A platysomid occurrence from the Tournaisian of Nova Scotia

The Hangenberg extinction has been hypothesized as a first order event in vertebrate evolution; however, information on the earliest Carboniferous vertebrate fauna, crucial in evaluating biodiversity changes, is scarce. Post-extinction recovery has been suggested as the driver of ray-finned fish (actinopterygian) richness increase and differentiation in the Carboniferous. Under this model, actinopterygian postcranial morphology differentiates in the second stage of their radiation. Here, we report on a platysomid occurrence from the Tournaisian of Nova Scotia, Canada. Despite long-standing taxonomic issues with deep-bodied actinopterygians, this specimen represents the earliest known occurrence of one such fish. Its presence in the earliest Carboniferous indicates that actinopterygians were already postcranially differentiated in the aftermath of the Hangenberg. Moreover, this specimen suggests that earliest Carboniferous actinopterygians used multiple locomotory modes; recent data from later Carboniferous taxa suggest that actinopterygian locomotory modes proliferated throughout the Carboniferous. Taken together, these data suggest that early Carboniferous actinopterygians were morphologically, ecologically, and functionally diverse.

www.nature.com/scientificreports/ remains generally poor. This is especially the case for Platysomus, the other putative Carboniferous origin 9 of deep-bodied actinopterygians. This genus has needed revision for a long time 11,13,25,26 because it is broadly inclusive of Palaeozoic deep-bodied actinopterygians. Because of this inclusiveness, Platysomus 11,26 occurs through most of the Palaeozoic (Viséan-Lopingian). Although the genus Platysomus does not include post-Palaeozoic taxa, a putative platysomid-bobasatraniid lineage 13,27 reaches the Mesozoic. Whereas the Viséan Platysomus superbus is the earliest described species of Platysomus 11 , the first Platysomus occurrence is controversial. Obruchev 28 lists '?Platysomus' among the early Tournaisian fish fauna of the Bystrianskaia Formation but includes no description, illustration, or specimen number. Zidek 13 is skeptical of it and Sallan and Coates 4 do not include it in their occurrences dataset. A better understanding of the first occurrence of platysomids (and deep-bodied actinopterygians more broadly) should help resolve the timing of key morphological, ecological, and functional changes in the Devonian-Carboniferous actinopterygian fauna. Here, we describe a platysomid specimen from the Tournaisian Horton Bluff Formation of Walton, Nova Scotia.

Material and methods
Institutional abbreviations. BWC . 2a) are contained in the Kennetcook Basin. In the area of Walton, these exposures are members of the Tournaisian Horton Group (Fig. 2b); these are succeeded unconformably by Viséan members of the Windsor Group to the South. West-northwest and East-northeast of Walton, along the South margin of the Minas Basin, the Horton Group is contacted by the Triassic Wolfville Formation. Although the contemporaneous locality of Blue Beach (Fig. 2b) has yielded an abundance of vertebrate fossils 29 , Walton has not previously been known as a fossiliferous locality. Initial efforts to recover additional fossils from Walton wharf were unsuccessful. However, recent prospecting by one of us (C.F.M) below the headland, a short distance towards the Walton lighthouse from the Walton wharf (Fig. 2a), has yielded additional vertebrate material preserved in a large slab. As this slab was ex-situ, the specific strata and locality cannot be discussed with precision. Given the close proximity of this slab to the Walton wharf, it seems likely that NSM 017.GF.017.001 is from the upper member of the Horton Bluff Formation rather than the middle member of the Cheverie Formation.
The specimen is preserved in a highly indurated grey silty sandstone. The matrix does not exhibit any reaction to acid. The lithology of the specimen is consistent with the lithology of the Horton Bluff Formation 29 and distinct from the red sandstones, evaporites, and carbonates of the Windsor Group and the brownish-red sandstones of the Wolfville Formation 30 . These Windsor Group and Wolfville Formation strata can be easily distinguished visually from the Horton Group in the Walton area.
Thus, the specimen is middle to late Tournaisian in age.

Description
Squamation. Flank scales. At least 30 pre-caudal scale rows are preserved. The flank scales (fs, Fig. 4) are generally dorsoventrally deep and arranged in nearly vertical rows (Fig. 3). Dorsally, the rows are nearly straight and directed slightly anteriorly, and curve again anteriorly towards the ventral margin. The curvature is more pronounced in more posterior rows; the posterior-most rows are nearly sigmoidal. Scale size generally increases anteriorly and medially ( In lateral line scales, the ornamentation takes a similar but more drastic curve around the lateral line notch. Dorsal to the level of the notch, this forces the ornamentation to form concentric, forward-opening arcs that straighten out in successive rows towards the anterior, sometimes pinching out ridges in the middle of the dorsal part of the scale. Ventral to the lateral line notch, the ornamentation straightens and becomes vertical.   ; these are longer than their distal neighbours. Each dorsal guard scale overlaps 2-3 lepidotrichia. The lepidotrichia vary from rectangular to nearly square from the anterior to posterior of the fin margin. All well-preserved lepidotrichia bear a single median ridge. A single dorsal fin radial (ra, Figs. 3 and 5) is displaced through the scale rows at the base of the dorsal fin. This is slender, but its distal end flares out. Broad, slightly elevated ridges running parallel to the left and right of the displaced radial may represent the position of other radials.
Anal fin. The preserved anal fin includes 56 fin rays (afr, Fig. 3, 4 and 6) in a series reaching the caudal peduncle. The dorsal most ray contacts the first three caudal basal fulcra (cbf, Figs. 3 and 7). The anal fin cannot be evaluated for the presence of fringing fulcra or lepidotrichial bifurcation because it is incomplete distally. The fin rays are segmented as many as 23 times without any evidence of bifurcation in the preserved extent.
The proximal-most lepidotrichia are overlapped by the anal guard scales (ags, Figs. 3, 4 and 6). Each anal guard scale (ags, Figs. 3, 4, and 6) overlaps 2-4 lepidotrichia. These are unsegmented and approximately twice the length of the next more distal lepidotrichium in the ray. Anteroventral lepidotrichia are rectangular and posteroventral lepidotrichia are nearly square. Lepidotrichia devoid of ornamentation, lepidotrichia with a single median ridge, and lepidotrichia with paired ridges on their posterior half are distributed irregularly.
In general, anteroventral lepidotrichia have more ridges than posterodorsal lepidotrichia, although the number of ridges varies within rays. The proximal-most lepidotrichia of the 30 dorsoposterior-most rays bear a single ridge of ornamentation, with lepidotrichia devoid of ornamentation occasionally present distally. The single ridge of ornamentation widens in proximal lepidotrichia anteroventrally, before splitting into paired ridges in the sixth proximal-most lepidotrichium of the 30th posterodorsal-most ray. Paired ridges are gradually found

Discussion
The characters of deep-bodied actinopterygians were recently reviewed by Sallan and Coates 9 . In previous analyses, character selection has emphasized the dermal skeleton and feeding structures, particularly the shape of the animal and its skull bones, and the dentition (e.g. 20 ). This character selection seems to reflect the limitations of the fossil record, as many deep-bodied actinopterygians (including this specimen) are preserved in flattened lateral aspect. However, changes in body depth may drive broad, linked changes in morphology (e.g. 34 , Fig. 152-155), affecting the independence of these characters. Significantly, Sallan and Coates 9 found that many characters used to evaluate the relationships of deep-bodied actinopterygians are homoplastic, dating back to Traquair's 10 typological argument that Platysomidae are specialized Palaeoniscidae.
The situation is especially dire for Platysomus. Zidek 13 (page 167) summed up the situation: "… it is obvious that without a revision of all the noted species it cannot be decided what is and what is not a Platysomus. " Clearly, this statement only makes sense in context of deep-bodied actinopterygians-many things are verifiably not Platysomus. But there is very little, if any, distinction in meaning between Platysomus and "platysomid". This is because platysomid genera cannot be distinguished from Platysomus (exemplified by Zidek's 13 transfer of Schaefferichthys to Platysomus) and because authors have continued to add species level diversity to Platysomus, noting the issues with the taxon, to avoid performing a genus-wide revision 13,26 . That this revision has been noted as necessary for more than a century e.g. ( 11,25 ) but has not yet been satisfactorily completed shows the difficulty and size of the problem. A further problem was created when Campbell and Phuoc 27 united Platysomus gibbosus, the type species of Platysomus, with Ebanaqua and deep-bodied Triassic taxa in the Bobasatraniiformes without taxonomic revision. This carries the implication that at least two potentially distinct evolutionary lineages (Carboniferous platysomids and bobasatraniids) are present in Platysomus 27 . We note that although Platysomus and platysomids have generally been treated as monophyletic in analyses incorporating Ebanaqua (Zidek 13 distinguishes between 'higher' and 'lower' platysomids within Platysomus and Sallan and Coates 9 list Platysomidae-Bobasatraniiformes as a potential independent origin of deep-bodied actinopterygians), this monophyly of Platysomus has not been rigorously tested in phylogenetic analysis 26 . The relationships of species contained in Platysomus are unclear and platysomids, properly defined, may be inclusive of taxa that are not strongly deep-bodied (c.f. Styracopterus, Fouldenia, and the Eurynotiformes). Thus, here we take a similar approach to previous authors: we note the need for a revision of Platysomus and emphasize comparisons among and between platysomids sensu lato. Future revision of Platysomus and platysomids sensu lato may produce a clear, monophyletic definition of Platysomus and a platysomid clade, necessitating a stricter approach. In the interim and for the remainder of this manuscript, we use platysomid only in its broadest sense.
Given these general issues in character selection and the taxonomic issues surrounding Platysomus, it is not surprising that character selection for platysomids is also thorny. As noted above, many previously used characters (e.g. phyllodont dentition) are significantly homoplastic 9 . Some homoplastic characters are still useful operationally: Sallan and Coates 9 contrast the heavy and highly imbricated fringing fulcra of eurynotiforms and the minute fringing fulcra of platysomids, which is useful when distinguishing between contemporaneous taxa. www.nature.com/scientificreports/ But minute fringing fulcra are also present in Discoserra 18 , so this is not an unambiguously platysomid character even among deep-bodied actinopterygians. Scale morphology and ornamentation appears significant. Sallan and Coates 9 distinguish between the central scale pegs of eurynotiforms and the anterodorsally placed pegs of Platysomus (closer to the condition in fusiform non-neopterygian actinopterygians). All members of Platysomus share a scale ornamentation that is usually described as fine, vertical striations 9 . However, this ornamentation can be more complex. In NSM 017. GF.017.001 specimen (as is figured for Platysomus striatus 35 , Plate 17), the ornamentation is slightly curved, pinching out some striae and opening space for others. This is similar to the condition of anterior scale ornamentation observed in some species of Mesopoma (i.e. M. smithsoni and M. pancheni) in which anterior striae curve posteriorly as they run ventrally, although the posterior scale ornamentation is distinct 17 .
Characters distinguishing the multiple lineages within Platysomus are better established but can also be problematic. Campbell and Phuoc 27 proposed 15 characters supporting their platysomid-bobasatraniid lineage, but Zidek 13 noted that many cannot be used because they are widespread among actinopterygians or because alternative states cannot be distinguished. Zidek 13 added several characters, but some also have problems: we cannot distinguish the parasagittal rows of caudal fulcra proposed as diagnostic for the platysomid-bobasatraniid from the pairs of caudal fulcra arranged in chevrons that are widespread in non-neopterygian actinopterygians 36 in the data given.
Because the overall shape body shape of NSM 017.GF.017.001 is most similar to members of Platysomus (Fig. 1b,c) and the scales bear distinctive sub-linear ridges of ornamentation (Fig. 4a,b), the specimen appears to be a platysomid. Unfortunately, characters of the skull and scales that might be useful in placing this specimen are not preserved because of specimen breakage and scale imbrication, respectively. Similarly, specimen preservation precludes evaluation of the characters defined by Campbell and Phuoc 27 and Zidek 13 for the platysomidbobasatraniid lineage ('higher' platysomids). No described platysomid-bobasatraniids have fringing fulcra on the leading edge of the caudal lobe, so the putative fringing fulcra in this specimen may be a point of difference.
More significantly, dorsal and anal guard scales are present in this specimen. Some specimens of Platysomus tenuistriatus (SM E4949 a and SM E4949b, Supplementary Table S1) also have anal guard scales (Fig. 8a,  ags) and an undescribed platysomid from Bear Gulch, FMNH PF 10,792 has both dorsal and anal guard scales (Fig. 8b, dgs and ags). This allows for comparison with other deep-bodied taxa, although this comparison is not comprehensive. Guard scales have not been described before in deep-bodied actinopterygians, despite their presence in Platysomus tenuistriatus, and may be underreported. Dorsal and anal guard scales are also present in Lineagruan spp. 20 ; however, this specimen differs from Lineagruan spp. in body depth and the absence of pectinate posterior scale margins.
The extensive differences between these platysomid specimens include the guard scales. Whereas 2-3 lepidotrichia contact each dorsal guard scale and 2-4 lepidotrichia contact each anal guard scale in NSM 017. GF.017.001, the ratio of guard scales to lepidotrichia is approximately 1:1 in Platysomus tenuistriatus and 1:2 in FMNH PF 10792. Nevertheless, the presence of dorsal and anal guard scales unites NSM 017.GF.017.001 and some members of Platysomus, reinforcing this specimen's identity as a platysomid. Furthermore, Platysomus tenuistriatus and FMNH PF 10792 are from the Viséan and Serpukhovian, respectively, so NSM 017.GF.017.001 appears more similar to close contemporaries than later Carboniferous platysomids.
The presence of a platysomid in the Tournaisian has consequences for our understanding of actinopterygian diversification and differentiation. This specimen likely predates the late Tournaisian Fouldenia, the oldest known eurynotiform taxon 9 , making it the earliest described deep-bodied actinopterygian. Although no morphometric analysis has been performed here, this early deep-bodied actinopterygian would seem to expand the morphospace occupation of Tournaisian actinopterygians along the post-cranial axis in the dataset of Sallan www.nature.com/scientificreports/ and Friedman 8 and weaken the interpretation that cranial disparity increases before post-cranial disparity in Palaeozoic actinopterygians. Locomotory strategies are important in this context of actinopterygian post-cranial disparity. Many of the wide range of locomotory strategies used by modern actinopterygians-including both body and/or caudal fin (BCF) locomotion and median and/or paired fin (MPF) locomotion-are first employed by Carboniferous actinopterygians. The elongate body-plan of Carboniferous tarasiids 37 evinces anguilliform (eel-like) locomotion 11 , an additional BCF locomotory mode. Fusiform non-neopterygian actinopterygian locomotion was long assessed as similar to chondrosteans 38,39 , as these taxa seem to share negatively buoyant armature, a heterocercal tail, and limited pectoral fin flexibility. Nevertheless, Coates and Tietjen 39 recently revealed a flexible "rowing" pectoral fin in Trawdenia planti, which indicates that pectoral fin based MPF locomotory modes were present in at least one fusiform Carboniferous actinopterygian.
Early Carboniferous deep-bodied actinopterygians also imply new locomotory strategies, since a laterally compressed, deep body is associated with unsteady swimming and low straight-line speed but high maneuverability 40,41 . Furcacaudiform thelodonts have been suggested as Devonian explorers of a deep bodyplan 41 , and the Devonian acanthodians Brochoadmones 42 , Ptomacanthus 43 , and Cassidiceps 44 have been described as deep-bodied, moderately deep-bodied, and relatively deep-bodied respectively. However, 'deep-bodied' is a subjective descriptor that encompasses a wide range of anatomy, and clearly these animals explore a deep bodyplan in a different way from actinopterygians. These taxa lack the long, posteriorly placed dorsal and anal fins which are ubiquitous in deep-bodied actinopterygians (although we note the confluence of the anal and caudal fins in Brochoadmones 42 ), and which modern deep-bodied actinopterygians undulate to effect ballistiform locomotion 40,41 . Although these modern taxa are highly specialized, the placement and flexibility of the anal and dorsal fins of Ebanaqua 27 and the bulky musculature of these fins interpreted in the description of Discoserra 18 suggest that Palaeozoic deep-bodied actinopterygians used some form of this gait in conjunction with other fins. Depending on the configuration of the pectoral fins and buoyancy 39 , the pectoral fins might be used for locomotion and fine-positional control 27 .
The deep body-plan and the configuration of the dorsal and anal fin in NSM 017.GF.017.001 is a strong signal indicating a more maneuverable locomotory strategy relative to fusiform contemporaries 40,41 . This does not imply the full exploration of a ballistiform gait, but the long dorsal and anal fins in the specimen and observed modifications for dorsal and anal fin locomotion in other Palaeozoic deep-bodied taxa 18,27 suggest that this MPF locomotory mode was used to some extent. Although the pectoral fins of this specimen are not preserved, the analysis of Coates and Tietjen 39 indicates that pectoral fin based modes of locomotion exist in Palaeozoic actinopterygians. Thus, the number of locomotory modes known for actinopterygians increases from one (a form of BCF locomotion) by the Famennian to three (BCF locomotion and MPF locomotion with the anal/dorsal fin and the pectoral fins) by the Tournaisian at the latest, and to at least five (with pectoral fin rowing in Trawdenia and anguilliform locomotion in tarasiids) later in the Carboniferous. This could support the hypothesis that the post-Devonian actinopterygian radiation encompasses locomotory differentiation. Since tight correlations between changes in feeding and locomotory structures have been previously observed in Actinopterygii 8,45 , one possibility is that locomotory changes opened new feeding opportunities in deep-bodied lineages during their invasion of new niches and ecospace as part of broader turnover among (especially durophagous 7 ) vertebrate predators 4 , congruent with the more general 'feeding-first' model considered by Sallan and Friedman 8 . This specimen and other Carboniferous taxa showing ecological and locomotory differentiation provide only a minimum age for these changes. Analysis of Devonian (especially Famennian) actinopterygian locomotory and feeding structures will be critical in evaluating the feeding-first hypothesis and understanding actinopterygian response to ecosystem change.
This occurrence might also provide an additional minimum age for actinopterygian diversification. When Platysomus spp. are included in phylogenetic analyses, they tend to be recovered as deeply nested in a radiation of post-Devonian actinopterygians (e.g. 20,21,23,24 ). Indeed, Giles et al. 21 recover Platysomus superbus as the stratigraphically oldest member of the actinopterygian crown, so its occurrence (~ 334 Myr) provides the minimum age of the actinopterygian crown in their fossil tree. This is younger than their preferred molecular tree, which places the appearance of the actinopterygian crown at 359.9 Myr. If the phylogenetic placement of Platysomus superbus in this analysis reflects platysomids more generally (Platysomus superbus is usually considered a nonbobasatraniid/'lower' platysomid 13 ), then the recovery of this Tournaisian platysomid helps resolve this discrepancy between the molecular and fossil tree. This supports the hypothesis that the appearance of the actinopterygian crown occurs later 21 than previously thought and is congruent with the view that a major actinopterygian radiation starts in the Early Carboniferous, post-Hangenberg 4,21,46 . However, it remains possible that critical divergences occur earlier and diversification starts pre-Hangenberg 23 .
Understanding the actinopterygian radiation requires evaluating diversification models (e.g. the explosive, long fuse, and short fuse models of Archibald and Deutschman 47 ), but this is difficult without a better understanding of actinopterygian interrelationships. The observation that an actinopterygian showing morphological and functional disparity is present by the Tournaisian at the latest might be helpful in this context. Genus-wide revision of Platysomus and phylogenetic analysis of deep-bodied actinopterygians in context of other Palaeozoic actinopterygians might further illuminate these problems.

Data availability
All data generated or analyzed during this study are included in this published article and its Supplementary Information files.