the posteroventral corner of the mandible. The angular and articular appear to be separate from one another, with the articular small and confined on the lateral side of the jaw, forming the facet for the quadrate. The quadrate is fairly broad, with the anterior and posterior edges forming an obtuse angle with one another. The posterior edge is strengthened with a ridge. The hyomandibula has a single dorsal head, with no notch between anterior and posterior points (Figs 4, 5). It is a tall bone, at least twice as tall as it is broad antero-posteriorly. There is a slight anterior flange visible in the paratype, giving the hyomandibula a sinuous anterior edge. The hyomandibula in this specimen is also clearly ornamented with rugose vertical lines. A symplectic is not visible in either specimen. The endopterygoid fills the orbit below the parasphenoid. The ectopterygoid is narrower and slightly angled in the middle (TMP 2015.011.0002, paratype). Although teeth are not visible on the ectopterygoid in the paratype, it bears several conical teeth, at least some of which are significantly smaller than those of the parasphenoid and jaws in the holotype (Fig. 4). In addition, there are three large teeth present in this area in TMP 2015.011.0003 (holotype; Fig. 4) just dorsal to the quadrate and articular; however, based on the size, these teeth are more likely to be from the basihyal or basibranchial bones. Branchiostegal bones are visible in both specimens. There are eight preserved in the holotype and paratype, but whether or not they are from a single side in either specimen is difficult to determine. The branchiostegals are slightly broader more posteriorly in the series, but none are very broad. In the holotype, the anterior ceratohyal is visible; it is fairly short and broad. Opercular series The opercle is the best preserved bone of the series in both the holotype and paratype. It is a rounded bone, shaped somewhat like a kidney bean, with a notch anterodorsally at the level of the facet for articulation with the hyomandibula (Figs 4, 5). There are striations radiating from this point to ornament the lateral surface of the bone. The shape and ornamentation of the opercle are almost identical to those of the other two species of †Joffrichthys. The opercle is about twice as high (dorsoventrally) as it is wide (antero-posteriorly). The preopercle is best preserved in TMP 2015.011.0002 (paratype; Fig. 5). It is deepest in the middle, with much narrower dorsal and anterior ends. The sensory canal appears to be under a flange on the ventral limb, and opens through pores on the dorsal limb. Only the anterior tip of the ventral limb is preserved in the holotype (TMP 2015.011.0003). Thin bone is present between the opercle and preopercle in both specimens but is not well preserved, with the thin bone being broken in many places; we interpret this as remains of both the interopercle and subopercle. Infraorbital region Remains of three infraorbital bones are preserved in the holotype (labelled as io2–3, and io4, and the dermosphenotic in Fig. 4), and four are preserved in the paratype (1, 2–3, 4 and the dermosphenotic; Fig. 5). One of the elements probably represents two fused infraorbitals as in †J. symmetropterus (Li & Wilson, 1996a: fig. 3) or some specimens of Hiodon (Hilton, 2002: fig. 36). Based only on size, we here identify the second preserved bone to be a fused infraorbital 2–3. The first infraorbital is small, with a low, broad, triangular shape (paratype; Fig. 5). Infraorbital 2–3 is greatly expanded, and is the largest bone of the series; it covers the posteroventral portion of the cheek from the orbit to the preopercle. Infraorbital 4 is about two-thirds the size of the preceding element in the series. Both of these bones have an enclosed sensory canal. The fifth infraorbital, the dermosphenotic, is preserved as a portion of a bone-enclosed tubular canal. This bone meets the pterotic in TMP 2015.011.0002 (paratype; Fig. 5), although it appears displaced, and is also displaced in the holotype (Fig. 4). There are no supraorbitals or antorbitals in the specimens. Vertebral column and predorsal bones The holotype (TMP 2015.011.0003) has 26 caudal centra, including two ural centra, and about 26 abdominal centra for a total count of about 52. The lateral surfaces of the centra are visible on the anteriormost centrum and nearly all of the centra in the caudal series (Fig. 6). These have a solid bone texture (with no small pits or struts) and two large lateral pits separated by a strut of bone, giving them the ‘H’ shape noted by Li & Wilson (1996a) for †J. symmetropterus. There appear to be 17 or 18 pairs of ribs, but because of their preservation a count is difficult. It is uncertain whether these articulate on the parapophyses or directly with the centra. Predorsal bones are visible in the holotype (Fig. 4); there are 11 long narrow bones anterior to the dorsal fin pterygiophores. Anterior left and right neural spines are not fused in the midline, as is the case in the other two species of †Joffrichthys (Li & Wilson, 1996a; Newbrey & Bozek, 2000). Figure 6. View largeDownload slide Photograph and interpretive drawing of the tail of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 6. View largeDownload slide Photograph and interpretive drawing of the tail of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Paired fins and girdles The cleithrum is mostly obscured by the opercle, but the posterior curved edge is visible in both specimens (Figs 4, 5). The cleithrum reaches anteriorly under the preopercle for about half the length of that bone. Details of the scapula and coracoid are not visible in either specimen. The supracleithrum is clear in TMP 2015.011.0002 (Fig. 5); it is about three times as long as it is wide, and broadens dorsally. The post-temporal is visible in the holotype (Fig. 4); the dorsal and ventral limbs are of equal length and almost parallel to one another. The pectoral fin has 15 rays in the holotype. The pelvic fin is preserved in the holotype TMP 2015.011.0003, but the pelvic girdle is not visible. There are six pelvic fin rays. Dorsal and anal fins The dorsal fin is not preserved in the paratype. In the holotype, the dorsal fin is positioned more anteriorly than the anal fin (Fig. 2), with a predorsal length of 0.5 SL and preanal length of 0.63. The insertion points of the first dorsal and anal fin pterygiophores are separated from one another by four vertebrae in the holotype. The dorsal fin rays are not all preserved, but based on those that are present and the pterygiophores, there appears to be four procurrent and 26 principal rays. There are about 21 pterygiophores supporting the dorsal fin. The anal fin in the holotype has three procurrent rays and 24 principal rays, or possibly four procurrent and 23 principal rays. Caudal skeleton and fin The caudal fin is preserved in the holotype (TMP 2015.011.0003; Fig. 6). The fin is supported ventrally by the haemal spines of the second and third preural centra, the parhypural and the hypurals. None of the rays of the dorsal lobe are supported by any neural spines. There are 18 principal rays (16 branched) forming a slightly forked tail fin. There are about seven dorsal and six ventral procurrent rays, giving a fin formula of vii,I,8,8,I,vi. There is a single neural spine on each of the preural centra (Fig. 6). The parhypural on the first preural centrum (pu1) is not much larger than the preceding haemal spine of the second preural centrum. The first ural centrum (diural terminology) in the holotype is slightly longer than the preural centra, and bears the first two hypurals. The much smaller second ural centrum bears five hypurals, for a total of seven hypurals in the caudal skeleton. There are a number of elements dorsal and posterodorsal to the two ural centra in the holotype that are difficult to interpret. The anteriormost appears free from, and slightly overlaps the first ural centrum and we identify that as the first uroneural. The next posterior element overlaps the dorsal part of the second ural centrum and so we feel quite confident that it is the second uroneural. Posterior to this, there is an element that is trapezoidal in shape, but appears to overlap the second uroneural (labelled with a question mark in Fig. 6), so it could be an abnormally shaped third uroneural, or it could potentially be a third ural centrum, but it seems to be lying lateral to the second ural centrum. Another possible alternative is that this bone is a urodermal; however, a urodermal has not been reported for living osteoglossomorphs. Posterior and slightly dorsal to this trapezoidal element is a long narrow bone that is ventral to the distal part of the second uroneural; therefore, we consider this to be another uroneural (the fourth if the trapezoidal bone is a uroneural, or the third uroneural if the trapezoidal bone is not a uroneural). All of these elements are quite closely associated with the ural centra, being ventral to the distal part of the long second uroneural; therefore, none are considered to be epurals. We cannot distinguish a separate element more dorsally positioned that could be an epural, so we believe there are no epurals present. Scales Cycloid scales are visible on the supraoccipital of the holotype, but not on the other skull bones. Scales on the dorsal surface are larger than those preserved on the abdomen near the pectoral girdle. The scales are large, circular to ovoid in shape, and with a central focus. They have 50 or more fine circuli, and are not reticulated. Comparisons among the species of †Joffrichthys In addition to the features listed in the diagnosis, there are a few other features that differ among the known specimens of the three species of †Joffrichthys. The shape of the dentary of the new species, †J. tanyourus, differs from that of the type specimen of †J. symmetropterus: in the type specimen (UALVP 23705), the dentary is shorter and more triangular than that of †J. tanyourus, the teeth are relatively shorter, and the symphysis is smaller. Based on published photographs (Newbrey & Bozek, 2000: figs. 1–2), †J. triangulpterus is similar to †J. symmetropterus in the shape of its dentary. Li & Wilson (1996a: fig. 6) illustrated a specimen of osteoglossomorph from Smoky Tower, Alberta; they considered it to belong to Joffrichthys but not necessarily to J. symmetropterus. This specimen, UALVP 34770, is similar to †J. tanyourus in having a dentary that appears relatively elongate and having taller teeth. However, in contrast to †J. tanyourus, only a single row of teeth appears to be present and the dentary appears deeper and with a larger symphysis. This morphology is also found in an isolated dentary preserved as an impression (UALVP 15069) that was referred to †Joffrichthys by Li & Wilson (1996a). The morphology of these dentaries allows us to exclude both UALVP 34770 and UALVP 15069 from †J. tanyourus. Instead, we find them most similar to the new genus described below, and suggest they represent that taxon instead (see section on Referred Specimen from Smokey Tower below). There is a single neural spine on the second preural centrum (pu2) in †J. tanyourus. Newbrey & Bozek (2000) reported two full spines on pu2 in †J. triangulpterus, whereas Li & Wilson (1996a) reported one long and a second short spine on †J. symmetropterus (although we cannot confirm this as the specimens of this latter do not have well-preserved caudal fins). The number of neural spines on pu2 may vary among species, or may be individual variation. The anal and dorsal fin ray numbers are equal in †J. symmetropterus, giving the species its name, with 24 rays in each, and the insertions of the first pterygiophore of each fin are separated by four centra. In †J. tanyourus, the fin insertions are also separated by four centra in the holotype (TMP 2015.011.0003), and the fin ray counts are similar to one another (23 or 24 rays in the anal fin and 26 in the dorsal fin). However, the first anal fin ray is positioned under the middle of the dorsal fin, not at the same level as the first dorsal fin ray as in †J. symmetropterus. In other words, the preanal and predorsal lengths in †J. symmetropterus are similar, but the predorsal length is clearly greater than the preanal length in †J. tanyourus. In †J. triangulpterus, the preanal length is also greater than the predorsal length, but the dorsal fin is larger with many more rays than the anal fin (27+ in the dorsal and 17–24 in the anal fin; Newbrey & Bozek, 2000), and there are more (8–9) centra between the insertion points of the first pterygiophores of each fin. As noted in the diagnosis, the number of vertebrae also varies among the three species, with more abdominal centra in the new species than the other two (Table 1). Ribs also vary slightly, with a few more pairs in †J. triangulpterus than the other species, and the predorsal bones number fewer in the new species than the other two (Table 1). Newbrey & Bozek (2000) indicated that the anterior neural spines of †J. triangulpterus were not fused in the midline, and said this differed from the condition in †J. symmetropterus. However, we believe they misunderstood Li & Wilson (1996a: p. 203) because these latter authors stated ‘on the caudals, bilaterally opposite neural spines are fused with each other’ indicating that was not the condition in the anterior vertebrae, and, in fact, after examining the specimens we can confirm that the anterior neural spines are paired, not fused. Therefore, this condition is found in all three species of †Joffrichthys. OSTEOGLOSSIFORMES Berg, 1940 Incertae Sedis †Lopadichthys gen. nov. Type and only known species: †Lopadichthys colwellae sp. nov. Diagnosis: As for type and only known species. Derivation of name: The genus is named for the fish being disk or plate shaped, from the Greek lopas or lopados meaning a dish or plate, and the ending ichthys, Greek for fish. Gender is masculine. †Lopadichthys colwellae sp. nov. Holotype: TMP 2015.011.0001, a complete fish preserved in left lateral view (Fig. 7). Figure 7. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 7. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Paratypes: TMP 2015.011.0004, a complete fish preserved in left lateral view (Fig. 8), and TMP 2015.011.0005, the posterior portion of a fish preserving the dorsal, anal and caudal fins and most of the body, but missing the head and anteroventral body, preserved in left lateral view (Fig. 9). Figure 8. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 8. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 9. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. Figure 9. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. Type locality and horizon: Paskapoo Formation, Calgary, Alberta, Canada; Palaeocene in age. Etymology: The specific epithet is in honour of Jane Colwell-Danis in recognition of her considerable contribution to vertebrate palaeontology in Alberta. She has published under the names Colwell, Danis and Colwell-Danis. Diagnosis: A deep-bodied osteoglossomorph fish with a head depth slightly greater than head length; two heads on the hyomandibula; striations on the opercle radiating from the facet for articulation with the hyomandibula; 45–47 vertebral centra with 19–22 abdominal; anteriormost centra with numerous pits forming an open, lacy bone texture; and short deep lower jaw with the level of articulation anterior to the orbit and with a high coronoid process on the dentary. Differs from members of the suborder Osteoglossoidei (caudal formula I,7,8,I) by having a caudal formula of I,8,8,I. Further differs from subgroups within Osteoglossoidei by lacking an elongate pectoral fin ray and reticulate scales (found in many osteoglossids); having abdominal pelvic fins and unexpanded pectoral fin (unlike the anteriorly placed pelvic fins and expanded pectoral fins of Pantodon); having a small anal fin not connected with the caudal fin (unlike the elongate anal fin confluent with the caudal fin found in notopterids); jaws terminal and unmodified (unlike the modifications of the jaws found in mormyrids). Differs from species of †Joffrichthys by having anteriormost abdominal and posterior caudal vertebral centra with many small pits and a network of bone forming a lacy structure, rather than ‘H-shaped’ (i.e. two large pits laterally separated by a strong bar of bone), and by having a relatively shorter and deeper caudal peduncle. Differs from Hiodontiformes by having a rounded dorsal border to the opercle (without the posterodorsal opercular spine as found in Hiodon and †Eohiodon). Differs from basal osteoglossomorphs by having six rays in the pelvic fin (instead of seven rays in †Wilsonichthys, †Shuleichthys, †Xixiaichthys and hiodontiforms). Description General body form This is a deep-bodied fish, with the greatest body depth, anterior to the dorsal fin origin (Figs 7–9), being greater than half (0.57–0.58) of the SL. The dorsal and anal fins are slightly falcate, and positioned posteriorly on the body, with the dorsal fin positioned more anteriorly than the anal fin. The caudal peduncle is short, and deeper than it is long. The caudal fin is gently forked. The head depth is about one-third of SL, and the head is deeper than it is long. The following description applies to the holotype, with the paratypes agreeing in details where they can be seen unless otherwise noted. Counts and measurements are given in Table 2. Table 2. Counts and measurements (in mm) for the three specimens of the †Lopadichthys colwellae gen. et sp. nov. TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – View Large Table 2. Counts and measurements (in mm) for the three specimens of the †Lopadichthys colwellae gen. et sp. nov. TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – View Large Neurocranium and skull roof The skull is best preserved in the holotype (TMP 2015.011.0001; Fig. 10). The nasal is long and narrow, with the width about one quarter of the length. The frontals are broader posteriorly than at the anterior tip, but with an expansion over the orbit (Figs 10, 11). The frontal is about 2.5–3 times as long as it is wide, with the anterior end angled so that the medial edge extends farther anteriorly than the lateral edge. The sensory canal is enclosed in bone in the frontal, and does not extend onto the parietal. The parietal is rectangular, about twice as wide (mediolaterally) as it is long (antero-posteriorly). The posterolateral corner of the frontal is notched behind the expansion over the orbit to receive the roughly square dermosphenotic. The pterotic is somewhat elongate, and bears a sensory canal that is at least partially open. The supraoccipital is a large bone, extending posteriorly to at least the level of the second vertebral centrum. There is no evidence of an extrascapular bone in the holotype, and the laterally placed temporal fossa is open between the parietal, pterotic, exoccipital and epioccipital bones (Fig. 10). However, a large irregularly shaped extrascapular bone covers this region in paratype TMP 2015.011.0004 (Fig. 11). Figure 10. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 10. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 11. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 11. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. The bones of the posterior braincase are partially visible, but appear displaced in the holotype (TMP 2015.011.0001). Although the limits of the bones in this area are difficult to interpret, we identify an exoccipital and epioccipital; we believe an intercalar was also present (Fig. 10) but no details can be discerned. In the dorsal part of the orbit, the orbitosphenoid and pterosphenoid are visible in paratype TMP 2015.011.0004 (Fig. 11). The parasphenoid is visible in the orbit, and bears large caniniform teeth along its visible length (Figs 10, 11). The level of the parasphenoid is significantly more ventral compared to the anterior end of the vertebral column which would articulate with the basioccipital. In this regard, this fish is similar to Hiodon, and we suggest it would have had a very tall posterior process to articulate with the basioccipital. A small square bone in the holotype and a roughly square bone in paratype TMP 2015.011.0004, both preserved anterior to the anterolateral corner of the frontal, are interpreted as lateral ethmoids. The difference in shape is attributed to preservational orientation of the bone. Jaws, branchial arches and suspensorium The jaws are relatively shorter and deeper in †L. colwellae compared to those of †J. tanyourus, but bear similar large caniniform teeth. It is uncertain whether or not a single tooth row is present, or multiple rows as in †J. tanyourus. The premaxilla has an anterior rounded articular process and a short ramus. There are five premaxillary teeth preserved in the holotype (TMP 2015.011.0001; Fig. 10), and four in the paratype (TMP 2015.011.0004; Fig. 11). The maxilla has a small, rounded anterior end, but the majority of the bone is deeper in lateral view. Neither specimen with a head preserves the maxilla completely, but the impression of the bone is preserved in TMP 2015.011.0004 (Fig. 11), although few teeth are present. No supramaxillary bones are present. The dentary (Figs 10, 11) bears seven teeth, and abruptly widens, posterior to the teeth, into the coronoid process that forms a distinct angle to the toothed portion of the bone. The angular and articular appear to be indistinguishably fused in lateral view, but a separate retroarticular is present at the ventral edge of that element. The triangular quadrate (Fig. 10) has an angle of only about 90°, and the notch for the symplectic is quite narrow. The hyomandibula has a notch in the dorsal edge (Figs 10, 11), giving it two articular heads. The process for articulation with the opercle is robust, and there is a robust anterior process (visible in the holotype TMP 2015.011.0001; Fig. 10). The symplectic is only visible in the holotype, and is not completely preserved; it is a fairly narrow bone. In paratype TMP 2015.011.0004 (Fig. 11), the anterior ceratohyal and the hypohyals are preserved. The anterior ceratohyal is quite short and deep. Remains of 12 branchiostegal rays are visible in this specimen; probably at least eight or nine belong to the left side, and the others possibly belong to the right side. In the holotype (Fig. 10), there are ten branchiostegal rays preserved and all appear to belong to the right side. Basibranchial teeth are visible below the right dentary in the holotype (Fig. 10); these are larger than the jaw teeth, and strongly curved. Opercular series The opercle is preserved mainly as an impression (Figs 10, 11). It has rounded dorsal and posterior margins, and the anterior and ventral edges of the opercle are also rounded. The opercle of the holotype (Fig. 10) is ornamented with striations that radiate from the articular facet for the hyomandibular process. The opercle is twice as high (dorsoventrally) as it is wide (antero-posteriorly). The preopercle is best preserved in the holotype (Fig. 10). The ventral limb is much more robust than the dorsal limb, but the bone is crushed and details are not clear. The subopercle has a slender portion visible under the posteroventral edge of the opercle (holotype TMP 2015.011.0001; Fig. 10). Infraorbital region The infraorbital bones are not well preserved. Remains of the first and second are present in the holotype (TMP 2015.011.0001; Fig. 10) The first is longer than deep, and the second is deeper and shorter than the first. The sensory canal of both appears to be enclosed in a bony tube. Remains of the presumed fourth infraorbital are found in paratype TMP 2015.011.0004; this bone would have been larger than the first and second, and covers the cheek to the anterior edge of the hyomandibula. The dermosphenotic (infraorbital 5) fits into the notch in the posterolateral corner of the frontal. There are no antorbital or supraorbital bones. Vertebral column and predorsal bones There are 45–47 centra, including the two ural centra. There may be an additional third ural centrum not included in the counts. The holotype (TMP 2015.011.0001) has 26 caudal and 19 abdominal centra, and the complete paratype (TMP 2015.011.0004) has 25 caudal and 22 abdominal centra. The vertebral centra are not well preserved in TMP 2015.011.0005. In the other two specimens, the anteriormost vertebral centra have multiple small pits ornamenting the bone, giving it a lacy appearance (Figs 10, 11). Further posterior in the column, the centra have fewer, larger pits. Neural arches are autogenous in the anterior centra. Parapophyses (Fig. 12), which can be seen on two of the anterior abdominal centra in TMP 2015.011.0001 (centra 6–7), are fused to the centra, and have a broad base (visible on centrum 7) and a thin dorsal edge (visible on centrum 6), similar to centra that Neuman & Brinkman (2005) referred to †Coriops. Also as in †Coriops, the rib articulates with the centrum posterior to the parapophysis. The number of ribs is difficult to determine, but there appear to have been 18 (TMP 2015.011.0004) or 20 (TMP 2015.011.0001) pairs. There are 13 long, thin, predorsal bones in both the holotype (TMP 2015.011.0001) and in TMP 2015.011.0004. Figure 12. View largeDownload slide Comparison of centra of †Lopadichthys and †Coriops. †Lopadichthys colwellae gen. et sp. nov., holotype TMP2015.011.0001, photographs and interpretive drawings of centra from the anterior (A), middle (B) and posterior (C) parts of the vertebral column. Three centra assigned to †Coriops in lateral (left) and anterior (right) views, UCMP 230710/V72210 (D), UCMP 276784/V77128 (E) TMP 1986.22.43 (F). Scale bars = 2 mm. Figure 12. View largeDownload slide Comparison of centra of †Lopadichthys and †Coriops. †Lopadichthys colwellae gen. et sp. nov., holotype TMP2015.011.0001, photographs and interpretive drawings of centra from the anterior (A), middle (B) and posterior (C) parts of the vertebral column. Three centra assigned to †Coriops in lateral (left) and anterior (right) views, UCMP 230710/V72210 (D), UCMP 276784/V77128 (E) TMP 1986.22.43 (F). Scale bars = 2 mm. Paired fins and girdles Little of the pectoral girdle can be seen. The dorsal and ventral limbs of the cleithrum form almost a 90° angle to one another, curving under the opercle (Fig. 11). Remains of the coracoid indicate that it had a robust, rounded dorsal edge. The dorsal arm of the post-temporal is twice as long as the ventral arm (Fig. 10); the dorsal arm broadens distally. There are 14 pectoral fin rays preserved in paratype TMP 2015.011.0004. The pelvic fin is positioned closer to the origin of the anal fin than it is to the pectoral girdle. It contains six rays. Dorsal and anal fins The dorsal fin is preserved in all specimens and has three procurrent rays and 26–27 principal rays; these are supported by at least 21 pterygiophores in the holotype (TMP 2015.011.0001) and at least 23 in paratype TMP 2015.011.0005. The anal fin has three (TMP 2015.011.0004) or four (holotype) procurrent rays and 22 (both paratypes) or 23 (holotype) principal rays. There are probably 23 pterygiophores supporting the anal fin. Caudal skeleton and fin The caudal fin and skeleton are well preserved on the holotype (TMP 2015.011.0001; Fig. 13) and both paratypes (Figs 14, 15). The fin is slightly forked with 18 principal rays (16 branched), seven dorsal procurrent and about ten ventral procurrent rays, giving a fin formula of vi,I,8,8,I,x in the holotype. The ventral procurrent rays in TMP 2015.11.0005 are not all clear, and this specimen has a formula of vi,I,8,8,I,v+. The parhypural is autogenous and is preserved slightly separated from the first preural centrum (pu1) in the holotype (Fig. 13), but articulates on that centrum in the paratypes (Figs 14, 15). The first ural centrum (u1) supports the first two hypurals (hy1 and hy2). The second ural centrum (u2) bears four hypurals, for a total of six hypurals in the caudal fin. All six hypurals are autogenous, and the upper four are roughly triangular with narrow proximal ends, while the first two are only slightly narrower at their proximal ends compared to the distal ends. The second preural centrum (pu2) has a single neural spine, but pu1 bears two neural spines in the holotype; whether there are two neural spines on pu1 in TMP 2015.011.0005 is difficult to see in the fossil (and we have figured it with a single neural spine; Fig. 15), but the silicone peel indicates there may have been two present. There is a single neural spine on pu1 in TMP 2015.011.0004 (Fig. 14). Figure 13. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 13. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 14. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 14. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 15. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. Figure 15. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. There are three elements between the posterior neural spine of pu1 and the sixth hypural in the holotype (Fig. 13) and also in TMP 2015.011.0004 (Fig. 14); the first two elements we initially identified as being one epural and one uroneural. The identification of epurals and uroneurals causes significant differences in the phylogenetic analysis (see Discussion below), and based on the phylogenetic analysis, it may be more reasonable that these are both uroneurals. The third element is unidentified, but could be a third ural centrum. This area of the caudal fin is less clear in the third specimen, TMP 2015.011.0005 (Fig. 15), but based on the silicone peel, there are four elements preserved, which we identify as two uroneurals and one epural, or three uroneurals, and a long procurrent ray. Scales Large, roughly circular cycloid scales with many fine circuli are visible on the specimens, but the edges of each are not clear. There are no reticulations on the scales. Referred specimen from Smoky Tower When Li & Wilson (1996a) described the species †J. symmetropterus based on material from the Joffre Bridge road cut (UALVP locality 56), they also referred an incomplete specimen (UALVP 34770) from the Smoky Tower locality of the Palaeocene Paskapoo Formation to the same genus but as an indeterminate species. The dentary of UALVP 34770 is relatively deeper than that of †J. symmetropterus and †J. tanyourus and the symphysis is relatively larger. In the proportions and the relative number and size of teeth, the dentary of UALVP 34770 is strikingly similar to the dentary of †Lopadichthys (Fig. 16). In addition, UALVP 34770 preserves almost complete dorsal and anal fins, and the base of the tail, which reveal that the caudal peduncle was quite short and deep, as in †Lopadichthys; the number of fin rays preserved does not contradict an affinity with †Lopadichthys. Based on the differences between UALVP 34770 and †Joffrichthys, and the similarities of UALVP 34770 with †Lopadichthys, we here refer the Smoky Tower specimen to †Lopadichthys, cf. †L. colwellae. An isolated dentary from the Joffre Bridge road cut locality, specimen UALVP 15069, was referred to †J. symmetropterus by Li & Wilson (1996a); we here reassign that specimen to †Lopadichthys. Figure 16. View largeDownload slide Comparative photographs of the lower jaws of several Palaeocene osteoglossomorphs. †Joffrichthys symmetropterus, holotype, UALVP 23705 (A), †Joffrichthys tanyourus sp. nov., holotype, TMP2015.11.0003 (B), †Lopadichthys colwellae gen. et sp. nov., paratype TMP2015.011.0004 (C), UALVP 34770, referred to †Joffrichthys sp., by Li & Wilson (1996), but here reassigned to †Lopadichthys cf. L. colwellae (D), UALVP 15069 assigned to †J. symmetropterus by Li & Wilson (1996), here reassigned to †Lopadichthys cf. L. colwellae (E), †Lopadichthys colwellae gen. et sp. nov. holotype TMP2015.011.0001 (F). Figure 16. View largeDownload slide Comparative photographs of the lower jaws of several Palaeocene osteoglossomorphs. †Joffrichthys symmetropterus, holotype, UALVP 23705 (A), †Joffrichthys tanyourus sp. nov., holotype, TMP2015.11.0003 (B), †Lopadichthys colwellae gen. et sp. nov., paratype TMP2015.011.0004 (C), UALVP 34770, referred to †Joffrichthys sp., by Li & Wilson (1996), but here reassigned to †Lopadichthys cf. L. colwellae (D), UALVP 15069 assigned to †J. symmetropterus by Li & Wilson (1996), here reassigned to †Lopadichthys cf. L. colwellae (E), †Lopadichthys colwellae gen. et sp. nov. holotype TMP2015.011.0001 (F). Phylogenetic analysis During the past two decades, a number of authors have been studying the phylogenetic relationships of the Osteoglossomorpha (e.g. Li et al., 1997; Hilton, 2003; Murray & Wilson, 2005; Zhang, 2006). These authors have been sharing a data set and adding and modifying it with new data from new fossil finds. The most recent iteration of the data set is that of Murray et al. (2016), who used the combined data sets of Hilton (2003) and Li et al. (1997) with additions and modifications as listed in Wilson & Murray (2008), and an additional taxon from Murray et al. (2010). For our first analysis, we used the data matrix of Murray et al. (2016) with modifications as noted in the Material and Methods, and further detailed below. The complete character list with character states can be found in Wilson & Murray (2008; Appendix 2) with many of the alternate character states figured and discussed in Hilton (2003). Hilton (2003) used a single outgroup, Elops, in his analysis. Li et al. (1997) included two outgroups, Clupeioidei and the ichthyodectiform †Cladocyclus gardneri. Our first analysis included all three of these taxa as has been done in previous analyses (e.g. Murray et al., 2016). For our second analysis, we removed †C. gardneri from the matrix, and replaced the coding of Clupeioidei from Li et al. (1997) with data for members of the two orders of Clupeomorpha, †Ellimmichthyiformes and Clupeiformes. Because osteoglossomorphs are basal teleosts, we also added data for Amia, as a non-teleost outgroup. The complete data matrix with our modifications to coding and taxa is given in Appendix 1. Modified coding and characters for the phylogenetic analysis Caudal fin ray counts: The overall similarity between †J. symmetropterus specimens and the new Paskapoo †J. tanyourus specimens led us to re-examine the former to determine why the caudal fins in these two species had different ray counts, with the former originally reported as having 15 branched principal rays (formula i,7,8,i) and the new material having 16 branched caudal fin rays (i,8,8,i). Re-examination of the holotype of †J. symmetropterus (UALVP 23705) and a second specimen preserving the tail (UALVP 31545) shows that a count of the caudal fin rays is difficult; however, we count eight branched rays in both the dorsal and ventral lobes of the fin, giving a formula of i,8,8,i (16 branched rays in total), not seven branched rays in the dorsal lobe as reported by Li & Wilson (1996a). Newbrey & Bozek (2000) also had difficulty counting caudal fin rays in the second described species of †Joffrichthys, †J. triangulpterus, because the upper lobe of the fin was incomplete. However, they considered this species to have a caudal fin ray formula of i,?,8,i [this is our reinterpretation of the formulae given in their table (2000: tab. 1) in which they have a formula of i+8+?+i for †J. triangulpterus and i+8 + 7+i for both †J. symmetropterus and †C. strickeri, indicating they have listed the ventral lobe of the fin first and the dorsal lobe second, as was done by Li & Wilson (1997), whereas we list the dorsal lobe first, following most other authors, e.g. Schultze & Arratia (1989)]. We modified the character state for number of caudal fin rays in †J. symmetropterus to reflect our reassessment. We also reassessed a species of presumed Osteoglossidae, †C. mahengeense, as it had conflicting characters: it has an elongate unbranched ray in the pectoral fin (otherwise found only in some species of the Osteoglossidae), but was also reported to have a caudal fin formula of i,8,8,i (Murray & Wilson, 2005), which is not found in osteoglossids. The caudal fin in this species is only preserved in two specimens, and in one of those (WM 378/96) only the ventral lobe of the fin is visible. Re-examination of a latex peel of the second specimen (WM 492/96) shows that the previously reported caudal formula is probably incorrect; it is more likely that there are only 15 branched principal rays in this species (i,7,8,i). We changed the coding for character 65 from state 1 (18 principal rays) to state 2 (17 or fewer principal rays) to reflect this. Epurals and uroneurals: In †Lopadichthys (TMP 2015.011.0001), there are two elements between the posterior neural spine of the first preural centrum and the sixth hypural that we identified in the description above as one epural and one uroneural; however, they may both be uroneurals, depending on the interpretation of the researcher. Taverne (1977, 1978) provided detailed osteological studies of a number of fossil and living osteoglossomorphs. Based on his examination of the caudal skeletons, he concluded that there was a single uroneural, no urodermals and no epural present in Osteoglossum, Scleropages, Heterotis, Arapaima, †Brychaetus, Pantodon and the notopterids. Hilton (2003), based on developmental evidence in Arapaima, determined that this taxon has a paired uroneural that has fused in the midline and no epural. Whether this is the case for all osteoglossomorphs cannot be determined without ontogenetic series for all the fossil forms. Taverne (1977) showed that Hiodon and †Eohiodon have both a single epural and three uroneurals. He further considered †Lycoptera to have a single epural, five uroneurals and a seventh element that could be either an epural or uroneural. Li & Wilson (1996a: fig. 5) in their drawing identified an element between the neural spine of the first ural centrum and hypural six as an epural in †J. symmetropterus. They listed two specimens for their interpretation of the caudal fin, UALVP 23705b (the holotype counterpart impression) and UALVP 37128b. We can see no element preserved in this position in either the part or counterpart of the holotype (UALVP 23705 a and b). In UALVP 37128 a and b, the caudal skeleton is not well preserved, and we cannot see the limits of individual bones and scales in this area. However, in another specimen, UALVP 31545b, which has a break right through the caudal fin skeleton, there is a partial thin element dorsal to the second ural centrum, which could be interpreted as an epural. Taverne (1978) considered †Singida to have a single epural and two uroneurals, but this was contradicted by Murray & Wilson (2005) based on better preserved material that showed †Singida to have two uroneurals, and no epurals or urodermals; the uroneurals are not fused in the midline, but clearly overlap the sides of the centra (Murray & Wilson, 2005). At least one well-preserved fossil taxon, †Shuleichthuys brachypteryx Murray et al., 2010, has both paired uroneurals (four) as well as a median element identified as an epural (Murray et al., 2010). The lack of developmental series in the many fossil species means we cannot use developmental data to determine what these elements are. As shown by the fossil †Shuleichthys as well as the extant Hiodon (Hilton, 2002), osteoglossomorphs may have both uroneurals and epurals present, so we cannot establish a priori what these elements may be in other fossil osteoglossomorphs. We are therefore left with a positional argument to determine what each element is, and thus we identify a uroneural as an element that is either clearly paired, or closely associated with the dorsal surface of the ural centra, whereas an epural is a median element positioned more dorsally in the caudal skeleton. The identification of the median element as an epural or uroneural causes significant changes in the phylogenetic analysis, but so does excluding the character of number of epurals (character 68). The single change of identifying †Lopadichthys as having two uroneurals compared to having one uroneural and one epural shifted several taxa in the analysis as described below. Similarly, excluding the character completely also changed the resultant tree significantly. Coverage of the cheek by infraorbitals 3 and 4: Hilton (2003) considered two of the characters used by Li & Wilson (1996b), their numbers 16 and 20, to be very similar: character 16 is the pterygoquadrate area posteroventral to the orbit being completely covered or not by the infraorbitals; character 20 is presence of a ‘cheek wall’ formed by the enlargement of the first through third infraorbital bones. Hilton (2003) considered both characters to refer to the enlargement of the posterior infraorbital bones (his character 25). We interpret the two characters slightly differently, and so retain them both. For the first character (our number 25; coverage of the pterygoquadrate area), we consider this character to be complete coverage if the posterior infraorbital bones cover all the area up to the anterior edge of the preopercle. For the second character (our number 81), we consider the cheek wall to be present if there is significant coverage by the infraorbitals, even if they do not meet the anterior edge of the preopercle. In the case of the new species of †Joffrichthys, we code a cheek wall as present, but full coverage of the pterygoquadrate area as absent. Modified parapophysis of first centrum: Forey & Hilton (2010) noted a modification to the first centrum in several osteoglossids that might be a useful synapomorphy for the group. They showed that in Heterotis niloticus (Cuvier, 1829), Arapaima gigas (Schinz, 1822), Osteoglossum bicirrhosum (Cuvier, 1829) and †Phareodus encaustus (Cope, 1871), the parapophysis of the first centrum is expanded or hypertrophied to reach under the occiput. We examined latex peels of the fossil osteoglossids †S. jacksonoides and †C. mahengeense to see if we could determine the condition of the bones in these fishes. For the most part, this area in these fossils is obscured by the overlying opercle. However, in one specimen of †S. jacksonoides (WM 315/96), we believe there is an expanded parapophysis on the first centrum that lies under the basioccipital (Fig. 17), somewhat similar to that figured for †Phareodus (Forey & Hilton, 2010: fig. 1D). This character has been added to the matrix and coded for those taxa in which it can be determined. In †J. symmetropterus, one specimen (the holotype, UALVP 23705) has the basioccipital visible; there is no evidence of a parapophysis extending from the first vertebra to the basioccipital. Figure 17. View largeDownload slide Photograph and interpretive drawing of the basioccipital and first centrum of †Singida jacksonoides latex peel of WM 315/96. Scale bar = 2 mm. Figure 17. View largeDownload slide Photograph and interpretive drawing of the basioccipital and first centrum of †Singida jacksonoides latex peel of WM 315/96. Scale bar = 2 mm. Recoding of †Joffrichthys symmetropterus: Based on our re-examination of the type material of †J. symmetropterus, we have changed the coding for 29 characters for this species as follows (indicated by the character number, followed in brackets by the old code and an arrow to the new code): 3(0➔1); 5(?➔0); 6(2➔1); 9(1➔?); 11(?➔1); 13(?➔1); 16(?➔0); 23(?➔1); 25(1➔0); 27(?➔1); 32(2➔?); 33(0➔1); 37(0➔1); 39(1➔0); 42(?➔2); 43(?➔1); 57(n/a➔0); 59(?➔0); 60(?➔0); 63(?➔0); 65(2➔1); 68(?➔2); 71(1➔?); 73(?➔2); 79(0 + 2➔0); 80(?➔0); 81(?➔1); 84(?➔1); 87(1➔0). Recoding of Elops: We recoded ten of the characters for Elops based on information in Ridewood (1904), Taverne (1974) and Schultze & Arratia (1989), which allowed us to provide character states for some characters that were previously missing data. The following changes were made (indicated by the character number, followed in brackets by the old code and an arrow to the new code): 1(0➔1); 11(0➔1); 20(0➔1); 37(0➔1); 80(?➔0); 81(?➔1); 82(?➔1); 84(?➔2); 86(1➔0); 87(2➔0). Exclusion of †Ostariostoma from Osteoglossomorpha †Ostariostoma wilseyi (Ostariostomidae) is represented by a single specimen from deposits in Montana that are either Late Cretaceous or Palaeocene in age (Schaeffer, 1949). When this fish was first described, Schaeffer (1949) noted the difficulty in determining its relationships or taxonomic identity, and placed it questionably with clupeomorph fishes (as Order?Isospondyli, Suborder?Clupeoidea). Grande & Cavender (1991) re-examined the holotype after additional preparation of the specimen; they also had a difficult time trying to determine its relationships. Eventually, they suggested it might be affiliated with osteoglossomorph fish partly because †Ostariostoma has a hiodontid-like anal fin, which is long and has a somewhat falcate posterior margin and anterior rays somewhat thicker than the posterior rays (as found in male Hiodon), as well as there being 18 principal caudal fin rays and no supramaxillae in †Ostariostoma and osteoglossomorphs. †Ostariostoma has large teeth on the jaws that are osteoglossomorph-like, but it lacks teeth on the parasphenoid, which is unlike osteoglossomorphs, and there is no evidence of teeth on the basibranchial bones, indicating this fish lacks the ‘tongue bite’ apparatus of osteoglossomorphs (see Hilton, 2001, for a discussion of this feature). The caudal fin of †Ostariostoma has a reported fin ray formula similar to hiodontiforms (i,8,8,i) and two ural centra as is common in osteoglossomorphs; however, these features may also be found separately in other fishes [e.g. two ural centra in Esox and Umbra (Fujita, 1990); 16 branched principal caudal fin rays in Polymixia, Percopsiformes and diplomystid catfishes (Nelson, 2006)]. Additionally, Wilson & Murray (2008) noted that the caudal fin ray count of †Ostariostoma is problematic; they noted this specimen has 18 principal rays but only 15 of them are branched with two lower unbranched rays, giving a formula of i,8,7,ii. As they noted, this may be an individual abnormality, but without additional specimens this cannot be determined. Taverne (1998) suggested †Ostariostoma should be considered an osteoglossomorph based on seven shared characters (numbered as in his publication): 2) the pars palatine being cartilaginous in the adult; 3) the maxilla loses the strong articulation or retains only a very thin connection with the autopalatine that is found in the other primitive teleosts; 4) loss of the supraorbital; 5) the temporal fossa shifts anterolaterally, often accompanied by hypertrophy of the fossa; 6) the loss of the anterior supramaxilla; 10) a reduction in the number of epurals to one or none; 12) a reduction in the number of principal caudal fin rays to 18 (16 branched). Taverne (1998) additionally noted that †Ostariostoma also has lost the posterior supramaxilla, has strong development of the opercle, and the subopercle is reduced but still elongated as in Osteoglossomorpha. Taverne (1998) further likened †Ostariosoma to Hiodontidae based on his characters 13 and 14: 13) first infraorbital elongated with loss of a large part or even the whole of its dermal bone component, becoming very narrow and mostly limited to its sensory canal tube; 14) hypertrophy of the temporal fossa which moves from its position on the back of the neurocranium to become placed completely on the lateral side of the skull; the fossa is no longer delimited by the exoccipital, epioccipital (= epiotic of Taverne, 1998) and pterotic, but instead by the parietal, epioccipital and pterotic; the fossa remains isolated from the cerebral cavity by a cartilaginous wing which serves as its base. Grande & Cavender (1991) also noted the similarity of †Ostariostoma to osteoglossomorphs based on the lack of supramaxillary bones and 18 principal rays in the caudal fin. However, the single known specimen of †Ostariostoma does not show the palatine (bringing characters 2 and 3 of Taverne (1998) into question), its antorbital is elongate but the dermal portion is not greatly reduced, the post-temporal fossa is laterally placed and bordered by the parietal, epiotic and post-temporal, but is not visibly enlarged, and the caudal fin is poorly preserved with an unidentified element that is potentially a second epural. The subopercle is narrow but elongate as noted by Taverne (1998). Schaeffer (1949) noted that the name †Ostariostoma was not meant to suggest an affinity with the Ostariophysi. Grande & Cavender (1991) also excluded †Ostariostoma from having any ostariophysan affinities because it shows no modification of the anterior vertebral elements. However, we suggest that it may be a basal member of the group in which the Weberian apparatus was at an early stage of development and modifications were not apparent in an articulated specimen as is the case in gonorynchiforms, where the first two centra are ribless and lack parapophyses. In some laterally preserved ostariophysans, the bones of the opercular and pectoral region are crushed on top of the anterior vertebrae concealing any modifications of these elements (e.g. Murray, 2003). Grande & Cavender (1991) actually did indicate that the first neural spine of †Ostariostoma shows some modification; in their figure (1991: fig. 2) the spine appears to be bifurcated posteriorly. The long, narrow body shape of Ostariostoma differs from the deep, football-shaped body of known Cretaceous and Paleogene osteoglossomorphs and is similar to that of gonorhynchiforms. A cast of the holotype of †O. wilseyi in the University of Alberta collections (UALVP 52610), as well as figures in Grande & Cavender (1991) show the vertebral centra to be longer than high, with a length twice their depth, then becoming slightly shorter and deeper for the last four preural centra in the column, with the exception of the two ural centra which are also about twice as long as they are high. The ribs either insert directly on the abdominal centra without parapophyses or fully cover the parapophyses so they are not visible in the articulated specimen. Additionally, and unlike osteoglossomorphs, †Ostariostoma apparently lacks intermuscular bones, and has only a few supraneural bones (shown in Grande & Cavender, 1991: figs. 1 and 2; not visible in our cast). We consider the vertebral column to be more like that of a gonorhynchiform fish rather than an osteoglossomorph fish (see also discussion in Brinkman et al., 2017). While we cannot conclusively determine the relationships of †Ostariostoma, we removed it from one of our analyses of Osteoglossomorpha to determine the effect this taxon may have on the analysis. Phylogenetic results First analysis The first analysis, using the data set from Murray et al. (2016) with the modified characters for Elops, †Joffrichthys and †Chauliopareion, and the additional character from Forey & Hilton (2010) (as discussed in the Material and Methods), resulted in six trees. The tree statistics from PAUP are: tree length (TL) = 242, consistency index (CI) = 0.434, retention index (RI) = 0.630 and rescaled consistency index (RC) = 0.273. The trees differ in the relative positions of the outgroups, Clupeioidei and Elops, and the relative positions of †Xixiaichthys, †Joffrichthys and †Wilsonichthys. The strict and 50% majority rule consensus trees are identical (Fig. 18). Figure 18. View largeDownload slide Consensus tree of six most parsimonious trees resulting from analysis 1. Both the strict consensus and 50% majority rule consensus are the same. Figure 18. View largeDownload slide Consensus tree of six most parsimonious trees resulting from analysis 1. Both the strict consensus and 50% majority rule consensus are the same. In all the trees, the clade †Paralycoptera + †Tanolepis is excluded from the Osteoglossomorpha. †Joffrichthys is recovered basal to a clade including the Osteoglossidae, notopterids, mormyroids and several other taxa, and in three trees †Joffrichthys forms the sister group to †Xixiaichthys. †Sinoglossus is placed as the sister group to Notopteridae in all trees. †Lopadichthys, the new genus from the Paskapoo Formation, is recovered in the same position in all trees, as the basal member of the clade uniting Notopteridae, †Sinoglossus, Mormyroidea, †Palaeonotopterus and †Ostariostoma. Second analysis using new outgroups The second phylogenetic analysis, using the matrix with the corrections to the data as in the first analysis, but with new outgroup taxa (†Ellimmichthyiformes, Clupeiformes and Amia) and including †Ostariostoma, resulted in three shortest trees of length 292 (Fig. 19). Tree statistics from PAUP are: TL = 292, CI = 0.404, RI = 0.594 and RC = 0.240. In all three trees, the clade †Paralycoptera + †Tanolepis is excluded from the Osteoglossomorpha, as in the first analysis. Figure 19. View largeDownload slide The three most parsimonious cladograms resulting from analysis 2. Figure 19. View largeDownload slide The three most parsimonious cladograms resulting from analysis 2. The three trees differ in the placement of †Lycoptera and the relationship of Hiodon + †Eohiodon to †Shuleichthys and †Wilsonichthys. In one tree (Fig. 19B), †Lycoptera is united in a clade with the other four taxa, †Shuleichthys is sister to †Wilsonichthys, and the two together form the sister group of Hiodon + †Eohiodon, with †Lycoptera the basal member of this group. In the other two trees, †Lycoptera is basal to all osteoglossomorphs except the †Paralycoptera + †Tanolepis clade, and †Wilsonichthys is either the sister group to †Shuleichthys (Fig. 19C), or is closer to the rest of the osteoglossomorphs than is †Shuleichthys (Fig. 19A). In both these trees, the Hiodon + †Eohiodon clade is more basal than †Shuleichthys and †Wilsonichthys. In all three trees, †Lopadichthys is recovered as the basal member of a clade including †Ostariostoma and the highly derived notopterids, and mormyroids; this clade is the same as recovered in the first analysis with the exclusion of †Sinoglossus, which in this second analysis is recovered as an osteoglossid, sister group to Heterotis. Two of the trees (Fig. 19A, C) are quite similar, with the only difference being whether †Shuleichthys and †Wilsonichthys were united as sister groups, or placed as successive lineages with †Shuleichthys more basal. The third tree is more different in that †Lycoptera, †Eohiodon + Hiodon and †Shuleichthys + †Wilsonichthys form a clade that is positioned between Elops and the Clupeomorpha, forming a polyphyletic Osteoglossomorpha even if †Paralycoptera and †Tanolepis are excluded from the superorder. Although the placement of the basal taxa in these three trees is not consistent, the more derived taxa are stable, and the relationships found are fairly similar to those of Murray et al. (2016), with the notable exception that †Joffrichthys is excluded from the Osteoglossidae, as would be expected based on our reinterpretation of the caudal skeleton. Third analysis excluding †Ostariostoma The third analysis was run with the corrected data matrix and the new outgroups but excluding †Ostariostoma from the analysis, on the consideration that it is not an osteoglossomorph (see above). This resulted in a single tree of 290 steps (Fig. 20). Tree statistics from PAUP are: TL = 290 steps, CI = 0.403, RI = 0.588 and RC = 0.237. The exclusion of †Ostariostoma resulted in the Clupeomorpha (†Ellimmichthyiformes + Clupeiformes) and Elops being embedded among the osteoglossomorph taxa, giving a polyphyletic Osteoglossomorpha. †Lopadichthys is recovered as the sister group of Elops, †Sinoglossus is removed from being a derived osteoglossid as in the second analysis (grouped with Arapaima and Heterotis) and instead is pushed basally to a position between Amia (the designated non-teleost outgroup) and all the other taxa (Elops, Clupeomorpha and all other osteoglossomorphs). The rest of the taxa are positioned fairly similarly to the previous analysis. Figure 20. View largeDownload slide The single most parsimonious cladogram resulting from analysis 3. Figure 20. View largeDownload slide The single most parsimonious cladogram resulting from analysis 3. Effect of the identification of epurals and uroneurals Although we initially identified a single epural and a single uroneural being present in the caudal fin of the holotype of †Lopadichthys (character 68, state 1), the difficulty in distinguishing epurals and uroneurals led us to examine the effect this character would have on the resulting tree if we had coded this taxon as state 2 (epurals absent), instead identifying the bones as two uroneurals. The character state for the number of uroneurals (character 66, state 1) did not change, as the state is ‘two or one.’ Changing character 68 from state 1 to 2 only in †Lopadichthys, with the rest of the data as in analysis 2, resulted in eight most parsimonious trees (TL = 291, CI = 0.405, RI = 0.0.596, RC = 0.242); in the consensus tree (Fig. 21), compared to analysis 2 (Fig. 19), Elops and †Lycoptera both move basally, †Shuleichthys and †Wilsonichthys form a clade with Hiodon and †Eohiodon (as in one tree in analysis 2), †Ostariostoma unites with the Clupeomorpha and †Xixiaichthys becomes the sister group to †Joffrichthys (Fig. 21). Figure 21. View largeDownload slide Majority (50%) rule consensus of eight most parsimonious trees resulting from analysis 2 with the state of character 68 (number of epurals) changed from one present to epurals absent for †Lopadichthys gen. nov. Figure 21. View largeDownload slide Majority (50%) rule consensus of eight most parsimonious trees resulting from analysis 2 with the state of character 68 (number of epurals) changed from one present to epurals absent for †Lopadichthys gen. nov. Changing character 68 from state 1 to 2, only in †Lopadichthys, with the rest of the data as in analysis 3 (excluding †Ostariostoma), resulted in a single tree (TL = 290, CI = 0.403, RI = 0.587, RC = 0.237), that is identical to the single tree found in analysis 3 (Fig. 20). Therefore, the coding of this character for this taxon makes no difference to the resulting tree if †Ostariostoma is not included in the analysis. Based on the differences in results for the second analysis, removing the character of number of epurals from the analysis might be warranted. However, if we use the data for analysis 3 (excluding †Ostariostoma) and exclude character 68, the result is 18 most parsimonious trees (TL = 283, CI = 0.406, RI = 0.592, RC = 0.241) which differ from the original analysis 3 by having Elops moving basally to take its expected position (as a basal teleost) between Amia and all the other taxa, †Sinoglossus going crownwards to become sister to Heterotis (as found in analysis 2), †Lopadichthys moving up the tree to be placed with the mormyroids and notopterids (as found in analysis 2) and †Shuleichthys leaving its sister position with †Wilsonichthys and instead forming a polytomy with (Hiodon + †Eohiodon) and (†Wilsonichthys + Clupeomorpha) (Fig. 22). Figure 22. View largeDownload slide Majority (50%) rule consensus of 18 most parsimonious trees resulting from analysis (†Ostariostoma excluded) with character 68 (number of epurals) excluded from the analysis. Figure 22. View largeDownload slide Majority (50%) rule consensus of 18 most parsimonious trees resulting from analysis (†Ostariostoma excluded) with character 68 (number of epurals) excluded from the analysis. DISCUSSION Phylogeny The results of the phylogenetic analyses reported here, in particular the polyphyletic nature of the Osteoglossomorpha found as a result of recoding of characters and inclusion of different outgroups, demonstrate that we do not yet have a robust phylogeny for this group. Much of the problem with the analyses of this group is that many characters that are considered useful for living osteoglossomorphs are rarely observable in the fossil members, such as the form of the basibranchial bones and associated toothplates. An assessment of the characters causing the different trees in the different analyses reveals that there is very little data contributing to some of the relationships found. For example, there is no unique character to support the inclusion of †Lopadichthys in a clade with Notopteridae, Mormyroidea and †Palaeonotopterus (analyses 1 and 2), nor is there any unique character to support the sister-group relationship of †Lopadichthys with Elops (analysis 3). However, all these four taxa share with †Lopadichthys the condition of character 1 (state 1), in which the temporal fossa is present and the exoccipital contributes to its border. †Ostariostoma is also part of this clade in analyses 1 and 2, but does not have this condition of the temporal fossa. However, it shares with mormyroids and notopterids the condition of the infraorbital canals having the sensory canal open in a gutter (character 24, state 1) rather than enclosed in a bony canal; †Lopadichthys does not have this condition of the sensory canal. Forey & Hilton (2010: appendix 1) listed and assessed the characters proposed as synapomorphies for the Osteoglossidae and subgroups within that family. The removal of †Joffrichthys from Osteoglossidae removes one of the previously anomalous characters for this genus, as Forey & Hilton (2010) had noted that †Joffrichthys lacked the reduced subopercle of Osteoglossidae. The phylogenetic relationships of the Osteoglossomorpha, using more or less the same morphological character data set, have been studied by numerous authors (e.g. Li et al., 1997; Hilton, 2003; Zhang, 2006; Murray & Wilson, 2005; Wilson & Murray, 2008; Murray et al., 2010). In the last iteration (Murray et al., 2016), there was increased resolution of the resultant tree compared to the previous analyses, but no great difference in placement of the taxa within the tree. However, a combined molecular and morphological analysis by Lavoué (2016), using an earlier iteration (from Murray et al., 2010) of the same morphological data set as used here, found some different relationships among taxa; in particular, Pantodon was excluded from the Osteoglossidae, and instead became part of a tricotomy with †Ostariostoma and all other Osteoglossiformes (Lavoué, 2016: fig. 2). As noted by Lavoué (2016), this placement of Pantodon has implications for the fossil genus †Singida, which has been placed close to Pantodon in some analyses. We here recovered Pantodon as sister to †Singida in our third analysis when †Ostariostoma was excluded (Fig. 20), but in the other two analyses, both of which included †Ostariostoma, Pantodon was recovered with Arapaima and Heterotis with or without †Sinoglossus (Figs 18, 19). The changes that result in the analyses with the recoding of a single character (number of epurals) demonstrate the difficulty of determining the phylogenetic relationships of fossil osteoglossomorphs. Without a developmental series, it is essentially impossible to determine the homologies among the taxa for the various elements in the caudal fin. Although this is somewhat discouraging, we can instead look at the crownwards part of the tree, and note that in this part of the tree the taxa remain relatively much more stable. It seems that it is the older, particularly the Cretaceous, material that is the most problematic. Exclusion of †Joffrichthys from Osteoglossiformes Li & Wilson (1996a) placed †Joffrichthys with the fossil genus †Sinoglossus and the recent genera Arapaima and Heterotis in the subfamily Heterotidinae, of the family Osteoglossidae. They justified the placement of †Joffrichthys within Osteoglossomorpha based on the presence of large teeth on the parasphenoid and basihyal, lack of a supraorbital and supramaxillary bones, and number of branched principal caudal fin rays being 16 or 15. We agree with these characters and †Joffrichthys is clearly an osteoglossomorph fish. However, Li & Wilson (1996a) further placed the genus within Osteoglossiformes, which is not supported by our examination of the specimens and our analysis. They based the ordinal designation of †Joffrichthys on the reduction of the number of hypurals to six, an oval or sub-semicircular shape of the opercle, absence of the uroneural, a developed neural spine on the first ural centrum and only 17 principal caudal fin rays (15 branched). While we agree with three features (the number of hypurals being 6, the opercle having an oval shape and the presence of a neural spine on the first ural centrum), the two others are problematic. Hilton (2003) discussed the identification of epurals and uroneurals in osteoglossomorphs including †Joffrichthys. As he noted, more developmental and histological studies of recent material are needed to determine the identity of the one or two bones lying dorsal to the ural centra in these fish, but they are likely uroneurals, as previously indicated by Taverne (1977, 1978) for many genera, with the second uroneural becoming almost indistinguishably fused with the dorsalmost hypural, at least in Arapaima (Taverne, 1977; Hilton, 2003) and Heterotis (Taverne, 1977). We followed Hilton (2003) in recognizing the single bone dorsal to the ural centra in †J. symmetropterus as a uroneural, contra Li & Wilson (1996a) who identified it as an epural. Wilson & Murray (2008) followed Hilton’s (2003) bone identities, which did not change the relationships of †J. symmetropterus from those found by Li & Wilson (1996a). The character of number of branched principal rays in the caudal fin is the primary reason we remove †Joffrichthys from the Osteoglossiformes. Sixteen branched principal rays in the caudal fin (18 total principal rays) have been accepted by authors as a synapomorphy of Osteoglossomorpha (e.g. Patterson & Rosen, 1977; Hilton, 2003). A reduction to 15 branched principal rays (giving a caudal formula of i,7,8,i) was given as a synapomorphy for Osteoglossiformes (Li & Wilson, 1996a) and is found only in Osteoglossidae and Notopteridae (Hilton, 2003; Wilson & Murray, 2008). Hiodontiforms, mormyrids and basal osteoglossomorphs, including the fossil genera †Shuleichthys (Murray et al., 2010) and †Wilsonichthys (Murray et al., 2016), retain 16 branched principal rays. If †Joffrichthys is not an osteoglossid, the synapomorphies it was thought to share with heterotidines would then be homoplasies or symplesiomorphies. These features are the length of the maxilla being relatively short, a relatively long ventral arm of the preopercle, a large trapezoidal second infraorbital, an anal fin as large as or only slightly smaller than the dorsal fin, and rounded dorsal and anal fins (Li & Wilson, 1996a). The last two characters were homoplastic among the relatively few taxa sampled in the analysis of Li & Wilson (1996a). The first two characters (proportions of the maxilla compared to the mandible and proportions of the two limbs of the preopercle) are the same in the non-osteoglossid †W. aridinsulensis as they are in †J. symmetropterus (DB Brinkman and AM Murray, pers. obs.), so are likely homoplasies as well. The final character, shape of the second infraorbital, is subjective. We consider there to be as much variation among the heterotidines (Li & Wilson, 1996a: figs. 3, 9) as there are similarities with other osteoglossomorphs (e.g. Hilton, 2003: fig. 21). Therefore, we do not consider any of these characters to be more important than the caudal fin ray number in determining the position of †Joffrichthys within Osteoglossomorpha. This is supported in all our phylogenetic analyses, in which †Joffrichthys is not included with the Osteoglossiformes. The caudal skeleton of Osteoglossomorpha The caudal skeleton of fishes is considered to provide a number of phylogenetically useful characters for higher groups within Teleostei (e.g. Schultze & Arratia, 1989), such as the number of principal caudal fin rays, as well as the number of hypurals, epurals and uroneurals, and the presence or absence of a full neural spine on the first preural and first ural centra. The caudal skeleton of osteoglossomorph fishes has been examined by a number of researchers who have discovered that there is significant variation within this group, as well as disagreement in the identification of elements (e.g. Greenwood, 1966; Schultze & Arratia, 1989; Hilton & Britz, 2010). The identification of epurals and uroneurals (see above) clearly can cause changes in the analysis. But the variation in other features might also cause problems. Although caudal fin ray counts are used to support the order Osteoglossiformes, there is variation in this character. Within the genus Hiodon, the caudal fin ray count varies greatly (Schultze & Arratia, 1989; Hilton & Britz, 2010), and within Osteoglossiformes, some taxa have the primitive caudal fin formula of 16 branched rays (Hilton & Britz, 2010). As noted by Hilton & Britz (2010), a simple count of fin rays may not be useful as a homologous character. Quite recently, Taverne (2016b) reported a Palaeocene osteoglossomorph tail skeleton from Angola, assigned to †Ridewoodichthys caheni. He assigned the specimen to Osteoglossidae, based on the caudal fin skeleton being similar to that of several osteoglossids including †Joffrichthys. If our exclusion of †Joffrichthys from Osteoglossiformes (and, therefore, also from Osteoglossidae) is correct, it would indicate that the caudal characters used by Taverne (2016b) to support inclusion of †Ridewoodichthys in Osteoglossidae are more widely distributed than previously thought, and therefore not indicative of osteoglossid relationships. Taverne (2016a) delineated several trends in the evolution of the caudal skeleton for osteoglossomorphs, including the reduction in numbers of hypurals, epurals and uroneurals, and increased fusion between hypurals and ural centra. However, osteoglossomorphs appear to show a mosaic of patterns in these elements with some taxa that are considered more highly derived displaying a mix of characters considered primitive and derived. For example, the living osteoglossid Scleropages has a caudal fin with advanced features such as fusion of the dorsal hypurals with each other and the second ural centrum, a single uroneural and no epurals, but also retains primitive features such as a full neural spine on the first preural centrum (e.g. Taverne, 1977; Zhang & Wilson, 2017). Hilton & Britz (2010) discuss the variation in the neural spine form and number on the first preural centrum, as well as variation in other characters. Taverne (2016b) suggested there were two separate evolutionary lineages of the caudal skeleton within Osteoglossiformes, one leading to modern osteoglossids, and a second leading to the Eocene African genera †Chauliopareion and †Singida. However, Taverne’s (2016b) interpretation of the caudal skeletons of these two African taxa differs from that of Murray & Wilson (2005) in both numbers of hypurals (Taverne indicated five for both taxa; Murray and Wilson were uncertain but suggested there may have been six in each) and uroneurals (Taverne indicated two uroneurals in each, but Murray and Wilson could not determine the number in †Chauliopareion). There is also confusion caused by the identification of elements, such as the epurals and uroneurals as noted above. Although much progress has been made on understanding the developmental history of the caudal skeleton in living osteoglossomorphs, better preserved fossils, such as the Cretaceous osteoglossomorph †Shuleichthys, which clearly preserves the hypurals, uroneurals, epurals and ural centra (Murray et al., 2010) are needed to help resolve the evolution of the osteoglossomorph caudal skeleton. Fossil record of osteoglossomorphs in North America Osteoglossomorph fishes that have been diagnosed on the basis of articulated specimens from Cretaceous and Palaeocene deposits of North America have all previously been included in the Osteoglossiformes or have been placed incertae sedis within the superorder. Articulated (and therefore diagnosable) members of the Hiodontiformes only appear in the fossil record in the Eocene. Fossil taxa previously considered to be osteoglossiforms from North America are the Cretaceous †Cretophareodus from Alberta, the Palaeocene †Joffrichthys, now with three species, two from Alberta and the other from North Dakota, and the Eocene †Phareodus (Green River deposits of the USA). Taxa that have been placed incertae sedis within the superorder are Chandlerichthys from the middle Cretaceous of Alaska, †Wilsonichthys from the late Maastrichtian of Alberta, as well as †Lopadichthys described here. The only species of North American fossil hiodontiforms that have been recognized include two or three species of †Eohiodon (depending on whether †E. falcatus is considered distinct from †E. woodroofi or not), and †Hiodon consteniorum (Hilton & Grande, 2008). In the analyses using new outgroups, the single tree resulting from our third analysis (excluding †Ostariostoma) and one tree from the second analysis (including †Ostariostoma) grouped the Chinese †Shuleichthys and the Canadian †Wilsonichthys in a clade with Hiodon + †Eohiodon. This result indicates that some of the taxa previously placed incertae sedis within the superorder may actually be better placed within Hiodontiformes. Although the current data are not strong enough to support this, if correct these taxa would provide part of the missing (Cretaceous) record for the order. Data on the early history of osteoglossomorphs in the Cretaceous and Palaeogene of North America provided by articulated specimens are supplemented by isolated elements from vertebrate microfossil sites. In addition to adding to our understanding of the diversity of osteoglossomorphs, this microfossil material helps to document more fully the stratigraphic and geographic distribution of the group. In some cases, the isolated osteoglossomorph elements can be referred to taxa based on comparison with the same elements in articulated specimens. In others, they indicate the presence of previously unrecognized fishes. One of the osteoglossomorph taxa from the Late Cretaceous of the Western Interior represented by articulated specimens, †Wilsonichthys, is also represented by isolated elements from vertebrate microfossil localities preserved in the same beds as the articulated specimens. These elements differ significantly from those of the two new Palaeocene taxa described above. Dentaries of †Wilsonichthys differ from those of †J. tanyourus and †Lopadichthys (Figs 16, 23) in being deeper and having a much deeper symphysis. Also, the sensory canal pores are located midway on the side of the dentary, rather than near the base, and the teeth are smaller and blunter. Abdominal centra differ from those of †Lopadichthys in having autogenous parapophyses. The rib articulates with the parapophysis, rather than in a pit on the centrum posterior to the parapophysis. Isolated elements from vertebrate microfossil localities also document the presence of otherwise unknown members of Osteoglossomorpha. Hiodontids were recognized from the Cretaceous of the Western Interior of North America on the basis of centra (Brinkman & Neuman, 2002; Brinkman et al., 2013, 2014). These were referred to Hiodontidae on the basis of apomorphic features of the anteriormost centrum, as well as a similarity in the general morphology of the more posterior abdominal centra. Tooth-bearing elements of hiodontids have not yet been identified, although a dentary from the late Campanian Dinosaur Park Formation with an arrangement of teeth similar to that of extant Hiodon may represent this family (Fig. 23F). Significantly, the proportions and arrangement of teeth on this dentary are also similar to †J. tanyourus. Figure 23. View largeDownload slide Comparative photographs of the lower jaws of several Cretaceous osteoglossomorphs. A, †Wilsonichthys aridinsulensis, holotype, TMP 2012.020.1493 from the Maastrichtian Scollard Formation, B, TMP 95.108.61, from Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, C, uncatalogued specimen in the collections of the Museum of Northern Arizona, from the late Turonian Smoky Hollow Member of the Straight Cliffs Formation, Utah, USA, D, †Coriops, TMP 1986.43.33 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, E, TMP 90.119.35A2 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, and F, TMP 2004.19.1 from Dinosaur Park Formation, Onefour, Alberta, here assigned to †Joffrichthys tanyourus sp. nov.. Scale bars = 2 mm. Figure 23. View largeDownload slide Comparative photographs of the lower jaws of several Cretaceous osteoglossomorphs. A, †Wilsonichthys aridinsulensis, holotype, TMP 2012.020.1493 from the Maastrichtian Scollard Formation, B, TMP 95.108.61, from Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, C, uncatalogued specimen in the collections of the Museum of Northern Arizona, from the late Turonian Smoky Hollow Member of the Straight Cliffs Formation, Utah, USA, D, †Coriops, TMP 1986.43.33 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, E, TMP 90.119.35A2 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, and F, TMP 2004.19.1 from Dinosaur Park Formation, Onefour, Alberta, here assigned to †Joffrichthys tanyourus sp. nov.. Scale bars = 2 mm. An additional osteoglossomorph taxon present in Late Cretaceous vertebrate microfossil assemblages, but not known from articulated material, is †Coriops. The genus †Coriops was erected by Estes (1969) on the basis of basibranchial toothplates with blunt crushing teeth. It was initially identified as a member of the Albulidae, which are similar to osteoglossomorphs in having basibranchial toothplates. However, Brinkman & Neuman (2002) recognized that the basibranchials of †Coriops are similar to those of extant osteoglossomorphs such as Scleropages in the presence of a network of bone forming the base of the toothplate; based on this they included †Coriops in the Osteoglossomorpha. Dentaries, vertebrae, basioccipitals and quadrates that occur in the same localities as the basibranchial elements were also referred to †Coriops by Brinkman & Neuman (2002) on the basis of comparison with extant osteoglossomorphs and size-frequency distributions. These elements are similar to those of †J. tanyourus and †Lopadichthys, supporting the identification of these elements as belonging to an osteoglossomorph. The dentaries of †Coriops (Fig. 23) are similar to those of †J. tanyourus (Fig. 16B) in the presence of multiple rows of teeth with the teeth of the lateral row being largest. The tooth row of †Lopadichthys (Fig. 16C, F) is not fully exposed, so it is uncertain whether or not multiple rows of teeth were present in that taxon. The dentaries of †Coriops differ from those of both †Lopadichthys and †J. tanyourus in being deeper and having a larger symphysis (Figs 16, 23). Premaxillae were referred to †Coriops on the basis of the presence of an arrangement of teeth similar to that of the dentary. These are similar to the new osteoglossomorphs described here in having a rounded anterior portion and a more rod-like posterior portion. Abdominal centra of †Coriops are similar to those of †Lopadichthys (Fig. 12) in that the anteriormost centrum lacks parapophyses and ribs are either absent or articulate with the ventralmost edge of the centrum, while in the more posterior abdominal centra long laterally directed parapophyses are present, and the ribs articulated with a large pit on the lateral surface of the centrum posterior to the parapophyses. Also, the neural arches of anterior centra are autogenous, and the dorsal edge of the parapophyses reaches the rib articular pit, and is very narrow in dorsal view. Neuman & Brinkman (2005) noted that the elements referred to †Coriops are similar to those of Hiodon in some features, suggesting a possible relationship between the two. The arrangement of teeth on the dentary of these two genera is similar in that the teeth of the outside row are largest and are set on the laterally facing portion of the jaw, the teeth of the innermost row are slightly smaller, and those of the intermediate rows are the smallest. The centra are similar in that the ribs articulate directly on the centrum posterior to the parapophyses. These similarities support the suggestion made above that at least some of the taxa from the Cretaceous and Palaeocene that were previously placed incertae sedis within the superorder may actually be better placed within Hiodontiformes and thus be part of a hiodontiform radiation. However, it is not possible to include the isolated elements assigned to †Coriops in a phylogenetic analysis, so additional specimens are necessary before the phylogenetic significance of these similarities can be fully evaluated. Distribution of osteoglossomorphs in the Cretaceous and Palaeogene of North America The data on diversity and distribution of osteoglossomorphs in North America based on articulated specimens are supplemented by data from isolated elements, and together these sources of information provide an understanding of the early history of osteoglossomorphs in North America that is more complete than either set of data on its own. The oldest osteoglossomorph in the non-marine deposits of North America is †Chandlerichthys from the mid-Cretaceous of Alaska. Isolated elements from vertebrate microfossil localities also occur in mid-Cretaceous beds in Utah (Brinkman et al., 2013), showing that osteoglossomorphs were widely distributed in North America at this time. Centra from the Cenomanian Cedar Mountain Formation of Utah indicate the presence of two osteoglossomorphs, †Coriops and a generically indeterminate hiodontid. †Coriops extends through the Late Cretaceous into the Palaeocene, although it has a strong latitudinal pattern and so is only seen in southern localities during times of relatively cool global temperatures (Brinkman et al., 2013). It is the most abundant, and one of the largest, teleosts in vertebrate microfossil localities in the late Campanian Belly River Group of Alberta. Hiodontid centra first occur in the Cenomanian Cedar Mountain Formation of Utah and extend to the late Maastrichtian of the Hell Creek Formation, where they are represented by two taxa. Fish centra from microvertebrate sites in the Late Cretaceous Nemget Formation of Mongolia have also been included in the Hiodontidae (Newbrey et al., 2013), providing evidence for interchange of non-marine aquatic vertebrates between Asia and North America in the Cretaceous. Isolated elements referred to†Wilsonichthys by Murray et al. (2016) show that this taxon first appears in the Turonian of Utah. The genus extends to the late Maastrichtian but is not yet known from the Palaeocene. †Cretophareodus is represented by a single articulated specimen, a partial skeleton preserved in an ironstone concretion from exposures of the Belly River Group in Dinosaur Provincial Park, Alberta. Centra are not preserved in this specimen, and the dentaries are incomplete, but based on the preserved portion of the dentaries, †Cretophareodus appears to be taxonomically distinct from †Coriops. No isolated elements have been referred to †Cretophareodus so no additional information on its stratigraphic and geographic distribution is available. Palaeogene osteoglossomorphs represented by articulated specimens include the Palaeocene †Joffrichthys, now with three species, two from Alberta and the other from North Dakota, our new taxon, †Lopadichthys, from the Palaeocene of Alberta, and the Eocene †Phareodus (Green River deposits). Although isolated centra demonstrate the presence of hiodontids in North America in the Cretaceous, no remains of hiodontids have been recovered from the Palaeocene. The previously reported presence of a hiodontid in the Palaeocene, based on isolated elements from the Paskapoo Formation that were referred to Hiodon by Wilson (1980), was later refuted with the understanding that these remains actually represent †Joffrichthys (Li & Wilson, 1996a). Hiodontids are absent in an assemblage of isolated fish elements from a vertebrate microfossil locality of early Eocene age (Divay & Murray, 2016). Thus, the remains of †Eohiodon falcatus from the late early Eocene beds of the Green River Formation provide the first record of the group in the Cenozoic of North America. They are also present in lake beds of middle and late Eocene age of Washington State, USA, and British Columbia, Canada (Li et al., 1997), and are represented by isolated elements from late Eocene–early Oligocene deposits of Saskatchewan (Divay & Murray, 2015, 2016), showing that after their reappearance in the fossil record in the early Eocene, they become widespread in North America. ACKNOWLEDGEMENTS We thank Mr Edgar Nernberg for bringing the fish block to the attention of D.K.Z., and to François Therrien for facilitating the transfer of data. We also thank Allan Lindoe for expert preparation of the specimens. Very helpful reviews from Eric Hilton and an anonymous reviewer allowed us to improve the manuscript. [Version of Record, published online 09 February 2018; http://zoobank.org/ urn:lsid:zoobank.org:pub:43235F0E- EFAF-482E-A578-CFE597B67CF3] REFERENCES Arratia G . 1997 . Basal teleosts and teleostean phylogeny . Palaeo Ichthyologica 7 : 5 – 168 . Berg L . 1940 . Classification of fishes both recent and fossil . Travaux de l’Institut Zoologique de l’Académie des Sciences de l’URSS. Moscow. Translated and reprinted in English, 1947. Ann Arbor, MI : J. W. Edwards . Brinkman DB , Neuman AG . 2002 . Teleost centra from uppermost Judith River Group (Dinosaur Park Formation, Campanian) of Alberta, Canada . Journal of Paleontology 76 : 138 – 155 . Google Scholar CrossRef Search ADS Brinkman DB , Newbrey MG , Neuman AG , Eaton JG . 2013 . Freshwater Osteichthyes from the Cenomanian to late Campanian of Grand Staircase–Escalante National Monument, Utah . In: Titus AL , Loewn MA , eds. At the top of the Grand Staircase: the Late Cretaceous of Southern Utah . Bloomington, IN : Indiana University Press , 195 – 236 . Brinkman DB , Neuman AG , Divay J . 2017 . Non-marine fishes of the late Santonian Milk River Formation of Alberta, Canada – evidence from vertebrate microfossil localities . Vertebrate Anatomy Morphology Palaeontology 3 : 7 – 46 . Brinkman DB , Newbrey MG , Neuman AG . 2014 . Diversity and paleoecology of actinopterygian fish from vertebrate microfossil localities of the Maastrichtian Hell Creek Formation of Montana . Geological Society of America Special Papers 503 : 247 – 270 . Google Scholar CrossRef Search ADS Carrigy MA . 1971 . Lithostratigraphy of the uppermost Cretaceous (Lance) and Paleocene strata of the Alberta Plains . Research Council of Alberta Bulletin 27 : 1 – 161 . Casier E . 1961 . Materiaux pour la faune ichthyologique Eocretacique du Congo . Annales du Musée Royal de l’Afrique Centrale, Série in–8, Sciences Géologiques, Tervuren 39 : 1 – 96 . Casier E . 1965 . Poissons fossiles de la Série du Kwanga (Congo) . Annales du Musée Royale de l’Afrique Centrale, Sciences Géologiques, Tervuren 50 : 1 – 64 . Cope ED . 1871 . On the fishes of the Tertiary shales of Green River, Wyoming Territory . In: Hayde EV , ed. Preliminary Report of the United States Geological Survey of Wyoming . Washington, DC : Government Printing Office , 425 – 431 . Cuvier G . 1829 . Le Règne Animal, distribué d’après son organisation, pour servir de base à l’histoire naturelle des animaux et d’introduction à l’anatomie comparée, 2nd edn . Paris : Déterville . Demchuk TD , Hills LV . 1991 . A re-examination of the Paskapoo Formation in the central Alberta Plains; the designation of three new members . Bulletin of Canadian Petroleum Geology 39 : 270 – 282 . Divay JD , Murray AM . 2015 . The late Eocene–early Oligocene ichthyofauna from the Eastend area of the Cypress Hills Formation, Saskatchewan, Canada . Journal of Vertebrate Paleontology 35 : e956877 . Google Scholar CrossRef Search ADS Divay JD , Murray AM . 2016 . An early Eocene fish fauna from the Bitter Creek area of the Wasatch Formation of southwestern Wyoming, U. S. A . Journal of Vertebrate Paleontology 36 : e1196211 . Google Scholar CrossRef Search ADS Estes R . 1969 . Two new Late Cretaceous fishes from Montana and Wyoming . Breviora 335 : 1 – 15 . Forey PL . 2004 . A three-dimensional skull of a primitive clupeomorph from the Cenomanian English Chalk, and implications for the evolution of the clupeomorph acusticolateralis system . In: Arratia G , Tintori A , eds. Mesozoic fishes 3 – systematics, paleoenvironments and biodiversity . München, Germany : Verlag Dr. Friedrich Pfeil , 405 – 427 . Forey PL , Hilton EJ . 2010 . Two new Tertiary osteoglossid fishes (Teleostei: Osteoglossomorpha) with notes on the history of the family . In: Elliott DK , Maisey JG , Yu X , Miao D , eds. Morphology, phylogeny and paleobiogeography of fossil fishes . München, Germany : Verlag Dr. Friedrich Pfeil , 215 – 246 . Forey PL , Yi L , Patterson C , Davies CE . 2003 . Fossil fishes from the Cenomanian (Upper Cretaceous) of Namoura, Lebanon . Journal of Systematic Palaeontology 1 : 227 – 330 . Google Scholar CrossRef Search ADS Fox RC . 1990 . The succession of Paleocene mammals in western Canada . In: Bown TM , Rose KD , eds. Dawn of the age of mammals in the northern part of the Rocky Mountain interior, North America . Geological Society of America Special Paper 243 : 51 – 70 . Google Scholar CrossRef Search ADS Fujita K . 1990 . The caudal skeleton of teleostean fishes . Tokyo, Japan : Tokai University Press . Grande L . 1985 . Recent and fossil clupeomorph fishes with materials for revision of the subgroups of clupeoids . Bulletin of the American Museum of Natural History 181 : 231 – 372 . Grande L . 1986 . The first articulated freshwater teleost fish from the Cretaceous of North America . Palaeontology 29 : 365 – 371 . Grande L , Bemis WE . 1998 . A comprehensive phylogenetic study of amiid fishes (Amiidae) based on comparative skeletal anatomy. An empirical search for interconnected patterns of natural history . Supplement to the Journal of Vertebrate Paleontology 18 : x + 690 pp . Grande L , Cavender TM . 1991 . Description and phylogenetic reassessment of the monotypic †Ostariostomidae (Teleostei) . Journal of Vertebrate Paleontology 11 : 405 – 416 . Google Scholar CrossRef Search ADS Greenwood PH , Patterson C . 1967 . A fossil osteoglossoid fish from Tanzania (E. Africa) . Zoological Journal of the Linnean Society 47 : 211 – 223 . Google Scholar CrossRef Search ADS Greenwood PH , Rosen DE , Weitzman SH , Myers GS . 1966 . Phyletic studies of teleostean fishes, with a provisional classification of living forms . Bulletin of the American Museum of Natural History 131 : 339 – 455 . Heckel JJ . 1856 . Beitrage zur Kenntniss der fossilen Fische Österreichs . Denkschriften der Kaizerlichen Akademie der Wissenschaften Mathematisch-naturwissenschaftliche Classe Wien 11 : 187 – 274 . Hilton EJ . 2001 . Tongue bite apparatus of osteoglossomorph fishes: variation of a character complex . Copeia 2001 : 372 – 381 . Google Scholar CrossRef Search ADS Hilton EJ . 2002 . Osteology of the extant North American fishes of the genus Hiodon Lesueur, 1818 (Teleostei: Osteoglossomorpha: Hiodontiformes) . Fieldiana 100 : 1 – 142 . Hilton EJ . 2003 . Comparative osteology and phylogenetic systematics of fossil and living bony-tongue fishes (Actinopterygii, Teleostei, Osteoglossomorpha) . Zoological Journal of the Linnean Society 137 : 1 – 100 . Google Scholar CrossRef Search ADS Hilton EJ , Britz R . 2010 . The caudal skeleton of osteoglossomorph fishes, revisited: comparisons, homologies, and characters . In: Nelson JS , Schultze HP , Wilson MVH , eds. Origin and Phylogenetic Interrelationships of Teleosts . München, Germany : Verlag Dr. Friedrich Pfeil , 219 – 237 . Hilton EJ , Grande L . 2008 . Fossil mooneyes (Teleostei: Hiodontiformes, Hiodontidae) from the Eocene of western North America, with a reassessment of their taxonomy . Geological Society Special Publications 295 : 221 – 251 . Google Scholar CrossRef Search ADS Hoffman GL , Stockey RA . 2000 . Geological setting and paleobotany of the Joffre Bridge Roadcut fossil locality (late Paleocene), Red Deer Valley, Alberta . Canadian Journal of Earth Sciences 36 : 2073 – 2084 . Google Scholar CrossRef Search ADS Jerzykiewicz T . 1997 . Stratigraphic framework of the uppermost Cretaceous to Paleocene strata of the Alberta Basin . Geological Survey of Canada Bulletin 510 : 121 . Krause DW . 1978 . Paleocene primates from western Canada . Canadian Journal of Earth Sciences 15 : 1250 – 1271 . Google Scholar CrossRef Search ADS Lavoué S . 2016 . Was Gondwanan breakup the cause of the intercontinental distribution of Osteoglossiformes? A time-calibrated phylogenetic test combining molecular, morphological, and paleontological evidence . Molecular Phylogenetics and Evolution 99 : 34 – 43 . Google Scholar CrossRef Search ADS PubMed Lerbekmo JF , Heaman LM , Baadsgaard H , Muehlenbachs K , Evans ME , Sweet AR . 2008 . Normal polarity magnetosubchrons in 24r and the age of the Paleocene-Eocene boundary . Canadian Journal of Earth Sciences 45 : 781 – 793 . Google Scholar CrossRef Search ADS Lerbekmo JF , Sweet AR . 2000 . Magnetostratigraphy and biostratigraphy of the continental Paleocene in the Calgary area, southwestern Alberta . Bulletin of Canadian Petroleum Geology 48 : 285 – 306 . Google Scholar CrossRef Search ADS Lerbekmo JF , Sweet AR . 2007 . Magnetobiostratigraphy of the continental Paleocene upper Coalspur and Paskapoo formations near Hinton, Alberta . Bulletin of Canadian Petroleum Geology 56 : 118 – 146 . Google Scholar CrossRef Search ADS Li G-Q . 1996 . A new species of Late Cretaceous osteoglossid (Teleostei) from the Oldman Formation of Alberta, Canada, and its phylogenetic relationships . In: Arratia G , Viohl G , eds. Mesozoic fishes – systematics and paleoecology . Munich, Germany : Verlag Dr. Friedrich Pfeil , 285 – 298 . Li G-Q , Wilson MVH . 1996a . The discovery of Heterotidinae (Teleostei: Osteoglossidae) from the Paleocene Paskapoo Formation of Alberta, Canada . Journal of Vertebrate Paleontology 16 : 198 – 209 . Google Scholar CrossRef Search ADS Li G-Q , Wilson MVH . 1996b . Phylogeny of Osteoglossomorpha . In: Stiassny MLJ , Parenti LR , Johnson GD , eds. Interrelationships of fishes . San Diego, CA : Academic Press , 163 – 174 . Li G-Q , Wilson MVH , Grande L . 1997 . Review of Eohiodon (Teleostei: Osteoglossomorpha) from western North America, with a phylogenetic reassessment of Hiodontidae . Journal of Paleontology 71 : 1109 – 1124 . Google Scholar CrossRef Search ADS Linnaeus C . 1766 . Systema naturae sive regna tria naturae, secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis, 12th edn . Holmiae : Laurentii Salvii . Maddison W , Maddison D . 2016 . Mesquite version 3.10 (build 765) software program . Copyright (c) 1997-2015. Müller J . 1845 . Über den Bau und die Grenzen der Ganoiden und über das natürlichen System der Fische . Abhandlungen Akademie der Wissenschaften, Berlin 1844 : 117 – 216 . Murray AM . 1996 . A new Paleocene genus and species of percopsid, †Massamorichthys wilsoni, (Paracanthopterygii) from Joffre Bridge, Alberta, Canada . Journal of Vertebrate Paleontology 16 : 642 – 652 . Google Scholar CrossRef Search ADS Murray AM . 2003 . A new Eocene citharinoid fish (Ostariophysi: Characiformes) from Tanzania . Journal of Vertebrate Paleontology 23 : 501 – 507 . Google Scholar CrossRef Search ADS Murray AM , Newbrey MG , Neuman AG , Brinkman DB . 2016 . New articulated osteoglossomorph from Late Cretaceous freshwater deposits (Maastrichtian, Scollard Formation) of Alberta, Canada . Journal of Vertebrate Paleontology 36 : e1120737 . Google Scholar CrossRef Search ADS Murray AM , Vernygora O , Japundžić S , Radovčić J , Wilson MVH , Bardack D , Grande T . 2017 . Relationships of the species of Armigatus (Clupeomorpha, Ellimmichthyiformes) and the description of a new species from the Cretaceous of Dalmatia, Croatia . Journal of Vertebrate Paleontology 36 : e1226851 . Google Scholar CrossRef Search ADS Murray AM , Wilson MVH . 2005 . Description of a new Eocene osteoglossid fish and additional information on †Singida jacksonoides Greenwood and Patterson, 1967 (Osteoglossomorpha), with an assessment of their phylogenetic relationships . Zoological Journal of the Linnean Society 144 : 213 – 228 . Google Scholar CrossRef Search ADS Murray AM , You H-L , Peng C . 2010 . A new Cretaceous osteoglossomorph fish from Gansu Province, China . Journal of Vertebrate Paleontology 30 : 322 – 332 . Google Scholar CrossRef Search ADS Nelson JS . 2006 . Fishes of the world, 3rd edn . Toronto, Ontario : John Wiley and Sons, Inc . Neuman AG , Brinkman DB . 2005 . Fishes of the fluvial beds . In: Currie PR , Kopplelhus EB , eds. Dinosaur Provincial Park: a spectacular ancient ecosystem revealed . Bloomington, IN : Indiana University Press , 167 – 185 . Newbrey MG , Bozek MA . 2000 . A new species of Joffrichthys (Teleostei: Osteoglossidae) from the Sentinel Butte Formation, (Paleocene) of North Dakota, USA . Journal of Vertebrate Paleontology 20 : 12 – 20 . Google Scholar CrossRef Search ADS Newbrey MG , Brinkman DB , Winkler DA , Freedman EA , Neuman AG , Fowler DW , Woodward HN . 2013 . Teleost centrum and jaw elements from the Upper Cretaceous Nemegt Formation (Campanian-Maastrichtian) of Mongolia and a re-identification of the fish centrum found with the theropod Raptorex kreigsteini . In: Arratia G , Schultze H-P , Wilson MVH , eds. Mesozoic fishes 5 – global diversity and evolution . München, Germany : Verlag Dr. Friedrich Pfeil , 291 – 303 . Patterson C , Rosen DE . 1977 . Review of ichthyodectiform and other Mesozoic teleost fishes and the theory and practice of classifying fossils . Bulletin of the American Museum of Natural History 158 : 83 – 172 . Ridewood WG . 1904 . On the cranial osteology of the fishes of the families Elopidae and Albulidae, with remarks on the morphology of the skull in the lower teleostean fishes generally . Proceedings of the Zoological Society of London 2 : 35 – 81 . Schaeffer B . 1949 . A teleost from the Livingston Formation of Montana . American Museum Novitates 1427 : 1 – 16 . Schinz HR . 1822 . Das Thierreich eingetheilt nach dem Bau der Thiere als Grundlage ihrer Naturgeschichte und der vergleichenden Anatomie . Mit vielen Zusätzenversehen von H. R. Schinz. Cotta, Stuttgart & Tübingen 2: i–xvi + 1–835 [a translation, with emendations, of Cuvier’s “Règne animal”; not seen]. Schultze H-P , Arratia G . 1989 . The composition of the caudal skeleton of teleosts (Actinopterygii: Osteichthyes) . Zoological Journal of the Linnean Society 97 : 189 – 231 . Google Scholar CrossRef Search ADS Sweet AR , Braman DR . 1992 . The K-T boundary and contiguous strata in western Canada, interactions between paleoenvironments and palynological assemblages . Cretaceous Research 13 : 31 – 79 . Google Scholar CrossRef Search ADS Swofford DL . 2002 . PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4.0a152 . Sunderland, MA : Sinauer Associates . Taverne L . 1974 . L’ostéologie d’Elops Linné, C., 1766 (Pisces Elopiformes) et son intérêt phylogénétique . Académie Royale de Belgique, Mémoirs de la Classe des Sciences, Collection in–8–2ieme série 41 : 1 – 96 . Taverne L . 1975 . Sur l’existence d’un poisson ostéoglossoïde fossil proche parent de l’actuel genre Heterotis dans le Crétacé moyen du Kwango (Zaïre) . Revue Zoologie Africain 89 : 964 – 968 . Taverne L . 1976a . Sur l’existence d’un second genre fossile de téléostéen ostéoglossomorphe dans le Cénomanien (Cretacé moyen) du Kwango (Zaïre) . Biologische Jaarboek, Dodonaea 44 : 311 – 317 . Taverne L . 1976b . Les téléostéens fossiles du Crétacé moyen de Kipala (Kwango, Zaire) . Annales Musée royal de l’Afrique Centrale, Série in–8, Sciences Géologiques,Tervuren 79 : 1 – 50 . Taverne L . 1977 . Ostéologie, phylogénèse et systématique des Téléostéens fossiles et actuels du super–ordre des Ostéoglossomorphes. Première partie. Ostéologie des genres Hiodon, Eohiodon, Lycoptera, Osteoglossum, Scleropages, Heterotis et Arapaima . Académie Royale de Belgique, Mémoires de la Classe des Sciences, Collection in–8°–2ieme série 42 : 1 – 234 . Taverne L . 1978 . Ostéologie, phylogénèse et systématique des Téléostéens fossiles et actuels du super–ordre des Ostéoglossomorphes. Deuxième partie. Ostéologie des genres Phareodus, Phareoides, Brychaetus, Musperia, Pantodon, Singida, Notopterus, Xenomystus et Papyrocranus . Académie Royale de Belgique, Mémoires de la Classe des Sciences, Collection in–8°–2ieme série 42 : 1 – 213 . Taverne L . 1979 . Ostéologie, phylogénèse et systématique des téléostéens fossiles et actuels du super–ordre des Ostéoglossomorphes. Troisième partie. Évolution des structures ostéologiques et conclusions générales relatives à la phylogénèse et à la systématique du super–ordre. Addendum . Académie Royale de Belgique, Mémoires de la Classe des Sciences, Collection in–8°–2ieme série 43 : 1 – 168 . Taverne L . 1998 . Les ostéoglossomorphes marins de l’Éocène du Monte Bolca (Italie): Monopteros Volta 1796, Thrissopterus Heckel, 1856 et Foreyichthys Taverne, 1979. Considérations sur la phylogénie des téléostéens ostéoglossomorphes . Museo di Historia Naturale, Verona Studi e Richerch sui Giacimenti Terziari di Bolca 7 : 67 – 158 . Taverne L . 2009 . New insights on the osteology and taxonomy of the osteoglossid fishes Phareodus, Brychaetus and Musperia (Teleostei, Osteoglossomorpha) . Bulletin de l’institut Royal des Sciences Naturelles de Belgique, Sciences de la Terre 79 : 175 – 190 . Taverne L . 2016a . Chanopsis lombardi (Teleostei, Osteoglossiformes) from the continental Lower Cretaceous of the Democratic Republic of Congo. Comments on the evolution of the caudal skeleton within osteoglossiform fishes . Geologica Belgica 19 : 291 – 301 . Google Scholar CrossRef Search ADS Taverne L . 2016b . New data on the osteoglossid fishes (Teleostei, Osteoglossiformes) from the marine Danian (Paleocene) of Landana (Cabinda Enclave, Angola) . Geo-Eco-Trop 40 : 297 – 304 . Volta GS . 1796 . Ittiolitologia Veronese del Museo Bozziano ora annesso a quello del Conte Giovambattista Gazola e di altri Gabinetti di Fossili Veronesi. Dalla Stamperia Giuliari, 2 volumes, 323 pp . [not seen]. Wilson MVH . 1980 . Oldest known Esox (Pisces: Esocidae), part of a new Paleocene teleost fauna from western Canada . Canadian Journal of Earth Sciences 17 : 307 – 312 . Google Scholar CrossRef Search ADS Wilson MVH , Murray AM . 2008 . Osteoglossomorpha: phylogeny, biogeography, and fossil record, and the significance of key African and Chinese fossil taxa . Geological Society Special Publications 295 : 185 – 219 . Google Scholar CrossRef Search ADS Wilson MVH , Williams RRG . 1991 . New Paleocene genus and species of smelt (Teleostei: Osmeridae) from freshwater deposits of the Paskapoo Formation, Alberta, Canada, and comments on osmerid phylogeny . Journal of Vertebrate Paleontology 11 : 434 – 451 . Google Scholar CrossRef Search ADS Zhang J-Y . 2006 . Phylogeny of Osteoglossomorpha . Vertebrata PalAsiatica 44 : 43 – 59 . Zhang J-Y , Wilson MVH . 2017 . First complete fossil Scleropages (Osteoglossomorpha) . Vertebrata Palasiatica 55 : 1 – 23 . APPENDIX 1 Character state matrix used in phylogenetic analysis based on several matrices with modifications as detailed in the Material and Methods section. Characters are listed in order from 1 to 88 by groups of ten. Amia  0?10123100 10000001?1 0010000000 0000000010 021100???2 1000000000  0000021000 0000?01020 01?21020 †Ellimmichthyiformes  3?00010001 1011100001 0?????0000 ?010011000 01???????? 1??0110000  0000010010 101??00?00 ?0?0?00? Clupeiformes  2000010011 1010100001 0010000000 0000011010 0111111??1 0001110100  0000010010 1011?00000 00?1?000 Elops  1000000110 1000000001 0000000100 0000001000 0000000000 0000020000  0000000010 0000?0?000 11?2?000 †Lycoptera  ??00110210 00???????? 000000?0?0 ?000000000 0??1??000? ????0000??  0?0010(0/1)100 002??00100 00?0?02? †Paralycoptera  0?00????00 0????????? ?1???0?0?? ?0001?0?00 0?????0??? ??????0???  ????200?00 10????0?00 01?1???? †Tanolepis  0000?1???0 0????????1 ?10??0???0 ?0000?0?00 0?????0??? ??????0???  ?????00?00 102???0200 01?1???? †Sinoglossus  ?000?2???? 10???????1 ?11??0???? ?000??0?11 0????????? ??????0???  ????2????0 ??2??0?221 1000??1? †Eohiodon  2000111210 10??00???1 010001?200 ?001011001 0001??010? 0???001000  0?001001(0/1)0 001??20100 00?0?12? Hiodon  2000111210 (0/1)000000001 0100011200 1001011001 0001000100 0000001000  01001001(0/1)(0/1) 0011120100 00?00120 †Joffrichthys symmetropterus  ??100132?1 100??0???1 01100010?? ??100?1001 021??????? ????000?00  0?001?1200 ?02??0?100 10?1??00 Heterotis  2220123001 1010100001 0110100111 0000111011 0200211??1 1100010101  0000211200 1220101121 10000021 Arapaima  3200123201 1010101011 0110100111 1000110011 0200201??2 1100000110  0000211200 1220111121 11000011 †Phareodus  ?221013?00 10010????1 010010001? ?210110001 0210??001? ????000??1  1?002?1?00 122??00121 0112?011 Pantodon  2210113200 1011101001 1100100001 1010210101 0100200011 101?001001  1010211200 1120101110 01120000 †Singida  2210103000 00???????1 01001000?? ?2102???01 0?1??????? ????0000?1  1010211200 102??00100 0102?001 Scleropages  3210023301 1110000101 0100100011 (0/1)210111001 0110210011 1000000101  1010211200 1220101021 0112000? Osteoglossum  3210023301 1110000101 0100100011 1210111001 0110210011 1000000101  1010211200 1220101021 01120001 †Ostariostoma  2?000???10 10???????1 0?0100???? ?3000?1?01 01???????? ????0011??  0?00110100 1?2??00100 00?0??0?  Mormyroidea  1000(0/1)022(0/1)0 10(0/1)011(2/3)111 1101022002 1000011111 13?(0/1)21(0/1)??2 (1/2)00100?011  0000110200 1121111200 00001(0/1)2? Notopteridae  11(1/2)0022(1/2)11 10(0/1)1001011 11010(0/2)3001 1200210001 1110110101 10(0/1)?100101  00?1221200 1021111211 00001000 †Palaeonotopterus  1000???210 100?00110? 1?????000? ?????????? ?????????? ????????1?  ?????????? ?????????? ???????0 †Chauliopareion  2010003?01 00???????1 01000000?? ?2101?1011 0?1??????? ????0010?1  10102????0 102??0?1?0 0000?11? †Xixiaichthys  0?100132?? 000??????1 ?1100??00? 0010210001 02????0??? 0???001?0?  0?00001200 001??00100 0002??1? †Shuleichthys  3010010210 100??0?0?1 1???0?100? 0(0/1)0001?0?1 0210??00?0 1???001??0  010010(0/1)100 001??00?00 00?1??2? †Wilsonichthys  2?100?0010 1????????1 1?0??0?10? ?0100?1011 0?????00?? ????0010??  0?00110111 001??00??? ?0?0???? †Joffrichthys tanyourus sp. nov.  20100132?0 100??0???1 0??0001001 02100?1000 021??????? ????000?00  0?00101200 002??00100 10?1??0? †Lopadicththys colwellae gen. et sp. nov.  10100002?0 10???????1 0?000001?? ??00??100? 021??????? ????000?0?  0??0110200 102??00100 10?0??0? Amia  0?10123100 10000001?1 0010000000 0000000010 021100???2 1000000000  0000021000 0000?01020 01?21020 †Ellimmichthyiformes  3?00010001 1011100001 0?????0000 ?010011000 01???????? 1??0110000  0000010010 101??00?00 ?0?0?00? Clupeiformes  2000010011 1010100001 0010000000 0000011010 0111111??1 0001110100  0000010010 1011?00000 00?1?000 Elops  1000000110 1000000001 0000000100 0000001000 0000000000 0000020000  0000000010 0000?0?000 11?2?000 †Lycoptera  ??00110210 00???????? 000000?0?0 ?000000000 0??1??000? ????0000??  0?0010(0/1)100 002??00100 00?0?02? †Paralycoptera  0?00????00 0????????? ?1???0?0?? ?0001?0?00 0?????0??? ??????0???  ????200?00 10????0?00 01?1???? †Tanolepis  0000?1???0 0????????1 ?10??0???0 ?0000?0?00 0?????0??? ??????0???  ?????00?00 102???0200 01?1???? †Sinoglossus  ?000?2???? 10???????1 ?11??0???? ?000??0?11 0????????? ??????0???  ????2????0 ??2??0?221 1000??1? †Eohiodon  2000111210 10??00???1 010001?200 ?001011001 0001??010? 0???001000  0?001001(0/1)0 001??20100 00?0?12? Hiodon  2000111210 (0/1)000000001 0100011200 1001011001 0001000100 0000001000  01001001(0/1)(0/1) 0011120100 00?00120 †Joffrichthys symmetropterus  ??100132?1 100??0???1 01100010?? ??100?1001 021??????? ????000?00  0?001?1200 ?02??0?100 10?1??00 Heterotis  2220123001 1010100001 0110100111 0000111011 0200211??1 1100010101  0000211200 1220101121 10000021 Arapaima  3200123201 1010101011 0110100111 1000110011 0200201??2 1100000110  0000211200 1220111121 11000011 †Phareodus  ?221013?00 10010????1 010010001? ?210110001 0210??001? ????000??1  1?002?1?00 122??00121 0112?011 Pantodon  2210113200 1011101001 1100100001 1010210101 0100200011 101?001001  1010211200 1120101110 01120000 †Singida  2210103000 00???????1 01001000?? ?2102???01 0?1??????? ????0000?1  1010211200 102??00100 0102?001 Scleropages  3210023301 1110000101 0100100011 (0/1)210111001 0110210011 1000000101  1010211200 1220101021 0112000? Osteoglossum  3210023301 1110000101 0100100011 1210111001 0110210011 1000000101  1010211200 1220101021 01120001 †Ostariostoma  2?000???10 10???????1 0?0100???? ?3000?1?01 01???????? ????0011??  0?00110100 1?2??00100 00?0??0?  Mormyroidea  1000(0/1)022(0/1)0 10(0/1)011(2/3)111 1101022002 1000011111 13?(0/1)21(0/1)??2 (1/2)00100?011  0000110200 1121111200 00001(0/1)2? Notopteridae  11(1/2)0022(1/2)11 10(0/1)1001011 11010(0/2)3001 1200210001 1110110101 10(0/1)?100101  00?1221200 1021111211 00001000 †Palaeonotopterus  1000???210 100?00110? 1?????000? ?????????? ?????????? ????????1?  ?????????? ?????????? ???????0 †Chauliopareion  2010003?01 00???????1 01000000?? ?2101?1011 0?1??????? ????0010?1  10102????0 102??0?1?0 0000?11? †Xixiaichthys  0?100132?? 000??????1 ?1100??00? 0010210001 02????0??? 0???001?0?  0?00001200 001??00100 0002??1? †Shuleichthys  3010010210 100??0?0?1 1???0?100? 0(0/1)0001?0?1 0210??00?0 1???001??0  010010(0/1)100 001??00?00 00?1??2? †Wilsonichthys  2?100?0010 1????????1 1?0??0?10? ?0100?1011 0?????00?? ????0010??  0?00110111 001??00??? ?0?0???? †Joffrichthys tanyourus sp. nov.  20100132?0 100??0???1 0??0001001 02100?1000 021??????? ????000?00  0?00101200 002??00100 10?1??0? †Lopadicththys colwellae gen. et sp. nov.  10100002?0 10???????1 0?000001?? ??00??100? 021??????? ????000?0?  0??0110200 102??00100 10?0??0? View Large Amia  0?10123100 10000001?1 0010000000 0000000010 021100???2 1000000000  0000021000 0000?01020 01?21020 †Ellimmichthyiformes  3?00010001 1011100001 0?????0000 ?010011000 01???????? 1??0110000  0000010010 101??00?00 ?0?0?00? Clupeiformes  2000010011 1010100001 0010000000 0000011010 0111111??1 0001110100  0000010010 1011?00000 00?1?000 Elops  1000000110 1000000001 0000000100 0000001000 0000000000 0000020000  0000000010 0000?0?000 11?2?000 †Lycoptera  ??00110210 00???????? 000000?0?0 ?000000000 0??1??000? ????0000??  0?0010(0/1)100 002??00100 00?0?02? †Paralycoptera  0?00????00 0????????? ?1???0?0?? ?0001?0?00 0?????0??? ??????0???  ????200?00 10????0?00 01?1???? †Tanolepis  0000?1???0 0????????1 ?10??0???0 ?0000?0?00 0?????0??? ??????0???  ?????00?00 102???0200 01?1???? †Sinoglossus  ?000?2???? 10???????1 ?11??0???? ?000??0?11 0????????? ??????0???  ????2????0 ??2??0?221 1000??1? †Eohiodon  2000111210 10??00???1 010001?200 ?001011001 0001??010? 0???001000  0?001001(0/1)0 001??20100 00?0?12? Hiodon  2000111210 (0/1)000000001 0100011200 1001011001 0001000100 0000001000  01001001(0/1)(0/1) 0011120100 00?00120 †Joffrichthys symmetropterus  ??100132?1 100??0???1 01100010?? ??100?1001 021??????? ????000?00  0?001?1200 ?02??0?100 10?1??00 Heterotis  2220123001 1010100001 0110100111 0000111011 0200211??1 1100010101  0000211200 1220101121 10000021 Arapaima  3200123201 1010101011 0110100111 1000110011 0200201??2 1100000110  0000211200 1220111121 11000011 †Phareodus  ?221013?00 10010????1 010010001? ?210110001 0210??001? ????000??1  1?002?1?00 122??00121 0112?011 Pantodon  2210113200 1011101001 1100100001 1010210101 0100200011 101?001001  1010211200 1120101110 01120000 †Singida  2210103000 00???????1 01001000?? ?2102???01 0?1??????? ????0000?1  1010211200 102??00100 0102?001 Scleropages  3210023301 1110000101 0100100011 (0/1)210111001 0110210011 1000000101  1010211200 1220101021 0112000? Osteoglossum  3210023301 1110000101 0100100011 1210111001 0110210011 1000000101  1010211200 1220101021 01120001 †Ostariostoma  2?000???10 10???????1 0?0100???? ?3000?1?01 01???????? ????0011??  0?00110100 1?2??00100 00?0??0?  Mormyroidea  1000(0/1)022(0/1)0 10(0/1)011(2/3)111 1101022002 1000011111 13?(0/1)21(0/1)??2 (1/2)00100?011  0000110200 1121111200 00001(0/1)2? Notopteridae  11(1/2)0022(1/2)11 10(0/1)1001011 11010(0/2)3001 1200210001 1110110101 10(0/1)?100101  00?1221200 1021111211 00001000 †Palaeonotopterus  1000???210 100?00110? 1?????000? ?????????? ?????????? ????????1?  ?????????? ?????????? ???????0 †Chauliopareion  2010003?01 00???????1 01000000?? ?2101?1011 0?1??????? ????0010?1  10102????0 102??0?1?0 0000?11? †Xixiaichthys  0?100132?? 000??????1 ?1100??00? 0010210001 02????0??? 0???001?0?  0?00001200 001??00100 0002??1? †Shuleichthys  3010010210 100??0?0?1 1???0?100? 0(0/1)0001?0?1 0210??00?0 1???001??0  010010(0/1)100 001??00?00 00?1??2? †Wilsonichthys  2?100?0010 1????????1 1?0??0?10? ?0100?1011 0?????00?? ????0010??  0?00110111 001??00??? ?0?0???? †Joffrichthys tanyourus sp. nov.  20100132?0 100??0???1 0??0001001 02100?1000 021??????? ????000?00  0?00101200 002??00100 10?1??0? †Lopadicththys colwellae gen. et sp. nov.  10100002?0 10???????1 0?000001?? ??00??100? 021??????? ????000?0?  0??0110200 102??00100 10?0??0? Amia  0?10123100 10000001?1 0010000000 0000000010 021100???2 1000000000  0000021000 0000?01020 01?21020 †Ellimmichthyiformes  3?00010001 1011100001 0?????0000 ?010011000 01???????? 1??0110000  0000010010 101??00?00 ?0?0?00? Clupeiformes  2000010011 1010100001 0010000000 0000011010 0111111??1 0001110100  0000010010 1011?00000 00?1?000 Elops  1000000110 1000000001 0000000100 0000001000 0000000000 0000020000  0000000010 0000?0?000 11?2?000 †Lycoptera  ??00110210 00???????? 000000?0?0 ?000000000 0??1??000? ????0000??  0?0010(0/1)100 002??00100 00?0?02? †Paralycoptera  0?00????00 0????????? ?1???0?0?? ?0001?0?00 0?????0??? ??????0???  ????200?00 10????0?00 01?1???? †Tanolepis  0000?1???0 0????????1 ?10??0???0 ?0000?0?00 0?????0??? ??????0???  ?????00?00 102???0200 01?1???? †Sinoglossus  ?000?2???? 10???????1 ?11??0???? ?000??0?11 0????????? ??????0???  ????2????0 ??2??0?221 1000??1? †Eohiodon  2000111210 10??00???1 010001?200 ?001011001 0001??010? 0???001000  0?001001(0/1)0 001??20100 00?0?12? Hiodon  2000111210 (0/1)000000001 0100011200 1001011001 0001000100 0000001000  01001001(0/1)(0/1) 0011120100 00?00120 †Joffrichthys symmetropterus  ??100132?1 100??0???1 01100010?? ??100?1001 021??????? ????000?00  0?001?1200 ?02??0?100 10?1??00 Heterotis  2220123001 1010100001 0110100111 0000111011 0200211??1 1100010101  0000211200 1220101121 10000021 Arapaima  3200123201 1010101011 0110100111 1000110011 0200201??2 1100000110  0000211200 1220111121 11000011 †Phareodus  ?221013?00 10010????1 010010001? ?210110001 0210??001? ????000??1  1?002?1?00 122??00121 0112?011 Pantodon  2210113200 1011101001 1100100001 1010210101 0100200011 101?001001  1010211200 1120101110 01120000 †Singida  2210103000 00???????1 01001000?? ?2102???01 0?1??????? ????0000?1  1010211200 102??00100 0102?001 Scleropages  3210023301 1110000101 0100100011 (0/1)210111001 0110210011 1000000101  1010211200 1220101021 0112000? Osteoglossum  3210023301 1110000101 0100100011 1210111001 0110210011 1000000101  1010211200 1220101021 01120001 †Ostariostoma  2?000???10 10???????1 0?0100???? ?3000?1?01 01???????? ????0011??  0?00110100 1?2??00100 00?0??0?  Mormyroidea  1000(0/1)022(0/1)0 10(0/1)011(2/3)111 1101022002 1000011111 13?(0/1)21(0/1)??2 (1/2)00100?011  0000110200 1121111200 00001(0/1)2? Notopteridae  11(1/2)0022(1/2)11 10(0/1)1001011 11010(0/2)3001 1200210001 1110110101 10(0/1)?100101  00?1221200 1021111211 00001000 †Palaeonotopterus  1000???210 100?00110? 1?????000? ?????????? ?????????? ????????1?  ?????????? ?????????? ???????0 †Chauliopareion  2010003?01 00???????1 01000000?? ?2101?1011 0?1??????? ????0010?1  10102????0 102??0?1?0 0000?11? †Xixiaichthys  0?100132?? 000??????1 ?1100??00? 0010210001 02????0??? 0???001?0?  0?00001200 001??00100 0002??1? †Shuleichthys  3010010210 100??0?0?1 1???0?100? 0(0/1)0001?0?1 0210??00?0 1???001??0  010010(0/1)100 001??00?00 00?1??2? †Wilsonichthys  2?100?0010 1????????1 1?0??0?10? ?0100?1011 0?????00?? ????0010??  0?00110111 001??00??? ?0?0???? †Joffrichthys tanyourus sp. nov.  20100132?0 100??0???1 0??0001001 02100?1000 021??????? ????000?00  0?00101200 002??00100 10?1??0? †Lopadicththys colwellae gen. et sp. nov.  10100002?0 10???????1 0?000001?? ??00??100? 021??????? ????000?0?  0??0110200 102??00100 10?0??0? View Large APPENDIX 2 Characters used in analysis. The list is from Wilson & Murray (2008), which was based on Hilton (2003) and Li et al. (1997), with the addition of character 88, from Forey & Hilton (2010). 1) Temporal fossa 0) absent 1) present, with the exoccipital making a contribution to the border 2) present, bordered by epioccipital, pterotic and parietal 3) present, bordered by epioccipital, and pterotic 2) Shape of extrascapular 0) expanded 1) reduced and irregularly shaped 2) reduced and tubular 3) Shape of frontal bones 0) anterior margin narrower than posterior margin 1) anterior margin about equal in width to posterior margin 2) anterior margin wider than posterior margin 4) Supraorbital shelf of frontal bone 0) absent 1) present 5) Length of frontal bone 0) over twice as long as parietal 1) less than twice as long as parietal 6) Relationship of nasal bones 0) some part separated by anterior portion of frontals 1) separated only by ethmoid bones 2) meet each other in midline 7) Nasal bones 0) tubular but not curved 1) tubular and strongly curved 2) gutter-like 3) flat and broad 8) Parasphenoid teeth 0) absent 1) small 2) large and found along the length of the parasphenoid 3) large and restricted to the basal portion of the parasphenoid 9) Basipterygoid process 0) absent 1) present 10) Supratemporal commissure passing through the parietals 0) absent 1) present 11) Supraorbital sensory canal 0) ending in parietal 1) ending in frontal 12) Orbitosphenoid 0) present 1) absent 13) Basisphenoid 0) present 1) absent 14) Basioccipital process of the parasphenoid 0) divided 1) median 15) Ventral occipital groove 0) present 1) absent 16) Intercalar 0) present 1) absent 17) Cranial nerve foramen/foramina 0) in the prootic 1) straddling the suture between the prootic and pterosphenoid 2) straddling the suture between the sphenotic and pterosphenoid 3) foramina separate from each other, straddling the suture between the prootic, sphenotic and the pterosphenoid (dorsally) and one straddling the suture between the prootic, pterosphenoid and parasphenoid (ventrally) 18) Suture between the parasphenoid and sphenotic 0) absent 1) present 19) Foramen for cranial nerve VI 0) opens within the prootic bridge 1) opens anterior to the prootic bridge 20) Supraorbital bone 0) present 1) absent 21) Otic and supraorbital sensory canal 0) in bony canals 1) partially or completely in grooves 22) Number of bones in the infraorbital series, not including the dermosphenotic or the antorbital if present 0) five 1) four 23) first infraorbital 0) ventral to orbit 1) anterior and ventral to orbit 24) Condition of the infraorbital sensory canal in at least some infraorbitals 0) enclosed in a bony canal 1) open in a gutter 25) Palatoquadrate area behind and below the orbit 0) not completely covered by the infraorbitals 1) completely covered by infraorbitals 26) Dermosphenotic 0) triangular 1) triradiate 2) tubular 27) Posterior extent of the fossa on the neurocranium for the hyomandibula 0) formed of the pterotic 1) formed of the pterotic and intercalar 2) formed of pterotic and exoccipital 3) formed of exoccipital and intercalar 28) Neurocranial heads of the hyomandibula 0) one head or two heads but continuous 1) two heads, separate 2) two heads, bridged 29) Anterior process (wing) of the hyomandibula that contacts the entopterygoid 0) absent 1) present 30) Bones of palatoquadrate 0) two lateral elements 1) one lateral element 2) one element, laterally and medially 31) Autopalatine bone 0) present 1) absent 32) Preopercular sensory canal 0) opens by pores the entire length of the canal 1) opens by pores ventrally and by a groove dorsally 2) opens by pores dorsally and a groove ventrally 3) opens by a groove the entire length of the canal 33) Opercle depth to width ratio 0) less than two 1) about two or greater than two 34) Posterodorsal spine on the opercle 0) absent 1) present 35) Subopercle bone 0) large and ventral to the opercle 1) small and anterior to the opercle 2) absent 36) Gular bone 0) present 1) absent 37) Ascending process of the premaxilla 0) well developed 1) only slightly developed if at all 38) Premaxillae 0) paired 1) median 39) Posterior portion of maxilla 0) lies on angular 1) lies on dentary 40)Supramaxillae 0) present 1) absent 41) Mandibular canal 0) enclosed in a bony tube 1) open in a groove 42) Posterior bones of the lower jaw 0) angular and retroarticular bones fused 1) angular and articular bones fused 2) all separate 3) all fused 43) Retroarticular bone 0) included in the articulation with the quadrate 1) excluded from the articulation with the quadrate 44) Medial wall of the Meckelian fossa of the lower jaw 0) present 1) absent 45) Bony elements associated with the second ventral gill arch 0) absent 1) present as autogenous elements 2) present as a bony process on the second hypobranchial 46) Toothplates associated with basibranchial 4 0) present 1) absent 47) Basihyal toothplate 0) present 1) absent 48) Basihyal toothplate 0) flat 1) with ventrally directed processes 49) Basibranchial toothplate and basihyal toothplate 0) separate 1) continuous 50) Basihyal 0) present and ossified 1) present and cartilaginous 2) absent 51) Hypohyals 0) two ossified pairs present 1) one ossified pair present 2) one ossified pair present but greatly reduced in size 52) Infrapharyngobranchial 3 0) undivided 1) divided into two elements 53) Infrapharyngobranchial 1 0) present 1) absent 54) Orientation of infrapharyngobranchial 1 0) proximal tip anteriorly directed 1) proximal tip posteriorly directed 55) Abdominal scutes 0) absent 1) present as paired structures 56) Epipleural bone 0) absent 1) only a few bones in anterior caudal region 2) present throughout abdominal and caudal region 57) Dorsal arm of the post-temporal bone 0) less than 1.5 times as long as the ventral arm 1) more than twice as long as the ventral arm 58) Lateral line that pierces the supracleithrum 0) present 1) absent 59) Cleithrum 0) with no or only a slight medial lamina 1) with a broad medial lamina 60) Coracoid fenestra 0) absent 1) present 61) First pectoral fin ray 0) normal 1) greatly expanded 62) Post-pelvic bone 0) absent 1) present 63) Pelvic bone 0) slender 1) possesses a thin deep lamella in dorsoventral plane 64) Posterior end of anal fin 0) separate from caudal fin 1) continuous with caudal fin 65) Number of principal caudal fin rays 0) 19 or more 1) 18 2) 17 or fewer 66) Uroneurals 0) three or more 1) two or one 2) absent 67) Neural spine on ural centrum 1 0) absent or rudimentary 1) one or more 68) Epurals (H68, L9) 0) two or three 1) one 2) absent 69) Neural spine on the first preural centrum 0) complete 1) rudimentary 2) absent 70) Number of neural spines on the second preural centrum 0) one 1) two 71) Number of hypurals 0) seven 1) six or fewer 72) Scales 0) no reticulate furrows 1) both radial and reticulate furrows present 2) reticulate furrows only present over entire scale 73) Pelvic fin ray number 0) more than seven 1) seven 2) six or fewer 74) Swimbladder-ear direct connection 0) absent 1) present 75) Intestine 0) coils to right of oesophagus and stomach 1) coils to left of oesophagus and stomach 76) Opercle shape dorsal to facet for articulation with hyomandibula 0) rounded 1) flattened or truncated 2) flattened with posterior recurved process 77) Upper hypurals and second ural 0) not fused 1) fused 78) Second infraorbital shape and size 0) more or less slender or tubular and small in size 1) triangular or rectangular and smaller then third infraorbital 2) expanded and equivalent in size to or larger than third infraorbital 79) Dorsal fin shape 0) base moderately long, fin triangular or falcate 1) base very short, much shorter than fin height, or fin absent 2) base moderately long to very long, fin with rounded outline anteriorly and posteriorly 80) Posterior rays of dorsal and anal fin 0) shorter than anterior ones 1) longer than or as long as anterior ones 81) ‘Cheek wall’ formed by enlargement of first to third infraorbitals 0) absent 1) present 82) Ventral part of preopercle 0) extending anteriorly to beneath orbit or to level of posterior edge of orbit 1) anteriorly does not reach level of orbit 83) Posterior edge of nasal when it is gutter-like or irregularly subrectangular 0) straight or slightly curved 1) strongly curved and extending backward 84) Angle of jaws 0) anterior to middle vertical line of orbit 1) between middle vertical line and posterior edge of orbit 2) behind orbit 85) Utriculus 0) connected with sacculus and lagena 1) completely separated from sacculus and lagena 86) Anal fin sexual dimorphism 0) absent 1) present 87) Ventral margin of opercle 0) rounded or pointed and narrower than midpoint of opercle 1) curved but not greatly narrowed compared to midpoint of opercle 2) flattened or only very slightly rounded 88) Parapophysis on the first centrum 0) not expanded or hypertrophied 1) expanded or hypertrophied to reach under the occiput © 2018 The Linnean Society of London, Zoological Journal of the Linnean Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Zoological Journal of the Linnean Society Oxford University Press

Two new Palaeocene osteoglossomorphs from Canada, with a reassessment of the relationships of the genus †Joffrichthys, and analysis of diversity from articulated versus microfossil material

Loading next page...
 
/lp/ou_press/two-new-palaeocene-osteoglossomorphs-from-canada-with-a-reassessment-asFsvFFiMp
Publisher
Oxford University Press
Copyright
© 2018 The Linnean Society of London, Zoological Journal of the Linnean Society
ISSN
0024-4082
eISSN
1096-3642
D.O.I.
10.1093/zoolinnean/zlx100
Publisher site
See Article on Publisher Site

Abstract

Abstract A single block containing five articulated osteoglossomorphs was recovered from the Paskapoo Formation of southern Alberta, during development of a residential community in Calgary. Two of the specimens represent a new species of †Joffrichthys, and the other three represent a new genus and species of osteoglossomorph. The discovery of a new species of †Joffrichthys led us to re-examine the type species and to recode many of the characters that have been used in phylogenetic analyses. In particular, we interpret the caudal skeleton of †Joffrichthys to have 16 branched principal rays, not 15, which indicates this genus does not belong in Osteoglossiformes, and removes it from the osteoglossid/heterotidine affiliations previously reported. We assessed the relationships of the two new taxa using a modified data matrix including new outgroups and corrected data, with and without the inclusion of †Ostariostoma. Our results show that †Joffrichthys is a basal member of the superorder, and not a member of the Heterotidinae, but the other new taxon is left incertae sedis in the superorder. We also provide data on the early history of osteoglossomorphs in North America provided by isolated elements from Cretaceous and Palaeocene microfossil sites that complement and supplement that provided by articulated specimens. Cretaceous, Heterotidinae, Hiodontiformes, Joffrichthys, Ostariostoma wilseyi, Osteoglossidae, Osteoglossomorpha INTRODUCTION Osteoglossomorpha is a basal teleostean lineage that may be the most primitive living lineage of Teleostei (e.g. see review in Arratia, 1997). Osteoglossomorph fishes are known from deposits as early as the Early Cretaceous and are still extant today. They have a greater diversity of lineages in the fossil record than they do in the modern fauna, although the modern fauna includes a much larger number of species, particularly within Mormyridae (freshwater elephantfishes). Despite many years of study, the phylogenetic relationships of the superorder are still not well resolved. The oldest fossil osteoglossomorphs are from Early Cretaceous deposits of China (Wilson & Murray, 2008; Murray, You & Peng, 2010). In North America, the oldest known is †Chandlerichthys strickeri Grande, 1986, from the mid-Cretaceous deposits of Alaska. This fish, preserved as a carbon film with few clear details, was allied with the Osteoglossomorpha by Grande (1986) and seems to fit this group well in the details that he could discern. Slightly younger are the Campanian †Cretophareodus alberticus Li, 1996, from the late Campanian Belly River Group of southern Alberta, and †Wilsonichthys aridinsulensis Murray et al., 2016, and a second, as yet undescribed, osteoglossomorph recently recovered from the late Maastrichtian Scollard Formation of central Alberta (Murray et al., 2016). All of these are deep to moderately deep-bodied forms with relatively large dorsal and anal fins, from freshwater deposits. Disarticulated fish elements from the Late Cretaceous vertebrate microfossil localities have also been referred to the Osteoglossomorpha based on comparison with recent and fossil members of the group (Brinkman & Neuman, 2002; Neuman & Brinkman, 2005; Brinkman et al., 2013; Brinkman, Newbrey & Neuman, 2014). These provide additional data on the diversity and distribution of the group, as well as relative abundance within assemblages. Palaeocene osteoglossomorphs in North America are less common than Cretaceous forms. †Ostariostoma wilseyi Schaeffer, 1949, from deposits of Late Cretaceous or early Palaeocene age of the Livingston Formation of Montana, was tentatively identified as a hiodontid; however, isolated elements referred to this taxon bring this identity into question, instead suggesting gonorhychiform relationships (Brinkman, Neuman & Divay, 2017). The only confirmed Palaeocene North American osteoglossomorph species are †Joffrichthys symmetropterus Li & Wilson, 1996a, from the Paskapoo Formation of Alberta, and †J. triangulpterus Newbrey & Bozek, 2000, from the Sentinel Butte Formation of North Dakota. Additionally, an indeterminate species of †Joffrichthys was reported from the Smoky Tower locality in north-central Alberta (Li & Wilson, 1996a). Although a hiodontid was reported by Wilson (1980) on the basis of isolated elements (scales, parasphenoids, maxillae and a partial tail), these remains, along with those Wilson (1980) assigned to Osteoglossidae, were all later referred to †Joffrichthys (Li & Wilson, 1996a). †Joffrichthys has been placed within the subfamily Heterotidinae (Osteoglossidae) since it was first described. This relationship implies a very long history for the subfamily and family, and, based on sister-group relationships, a very long history for many other extant osteoglossomorph lineages. We here describe two new osteoglossomorph taxa from the Paskapoo Formation of Alberta, and assess the relationships of them and the genus †Joffrichthys, based on a reinterpretation of the type species. With the additional data on the morphology of early osteoglossomorphs provided by these specimens, we also evaluate the vertebrate microfossil remains from the Late Cretaceous previously referred to the Osteoglossomorpha and review the data they provide on the diversity and distribution of the group within North America during the Late Cretaceous. Geology The Palaeocene beds of west-central Alberta consist of fluviolacustrine sediments deposited by rivers flowing in an easterly direction (Hoffman & Stockey, 2000). Four lithological units of Palaeocene age in Alberta are currently recognized. These are the upper (Palaeocene) part of the Scollard Formation, upper (Palaeocene) part of the Willow Creek Formation, and the exclusively Palaeocene Paskapoo and Porcupine Hills formations. The Scollard Formation and Willow Creek Formation both contain the Cretaceous/Palaeogene (K/Pg) boundary, which divides the units into Cretaceous and Palaeocene parts (Sweet & Braman, 1992). Although the Paskapoo Formation has produced many articulated fishes, articulated specimens of teleosts have not been recovered from the Palaeocene part of the Scollard and Willow Creek formations. Vertebrate microfossil localities containing isolated fish elements are present, but these have not yet been thoroughly studied. The Paskapoo Formation is widespread in western Alberta, extending from Calgary in the south to the Hinton area north-west of Edmonton. The formation was deposited by low energy fluvial systems under humid conditions, with associated overbank environments including ponds and oxbow lakes (Hoffman & Stockey, 2000). The rivers depositing Paskapoo sediments trended slightly south of east (Carrigy, 1971). Many of the fish from the Paskapoo Formation have been recovered from outcrops (Wilson, 1980). A particularly significant locality is the Joffre Bridge road cut, which preserved a mass death layer containing hundreds of fishes. Actinopterygian fishes recovered from the mass death layer include an amiid, osmerid (†Speirsaenigma lindoei), percopsid (†Massamorichthys wilsoni) and the osteoglossomorph †J. symmetropterus (Wilson & Williams, 1991; Li & Wilson, 1996a; Murray, 1996). The Porcupine Hills Formation was deposited in seasonally dry conditions by rivers that trended slightly east of north. It differs from the Paskapoo Formation in its mineralogical composition and the presence of caliche (Carrigy, 1971). Strata of the Porcupine Hills Formation disconformably overlie those of the early Palaeocene levels of the Scollard Formation in the Calgary region and the Willow Creek Formation farther south (Jerzykiewicz, 1997; Lerbekmo & Sweet, 2000). Both the Porcupine Hills Formation and the Paskapoo Formation are present in Calgary, with the Paskapoo overlying the Porcupine Hills Formation. Three mass death assemblages of gar fish have been recovered from Calgary and area, although it is uncertain whether these are from the Paskapoo Formation or the Porcupine Hills Formation. One of these, TMP 2013.009.0001, a specimen in the Royal Tyrrell Museum, is exceptional in that it preserves the remains of 24 fully articulated three-dimensionally preserved fish. However, articulated teleosts have not previously been recovered from the Palaeocene beds in Calgary or from the Porcupine Hills Formation further south. The material described here came from Calgary and was found during the development of a subdivision on the north edge of the city. The large block contains the remains of five osteoglossomorph fishes (Fig. 1). The block was recovered by Mr Edgar Nernberg of Calgary during a routine excavation of a basement with a backhoe in mid-March 2015. Two weeks later he contacted Dr Darla Zelenitsky at the University of Calgary about the discovery; she then reported it to Dr Francois Therrien at the Royal Tyrrell Museum where the specimen is now curated. Zelenitsky visited the site at the time of the report, although the location where the osteoglossomorph block was found had been fully built over with a home basement. Small blocks and rubble of sandstone similar to that of the osteoglossomorph block were present immediately around the constructed basement but contained no visible fossils. Since the block was recovered from a relatively high elevation in the city, it is interpreted as having come from the upper portion of the Paskapoo Formation. The age of the Paskapoo Formation in Calgary has been estimated by magnetostratigraphic (Lerbekmo & Sweet, 2000, 2007), radiometric (Lerbekmo et al., 2008) and biostratigraphic studies (Krause, 1978; Fox, 1990; Demchuk & Hills, 1991). For the latter, both palynology and mammals indicate that the base of the Paskapoo Formation in the Calgary area is middle Palaeocene and the upper beds are late Palaeocene in age. Thus, the block is assumed to be from the late Palaeocene. Figure 1. View largeDownload slide Photograph of the whole block from the Paskapoo Formation showing the relative positions of the five specimens. †Joffrichthys tanyourus sp. nov., holotype TMP 2015.011.0003 (A) and paratype TMP 2015.011.0002 (B); †Lopadichthys colwellae gen. et sp. nov., holotype TMP 2015.011.0001 (C), paratype TMP 2015.011.0004 (D) and paratype TMP 2015.011.0005 (E). Fish are oriented with the dorsal surface towards the top of the page (B, C), with the dorsal surface towards the bottom of the page (A), with the dorsal surface to the left (D) and with the dorsal surface to the right (E). Scale bar is in centimetres. Figure 1. View largeDownload slide Photograph of the whole block from the Paskapoo Formation showing the relative positions of the five specimens. †Joffrichthys tanyourus sp. nov., holotype TMP 2015.011.0003 (A) and paratype TMP 2015.011.0002 (B); †Lopadichthys colwellae gen. et sp. nov., holotype TMP 2015.011.0001 (C), paratype TMP 2015.011.0004 (D) and paratype TMP 2015.011.0005 (E). Fish are oriented with the dorsal surface towards the top of the page (B, C), with the dorsal surface towards the bottom of the page (A), with the dorsal surface to the left (D) and with the dorsal surface to the right (E). Scale bar is in centimetres. MATERIAL AND METHODS The new osteoglossomorph material described here is preserved on a bedding plane of a single large block held in the collections of the Royal Tyrrell Museum of Palaeontology (TMP), Alberta, Canada. Five fishes are present; three are complete and two are partial specimens, one of which preserves the anteroventral half of the body, and the other preserves the posterior two-thirds of the body. The block containing the five fishes (Fig. 1) with the individual fish specimens designated by the numbers TMP 2015.011.0001, TMP 2015.011.0002, TMP 2015.011.003, TMP 2015.011.0004 and TMP 2015.011.0005. Comparative fossil and extant material examined is from the TMP, the Royal Ontario Museum (ROM), Toronto, Ontario; University of Alberta Museum of Zoology (UAMZ), and University of Alberta Laboratory for Vertebrate Palaeontology (UALVP) Edmonton, Alberta; University of California Museum of Paleontology (UCMP) Berkeley, California; and the National Museum of Tanzania (latex peels of specimens collected by the Wembere Manonga palaeontological expedition, with catalogue numbers prefixed WM). A dagger symbol (†) is used to indicate fossil taxa. The new Paskapoo osteoglossomorph fossil material was prepared by Allan Lindoe, UALVP, using a microscribe and needles. To highlight the relief, specimens were coated with ammonium chloride prior to taking photographs. Silicone peels were made of the specimens and used to confirm some anatomical characters after descriptions were made based on the original fossil material. Comparative material examined †Joffrichthys symmetropterus UALVP 23705 (holotype), UALVP 31545, 37128; †J. triangulpterus UALVP 51921 (cast of Field Museum specimen PF12171a,b as †J. symmetropterus); †Joffrichthys sp. UALVP 34770; †Singida jacksonoides Greenwood & Patterson, 1967, WM 241/96, WM 298/96, WM 314/96, WM 315/96, WM 536/96; †Chauliopareion mahengeense Murray & Wilson, 2005, WM 378/96, WM 492/96; †W. aridinsulensis TMP 2012.020.1493 (holotype), TMP 2012.020.1498; †O. wilseyi Shaeffer, 1949, UALVP 52610 (cast of the holotype, Princeton University Geological Museum specimen PU14728); Amia calva Linnaeus, 1766, UAMZ 1260 and three unnumbered skulls. Phylogenetic analysis The phylogenetic analysis is based on a data set for Osteoglossomorpha that has been used in a number of publications, with a few modifications. The original data matrices are from Li, Wilson & Grande (1997) and Hilton (2003). The characters were assessed for overlap and clarity and combined in a single matrix by Wilson & Murray (2008); the character list we use here is from that publication. Additional taxa were added by Wilson & Murray (2008), as well as Murray et al. (2010; †Shuleichthys brachypteryx from the Cretaceous of China) and Murray et al. (2016; †W. aridinsulensis from the Maastrichtian of Alberta). Caudal fin ray counts for †J. symmetropterus and †C. mahengeense were reassessed and modified after examination of the holotype and referred material for the former and latex peels of the latter (see Discussion). We also modified some of the data for Elops, making changes based on Ridewood (1904), Taverne (1974) and Schultze & Arratia (1989); many of these changes were providing a code for which there was previously a question mark in the data matrix. This corrected data matrix was used for our first analysis, to compare the position of the new taxa based on the previously used data matrix. A number of named fossil taxa have not been included in our, nor a number of other (see Wilson & Murray, 2008), phylogenetic analyses. In particular, African genera such as †Ridewoodichthys Taverne, 2009 (based on disarticulated jaw bones and a caudal skeleton), †Paradercetis Casier, 1965 (based on a partial skull), and †Chanopsis Casier, 1961, have been excluded, but all were placed in the more derived family Osteoglossidae by Taverne (1975, 1976a, 2009, 2016a) and so should not affect the relationships of more basal members of the superorder examined here. We also excluded †Kipalaichthys Casier, 1965, which was considered an osteoglossomorph of uncertain affinities by Taverne (1976a, b) and is not well known for the skull. Several Asian taxa have been excluded; these were not readily available to us but future plans of several researchers include detailed reanalyses of these taxa. Three European genera from Eocene marine deposits of Italy (Taverne, 1998) named for articulated fossil material are also not included in our analyses. Taverne (1998) considered one to belong to the family Osteoglossidae (†Thrissopterus Heckel, 1856), and reported another (†Monopteros Volta, 1796) as lacking the caudal skeleton and both that taxon and Foreyichthys Taverne, 1979 as lacking many skull elements; this makes it difficult to code either for the phylogenetic analysis. In the second analysis, we assessed the taxa used in the previous data matrix and replaced the outgroup taxa, so that we could reduce the amount of missing data. We removed the ichthyodectiform †Cladocyclus, and replaced it with the extant, non-teleost A. calva, which we coded based on the text and figures of Grande & Bemis (1998) and personal examination of skeletal material. With the inclusion of Amia, we changed character 68 (number of epurals) state 0 from ‘two or three’ to ‘two or more’, and we modified character 71 (number of hypurals) state 0 from ‘seven’ to ‘seven or more’ to accommodate the more numerous epurals and hypurals of Amia. We also removed the data for Clupeoidei that were taken from Li et al. (1997), and instead coded for the extinct clupeomorph order †Ellimmichthyiformes based on †Armigatus dalmaticus Murray et al., 2017 and †A. namourensis Forey et al., 2003, with additional data from a three-dimensional skull of †Diplomystus (from Forey, 2004). We also included Clupeiformes, the extant clupeomorph order, based on Dorosoma (data taken from Grande, 1985). †Eohiodon is considered here to be a valid taxon following Murray et al. (2010), and not subsumed into Hiodon as supported by Hilton & Grande (2008). We combined the data for the three mormyrids into a single polymorphic taxon and did the same for the notopterids, following Murray et al. (2010), and also excluded †Ostariostoma from one of the analyses as discussed below. A single new character was also added to the matrix. This character, the expansion of the parapophysis on the first centrum, was noted by Forey & Hilton (2010). Based on their work, character 88 is formulated as: parapophysis on first centrum: 0) not expanded or hypertrophied; 1) expanded or hypertrophied to reach under the occiput. This feature is discussed further in the Phylogenetic Analysis section. The final character state matrix for this modified data set is given in Appendix 1, with a list of the characters and states in Appendix 2. The original and new data matrices were manipulated in Mesquite v. 3.10 build 765 (Maddison & Maddison, 2016) and trees were visualized in the same software. Data were analysed in PAUP v. 4.0a152 (Swofford, 2002). We used parsimony analysis with an heuristic search, using simple stepwise addition and TBR branch swapping, with all characters unordered and unweighted for all analyses. Anatomical abbreviations aa, anguloarticular; ang, angular; art, articular; boc, basioccipital; brst, branchiostegal rays; ch, anterior ceratohyal; c, centrum; cl, cleithrum; cor, coracoid; den, dentary; ds, dermosphenotic; ect, ectopterygoid; end, endopterygoid; ep, epural; epi, epioccipital; es, extrascapular; exo, exoccipital; fr, frontal; hh, hypohyal; hy, hypural; hyo, hyomandibula; ic, intercalar; io, infraorbital; iop, interopercle; l, left; le, lateral ethmoid; m, mesethmoid; met, metapterygoid; mx, maxilla; n, neural arch; na, nasal; naap, neural arch articulation pit; nsp, neural spine; op, opercle; os, orbitosphenoid; pa, parietal; par, parapophysis; pd, predorsal bone; ph, parhypural; pmx, premaxilla; pop, preopercle; ps, parasphenoid; pto, pterotic; pts, pterosphenoid; ptt, post-temporal; pu, preural centrum; q, quadrate; r, right; ra, retroarticular; rap, rib articulation pit; scl, supracleithrum; soc, supraoccipital; sop, subopercle; sym, symplectic; tb, toothed bone of pharyngeal arches; un, uroneural; u, ural centrum. SYSTEMATIC PALAEONTOLOGY Teleostei Müller, 1845 Osteoglossomorpha Greenwood, Rosen, Weitzman, & Myers, 1966 incertae sedis †Joffrichthys Li & Wilson, 1996a Included species: †Joffrichthys symmetropterus Li & Wilson, 1996a; †J. triangulpterus Newbrey & Bozek, 2000. Emended diagnosis: Deep-bodied osteoglossomorph fishes with six hypurals, no epurals, one distinct uroneural, six pelvic fin rays, kidney-shaped opercle ornamented laterally with striations radiating from area of facet for articulation with hyomandibula, and large dorsal and anal fins positioned posteriorly. Differs from Osteoglossidae and Notopteridae by having 18 principal rays (16 branched) in the caudal fin; from Notopteridae by lacking a long anal fin confluent with the caudal fin, and from Mormyridae by lacking the rounded snout, elongate jaw bones and other specializations of those families; from Hiodontiformes by lacking the posteriorly recurved spine on the opercle; and from basal members of the superorder by having the supraorbital sensory canal ending in the frontal not the parietal (which is similar to the more derived Osteoglossidae), and having the dorsal and ventral arms of the post-temporal equal in length (this last feature is probable, but not clearly visible in the type species). †Joffrichthys tanyourus sp. nov. Holotype: TMP 2015.011.0003, a complete fish preserved in right lateral view (Fig. 2). Figure 2. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 2. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Paratype: TMP 2015.011.0002, the anteroventral portion of a fish preserving the head and ventral body, missing the dorsal and caudal regions, preserved in right lateral view (Fig. 3). Figure 3. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. Figure 3. View largeDownload slide Photograph of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. Type locality and age: Paskapoo Formation, Calgary, Alberta, Canada; late Palaeocene in age. Etymology: The specific epithet is for the Greek tany meaning long and oura meaning tail. The two previously described species are named for the shape of their dorsal and anal fins, but we name this third species for having a longer caudal peduncle than the others. Diagnosis: Differs from †J. symmetropterus and †J. triangulpterus by having 26 abdominal vertebrae (compared to 22 in †J. symmetropterus and 23–25 in †J. triangulpterus), falcate dorsal and anal fin (compared to rounded fin margins in †J. symmetropterus and triangular fins in †J. triangulpterus), 11 predorsal (supraneural) bones (compared to 16 in †J. triangulpterus and 18 in †J. symmetropterus). Further differs from †J. triangulpterus by having the pelvic fin positioned near to the centre point between the origins of the pectoral and anal fins rather than closer to the anal fin and having about four centra between the insertion of the first anal and dorsal fin pterygiophores, rather than 8–9 centra. Further differs from †J. symmetropterus by having the caudal peduncle longer than deep instead of deeper than long (unknown in †J. triangulpterus). Description General body form This is a moderately deep-bodied fish, with a terete shape (Figs 2, 3). Both specimens are preserved in lateral view, indicating this species was probably laterally compressed in life. The complete specimen (holotype) is 205 mm in total length, and the paratype (incomplete specimen) would have been slightly smaller. The anal and dorsal fins are positioned posteriorly, with the dorsal fin inserting more anteriorly than the anal fin. Both fins are falcate, with the anterior rays being the longest, then the fin steps down to a shorter depth posterior to the midpoint. The caudal peduncle is longer than it is deep. The head is triangular in lateral view, much deeper posteriorly than at the jaws, and is about one quarter of the standard length (SL) in the holotype. The following description applies to the holotype, with the paratype agreeing in details where they can be seen unless otherwise noted. Counts and measurements are given in Table 1, along with those for specimens of the two other species. Table 1. Counts and measurements (in mm) for the two specimens of †Joffrichthys tanyourus sp. nov., J. triangulpterus (taken from Newbrey & Bozak, 2000) and J. symmetropterus †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 Measurements are in millimetres. ? indicates some uncertainty in the data reported here. *These counts and measurements were made by the authors and some data differ from those reported by Li & Wilson (1996a). †Counts and measurements for this specimen were made by adding numbers from the part and counterpart which preserve different parts of the fish. View Large Table 1. Counts and measurements (in mm) for the two specimens of †Joffrichthys tanyourus sp. nov., J. triangulpterus (taken from Newbrey & Bozak, 2000) and J. symmetropterus †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 †J. tanyourus sp. nov. (TMP) †J. triangulpterus (ND) †J. symmetropterus* (UALVP) 2015.011.0003 2015.011.0002 98-1. 298-1.1 23705 31545 37128† Total length (TL) 205 – – – 108 79 – Standard length (SL) 169 – 58 102 89 73 48 Head length (HL) 40 4.0 21 33 25 22 13 Head depth (HD) 47 5.3 19 33 29 21 16 Body depth (BD) 71 – 30 56 44 32 25 Caudal peduncle length (CPL) 22 – 11 20 10 8 9 Caudal peduncle depth (CPD) 17 – 9 13 10 6 8 Anal fin base length (AFL) 45 – 12 27 33 24 15 Dorsal fin base length (DFL) 58 – 20 32 29 15 Preanal length (PAL) 107 9.8 42 75 54 46 32 Predorsal length (PDL) 84 – 34 60 54 44 29 Prepelvic length (PVL) 73 6.4 34 41 35 21 HL/SL 0.24 – 0.36 0.32 0.28 0.28 0.27 HD/SL 0.28 – 0.33 0.32 0.33 0.27 0.33 HL/HD 0.85 0.75 1.11 1.00 0.86 1.05 0.81 BD/SL 0.42 – 0.52 0.55 0.49 0.41 0.52 CPL/CPD 1.29 – 1.22 1.54 1.00 1.33 1.13 PAL/SL 0.63 – 0.72 0.74 0.61 0.58 0.67 PDL/SL 0.50 – 0.59 0.59 0.61 0.56 0.60 PVL/SL 0.43 – 0.59 0.46 0.44 0.44 Dorsal fin rays ?iv,26 – ii,28 ii,25 ii,25 Dorsal fin pterygiophores ?21 – 29 24 22 ?23 Anal fin rays iii,24 – iv,17 ~21 iii,25 iii,26 iii,24 Anal fin pterygiophores – – 21 20 25 24 ?22 Number of centra between first anal and dorsal fin pterygiophore insertions 4 – 8–9 ?4 4 5 Pectoral fin rays 15 – – 9+ 9+ 14 14 Pelvic fin rays 6 6 – – – ?6 6 Caudal fin rays i88i – – i?8i i88i (?)i88i ??8i Caudal centra (incl. u1 and u2) 26 – 25 25 28 28 ?26 Abdominal centra 26 – 25 23 23 ?22 – Total centra 52 – 50 48 51 50 – Branchiostegal rays 7 8 8 5+ ?7 8 – Pairs of ribs 17 or 18 – 21 19 19 or 20 ?17 – Predorsal bones 11 – 16 16 19 17 ?18 Measurements are in millimetres. ? indicates some uncertainty in the data reported here. *These counts and measurements were made by the authors and some data differ from those reported by Li & Wilson (1996a). †Counts and measurements for this specimen were made by adding numbers from the part and counterpart which preserve different parts of the fish. View Large Neurocranium and skull roof The head is preserved in both TMP 2015.011.0003 (holotype; Fig. 4) and specimen TMP 2015.011.0002 (Fig. 5). A mesethmoid is present but no details can be determined. The nasal is a broad, flat, bone, with a notch and expansion anteriorly (Fig. 4); it is about twice as long as it is wide. The left and right nasal bones are completely separated by the mesethmoid. There appears to be a canal in the nasal extending from the posteromedial corner to the midpoint of the lateral edge. The frontal is roughly trapezoidal, widening gradually posteriorly; there is no distinct anterior expansion as found in †Phareodus and †Brychaetus (anterior supraorbital shelf of Hilton, 2003: fig. 11). The sensory canal on the frontal is enclosed in bone. The frontals are more than twice as long (antero-posterior length) as they are broad (medial-lateral width). The parietals are about one-third as long as the frontals. They are roughly rectangular in shape and do not bear a sensory canal. The supraoccipital crest is high and prolonged posteriorly. There is a large, roughly square, extrascapular bone positioned over the area between the parietal and the pterotic (Fig. 5). The pterotic is a long bone, reaching from the posterior portion of the frontal to the posterior end of the hyomandibular head. It bears an enclosed canal and a dorsal flange in its midpoint (Fig. 5). Figure 4. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 4. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 5. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. Figure 5. View largeDownload slide Photograph and interpretive drawing of the head of †Joffrichthys tanyourus sp. nov. paratype TMP 2015.011.0002. Scale bar = 1 cm. The supraoccipital is large, and angles dorsally above the level of the rest of the skull. Below this area in both specimens are bony remains that we interpret as the basioccipital and epioccipitals, with an additional unidentified bone in the paratype (Fig. 5), that could be a portion of the post-temporal. Ventrally, the parasphenoid bears large caniniform teeth along all of its visible length (Figs 4, 5). Posteriorly, the bone angles dorsally (Fig. 5). Based on the position of the visible part of the parasphenoid, which is ventral compared to the basioccipital condyle and vertebral column, either the posterior portion of the parasphenoid or the anterior portion of the basioccipital, or both, must extend significantly dorsoventrally in order for the two bones to meet one another. Jaws, suspensorium and branchial arches The terminal jaws are robust and large, bearing large caniniform teeth (Figs 4, 5). The articulation between the mandible and quadrate is positioned under the middle of the orbit. The anterior end of the premaxilla is enlarged into a low, rounded, ascending process that extends for about half the length of the bone. The premaxilla bears multiple rows of teeth with the teeth in the outside row being the largest (best seen in TMP 2015.011.0003, holotype, but not visible in the figures). There are six premaxillary teeth in the external row of teeth in the paratype (Fig. 5), and four teeth with several tooth sockets preserved in the holotype (TMP 2015.011.0003; Fig. 4). The maxilla appears deep in the paratype, but quite narrow anteriorly in the holotype; we interpret this as differences in the orientation of the preserved elements and incomplete preservation of the maxilla in the holotype. The anterior head of the maxilla narrows to articulate with the premaxilla, and the posterior end of the bone is gently rounded. A single row of 11 (TMP 2015.011.0002, paratype) or 12 (TMP 2015.011.0003 holotype) teeth is preserved on the maxilla. There are no supramaxillae. The dentary is relatively elongate and slender compared with †J. symmetropterus and has a relatively smaller symphysis. The ventral edge is slightly curved. An enclosed mandibular sensory canal that opens through four or more pores is located near the ventral edge of the bone. Similarly to the premaxilla, multiple rows of teeth are present on the dentary with the outer row of teeth being the largest (best seen in the holotype). The retroarticular is clearly a separate ossification, and is confined to the posteroventral corner of the mandible. The angular and articular appear to be separate from one another, with the articular small and confined on the lateral side of the jaw, forming the facet for the quadrate. The quadrate is fairly broad, with the anterior and posterior edges forming an obtuse angle with one another. The posterior edge is strengthened with a ridge. The hyomandibula has a single dorsal head, with no notch between anterior and posterior points (Figs 4, 5). It is a tall bone, at least twice as tall as it is broad antero-posteriorly. There is a slight anterior flange visible in the paratype, giving the hyomandibula a sinuous anterior edge. The hyomandibula in this specimen is also clearly ornamented with rugose vertical lines. A symplectic is not visible in either specimen. The endopterygoid fills the orbit below the parasphenoid. The ectopterygoid is narrower and slightly angled in the middle (TMP 2015.011.0002, paratype). Although teeth are not visible on the ectopterygoid in the paratype, it bears several conical teeth, at least some of which are significantly smaller than those of the parasphenoid and jaws in the holotype (Fig. 4). In addition, there are three large teeth present in this area in TMP 2015.011.0003 (holotype; Fig. 4) just dorsal to the quadrate and articular; however, based on the size, these teeth are more likely to be from the basihyal or basibranchial bones. Branchiostegal bones are visible in both specimens. There are eight preserved in the holotype and paratype, but whether or not they are from a single side in either specimen is difficult to determine. The branchiostegals are slightly broader more posteriorly in the series, but none are very broad. In the holotype, the anterior ceratohyal is visible; it is fairly short and broad. Opercular series The opercle is the best preserved bone of the series in both the holotype and paratype. It is a rounded bone, shaped somewhat like a kidney bean, with a notch anterodorsally at the level of the facet for articulation with the hyomandibula (Figs 4, 5). There are striations radiating from this point to ornament the lateral surface of the bone. The shape and ornamentation of the opercle are almost identical to those of the other two species of †Joffrichthys. The opercle is about twice as high (dorsoventrally) as it is wide (antero-posteriorly). The preopercle is best preserved in TMP 2015.011.0002 (paratype; Fig. 5). It is deepest in the middle, with much narrower dorsal and anterior ends. The sensory canal appears to be under a flange on the ventral limb, and opens through pores on the dorsal limb. Only the anterior tip of the ventral limb is preserved in the holotype (TMP 2015.011.0003). Thin bone is present between the opercle and preopercle in both specimens but is not well preserved, with the thin bone being broken in many places; we interpret this as remains of both the interopercle and subopercle. Infraorbital region Remains of three infraorbital bones are preserved in the holotype (labelled as io2–3, and io4, and the dermosphenotic in Fig. 4), and four are preserved in the paratype (1, 2–3, 4 and the dermosphenotic; Fig. 5). One of the elements probably represents two fused infraorbitals as in †J. symmetropterus (Li & Wilson, 1996a: fig. 3) or some specimens of Hiodon (Hilton, 2002: fig. 36). Based only on size, we here identify the second preserved bone to be a fused infraorbital 2–3. The first infraorbital is small, with a low, broad, triangular shape (paratype; Fig. 5). Infraorbital 2–3 is greatly expanded, and is the largest bone of the series; it covers the posteroventral portion of the cheek from the orbit to the preopercle. Infraorbital 4 is about two-thirds the size of the preceding element in the series. Both of these bones have an enclosed sensory canal. The fifth infraorbital, the dermosphenotic, is preserved as a portion of a bone-enclosed tubular canal. This bone meets the pterotic in TMP 2015.011.0002 (paratype; Fig. 5), although it appears displaced, and is also displaced in the holotype (Fig. 4). There are no supraorbitals or antorbitals in the specimens. Vertebral column and predorsal bones The holotype (TMP 2015.011.0003) has 26 caudal centra, including two ural centra, and about 26 abdominal centra for a total count of about 52. The lateral surfaces of the centra are visible on the anteriormost centrum and nearly all of the centra in the caudal series (Fig. 6). These have a solid bone texture (with no small pits or struts) and two large lateral pits separated by a strut of bone, giving them the ‘H’ shape noted by Li & Wilson (1996a) for †J. symmetropterus. There appear to be 17 or 18 pairs of ribs, but because of their preservation a count is difficult. It is uncertain whether these articulate on the parapophyses or directly with the centra. Predorsal bones are visible in the holotype (Fig. 4); there are 11 long narrow bones anterior to the dorsal fin pterygiophores. Anterior left and right neural spines are not fused in the midline, as is the case in the other two species of †Joffrichthys (Li & Wilson, 1996a; Newbrey & Bozek, 2000). Figure 6. View largeDownload slide Photograph and interpretive drawing of the tail of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Figure 6. View largeDownload slide Photograph and interpretive drawing of the tail of †Joffrichthys tanyourus sp. nov. holotype TMP 2015.011.0003. Scale bar = 1 cm. Paired fins and girdles The cleithrum is mostly obscured by the opercle, but the posterior curved edge is visible in both specimens (Figs 4, 5). The cleithrum reaches anteriorly under the preopercle for about half the length of that bone. Details of the scapula and coracoid are not visible in either specimen. The supracleithrum is clear in TMP 2015.011.0002 (Fig. 5); it is about three times as long as it is wide, and broadens dorsally. The post-temporal is visible in the holotype (Fig. 4); the dorsal and ventral limbs are of equal length and almost parallel to one another. The pectoral fin has 15 rays in the holotype. The pelvic fin is preserved in the holotype TMP 2015.011.0003, but the pelvic girdle is not visible. There are six pelvic fin rays. Dorsal and anal fins The dorsal fin is not preserved in the paratype. In the holotype, the dorsal fin is positioned more anteriorly than the anal fin (Fig. 2), with a predorsal length of 0.5 SL and preanal length of 0.63. The insertion points of the first dorsal and anal fin pterygiophores are separated from one another by four vertebrae in the holotype. The dorsal fin rays are not all preserved, but based on those that are present and the pterygiophores, there appears to be four procurrent and 26 principal rays. There are about 21 pterygiophores supporting the dorsal fin. The anal fin in the holotype has three procurrent rays and 24 principal rays, or possibly four procurrent and 23 principal rays. Caudal skeleton and fin The caudal fin is preserved in the holotype (TMP 2015.011.0003; Fig. 6). The fin is supported ventrally by the haemal spines of the second and third preural centra, the parhypural and the hypurals. None of the rays of the dorsal lobe are supported by any neural spines. There are 18 principal rays (16 branched) forming a slightly forked tail fin. There are about seven dorsal and six ventral procurrent rays, giving a fin formula of vii,I,8,8,I,vi. There is a single neural spine on each of the preural centra (Fig. 6). The parhypural on the first preural centrum (pu1) is not much larger than the preceding haemal spine of the second preural centrum. The first ural centrum (diural terminology) in the holotype is slightly longer than the preural centra, and bears the first two hypurals. The much smaller second ural centrum bears five hypurals, for a total of seven hypurals in the caudal skeleton. There are a number of elements dorsal and posterodorsal to the two ural centra in the holotype that are difficult to interpret. The anteriormost appears free from, and slightly overlaps the first ural centrum and we identify that as the first uroneural. The next posterior element overlaps the dorsal part of the second ural centrum and so we feel quite confident that it is the second uroneural. Posterior to this, there is an element that is trapezoidal in shape, but appears to overlap the second uroneural (labelled with a question mark in Fig. 6), so it could be an abnormally shaped third uroneural, or it could potentially be a third ural centrum, but it seems to be lying lateral to the second ural centrum. Another possible alternative is that this bone is a urodermal; however, a urodermal has not been reported for living osteoglossomorphs. Posterior and slightly dorsal to this trapezoidal element is a long narrow bone that is ventral to the distal part of the second uroneural; therefore, we consider this to be another uroneural (the fourth if the trapezoidal bone is a uroneural, or the third uroneural if the trapezoidal bone is not a uroneural). All of these elements are quite closely associated with the ural centra, being ventral to the distal part of the long second uroneural; therefore, none are considered to be epurals. We cannot distinguish a separate element more dorsally positioned that could be an epural, so we believe there are no epurals present. Scales Cycloid scales are visible on the supraoccipital of the holotype, but not on the other skull bones. Scales on the dorsal surface are larger than those preserved on the abdomen near the pectoral girdle. The scales are large, circular to ovoid in shape, and with a central focus. They have 50 or more fine circuli, and are not reticulated. Comparisons among the species of †Joffrichthys In addition to the features listed in the diagnosis, there are a few other features that differ among the known specimens of the three species of †Joffrichthys. The shape of the dentary of the new species, †J. tanyourus, differs from that of the type specimen of †J. symmetropterus: in the type specimen (UALVP 23705), the dentary is shorter and more triangular than that of †J. tanyourus, the teeth are relatively shorter, and the symphysis is smaller. Based on published photographs (Newbrey & Bozek, 2000: figs. 1–2), †J. triangulpterus is similar to †J. symmetropterus in the shape of its dentary. Li & Wilson (1996a: fig. 6) illustrated a specimen of osteoglossomorph from Smoky Tower, Alberta; they considered it to belong to Joffrichthys but not necessarily to J. symmetropterus. This specimen, UALVP 34770, is similar to †J. tanyourus in having a dentary that appears relatively elongate and having taller teeth. However, in contrast to †J. tanyourus, only a single row of teeth appears to be present and the dentary appears deeper and with a larger symphysis. This morphology is also found in an isolated dentary preserved as an impression (UALVP 15069) that was referred to †Joffrichthys by Li & Wilson (1996a). The morphology of these dentaries allows us to exclude both UALVP 34770 and UALVP 15069 from †J. tanyourus. Instead, we find them most similar to the new genus described below, and suggest they represent that taxon instead (see section on Referred Specimen from Smokey Tower below). There is a single neural spine on the second preural centrum (pu2) in †J. tanyourus. Newbrey & Bozek (2000) reported two full spines on pu2 in †J. triangulpterus, whereas Li & Wilson (1996a) reported one long and a second short spine on †J. symmetropterus (although we cannot confirm this as the specimens of this latter do not have well-preserved caudal fins). The number of neural spines on pu2 may vary among species, or may be individual variation. The anal and dorsal fin ray numbers are equal in †J. symmetropterus, giving the species its name, with 24 rays in each, and the insertions of the first pterygiophore of each fin are separated by four centra. In †J. tanyourus, the fin insertions are also separated by four centra in the holotype (TMP 2015.011.0003), and the fin ray counts are similar to one another (23 or 24 rays in the anal fin and 26 in the dorsal fin). However, the first anal fin ray is positioned under the middle of the dorsal fin, not at the same level as the first dorsal fin ray as in †J. symmetropterus. In other words, the preanal and predorsal lengths in †J. symmetropterus are similar, but the predorsal length is clearly greater than the preanal length in †J. tanyourus. In †J. triangulpterus, the preanal length is also greater than the predorsal length, but the dorsal fin is larger with many more rays than the anal fin (27+ in the dorsal and 17–24 in the anal fin; Newbrey & Bozek, 2000), and there are more (8–9) centra between the insertion points of the first pterygiophores of each fin. As noted in the diagnosis, the number of vertebrae also varies among the three species, with more abdominal centra in the new species than the other two (Table 1). Ribs also vary slightly, with a few more pairs in †J. triangulpterus than the other species, and the predorsal bones number fewer in the new species than the other two (Table 1). Newbrey & Bozek (2000) indicated that the anterior neural spines of †J. triangulpterus were not fused in the midline, and said this differed from the condition in †J. symmetropterus. However, we believe they misunderstood Li & Wilson (1996a: p. 203) because these latter authors stated ‘on the caudals, bilaterally opposite neural spines are fused with each other’ indicating that was not the condition in the anterior vertebrae, and, in fact, after examining the specimens we can confirm that the anterior neural spines are paired, not fused. Therefore, this condition is found in all three species of †Joffrichthys. OSTEOGLOSSIFORMES Berg, 1940 Incertae Sedis †Lopadichthys gen. nov. Type and only known species: †Lopadichthys colwellae sp. nov. Diagnosis: As for type and only known species. Derivation of name: The genus is named for the fish being disk or plate shaped, from the Greek lopas or lopados meaning a dish or plate, and the ending ichthys, Greek for fish. Gender is masculine. †Lopadichthys colwellae sp. nov. Holotype: TMP 2015.011.0001, a complete fish preserved in left lateral view (Fig. 7). Figure 7. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 7. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Paratypes: TMP 2015.011.0004, a complete fish preserved in left lateral view (Fig. 8), and TMP 2015.011.0005, the posterior portion of a fish preserving the dorsal, anal and caudal fins and most of the body, but missing the head and anteroventral body, preserved in left lateral view (Fig. 9). Figure 8. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 8. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 9. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. Figure 9. View largeDownload slide Photograph of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. Type locality and horizon: Paskapoo Formation, Calgary, Alberta, Canada; Palaeocene in age. Etymology: The specific epithet is in honour of Jane Colwell-Danis in recognition of her considerable contribution to vertebrate palaeontology in Alberta. She has published under the names Colwell, Danis and Colwell-Danis. Diagnosis: A deep-bodied osteoglossomorph fish with a head depth slightly greater than head length; two heads on the hyomandibula; striations on the opercle radiating from the facet for articulation with the hyomandibula; 45–47 vertebral centra with 19–22 abdominal; anteriormost centra with numerous pits forming an open, lacy bone texture; and short deep lower jaw with the level of articulation anterior to the orbit and with a high coronoid process on the dentary. Differs from members of the suborder Osteoglossoidei (caudal formula I,7,8,I) by having a caudal formula of I,8,8,I. Further differs from subgroups within Osteoglossoidei by lacking an elongate pectoral fin ray and reticulate scales (found in many osteoglossids); having abdominal pelvic fins and unexpanded pectoral fin (unlike the anteriorly placed pelvic fins and expanded pectoral fins of Pantodon); having a small anal fin not connected with the caudal fin (unlike the elongate anal fin confluent with the caudal fin found in notopterids); jaws terminal and unmodified (unlike the modifications of the jaws found in mormyrids). Differs from species of †Joffrichthys by having anteriormost abdominal and posterior caudal vertebral centra with many small pits and a network of bone forming a lacy structure, rather than ‘H-shaped’ (i.e. two large pits laterally separated by a strong bar of bone), and by having a relatively shorter and deeper caudal peduncle. Differs from Hiodontiformes by having a rounded dorsal border to the opercle (without the posterodorsal opercular spine as found in Hiodon and †Eohiodon). Differs from basal osteoglossomorphs by having six rays in the pelvic fin (instead of seven rays in †Wilsonichthys, †Shuleichthys, †Xixiaichthys and hiodontiforms). Description General body form This is a deep-bodied fish, with the greatest body depth, anterior to the dorsal fin origin (Figs 7–9), being greater than half (0.57–0.58) of the SL. The dorsal and anal fins are slightly falcate, and positioned posteriorly on the body, with the dorsal fin positioned more anteriorly than the anal fin. The caudal peduncle is short, and deeper than it is long. The caudal fin is gently forked. The head depth is about one-third of SL, and the head is deeper than it is long. The following description applies to the holotype, with the paratypes agreeing in details where they can be seen unless otherwise noted. Counts and measurements are given in Table 2. Table 2. Counts and measurements (in mm) for the three specimens of the †Lopadichthys colwellae gen. et sp. nov. TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – View Large Table 2. Counts and measurements (in mm) for the three specimens of the †Lopadichthys colwellae gen. et sp. nov. TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – TMP 2015.011.0001 TMP 2015.011.0004 TMP 2015.011.0005 Total length (TL) 227 175 – Standard length (SL) 192 144 – Head length (HL) 49 46 – Head depth (HD) 70 50 – Body depth (BD) 109 83 78 Caudal peduncle length (CPL) 21 15 18 Caudal peduncle depth (CPD) 28 21 19 Anal fin base length (AFL) 61 44 37 Dorsal fin base length (DFL) 70 46 48 Preanal length (PAL) 113 86 – Predorsal length (PDL) 112 86 – Prepelvic length (PPL) 82 57 – HL/SL 0.26 0.32 – HD/SL 0.36 0.35 – HL/HD 0.7 0.92 – BD/SL 0.57 0.58 – CPL/CPD 0.75 0.71 0.95 PAL/SL 0.59 0.60 – PDL/SL 0.58 0.60 – PPL/SL 0.43 0.40 – Dorsal fin rays iii,26 iii,26 iii,27 Dorsal fin pterygiophores ?21 – ?23 Anal fin rays iv,23 iii,22 ?,22 Anal fin pterygiophores 23 ?23 ?22 Pectoral fin rays – 16 – Pelvic fin rays 6 6 – Caudal fin rays I,8,8,I I,8,8,I I,8,8,I Caudal centra (incl. u1 and u2) 26 25 24 Abdominal centra 19 22 – Total centra 45 47 – Predorsal bones 7+ ?8 – Branchiostegal rays 9 10 – View Large Neurocranium and skull roof The skull is best preserved in the holotype (TMP 2015.011.0001; Fig. 10). The nasal is long and narrow, with the width about one quarter of the length. The frontals are broader posteriorly than at the anterior tip, but with an expansion over the orbit (Figs 10, 11). The frontal is about 2.5–3 times as long as it is wide, with the anterior end angled so that the medial edge extends farther anteriorly than the lateral edge. The sensory canal is enclosed in bone in the frontal, and does not extend onto the parietal. The parietal is rectangular, about twice as wide (mediolaterally) as it is long (antero-posteriorly). The posterolateral corner of the frontal is notched behind the expansion over the orbit to receive the roughly square dermosphenotic. The pterotic is somewhat elongate, and bears a sensory canal that is at least partially open. The supraoccipital is a large bone, extending posteriorly to at least the level of the second vertebral centrum. There is no evidence of an extrascapular bone in the holotype, and the laterally placed temporal fossa is open between the parietal, pterotic, exoccipital and epioccipital bones (Fig. 10). However, a large irregularly shaped extrascapular bone covers this region in paratype TMP 2015.011.0004 (Fig. 11). Figure 10. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 10. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 11. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 11. View largeDownload slide Photograph and interpretive drawing of the head of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. The bones of the posterior braincase are partially visible, but appear displaced in the holotype (TMP 2015.011.0001). Although the limits of the bones in this area are difficult to interpret, we identify an exoccipital and epioccipital; we believe an intercalar was also present (Fig. 10) but no details can be discerned. In the dorsal part of the orbit, the orbitosphenoid and pterosphenoid are visible in paratype TMP 2015.011.0004 (Fig. 11). The parasphenoid is visible in the orbit, and bears large caniniform teeth along its visible length (Figs 10, 11). The level of the parasphenoid is significantly more ventral compared to the anterior end of the vertebral column which would articulate with the basioccipital. In this regard, this fish is similar to Hiodon, and we suggest it would have had a very tall posterior process to articulate with the basioccipital. A small square bone in the holotype and a roughly square bone in paratype TMP 2015.011.0004, both preserved anterior to the anterolateral corner of the frontal, are interpreted as lateral ethmoids. The difference in shape is attributed to preservational orientation of the bone. Jaws, branchial arches and suspensorium The jaws are relatively shorter and deeper in †L. colwellae compared to those of †J. tanyourus, but bear similar large caniniform teeth. It is uncertain whether or not a single tooth row is present, or multiple rows as in †J. tanyourus. The premaxilla has an anterior rounded articular process and a short ramus. There are five premaxillary teeth preserved in the holotype (TMP 2015.011.0001; Fig. 10), and four in the paratype (TMP 2015.011.0004; Fig. 11). The maxilla has a small, rounded anterior end, but the majority of the bone is deeper in lateral view. Neither specimen with a head preserves the maxilla completely, but the impression of the bone is preserved in TMP 2015.011.0004 (Fig. 11), although few teeth are present. No supramaxillary bones are present. The dentary (Figs 10, 11) bears seven teeth, and abruptly widens, posterior to the teeth, into the coronoid process that forms a distinct angle to the toothed portion of the bone. The angular and articular appear to be indistinguishably fused in lateral view, but a separate retroarticular is present at the ventral edge of that element. The triangular quadrate (Fig. 10) has an angle of only about 90°, and the notch for the symplectic is quite narrow. The hyomandibula has a notch in the dorsal edge (Figs 10, 11), giving it two articular heads. The process for articulation with the opercle is robust, and there is a robust anterior process (visible in the holotype TMP 2015.011.0001; Fig. 10). The symplectic is only visible in the holotype, and is not completely preserved; it is a fairly narrow bone. In paratype TMP 2015.011.0004 (Fig. 11), the anterior ceratohyal and the hypohyals are preserved. The anterior ceratohyal is quite short and deep. Remains of 12 branchiostegal rays are visible in this specimen; probably at least eight or nine belong to the left side, and the others possibly belong to the right side. In the holotype (Fig. 10), there are ten branchiostegal rays preserved and all appear to belong to the right side. Basibranchial teeth are visible below the right dentary in the holotype (Fig. 10); these are larger than the jaw teeth, and strongly curved. Opercular series The opercle is preserved mainly as an impression (Figs 10, 11). It has rounded dorsal and posterior margins, and the anterior and ventral edges of the opercle are also rounded. The opercle of the holotype (Fig. 10) is ornamented with striations that radiate from the articular facet for the hyomandibular process. The opercle is twice as high (dorsoventrally) as it is wide (antero-posteriorly). The preopercle is best preserved in the holotype (Fig. 10). The ventral limb is much more robust than the dorsal limb, but the bone is crushed and details are not clear. The subopercle has a slender portion visible under the posteroventral edge of the opercle (holotype TMP 2015.011.0001; Fig. 10). Infraorbital region The infraorbital bones are not well preserved. Remains of the first and second are present in the holotype (TMP 2015.011.0001; Fig. 10) The first is longer than deep, and the second is deeper and shorter than the first. The sensory canal of both appears to be enclosed in a bony tube. Remains of the presumed fourth infraorbital are found in paratype TMP 2015.011.0004; this bone would have been larger than the first and second, and covers the cheek to the anterior edge of the hyomandibula. The dermosphenotic (infraorbital 5) fits into the notch in the posterolateral corner of the frontal. There are no antorbital or supraorbital bones. Vertebral column and predorsal bones There are 45–47 centra, including the two ural centra. There may be an additional third ural centrum not included in the counts. The holotype (TMP 2015.011.0001) has 26 caudal and 19 abdominal centra, and the complete paratype (TMP 2015.011.0004) has 25 caudal and 22 abdominal centra. The vertebral centra are not well preserved in TMP 2015.011.0005. In the other two specimens, the anteriormost vertebral centra have multiple small pits ornamenting the bone, giving it a lacy appearance (Figs 10, 11). Further posterior in the column, the centra have fewer, larger pits. Neural arches are autogenous in the anterior centra. Parapophyses (Fig. 12), which can be seen on two of the anterior abdominal centra in TMP 2015.011.0001 (centra 6–7), are fused to the centra, and have a broad base (visible on centrum 7) and a thin dorsal edge (visible on centrum 6), similar to centra that Neuman & Brinkman (2005) referred to †Coriops. Also as in †Coriops, the rib articulates with the centrum posterior to the parapophysis. The number of ribs is difficult to determine, but there appear to have been 18 (TMP 2015.011.0004) or 20 (TMP 2015.011.0001) pairs. There are 13 long, thin, predorsal bones in both the holotype (TMP 2015.011.0001) and in TMP 2015.011.0004. Figure 12. View largeDownload slide Comparison of centra of †Lopadichthys and †Coriops. †Lopadichthys colwellae gen. et sp. nov., holotype TMP2015.011.0001, photographs and interpretive drawings of centra from the anterior (A), middle (B) and posterior (C) parts of the vertebral column. Three centra assigned to †Coriops in lateral (left) and anterior (right) views, UCMP 230710/V72210 (D), UCMP 276784/V77128 (E) TMP 1986.22.43 (F). Scale bars = 2 mm. Figure 12. View largeDownload slide Comparison of centra of †Lopadichthys and †Coriops. †Lopadichthys colwellae gen. et sp. nov., holotype TMP2015.011.0001, photographs and interpretive drawings of centra from the anterior (A), middle (B) and posterior (C) parts of the vertebral column. Three centra assigned to †Coriops in lateral (left) and anterior (right) views, UCMP 230710/V72210 (D), UCMP 276784/V77128 (E) TMP 1986.22.43 (F). Scale bars = 2 mm. Paired fins and girdles Little of the pectoral girdle can be seen. The dorsal and ventral limbs of the cleithrum form almost a 90° angle to one another, curving under the opercle (Fig. 11). Remains of the coracoid indicate that it had a robust, rounded dorsal edge. The dorsal arm of the post-temporal is twice as long as the ventral arm (Fig. 10); the dorsal arm broadens distally. There are 14 pectoral fin rays preserved in paratype TMP 2015.011.0004. The pelvic fin is positioned closer to the origin of the anal fin than it is to the pectoral girdle. It contains six rays. Dorsal and anal fins The dorsal fin is preserved in all specimens and has three procurrent rays and 26–27 principal rays; these are supported by at least 21 pterygiophores in the holotype (TMP 2015.011.0001) and at least 23 in paratype TMP 2015.011.0005. The anal fin has three (TMP 2015.011.0004) or four (holotype) procurrent rays and 22 (both paratypes) or 23 (holotype) principal rays. There are probably 23 pterygiophores supporting the anal fin. Caudal skeleton and fin The caudal fin and skeleton are well preserved on the holotype (TMP 2015.011.0001; Fig. 13) and both paratypes (Figs 14, 15). The fin is slightly forked with 18 principal rays (16 branched), seven dorsal procurrent and about ten ventral procurrent rays, giving a fin formula of vi,I,8,8,I,x in the holotype. The ventral procurrent rays in TMP 2015.11.0005 are not all clear, and this specimen has a formula of vi,I,8,8,I,v+. The parhypural is autogenous and is preserved slightly separated from the first preural centrum (pu1) in the holotype (Fig. 13), but articulates on that centrum in the paratypes (Figs 14, 15). The first ural centrum (u1) supports the first two hypurals (hy1 and hy2). The second ural centrum (u2) bears four hypurals, for a total of six hypurals in the caudal fin. All six hypurals are autogenous, and the upper four are roughly triangular with narrow proximal ends, while the first two are only slightly narrower at their proximal ends compared to the distal ends. The second preural centrum (pu2) has a single neural spine, but pu1 bears two neural spines in the holotype; whether there are two neural spines on pu1 in TMP 2015.011.0005 is difficult to see in the fossil (and we have figured it with a single neural spine; Fig. 15), but the silicone peel indicates there may have been two present. There is a single neural spine on pu1 in TMP 2015.011.0004 (Fig. 14). Figure 13. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 13. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. holotype TMP 2015.011.0001. Scale bar = 1 cm. Figure 14. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 14. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0004. Scale bar = 1 cm. Figure 15. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. Figure 15. View largeDownload slide Photograph and interpretive drawing of the tail of †Lopadichthys colwellae gen. et sp. nov. paratype TMP 2015.011.0005. Scale bar = 1 cm. There are three elements between the posterior neural spine of pu1 and the sixth hypural in the holotype (Fig. 13) and also in TMP 2015.011.0004 (Fig. 14); the first two elements we initially identified as being one epural and one uroneural. The identification of epurals and uroneurals causes significant differences in the phylogenetic analysis (see Discussion below), and based on the phylogenetic analysis, it may be more reasonable that these are both uroneurals. The third element is unidentified, but could be a third ural centrum. This area of the caudal fin is less clear in the third specimen, TMP 2015.011.0005 (Fig. 15), but based on the silicone peel, there are four elements preserved, which we identify as two uroneurals and one epural, or three uroneurals, and a long procurrent ray. Scales Large, roughly circular cycloid scales with many fine circuli are visible on the specimens, but the edges of each are not clear. There are no reticulations on the scales. Referred specimen from Smoky Tower When Li & Wilson (1996a) described the species †J. symmetropterus based on material from the Joffre Bridge road cut (UALVP locality 56), they also referred an incomplete specimen (UALVP 34770) from the Smoky Tower locality of the Palaeocene Paskapoo Formation to the same genus but as an indeterminate species. The dentary of UALVP 34770 is relatively deeper than that of †J. symmetropterus and †J. tanyourus and the symphysis is relatively larger. In the proportions and the relative number and size of teeth, the dentary of UALVP 34770 is strikingly similar to the dentary of †Lopadichthys (Fig. 16). In addition, UALVP 34770 preserves almost complete dorsal and anal fins, and the base of the tail, which reveal that the caudal peduncle was quite short and deep, as in †Lopadichthys; the number of fin rays preserved does not contradict an affinity with †Lopadichthys. Based on the differences between UALVP 34770 and †Joffrichthys, and the similarities of UALVP 34770 with †Lopadichthys, we here refer the Smoky Tower specimen to †Lopadichthys, cf. †L. colwellae. An isolated dentary from the Joffre Bridge road cut locality, specimen UALVP 15069, was referred to †J. symmetropterus by Li & Wilson (1996a); we here reassign that specimen to †Lopadichthys. Figure 16. View largeDownload slide Comparative photographs of the lower jaws of several Palaeocene osteoglossomorphs. †Joffrichthys symmetropterus, holotype, UALVP 23705 (A), †Joffrichthys tanyourus sp. nov., holotype, TMP2015.11.0003 (B), †Lopadichthys colwellae gen. et sp. nov., paratype TMP2015.011.0004 (C), UALVP 34770, referred to †Joffrichthys sp., by Li & Wilson (1996), but here reassigned to †Lopadichthys cf. L. colwellae (D), UALVP 15069 assigned to †J. symmetropterus by Li & Wilson (1996), here reassigned to †Lopadichthys cf. L. colwellae (E), †Lopadichthys colwellae gen. et sp. nov. holotype TMP2015.011.0001 (F). Figure 16. View largeDownload slide Comparative photographs of the lower jaws of several Palaeocene osteoglossomorphs. †Joffrichthys symmetropterus, holotype, UALVP 23705 (A), †Joffrichthys tanyourus sp. nov., holotype, TMP2015.11.0003 (B), †Lopadichthys colwellae gen. et sp. nov., paratype TMP2015.011.0004 (C), UALVP 34770, referred to †Joffrichthys sp., by Li & Wilson (1996), but here reassigned to †Lopadichthys cf. L. colwellae (D), UALVP 15069 assigned to †J. symmetropterus by Li & Wilson (1996), here reassigned to †Lopadichthys cf. L. colwellae (E), †Lopadichthys colwellae gen. et sp. nov. holotype TMP2015.011.0001 (F). Phylogenetic analysis During the past two decades, a number of authors have been studying the phylogenetic relationships of the Osteoglossomorpha (e.g. Li et al., 1997; Hilton, 2003; Murray & Wilson, 2005; Zhang, 2006). These authors have been sharing a data set and adding and modifying it with new data from new fossil finds. The most recent iteration of the data set is that of Murray et al. (2016), who used the combined data sets of Hilton (2003) and Li et al. (1997) with additions and modifications as listed in Wilson & Murray (2008), and an additional taxon from Murray et al. (2010). For our first analysis, we used the data matrix of Murray et al. (2016) with modifications as noted in the Material and Methods, and further detailed below. The complete character list with character states can be found in Wilson & Murray (2008; Appendix 2) with many of the alternate character states figured and discussed in Hilton (2003). Hilton (2003) used a single outgroup, Elops, in his analysis. Li et al. (1997) included two outgroups, Clupeioidei and the ichthyodectiform †Cladocyclus gardneri. Our first analysis included all three of these taxa as has been done in previous analyses (e.g. Murray et al., 2016). For our second analysis, we removed †C. gardneri from the matrix, and replaced the coding of Clupeioidei from Li et al. (1997) with data for members of the two orders of Clupeomorpha, †Ellimmichthyiformes and Clupeiformes. Because osteoglossomorphs are basal teleosts, we also added data for Amia, as a non-teleost outgroup. The complete data matrix with our modifications to coding and taxa is given in Appendix 1. Modified coding and characters for the phylogenetic analysis Caudal fin ray counts: The overall similarity between †J. symmetropterus specimens and the new Paskapoo †J. tanyourus specimens led us to re-examine the former to determine why the caudal fins in these two species had different ray counts, with the former originally reported as having 15 branched principal rays (formula i,7,8,i) and the new material having 16 branched caudal fin rays (i,8,8,i). Re-examination of the holotype of †J. symmetropterus (UALVP 23705) and a second specimen preserving the tail (UALVP 31545) shows that a count of the caudal fin rays is difficult; however, we count eight branched rays in both the dorsal and ventral lobes of the fin, giving a formula of i,8,8,i (16 branched rays in total), not seven branched rays in the dorsal lobe as reported by Li & Wilson (1996a). Newbrey & Bozek (2000) also had difficulty counting caudal fin rays in the second described species of †Joffrichthys, †J. triangulpterus, because the upper lobe of the fin was incomplete. However, they considered this species to have a caudal fin ray formula of i,?,8,i [this is our reinterpretation of the formulae given in their table (2000: tab. 1) in which they have a formula of i+8+?+i for †J. triangulpterus and i+8 + 7+i for both †J. symmetropterus and †C. strickeri, indicating they have listed the ventral lobe of the fin first and the dorsal lobe second, as was done by Li & Wilson (1997), whereas we list the dorsal lobe first, following most other authors, e.g. Schultze & Arratia (1989)]. We modified the character state for number of caudal fin rays in †J. symmetropterus to reflect our reassessment. We also reassessed a species of presumed Osteoglossidae, †C. mahengeense, as it had conflicting characters: it has an elongate unbranched ray in the pectoral fin (otherwise found only in some species of the Osteoglossidae), but was also reported to have a caudal fin formula of i,8,8,i (Murray & Wilson, 2005), which is not found in osteoglossids. The caudal fin in this species is only preserved in two specimens, and in one of those (WM 378/96) only the ventral lobe of the fin is visible. Re-examination of a latex peel of the second specimen (WM 492/96) shows that the previously reported caudal formula is probably incorrect; it is more likely that there are only 15 branched principal rays in this species (i,7,8,i). We changed the coding for character 65 from state 1 (18 principal rays) to state 2 (17 or fewer principal rays) to reflect this. Epurals and uroneurals: In †Lopadichthys (TMP 2015.011.0001), there are two elements between the posterior neural spine of the first preural centrum and the sixth hypural that we identified in the description above as one epural and one uroneural; however, they may both be uroneurals, depending on the interpretation of the researcher. Taverne (1977, 1978) provided detailed osteological studies of a number of fossil and living osteoglossomorphs. Based on his examination of the caudal skeletons, he concluded that there was a single uroneural, no urodermals and no epural present in Osteoglossum, Scleropages, Heterotis, Arapaima, †Brychaetus, Pantodon and the notopterids. Hilton (2003), based on developmental evidence in Arapaima, determined that this taxon has a paired uroneural that has fused in the midline and no epural. Whether this is the case for all osteoglossomorphs cannot be determined without ontogenetic series for all the fossil forms. Taverne (1977) showed that Hiodon and †Eohiodon have both a single epural and three uroneurals. He further considered †Lycoptera to have a single epural, five uroneurals and a seventh element that could be either an epural or uroneural. Li & Wilson (1996a: fig. 5) in their drawing identified an element between the neural spine of the first ural centrum and hypural six as an epural in †J. symmetropterus. They listed two specimens for their interpretation of the caudal fin, UALVP 23705b (the holotype counterpart impression) and UALVP 37128b. We can see no element preserved in this position in either the part or counterpart of the holotype (UALVP 23705 a and b). In UALVP 37128 a and b, the caudal skeleton is not well preserved, and we cannot see the limits of individual bones and scales in this area. However, in another specimen, UALVP 31545b, which has a break right through the caudal fin skeleton, there is a partial thin element dorsal to the second ural centrum, which could be interpreted as an epural. Taverne (1978) considered †Singida to have a single epural and two uroneurals, but this was contradicted by Murray & Wilson (2005) based on better preserved material that showed †Singida to have two uroneurals, and no epurals or urodermals; the uroneurals are not fused in the midline, but clearly overlap the sides of the centra (Murray & Wilson, 2005). At least one well-preserved fossil taxon, †Shuleichthuys brachypteryx Murray et al., 2010, has both paired uroneurals (four) as well as a median element identified as an epural (Murray et al., 2010). The lack of developmental series in the many fossil species means we cannot use developmental data to determine what these elements are. As shown by the fossil †Shuleichthys as well as the extant Hiodon (Hilton, 2002), osteoglossomorphs may have both uroneurals and epurals present, so we cannot establish a priori what these elements may be in other fossil osteoglossomorphs. We are therefore left with a positional argument to determine what each element is, and thus we identify a uroneural as an element that is either clearly paired, or closely associated with the dorsal surface of the ural centra, whereas an epural is a median element positioned more dorsally in the caudal skeleton. The identification of the median element as an epural or uroneural causes significant changes in the phylogenetic analysis, but so does excluding the character of number of epurals (character 68). The single change of identifying †Lopadichthys as having two uroneurals compared to having one uroneural and one epural shifted several taxa in the analysis as described below. Similarly, excluding the character completely also changed the resultant tree significantly. Coverage of the cheek by infraorbitals 3 and 4: Hilton (2003) considered two of the characters used by Li & Wilson (1996b), their numbers 16 and 20, to be very similar: character 16 is the pterygoquadrate area posteroventral to the orbit being completely covered or not by the infraorbitals; character 20 is presence of a ‘cheek wall’ formed by the enlargement of the first through third infraorbital bones. Hilton (2003) considered both characters to refer to the enlargement of the posterior infraorbital bones (his character 25). We interpret the two characters slightly differently, and so retain them both. For the first character (our number 25; coverage of the pterygoquadrate area), we consider this character to be complete coverage if the posterior infraorbital bones cover all the area up to the anterior edge of the preopercle. For the second character (our number 81), we consider the cheek wall to be present if there is significant coverage by the infraorbitals, even if they do not meet the anterior edge of the preopercle. In the case of the new species of †Joffrichthys, we code a cheek wall as present, but full coverage of the pterygoquadrate area as absent. Modified parapophysis of first centrum: Forey & Hilton (2010) noted a modification to the first centrum in several osteoglossids that might be a useful synapomorphy for the group. They showed that in Heterotis niloticus (Cuvier, 1829), Arapaima gigas (Schinz, 1822), Osteoglossum bicirrhosum (Cuvier, 1829) and †Phareodus encaustus (Cope, 1871), the parapophysis of the first centrum is expanded or hypertrophied to reach under the occiput. We examined latex peels of the fossil osteoglossids †S. jacksonoides and †C. mahengeense to see if we could determine the condition of the bones in these fishes. For the most part, this area in these fossils is obscured by the overlying opercle. However, in one specimen of †S. jacksonoides (WM 315/96), we believe there is an expanded parapophysis on the first centrum that lies under the basioccipital (Fig. 17), somewhat similar to that figured for †Phareodus (Forey & Hilton, 2010: fig. 1D). This character has been added to the matrix and coded for those taxa in which it can be determined. In †J. symmetropterus, one specimen (the holotype, UALVP 23705) has the basioccipital visible; there is no evidence of a parapophysis extending from the first vertebra to the basioccipital. Figure 17. View largeDownload slide Photograph and interpretive drawing of the basioccipital and first centrum of †Singida jacksonoides latex peel of WM 315/96. Scale bar = 2 mm. Figure 17. View largeDownload slide Photograph and interpretive drawing of the basioccipital and first centrum of †Singida jacksonoides latex peel of WM 315/96. Scale bar = 2 mm. Recoding of †Joffrichthys symmetropterus: Based on our re-examination of the type material of †J. symmetropterus, we have changed the coding for 29 characters for this species as follows (indicated by the character number, followed in brackets by the old code and an arrow to the new code): 3(0➔1); 5(?➔0); 6(2➔1); 9(1➔?); 11(?➔1); 13(?➔1); 16(?➔0); 23(?➔1); 25(1➔0); 27(?➔1); 32(2➔?); 33(0➔1); 37(0➔1); 39(1➔0); 42(?➔2); 43(?➔1); 57(n/a➔0); 59(?➔0); 60(?➔0); 63(?➔0); 65(2➔1); 68(?➔2); 71(1➔?); 73(?➔2); 79(0 + 2➔0); 80(?➔0); 81(?➔1); 84(?➔1); 87(1➔0). Recoding of Elops: We recoded ten of the characters for Elops based on information in Ridewood (1904), Taverne (1974) and Schultze & Arratia (1989), which allowed us to provide character states for some characters that were previously missing data. The following changes were made (indicated by the character number, followed in brackets by the old code and an arrow to the new code): 1(0➔1); 11(0➔1); 20(0➔1); 37(0➔1); 80(?➔0); 81(?➔1); 82(?➔1); 84(?➔2); 86(1➔0); 87(2➔0). Exclusion of †Ostariostoma from Osteoglossomorpha †Ostariostoma wilseyi (Ostariostomidae) is represented by a single specimen from deposits in Montana that are either Late Cretaceous or Palaeocene in age (Schaeffer, 1949). When this fish was first described, Schaeffer (1949) noted the difficulty in determining its relationships or taxonomic identity, and placed it questionably with clupeomorph fishes (as Order?Isospondyli, Suborder?Clupeoidea). Grande & Cavender (1991) re-examined the holotype after additional preparation of the specimen; they also had a difficult time trying to determine its relationships. Eventually, they suggested it might be affiliated with osteoglossomorph fish partly because †Ostariostoma has a hiodontid-like anal fin, which is long and has a somewhat falcate posterior margin and anterior rays somewhat thicker than the posterior rays (as found in male Hiodon), as well as there being 18 principal caudal fin rays and no supramaxillae in †Ostariostoma and osteoglossomorphs. †Ostariostoma has large teeth on the jaws that are osteoglossomorph-like, but it lacks teeth on the parasphenoid, which is unlike osteoglossomorphs, and there is no evidence of teeth on the basibranchial bones, indicating this fish lacks the ‘tongue bite’ apparatus of osteoglossomorphs (see Hilton, 2001, for a discussion of this feature). The caudal fin of †Ostariostoma has a reported fin ray formula similar to hiodontiforms (i,8,8,i) and two ural centra as is common in osteoglossomorphs; however, these features may also be found separately in other fishes [e.g. two ural centra in Esox and Umbra (Fujita, 1990); 16 branched principal caudal fin rays in Polymixia, Percopsiformes and diplomystid catfishes (Nelson, 2006)]. Additionally, Wilson & Murray (2008) noted that the caudal fin ray count of †Ostariostoma is problematic; they noted this specimen has 18 principal rays but only 15 of them are branched with two lower unbranched rays, giving a formula of i,8,7,ii. As they noted, this may be an individual abnormality, but without additional specimens this cannot be determined. Taverne (1998) suggested †Ostariostoma should be considered an osteoglossomorph based on seven shared characters (numbered as in his publication): 2) the pars palatine being cartilaginous in the adult; 3) the maxilla loses the strong articulation or retains only a very thin connection with the autopalatine that is found in the other primitive teleosts; 4) loss of the supraorbital; 5) the temporal fossa shifts anterolaterally, often accompanied by hypertrophy of the fossa; 6) the loss of the anterior supramaxilla; 10) a reduction in the number of epurals to one or none; 12) a reduction in the number of principal caudal fin rays to 18 (16 branched). Taverne (1998) additionally noted that †Ostariostoma also has lost the posterior supramaxilla, has strong development of the opercle, and the subopercle is reduced but still elongated as in Osteoglossomorpha. Taverne (1998) further likened †Ostariosoma to Hiodontidae based on his characters 13 and 14: 13) first infraorbital elongated with loss of a large part or even the whole of its dermal bone component, becoming very narrow and mostly limited to its sensory canal tube; 14) hypertrophy of the temporal fossa which moves from its position on the back of the neurocranium to become placed completely on the lateral side of the skull; the fossa is no longer delimited by the exoccipital, epioccipital (= epiotic of Taverne, 1998) and pterotic, but instead by the parietal, epioccipital and pterotic; the fossa remains isolated from the cerebral cavity by a cartilaginous wing which serves as its base. Grande & Cavender (1991) also noted the similarity of †Ostariostoma to osteoglossomorphs based on the lack of supramaxillary bones and 18 principal rays in the caudal fin. However, the single known specimen of †Ostariostoma does not show the palatine (bringing characters 2 and 3 of Taverne (1998) into question), its antorbital is elongate but the dermal portion is not greatly reduced, the post-temporal fossa is laterally placed and bordered by the parietal, epiotic and post-temporal, but is not visibly enlarged, and the caudal fin is poorly preserved with an unidentified element that is potentially a second epural. The subopercle is narrow but elongate as noted by Taverne (1998). Schaeffer (1949) noted that the name †Ostariostoma was not meant to suggest an affinity with the Ostariophysi. Grande & Cavender (1991) also excluded †Ostariostoma from having any ostariophysan affinities because it shows no modification of the anterior vertebral elements. However, we suggest that it may be a basal member of the group in which the Weberian apparatus was at an early stage of development and modifications were not apparent in an articulated specimen as is the case in gonorynchiforms, where the first two centra are ribless and lack parapophyses. In some laterally preserved ostariophysans, the bones of the opercular and pectoral region are crushed on top of the anterior vertebrae concealing any modifications of these elements (e.g. Murray, 2003). Grande & Cavender (1991) actually did indicate that the first neural spine of †Ostariostoma shows some modification; in their figure (1991: fig. 2) the spine appears to be bifurcated posteriorly. The long, narrow body shape of Ostariostoma differs from the deep, football-shaped body of known Cretaceous and Paleogene osteoglossomorphs and is similar to that of gonorhynchiforms. A cast of the holotype of †O. wilseyi in the University of Alberta collections (UALVP 52610), as well as figures in Grande & Cavender (1991) show the vertebral centra to be longer than high, with a length twice their depth, then becoming slightly shorter and deeper for the last four preural centra in the column, with the exception of the two ural centra which are also about twice as long as they are high. The ribs either insert directly on the abdominal centra without parapophyses or fully cover the parapophyses so they are not visible in the articulated specimen. Additionally, and unlike osteoglossomorphs, †Ostariostoma apparently lacks intermuscular bones, and has only a few supraneural bones (shown in Grande & Cavender, 1991: figs. 1 and 2; not visible in our cast). We consider the vertebral column to be more like that of a gonorhynchiform fish rather than an osteoglossomorph fish (see also discussion in Brinkman et al., 2017). While we cannot conclusively determine the relationships of †Ostariostoma, we removed it from one of our analyses of Osteoglossomorpha to determine the effect this taxon may have on the analysis. Phylogenetic results First analysis The first analysis, using the data set from Murray et al. (2016) with the modified characters for Elops, †Joffrichthys and †Chauliopareion, and the additional character from Forey & Hilton (2010) (as discussed in the Material and Methods), resulted in six trees. The tree statistics from PAUP are: tree length (TL) = 242, consistency index (CI) = 0.434, retention index (RI) = 0.630 and rescaled consistency index (RC) = 0.273. The trees differ in the relative positions of the outgroups, Clupeioidei and Elops, and the relative positions of †Xixiaichthys, †Joffrichthys and †Wilsonichthys. The strict and 50% majority rule consensus trees are identical (Fig. 18). Figure 18. View largeDownload slide Consensus tree of six most parsimonious trees resulting from analysis 1. Both the strict consensus and 50% majority rule consensus are the same. Figure 18. View largeDownload slide Consensus tree of six most parsimonious trees resulting from analysis 1. Both the strict consensus and 50% majority rule consensus are the same. In all the trees, the clade †Paralycoptera + †Tanolepis is excluded from the Osteoglossomorpha. †Joffrichthys is recovered basal to a clade including the Osteoglossidae, notopterids, mormyroids and several other taxa, and in three trees †Joffrichthys forms the sister group to †Xixiaichthys. †Sinoglossus is placed as the sister group to Notopteridae in all trees. †Lopadichthys, the new genus from the Paskapoo Formation, is recovered in the same position in all trees, as the basal member of the clade uniting Notopteridae, †Sinoglossus, Mormyroidea, †Palaeonotopterus and †Ostariostoma. Second analysis using new outgroups The second phylogenetic analysis, using the matrix with the corrections to the data as in the first analysis, but with new outgroup taxa (†Ellimmichthyiformes, Clupeiformes and Amia) and including †Ostariostoma, resulted in three shortest trees of length 292 (Fig. 19). Tree statistics from PAUP are: TL = 292, CI = 0.404, RI = 0.594 and RC = 0.240. In all three trees, the clade †Paralycoptera + †Tanolepis is excluded from the Osteoglossomorpha, as in the first analysis. Figure 19. View largeDownload slide The three most parsimonious cladograms resulting from analysis 2. Figure 19. View largeDownload slide The three most parsimonious cladograms resulting from analysis 2. The three trees differ in the placement of †Lycoptera and the relationship of Hiodon + †Eohiodon to †Shuleichthys and †Wilsonichthys. In one tree (Fig. 19B), †Lycoptera is united in a clade with the other four taxa, †Shuleichthys is sister to †Wilsonichthys, and the two together form the sister group of Hiodon + †Eohiodon, with †Lycoptera the basal member of this group. In the other two trees, †Lycoptera is basal to all osteoglossomorphs except the †Paralycoptera + †Tanolepis clade, and †Wilsonichthys is either the sister group to †Shuleichthys (Fig. 19C), or is closer to the rest of the osteoglossomorphs than is †Shuleichthys (Fig. 19A). In both these trees, the Hiodon + †Eohiodon clade is more basal than †Shuleichthys and †Wilsonichthys. In all three trees, †Lopadichthys is recovered as the basal member of a clade including †Ostariostoma and the highly derived notopterids, and mormyroids; this clade is the same as recovered in the first analysis with the exclusion of †Sinoglossus, which in this second analysis is recovered as an osteoglossid, sister group to Heterotis. Two of the trees (Fig. 19A, C) are quite similar, with the only difference being whether †Shuleichthys and †Wilsonichthys were united as sister groups, or placed as successive lineages with †Shuleichthys more basal. The third tree is more different in that †Lycoptera, †Eohiodon + Hiodon and †Shuleichthys + †Wilsonichthys form a clade that is positioned between Elops and the Clupeomorpha, forming a polyphyletic Osteoglossomorpha even if †Paralycoptera and †Tanolepis are excluded from the superorder. Although the placement of the basal taxa in these three trees is not consistent, the more derived taxa are stable, and the relationships found are fairly similar to those of Murray et al. (2016), with the notable exception that †Joffrichthys is excluded from the Osteoglossidae, as would be expected based on our reinterpretation of the caudal skeleton. Third analysis excluding †Ostariostoma The third analysis was run with the corrected data matrix and the new outgroups but excluding †Ostariostoma from the analysis, on the consideration that it is not an osteoglossomorph (see above). This resulted in a single tree of 290 steps (Fig. 20). Tree statistics from PAUP are: TL = 290 steps, CI = 0.403, RI = 0.588 and RC = 0.237. The exclusion of †Ostariostoma resulted in the Clupeomorpha (†Ellimmichthyiformes + Clupeiformes) and Elops being embedded among the osteoglossomorph taxa, giving a polyphyletic Osteoglossomorpha. †Lopadichthys is recovered as the sister group of Elops, †Sinoglossus is removed from being a derived osteoglossid as in the second analysis (grouped with Arapaima and Heterotis) and instead is pushed basally to a position between Amia (the designated non-teleost outgroup) and all the other taxa (Elops, Clupeomorpha and all other osteoglossomorphs). The rest of the taxa are positioned fairly similarly to the previous analysis. Figure 20. View largeDownload slide The single most parsimonious cladogram resulting from analysis 3. Figure 20. View largeDownload slide The single most parsimonious cladogram resulting from analysis 3. Effect of the identification of epurals and uroneurals Although we initially identified a single epural and a single uroneural being present in the caudal fin of the holotype of †Lopadichthys (character 68, state 1), the difficulty in distinguishing epurals and uroneurals led us to examine the effect this character would have on the resulting tree if we had coded this taxon as state 2 (epurals absent), instead identifying the bones as two uroneurals. The character state for the number of uroneurals (character 66, state 1) did not change, as the state is ‘two or one.’ Changing character 68 from state 1 to 2 only in †Lopadichthys, with the rest of the data as in analysis 2, resulted in eight most parsimonious trees (TL = 291, CI = 0.405, RI = 0.0.596, RC = 0.242); in the consensus tree (Fig. 21), compared to analysis 2 (Fig. 19), Elops and †Lycoptera both move basally, †Shuleichthys and †Wilsonichthys form a clade with Hiodon and †Eohiodon (as in one tree in analysis 2), †Ostariostoma unites with the Clupeomorpha and †Xixiaichthys becomes the sister group to †Joffrichthys (Fig. 21). Figure 21. View largeDownload slide Majority (50%) rule consensus of eight most parsimonious trees resulting from analysis 2 with the state of character 68 (number of epurals) changed from one present to epurals absent for †Lopadichthys gen. nov. Figure 21. View largeDownload slide Majority (50%) rule consensus of eight most parsimonious trees resulting from analysis 2 with the state of character 68 (number of epurals) changed from one present to epurals absent for †Lopadichthys gen. nov. Changing character 68 from state 1 to 2, only in †Lopadichthys, with the rest of the data as in analysis 3 (excluding †Ostariostoma), resulted in a single tree (TL = 290, CI = 0.403, RI = 0.587, RC = 0.237), that is identical to the single tree found in analysis 3 (Fig. 20). Therefore, the coding of this character for this taxon makes no difference to the resulting tree if †Ostariostoma is not included in the analysis. Based on the differences in results for the second analysis, removing the character of number of epurals from the analysis might be warranted. However, if we use the data for analysis 3 (excluding †Ostariostoma) and exclude character 68, the result is 18 most parsimonious trees (TL = 283, CI = 0.406, RI = 0.592, RC = 0.241) which differ from the original analysis 3 by having Elops moving basally to take its expected position (as a basal teleost) between Amia and all the other taxa, †Sinoglossus going crownwards to become sister to Heterotis (as found in analysis 2), †Lopadichthys moving up the tree to be placed with the mormyroids and notopterids (as found in analysis 2) and †Shuleichthys leaving its sister position with †Wilsonichthys and instead forming a polytomy with (Hiodon + †Eohiodon) and (†Wilsonichthys + Clupeomorpha) (Fig. 22). Figure 22. View largeDownload slide Majority (50%) rule consensus of 18 most parsimonious trees resulting from analysis (†Ostariostoma excluded) with character 68 (number of epurals) excluded from the analysis. Figure 22. View largeDownload slide Majority (50%) rule consensus of 18 most parsimonious trees resulting from analysis (†Ostariostoma excluded) with character 68 (number of epurals) excluded from the analysis. DISCUSSION Phylogeny The results of the phylogenetic analyses reported here, in particular the polyphyletic nature of the Osteoglossomorpha found as a result of recoding of characters and inclusion of different outgroups, demonstrate that we do not yet have a robust phylogeny for this group. Much of the problem with the analyses of this group is that many characters that are considered useful for living osteoglossomorphs are rarely observable in the fossil members, such as the form of the basibranchial bones and associated toothplates. An assessment of the characters causing the different trees in the different analyses reveals that there is very little data contributing to some of the relationships found. For example, there is no unique character to support the inclusion of †Lopadichthys in a clade with Notopteridae, Mormyroidea and †Palaeonotopterus (analyses 1 and 2), nor is there any unique character to support the sister-group relationship of †Lopadichthys with Elops (analysis 3). However, all these four taxa share with †Lopadichthys the condition of character 1 (state 1), in which the temporal fossa is present and the exoccipital contributes to its border. †Ostariostoma is also part of this clade in analyses 1 and 2, but does not have this condition of the temporal fossa. However, it shares with mormyroids and notopterids the condition of the infraorbital canals having the sensory canal open in a gutter (character 24, state 1) rather than enclosed in a bony canal; †Lopadichthys does not have this condition of the sensory canal. Forey & Hilton (2010: appendix 1) listed and assessed the characters proposed as synapomorphies for the Osteoglossidae and subgroups within that family. The removal of †Joffrichthys from Osteoglossidae removes one of the previously anomalous characters for this genus, as Forey & Hilton (2010) had noted that †Joffrichthys lacked the reduced subopercle of Osteoglossidae. The phylogenetic relationships of the Osteoglossomorpha, using more or less the same morphological character data set, have been studied by numerous authors (e.g. Li et al., 1997; Hilton, 2003; Zhang, 2006; Murray & Wilson, 2005; Wilson & Murray, 2008; Murray et al., 2010). In the last iteration (Murray et al., 2016), there was increased resolution of the resultant tree compared to the previous analyses, but no great difference in placement of the taxa within the tree. However, a combined molecular and morphological analysis by Lavoué (2016), using an earlier iteration (from Murray et al., 2010) of the same morphological data set as used here, found some different relationships among taxa; in particular, Pantodon was excluded from the Osteoglossidae, and instead became part of a tricotomy with †Ostariostoma and all other Osteoglossiformes (Lavoué, 2016: fig. 2). As noted by Lavoué (2016), this placement of Pantodon has implications for the fossil genus †Singida, which has been placed close to Pantodon in some analyses. We here recovered Pantodon as sister to †Singida in our third analysis when †Ostariostoma was excluded (Fig. 20), but in the other two analyses, both of which included †Ostariostoma, Pantodon was recovered with Arapaima and Heterotis with or without †Sinoglossus (Figs 18, 19). The changes that result in the analyses with the recoding of a single character (number of epurals) demonstrate the difficulty of determining the phylogenetic relationships of fossil osteoglossomorphs. Without a developmental series, it is essentially impossible to determine the homologies among the taxa for the various elements in the caudal fin. Although this is somewhat discouraging, we can instead look at the crownwards part of the tree, and note that in this part of the tree the taxa remain relatively much more stable. It seems that it is the older, particularly the Cretaceous, material that is the most problematic. Exclusion of †Joffrichthys from Osteoglossiformes Li & Wilson (1996a) placed †Joffrichthys with the fossil genus †Sinoglossus and the recent genera Arapaima and Heterotis in the subfamily Heterotidinae, of the family Osteoglossidae. They justified the placement of †Joffrichthys within Osteoglossomorpha based on the presence of large teeth on the parasphenoid and basihyal, lack of a supraorbital and supramaxillary bones, and number of branched principal caudal fin rays being 16 or 15. We agree with these characters and †Joffrichthys is clearly an osteoglossomorph fish. However, Li & Wilson (1996a) further placed the genus within Osteoglossiformes, which is not supported by our examination of the specimens and our analysis. They based the ordinal designation of †Joffrichthys on the reduction of the number of hypurals to six, an oval or sub-semicircular shape of the opercle, absence of the uroneural, a developed neural spine on the first ural centrum and only 17 principal caudal fin rays (15 branched). While we agree with three features (the number of hypurals being 6, the opercle having an oval shape and the presence of a neural spine on the first ural centrum), the two others are problematic. Hilton (2003) discussed the identification of epurals and uroneurals in osteoglossomorphs including †Joffrichthys. As he noted, more developmental and histological studies of recent material are needed to determine the identity of the one or two bones lying dorsal to the ural centra in these fish, but they are likely uroneurals, as previously indicated by Taverne (1977, 1978) for many genera, with the second uroneural becoming almost indistinguishably fused with the dorsalmost hypural, at least in Arapaima (Taverne, 1977; Hilton, 2003) and Heterotis (Taverne, 1977). We followed Hilton (2003) in recognizing the single bone dorsal to the ural centra in †J. symmetropterus as a uroneural, contra Li & Wilson (1996a) who identified it as an epural. Wilson & Murray (2008) followed Hilton’s (2003) bone identities, which did not change the relationships of †J. symmetropterus from those found by Li & Wilson (1996a). The character of number of branched principal rays in the caudal fin is the primary reason we remove †Joffrichthys from the Osteoglossiformes. Sixteen branched principal rays in the caudal fin (18 total principal rays) have been accepted by authors as a synapomorphy of Osteoglossomorpha (e.g. Patterson & Rosen, 1977; Hilton, 2003). A reduction to 15 branched principal rays (giving a caudal formula of i,7,8,i) was given as a synapomorphy for Osteoglossiformes (Li & Wilson, 1996a) and is found only in Osteoglossidae and Notopteridae (Hilton, 2003; Wilson & Murray, 2008). Hiodontiforms, mormyrids and basal osteoglossomorphs, including the fossil genera †Shuleichthys (Murray et al., 2010) and †Wilsonichthys (Murray et al., 2016), retain 16 branched principal rays. If †Joffrichthys is not an osteoglossid, the synapomorphies it was thought to share with heterotidines would then be homoplasies or symplesiomorphies. These features are the length of the maxilla being relatively short, a relatively long ventral arm of the preopercle, a large trapezoidal second infraorbital, an anal fin as large as or only slightly smaller than the dorsal fin, and rounded dorsal and anal fins (Li & Wilson, 1996a). The last two characters were homoplastic among the relatively few taxa sampled in the analysis of Li & Wilson (1996a). The first two characters (proportions of the maxilla compared to the mandible and proportions of the two limbs of the preopercle) are the same in the non-osteoglossid †W. aridinsulensis as they are in †J. symmetropterus (DB Brinkman and AM Murray, pers. obs.), so are likely homoplasies as well. The final character, shape of the second infraorbital, is subjective. We consider there to be as much variation among the heterotidines (Li & Wilson, 1996a: figs. 3, 9) as there are similarities with other osteoglossomorphs (e.g. Hilton, 2003: fig. 21). Therefore, we do not consider any of these characters to be more important than the caudal fin ray number in determining the position of †Joffrichthys within Osteoglossomorpha. This is supported in all our phylogenetic analyses, in which †Joffrichthys is not included with the Osteoglossiformes. The caudal skeleton of Osteoglossomorpha The caudal skeleton of fishes is considered to provide a number of phylogenetically useful characters for higher groups within Teleostei (e.g. Schultze & Arratia, 1989), such as the number of principal caudal fin rays, as well as the number of hypurals, epurals and uroneurals, and the presence or absence of a full neural spine on the first preural and first ural centra. The caudal skeleton of osteoglossomorph fishes has been examined by a number of researchers who have discovered that there is significant variation within this group, as well as disagreement in the identification of elements (e.g. Greenwood, 1966; Schultze & Arratia, 1989; Hilton & Britz, 2010). The identification of epurals and uroneurals (see above) clearly can cause changes in the analysis. But the variation in other features might also cause problems. Although caudal fin ray counts are used to support the order Osteoglossiformes, there is variation in this character. Within the genus Hiodon, the caudal fin ray count varies greatly (Schultze & Arratia, 1989; Hilton & Britz, 2010), and within Osteoglossiformes, some taxa have the primitive caudal fin formula of 16 branched rays (Hilton & Britz, 2010). As noted by Hilton & Britz (2010), a simple count of fin rays may not be useful as a homologous character. Quite recently, Taverne (2016b) reported a Palaeocene osteoglossomorph tail skeleton from Angola, assigned to †Ridewoodichthys caheni. He assigned the specimen to Osteoglossidae, based on the caudal fin skeleton being similar to that of several osteoglossids including †Joffrichthys. If our exclusion of †Joffrichthys from Osteoglossiformes (and, therefore, also from Osteoglossidae) is correct, it would indicate that the caudal characters used by Taverne (2016b) to support inclusion of †Ridewoodichthys in Osteoglossidae are more widely distributed than previously thought, and therefore not indicative of osteoglossid relationships. Taverne (2016a) delineated several trends in the evolution of the caudal skeleton for osteoglossomorphs, including the reduction in numbers of hypurals, epurals and uroneurals, and increased fusion between hypurals and ural centra. However, osteoglossomorphs appear to show a mosaic of patterns in these elements with some taxa that are considered more highly derived displaying a mix of characters considered primitive and derived. For example, the living osteoglossid Scleropages has a caudal fin with advanced features such as fusion of the dorsal hypurals with each other and the second ural centrum, a single uroneural and no epurals, but also retains primitive features such as a full neural spine on the first preural centrum (e.g. Taverne, 1977; Zhang & Wilson, 2017). Hilton & Britz (2010) discuss the variation in the neural spine form and number on the first preural centrum, as well as variation in other characters. Taverne (2016b) suggested there were two separate evolutionary lineages of the caudal skeleton within Osteoglossiformes, one leading to modern osteoglossids, and a second leading to the Eocene African genera †Chauliopareion and †Singida. However, Taverne’s (2016b) interpretation of the caudal skeletons of these two African taxa differs from that of Murray & Wilson (2005) in both numbers of hypurals (Taverne indicated five for both taxa; Murray and Wilson were uncertain but suggested there may have been six in each) and uroneurals (Taverne indicated two uroneurals in each, but Murray and Wilson could not determine the number in †Chauliopareion). There is also confusion caused by the identification of elements, such as the epurals and uroneurals as noted above. Although much progress has been made on understanding the developmental history of the caudal skeleton in living osteoglossomorphs, better preserved fossils, such as the Cretaceous osteoglossomorph †Shuleichthys, which clearly preserves the hypurals, uroneurals, epurals and ural centra (Murray et al., 2010) are needed to help resolve the evolution of the osteoglossomorph caudal skeleton. Fossil record of osteoglossomorphs in North America Osteoglossomorph fishes that have been diagnosed on the basis of articulated specimens from Cretaceous and Palaeocene deposits of North America have all previously been included in the Osteoglossiformes or have been placed incertae sedis within the superorder. Articulated (and therefore diagnosable) members of the Hiodontiformes only appear in the fossil record in the Eocene. Fossil taxa previously considered to be osteoglossiforms from North America are the Cretaceous †Cretophareodus from Alberta, the Palaeocene †Joffrichthys, now with three species, two from Alberta and the other from North Dakota, and the Eocene †Phareodus (Green River deposits of the USA). Taxa that have been placed incertae sedis within the superorder are Chandlerichthys from the middle Cretaceous of Alaska, †Wilsonichthys from the late Maastrichtian of Alberta, as well as †Lopadichthys described here. The only species of North American fossil hiodontiforms that have been recognized include two or three species of †Eohiodon (depending on whether †E. falcatus is considered distinct from †E. woodroofi or not), and †Hiodon consteniorum (Hilton & Grande, 2008). In the analyses using new outgroups, the single tree resulting from our third analysis (excluding †Ostariostoma) and one tree from the second analysis (including †Ostariostoma) grouped the Chinese †Shuleichthys and the Canadian †Wilsonichthys in a clade with Hiodon + †Eohiodon. This result indicates that some of the taxa previously placed incertae sedis within the superorder may actually be better placed within Hiodontiformes. Although the current data are not strong enough to support this, if correct these taxa would provide part of the missing (Cretaceous) record for the order. Data on the early history of osteoglossomorphs in the Cretaceous and Palaeogene of North America provided by articulated specimens are supplemented by isolated elements from vertebrate microfossil sites. In addition to adding to our understanding of the diversity of osteoglossomorphs, this microfossil material helps to document more fully the stratigraphic and geographic distribution of the group. In some cases, the isolated osteoglossomorph elements can be referred to taxa based on comparison with the same elements in articulated specimens. In others, they indicate the presence of previously unrecognized fishes. One of the osteoglossomorph taxa from the Late Cretaceous of the Western Interior represented by articulated specimens, †Wilsonichthys, is also represented by isolated elements from vertebrate microfossil localities preserved in the same beds as the articulated specimens. These elements differ significantly from those of the two new Palaeocene taxa described above. Dentaries of †Wilsonichthys differ from those of †J. tanyourus and †Lopadichthys (Figs 16, 23) in being deeper and having a much deeper symphysis. Also, the sensory canal pores are located midway on the side of the dentary, rather than near the base, and the teeth are smaller and blunter. Abdominal centra differ from those of †Lopadichthys in having autogenous parapophyses. The rib articulates with the parapophysis, rather than in a pit on the centrum posterior to the parapophysis. Isolated elements from vertebrate microfossil localities also document the presence of otherwise unknown members of Osteoglossomorpha. Hiodontids were recognized from the Cretaceous of the Western Interior of North America on the basis of centra (Brinkman & Neuman, 2002; Brinkman et al., 2013, 2014). These were referred to Hiodontidae on the basis of apomorphic features of the anteriormost centrum, as well as a similarity in the general morphology of the more posterior abdominal centra. Tooth-bearing elements of hiodontids have not yet been identified, although a dentary from the late Campanian Dinosaur Park Formation with an arrangement of teeth similar to that of extant Hiodon may represent this family (Fig. 23F). Significantly, the proportions and arrangement of teeth on this dentary are also similar to †J. tanyourus. Figure 23. View largeDownload slide Comparative photographs of the lower jaws of several Cretaceous osteoglossomorphs. A, †Wilsonichthys aridinsulensis, holotype, TMP 2012.020.1493 from the Maastrichtian Scollard Formation, B, TMP 95.108.61, from Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, C, uncatalogued specimen in the collections of the Museum of Northern Arizona, from the late Turonian Smoky Hollow Member of the Straight Cliffs Formation, Utah, USA, D, †Coriops, TMP 1986.43.33 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, E, TMP 90.119.35A2 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, and F, TMP 2004.19.1 from Dinosaur Park Formation, Onefour, Alberta, here assigned to †Joffrichthys tanyourus sp. nov.. Scale bars = 2 mm. Figure 23. View largeDownload slide Comparative photographs of the lower jaws of several Cretaceous osteoglossomorphs. A, †Wilsonichthys aridinsulensis, holotype, TMP 2012.020.1493 from the Maastrichtian Scollard Formation, B, TMP 95.108.61, from Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, C, uncatalogued specimen in the collections of the Museum of Northern Arizona, from the late Turonian Smoky Hollow Member of the Straight Cliffs Formation, Utah, USA, D, †Coriops, TMP 1986.43.33 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, E, TMP 90.119.35A2 from the Dinosaur Park Formation, Dinosaur Provincial Park, Alberta, and F, TMP 2004.19.1 from Dinosaur Park Formation, Onefour, Alberta, here assigned to †Joffrichthys tanyourus sp. nov.. Scale bars = 2 mm. An additional osteoglossomorph taxon present in Late Cretaceous vertebrate microfossil assemblages, but not known from articulated material, is †Coriops. The genus †Coriops was erected by Estes (1969) on the basis of basibranchial toothplates with blunt crushing teeth. It was initially identified as a member of the Albulidae, which are similar to osteoglossomorphs in having basibranchial toothplates. However, Brinkman & Neuman (2002) recognized that the basibranchials of †Coriops are similar to those of extant osteoglossomorphs such as Scleropages in the presence of a network of bone forming the base of the toothplate; based on this they included †Coriops in the Osteoglossomorpha. Dentaries, vertebrae, basioccipitals and quadrates that occur in the same localities as the basibranchial elements were also referred to †Coriops by Brinkman & Neuman (2002) on the basis of comparison with extant osteoglossomorphs and size-frequency distributions. These elements are similar to those of †J. tanyourus and †Lopadichthys, supporting the identification of these elements as belonging to an osteoglossomorph. The dentaries of †Coriops (Fig. 23) are similar to those of †J. tanyourus (Fig. 16B) in the presence of multiple rows of teeth with the teeth of the lateral row being largest. The tooth row of †Lopadichthys (Fig. 16C, F) is not fully exposed, so it is uncertain whether or not multiple rows of teeth were present in that taxon. The dentaries of †Coriops differ from those of both †Lopadichthys and †J. tanyourus in being deeper and having a larger symphysis (Figs 16, 23). Premaxillae were referred to †Coriops on the basis of the presence of an arrangement of teeth similar to that of the dentary. These are similar to the new osteoglossomorphs described here in having a rounded anterior portion and a more rod-like posterior portion. Abdominal centra of †Coriops are similar to those of †Lopadichthys (Fig. 12) in that the anteriormost centrum lacks parapophyses and ribs are either absent or articulate with the ventralmost edge of the centrum, while in the more posterior abdominal centra long laterally directed parapophyses are present, and the ribs articulated with a large pit on the lateral surface of the centrum posterior to the parapophyses. Also, the neural arches of anterior centra are autogenous, and the dorsal edge of the parapophyses reaches the rib articular pit, and is very narrow in dorsal view. Neuman & Brinkman (2005) noted that the elements referred to †Coriops are similar to those of Hiodon in some features, suggesting a possible relationship between the two. The arrangement of teeth on the dentary of these two genera is similar in that the teeth of the outside row are largest and are set on the laterally facing portion of the jaw, the teeth of the innermost row are slightly smaller, and those of the intermediate rows are the smallest. The centra are similar in that the ribs articulate directly on the centrum posterior to the parapophyses. These similarities support the suggestion made above that at least some of the taxa from the Cretaceous and Palaeocene that were previously placed incertae sedis within the superorder may actually be better placed within Hiodontiformes and thus be part of a hiodontiform radiation. However, it is not possible to include the isolated elements assigned to †Coriops in a phylogenetic analysis, so additional specimens are necessary before the phylogenetic significance of these similarities can be fully evaluated. Distribution of osteoglossomorphs in the Cretaceous and Palaeogene of North America The data on diversity and distribution of osteoglossomorphs in North America based on articulated specimens are supplemented by data from isolated elements, and together these sources of information provide an understanding of the early history of osteoglossomorphs in North America that is more complete than either set of data on its own. The oldest osteoglossomorph in the non-marine deposits of North America is †Chandlerichthys from the mid-Cretaceous of Alaska. Isolated elements from vertebrate microfossil localities also occur in mid-Cretaceous beds in Utah (Brinkman et al., 2013), showing that osteoglossomorphs were widely distributed in North America at this time. Centra from the Cenomanian Cedar Mountain Formation of Utah indicate the presence of two osteoglossomorphs, †Coriops and a generically indeterminate hiodontid. †Coriops extends through the Late Cretaceous into the Palaeocene, although it has a strong latitudinal pattern and so is only seen in southern localities during times of relatively cool global temperatures (Brinkman et al., 2013). It is the most abundant, and one of the largest, teleosts in vertebrate microfossil localities in the late Campanian Belly River Group of Alberta. Hiodontid centra first occur in the Cenomanian Cedar Mountain Formation of Utah and extend to the late Maastrichtian of the Hell Creek Formation, where they are represented by two taxa. Fish centra from microvertebrate sites in the Late Cretaceous Nemget Formation of Mongolia have also been included in the Hiodontidae (Newbrey et al., 2013), providing evidence for interchange of non-marine aquatic vertebrates between Asia and North America in the Cretaceous. Isolated elements referred to†Wilsonichthys by Murray et al. (2016) show that this taxon first appears in the Turonian of Utah. The genus extends to the late Maastrichtian but is not yet known from the Palaeocene. †Cretophareodus is represented by a single articulated specimen, a partial skeleton preserved in an ironstone concretion from exposures of the Belly River Group in Dinosaur Provincial Park, Alberta. Centra are not preserved in this specimen, and the dentaries are incomplete, but based on the preserved portion of the dentaries, †Cretophareodus appears to be taxonomically distinct from †Coriops. No isolated elements have been referred to †Cretophareodus so no additional information on its stratigraphic and geographic distribution is available. Palaeogene osteoglossomorphs represented by articulated specimens include the Palaeocene †Joffrichthys, now with three species, two from Alberta and the other from North Dakota, our new taxon, †Lopadichthys, from the Palaeocene of Alberta, and the Eocene †Phareodus (Green River deposits). Although isolated centra demonstrate the presence of hiodontids in North America in the Cretaceous, no remains of hiodontids have been recovered from the Palaeocene. The previously reported presence of a hiodontid in the Palaeocene, based on isolated elements from the Paskapoo Formation that were referred to Hiodon by Wilson (1980), was later refuted with the understanding that these remains actually represent †Joffrichthys (Li & Wilson, 1996a). Hiodontids are absent in an assemblage of isolated fish elements from a vertebrate microfossil locality of early Eocene age (Divay & Murray, 2016). Thus, the remains of †Eohiodon falcatus from the late early Eocene beds of the Green River Formation provide the first record of the group in the Cenozoic of North America. They are also present in lake beds of middle and late Eocene age of Washington State, USA, and British Columbia, Canada (Li et al