Oechsleria unterfeldensis, gen. et sp. nov., a sailfin velifer fish (Lampridiformes, Veliferidae) from the Oligocene of the Unterfeld (“Frauenweiler”) clay pit

These are the first and so far only records of a sailfin velifer fish from the Lower Oligocene of the Unterfeld (“Frauenweiler”) clay pit at Rauenberg (S Germany), and only the fourth fossil skeletal finds of this group worldwide. The new genus and species †Oechsleria unterfeldensis is described in detail, diagnosed, and compared to other fossil and extant representatives of the Veliferidae. It appears to be a comparatively small-sized fish, which differs from the other representatives of this family, apart from body proportions, by having a lower number of vertebrae and dorsal-fin rays, the absence of massive spines in the dorsal and anal fins, four anal-fin pterygiophores in front of the anteriormost haemal spine, and a different morphology and size of various skeletal elements. Amongst others, the following character states are of relevance: a supraoccipital that is laterally sculptured by strong surface ridges; a weak ascending process of the premaxilla that is shorter than in the other veliferid taxa; compound (fused) anteriormost dorsal- and anal-fin pterygiophores, together with spineless dorsal and anal fins with unornamented rays (no spinules), of which the majority are bilaterally paired and both halves are fused only proximally but separate and segmented distally; a short coracoid that does not reach the ventral body margin; a broad and anterodorsally directed pelvic bone that bears a well-developed postpelvic process, and probably also the presence of a short pointed neural spine (in contrast to a distally blunt one and/or a low crest) on the second preural centrum. All in all, the new fossil records remarkably increase the known biodiversity of the Veliferidae. They also expand the known palaeogeographic range of this family as far as to the Western Paratethys. With reference to the occurrence and life habits as deduced from the extant forms, they seem to be another Indo-Pacific (respectively Palaeo-Mediterranean, when referring to the fossil forms) element of the Grube Unterfeld fish fauna with a preference for deeper waters and affinities to temperate to tropical climates.


Introduction
Extant Veliferidae (velifers or sailfin moonfishes) are a small group of pelagic marine fishes that is considered to be the sister group of all Lampridiformes (e.g., Olney et al. 1993;Wiley et al. 1998;Miya et al. 2007). They inhabit near-shore waters of the Indian Ocean and the western part of the mid-Pacific, where they are comparatively rare. Morphologically, they are characterized by a deep and laterally compressed body and long dorsal and anal fins that are each covered at their base by a well-developed scaly sheath (Nelson et al. 2016). There are two extant species: the Spinyfin velifer, Metavelifer multiradiatus (Regan, 1907a), and the Sailfin velifer, Velifer hypselopterus Bleeker, 1879. They differ (a.a.) in the presence of numerous spiny rays in the dorsal and anal fins in the first mentioned species, whilst these are lacking in the second one (Heemstra 1986).
Accordingly, the present work provides significant additions to this extremely sparse fossil record. The finds described herein increase the known biodiversity of the family Veliferidae and also close a gap in their known temporal distribution. With their originating from the Grube Unterfeld locality at Rauenberg (Baden-Württemberg, S Germany), which means from a fossil site within the Western Paratethys, they also considerably expand the known palaeogeographic range of Veliferidae.
Grube Unterfeld is the last fossil site in the Upper Rhine Rift Valley that has been accessible for palaeontological excavations in recent times. It became well known from the numerous plants, invertebrates, and vertebrates that were recovered from there and reported on and described in various popular and scientific contributions (e.g., Maxwell et al. 2016). It is especially famous for partly very rare finds of cartilaginous and bony fossil fish (e.g., Hovestadt and Hovestadt-Euler, 1999;Micklich et al. 2009Micklich et al. , 2016Micklich et al. , 2017Hovestadt et al. 2010). More recent investigations (Monsch and Micklich 2018) clearly pointed out that this extraordinary potential still is far from becoming exhausted. This is particularly underscored by the find of the fossil sailfish portrayed herein. Unfortunately, the Grube Unterfeld fossil site and the restricted area that has remained accessible for excavations, respectively, have fallen into a rather neglected state over the past few years, and a restart of the excavations is expected to be particularly challenging. In any case, there still is some hope that some recent plans (as of July 2021) to this effect will finally succeed.
For more detailed information concerning the geology and stratigraphy of the Grube Unterfeld ("Frauenweiler") locality, see Micklich and Hildebrandt (2010) as well as Maxwell et al. (2016).
Specimens SMNS 87457/103 and 80738/81 (part and counterpart, Fig. 1), were found by the amateur palaeontologist Rudolf Metzler of Weinsberg-Grantschen (close to Heilbronn, S Germany), who also prepared SMNS 87457/103. SMNS 80738/81 was recently prepared by the senior author. According to the finder (R. Metzler, pers. comm.), both parts were recovered in 1994 from the more eastern part of the clay pit in a horizon that lies directly above the lowest layer with septarian nodules (uppermost layer of stratigraphical unit Hdt 23 sensu Micklich and Hildebrandt 2010, layer 23 in the nomenclature formerly used by amateur palaeontologists). In 2001, 1996 respectively, they were sold to the SMNS. HLMD-SMFF 18a, b (part and counterpart, Fig. 2) were found and prepared by the senior author in November 1992. They had been recovered from the northern slope of the pit in layers at some distance above the lowest layer with septarian nodules (probably Hdt 14 or Hdt 15 sensu Micklich and Hildebrandt 2010, layer 18 or "Barsch-Schicht" in the nomenclature formerly used by amateur palaeontologists, respectively).

Preparation
All fossil specimens referred to herein were prepared using the transfer method (for details, see Micklich and Drobek 2007). Radiographs, a photograph of a cleared and stained specimen, and a CT-Scan were used for the extant reference specimens.

Investigation
All specimens as well as the radiographs were examined and documented with a Wild M3Z stereomicroscope with an attached photographic device. Sketch drawings and reconstructions were prepared using the Adobe Photoshop 2020
The orientation of bony processes and articulations follow the terminology of Barel et al. (1976). This, e.g., means that the ending "…ad" always indicates that the described structure points in the direction of the respective element, but that it does not necessarily belong to that element (e.g., "hyomandibulad" means that this structure/element is orientated in the direction of the hyomandibula, but is not located on the latter).
Etymology. Named in honour of the amateur palaeontologists Annette and Harald Oechsler (Waghäusel) for their great merits for this fossil site. Gender feminine.
Diagnosis. Comparatively small-sized veliferid fish with a large head that can be distinguished from all other representative of this family, apart from a different morphology and size of various skeletal elements, by the combination of the following characters, respectively character states: supraoccipital laterally sculptured with strong surface ridges; a comparatively short and weak ascending process of the premaxilla; compound (fused) anteriormost dorsal-fin pterygiophore, spineless dorsal and anal fins with unornamented rays (no spinules); a short coracoid that does not reach the ventral body margin; an uppermost pectoral radial fused to scapula; presence of four anal-fin pterygiophores in front of the anteriormost haemal spine, and probably also the presence of a short and pointed neural spine (in contrast to a distally blunt one and/or a low crest) on second preural centrum. 29-30 vertebrae, of which 19 are caudal ones, up to 9-13 pectoralfin rays, up to 7 pelvic-fin rays, and about 32 dorsal-fin rays, of which first one is shortened and the third one is the longest. Caudal fin with six to seven (possibly up to nine) dorsal procurrent rays, at least six ventral procurrent rays, and 10 + 9 principal rays. Caudal fin endoskeleton with six hypurals, of which the third and fourth are fused to second ural centrum, three epurals, and two uroneurals. Anal fin with about 29 rays of which the sixth and seventh are the longest; spineless pelvic fin; basipterygium anterodorsally orientated, with a well-developed postpelvic process. Scales large and cycloid.
Diagnosis. As for the genus.

General body proportions
With a maximum TL of (estimated) 92.5 mm and an estimated maximum SL of 74.8 mm, the new Grube Unterfeld specimens are smaller in size than the other known fossil and extant forms. The body is laterally compressed and deeper than in most other veliferids. With a HL of 32.8% SL, the head is larger than in †Wettonius angeloi, †Veronavelifer sorbinii (Table 1: HL 25.8% SL, 29.8% SL respectively), and in the extant species, but somewhat smaller than in †Nardovelifer altipinnis (Table 1: HL 41.7% SL). The dorsal fin is more posteriorly located than in †Wettonius angeloi, but more anteriorly than in the other species. The anal fin inserts more posteriorly than in †Wettonius angeloi and Metavelifer multiradiatus, but (partially rather much) more anteriorly than in the other species. The caudal peduncle is deeper than in all other veliferids, but with a span of about 35.7% SL, the caudal fin is smaller than those of †Wettonius angeloi and Velifer multiradiatus, larger than those of †Veronavelifer sorbinii and Metavelifer multiradiatus, and of a similar size as that of †Nardovelifer altipinnis. The scales are large.
Concrete size information on the other species is wanting, but the scales of at least †Nardovelifer altipinnis and Velifer hypselopterus seem to be smaller.
A comparative synopsis of some body proportions of †Oechsleria unterfeldensis and both the fossil and extant veliferid species is given in Table 1.

Neurocranium
This portion of the skull is only preserved in SMNS 80738/81, where most of the respective bones are severely crushed laterally (Fig. 3). As far as can be estimated from these rather poor remains, there may be a NL of about 67% of the HL. According to the fossils, this is significantly less than in †Wettonius angeloi (about 90%; Carnevale and Bannikov 2018: Fig. 2). As for the extant forms, this is also a lower proportion than in Metavelifer multiradiatus (about 79%; radiograph of USMM 394,440), but within the range of Velifer hypselopterus (63-77%; Regan 1907b: Fig. 167;Gregory 1933: Fig. 173; c & s specimen AMS 21840020; CT scan of MNHN-IC-1982-0025). Large parts of the supraoccipital are more or less sufficiently preserved. They clearly exhibit a lateral ornamentation that consists of several antero-dorsal ridges that very much resembles the one described and figured for †Wettonius angeloi (Carnevale and Bannikov 2018). The outline of the forehead of †Oechsleria unterfeldensis seems to be rather steep and encloses an angle of about 70° with the neurocranial base. It is thus slightly steeper than the one of †Wettonius angeloi (about 62°; Carnevale and Bannikov 2018: Fig. 2), whereas it is distinctly steeper than in both †Veronavelifer sorbinii and †Nardovelifer altipinnis (35-38°; Bannikov 1991; Fig. 1; Sorbini and Sorbini 1999; Fig. 1). Once again, it is rather closely similar to the extant Velifer hypselopterus (65-75° according to the

Infraorbital bones and sclerotic ring
Of these elements, only one single infraorbital is preserved in specimen SMNK-87457/103 (Fig. 3). It is subrectangular in shape, and its sensory canal runs in an open groove.

Jaws, palatoquadrate apparatus and opercular series (Figs. 3, 4)
The premaxilla seems to be toothless (Fig. 3). It has a comparatively weakly developed and short ascending process that is slightly less than 1.3 times longer than the alveolar one. This is distinctively less than in the other fossil and extant species (between 2.05 in †Nardovelifer altipinnis and up to 5.34 in Metavelifer multiradiatus; see Sorbini 1999 andOlney et al. 1993: Fig. 5). Both processes enclose an angle of more than 110°. A postmaxillary process is not discernible in the available materials. The maxilla is robust and has an almost straight longitudinal axis (Fig. 4c). It measures slightly more than 36% of the HL. Only the anterior articular process (nomenclature sensu Keivany 2014) is completely preserved, but according to the visible remains, there also was a posterodorsally orientated and distally somewhat pointed posterior articular process, which medially wraps around the ascending process of the premaxilla. There is no supramaxilla. The lower jaw is contained about 2.2 times in the head length. The dentary is toothless (Fig. 4e). The coronoid and the ventral processes enclose an angle of less than 60°. The maximum extension of the angulo-articular accounts for more than 64% of the complete lower jaw length. It is broadened dorsoventrally, and completely fills the space between the coronoid process and the ventral one of the dentary. The lower jaw articulation facet is located close to its posteriormost end. The retroarticular is rounded-squarish in outline and comparatively small: its maximum diameter reaches about 9.4% of the complete mandibular length. As for the palatoquadrate apparatus, only parts of the ectopterygoid, the quadrate, and the symplectic are preserved (Figs. 3, 4g). The ectopterygoid is elongate with a straight anterior and a broadened and rounded posterior margin. The quadrate has a maximum length of slightly more than 25% of the HL. It is posteriorly broadened with dorsal and ventral margins that enclose an angle of slightly less than 60°. The symplectic scarcely protrudes beyond the margin of the quadrate.
Of the opercular series, only the preopercle and the opercle are preserved (Figs. 3, 4i, 11a, b). The first has a maximum extension of about 50.7% of the HL. It is sickleshaped with a smooth posterior margin. The sensory canal is covered by a lateral ledge. The opercle has a more or less straight anterior margin and is rounded out posteriorly. Its maximum depth amounts to 60-70% of the HL, and it is 1.7 times deeper than wide.

Hyoid apparatus (Figs. 3, 4g)
The hyomandibula is an elongate bone, with a maximum extension of about 54.2% of the head length. It has welldeveloped and well-differentiated anterior, dorsal, and operculad processes. The maximum extension of the complete articulation area makes up slightly less than one third of the complete hyomandibular length.
No other skull elements are preserved. A reconstruction of the skull together with the shoulder girdle is given in Fig. 5.
Vertebral column (Fig. 6) There are 11-12 abdominal and 19 caudal vertebrae, including the second ural centrum. All of them are subrectangular in shape and become a little shorter anteriorly and distinctively shorter at the end. Generally, the vertebral column shows an almost straight course, and it is only very slightly dorsally convex in the abdominal region. Most abdominal vertebrae are crushed and poorly preserved. The caudal ones bear a characteristic ornamentation, which consists of longitudinal ridges and several grooves of different sizes and shapes. This ornamentation seems to differ between specimens HLMD-SMFF 18a, b, and SMNS 80738/81 with the lateral surface ridges being more rounded in the former. Nevertheless, the vertebrae are poorly preserved there as well, so that these perceived differences also may simply be artefacts of preservation. Pre-and postzygapophyses are not noticeable, probably also due to the imperfect state of preservation. The anteriormost neural arch/spine seems to be autogenous; at least, a seemingly corresponding structure is visible above the respective portion of the vertebral column. The parapophyses increase in length posteriorly in the series; the haemal spines are all longer than the neural ones, with the anterior ones being longest. Both are only slightly inclined posteriorly. In HLMD-SMFF 18b, the first haemal spine is completely missing and was obviously lost during fossilization together with the respective vertebral centrum. It is preserved in SMNS 87457/103), however, where its upper, and especially its medial part, so tightly adjoins the second haemal spine, that it can hardly be distinguished from the latter. Most ribs are broken, so that their number cannot be clearly determined. Nevertheless, they seem to have been slender, clearly inclined posteriorly, and comparatively short, extending over the upper half of the abdominal cavity only slightly. The anterior ones are almost equal in length; the posteriormost ones are distinctly shorter. Epineurals are visible above the anteriormost nine abdominal vertebrae. They are poorly preserved, but obviously articulated to the neural arches, respectively spines. In HLMD-SMFF 18b, epineural remains are also visible above the fourth to sixth caudal vertebrae. It is highly probable that this was not their original position and they were posteriorly displaced during fossilization, though. Epipleurals are altogether absent.
Caudal fin and caudal-fin supports (Figs. 7, 8, 9) With a span of about 35.7% of the SL, the caudal fin of †Oechsleria unterfeldensis is smaller than those of †Wettonius angeloi and Velifer multiradiatus, but larger than the ones of †Veronavelifer sorbinii and Metavelifer multiradiatus (Fig. 7). There are up to nine dorsal procurrent rays and at least six ventral ones. The dorsal lobe consists of 10 principal rays, of which nine are branched (two times at least), and the ventral lobe of nine principal rays, of which eight are branched (also at least two times). The first branching takes place at slightly less than 50% of the length in the dorsalmost and ventralmost principal rays and the second one at about 76%.
The caudal fin is supported by six vertebrae in total (Figs. 8,9). Of these, the first preural centrum and the first ural centrum are fused to each other. Together with the second ural centrum, this section is distinctly more strongly inclined dorsally than in, e.g., †Wettonius angeloi and Velifer hypselopterus. Furthermore, these three centra together are longer than deep, whilst they are deeper than long in the afore-mentioned species. With a few exceptions, the caudalfin endoskeleton otherwise corresponds to that known from the extant Velifer hypselopterus (e.g., Gosline 1961;Rosen 1973;Oelschläger 1974;Davesne et al. 2014). There are six hypurals. In contrast to †Wettonius angeloi, the first and the second ones do not seem to be fused to each other. The third and the fourth ones, however, are fused, and also to the second ural centrum in SMNS 87457/103 (Figs. 8a, b). This is considered an important character state in the phylogenetic analysis of extant Lampridiformes (Davesne et al. 2014). It furthermore would be a point of distinction, at least compared to †Wettonius angeloi, †Veronavelifer sorbinii and Velifer hypselopterus, where these elements are only basally fused. The situation may be a little different in HLMD-SMFF 18b, where both of them do not seem to be fused to each other, at least not in their proximal portions (Figs. 8c,  d). The fifth and sixth hypurals are autogenous and separated from each other. Furthermore, there are two uroneurals and probably three epurals (that are at least visible as positive prints in the resin matrix of HLMD-SMFF 18a). The neural spine of the second preural centrum seems to be fused to its centrum, and this may be the same with the haemal spines of the second and third preural centra. The neural spine of the second preural vertebra seems to be distally pointed and not reduced to a low crest as in other veliferids. However, the possibility cannot be completely excluded that this spine-like appearance is an artefact, due to a loss of the more lamellar posterior part of this bone during fossilisation.
Certain deviations stand out in the composition of the caudal endoskeleton in SMNS 80738/81 and HLMD-SMFF 18b (Figs. 8a, b versus 8c, d). In the first specimen, the uroneurals are, e.g., inserted at almost the same level and lie close to each other and directly above and in parallel to the dorsal margins of the posteriormost vertebrae (compound preural and ural centrum one, together with the second ural centrum), whereas they insert right after each other and are dorsally inclined from the latter in the second specimen. Nevertheless, and especially when considering the overall rather poor state of preservation of these specimens, it is more likely that these differences are merely taphonomical artefacts.
A reconstruction of the presumed most typical composition of the caudal-fin endoskeleton is given in Fig. 9.

Dorsal fin and dorsal-fin supports (Figs. 3, 10)
In SMNS 80738/81, a small, elongate element is located directly in front of the above-mentioned, presumed autogenous, neural arch/spine. This may be the remainder of a supraneural, which would then also mean that †Oechsleria unterfeldensis has only one such element, in contrast to most fossil and extant Veliferidae (with the exception of Metavelifer multiradiatus), in which there are two. The dorsal fin originates shortly behind the supraoccipital. It has a long base that extends over slightly more than 72% of the SL. The first dorsal-fin pterygiophore seems to be a compound structure (resulting from the fusion of formerly two such elements) that inserts anterior to the first neural spine. Most other pterygiophores (except the posteriormost ones) seem to be distributed over the interneural spaces at a one-to-one ratio. Those in the anterior and middle parts of the fin base are slightly inclined anteriorly (which is probably an artefact of preservation), whereas those in the posteriormost portion are inclined posteriorly.
The dorsal fin consists of about 32 segmented rays, of which the first one articulates on the first pterygiophore in supernumerary association. There are no spines. The majority of the lepidotrichia are bilaterally paired, and both halves are only proximally fused but separate and segmented distally, which makes it difficult to determine their total number. The segmentation starts at some distance from their bases in the anteriormost rays, but proceeds more proximally in the middle and posterior portions of the fin. The first two dorsal-fin rays are shortened; the third and fourth are the longest, reaching about 51% of the SL.
Anal fin and anal-fin supports (Fig. 11) The first anal-fin ray is inserted vertically under the third to last abdominal vertebra and well behind the origin of the dorsal fin, whereas it articulates more anteriorly (beneath the sixth to 15th abdominal vertebra) in some of the other veliferid species. The base of the anal fin is shorter than that of the dorsal fin and only accounts for slightly more than 54% of the SL. The fin is preceded by a roughly triangular structure that cannot be clearly identified as either a modified (fulcral) scale or a minuscule spine (Figs. 11a,  b). If it is not the latter, only the shortened first ray/spine would articulate on the first pterygiophore in supernumerary association. As in the dorsal fin, most of the anal-fin lepidotrichia are bilaterally paired, and both halves are fused only proximally and separate and segmented distally, and, therefore, difficult to count. Probably, there was a total of 29 anal-fin rays. In contrast to †Wettonius angeloi, the first two rays are shortened rather than elongated. Furthermore, it is not entirely clear whether the anteriormost "true" ray of †Oechsleria unterfeldensis is segmented and, therefore, does not qualify to be termed a spine here. All other rays are clearly segmented, with the segmentation starting at a little greater distance from the base than in the dorsal-fin rays. The sixth and seventh rays are the longest, equalling almost 36% of the SL.
Four anterior anal-fin pterygiophores are situated in front of the anteriormost haemal spine (see Diagnosis), and most other pterygiophores (except the posteriormost ones) seem to be distributed below the interhaemal spaces at a one-toone ratio. The five anteriormost pterygiophores (of which the first one seems to be a compound structure that resulted from the fusion of formerly two such elements as in the dorsal fin) are strongly inclined anteriorly. All the others are inclined posteriorly with the posteriormost ones being in an almost horizontal orientation.
Apparently, specimens SMNS 80738/81 and HLMD-SMFF 18b differ considerably in the number of pterygiophores that precede the anteriormost haemal spine (Fig. 11c-f). Whilst these number about three in the first mentioned specimen, there seems to be a much higher number (up to seven) in the latter. Such an exceptional deviation has never been reported from any other fossil or extant veliferid species (e.g., see discussion and figures in Sorbini and Sorbini 1999), and is herein regarded as an artefact of preservation, having resulted from a loss of the first caudal vertebra and its haemal spine (see also the discussion later on).

Shoulder girdle and pectoral fin (Figs. 3, 5)
The posttemporal is small. Its ventral arm is partially concealed beneath the opercle, but may have enclosed an angle of about 53° with its dorsal counterpart. It cannot be stated whether the latter was firmly fused to the exoccipital by connective tissue. The supracleithrum is an elongate and rather slender element; its maximum width is contained more Fig. 7 †Oechsleria unterfeldensis gen. et. sp. nov., details of caudalfin rays. a SMNS 87457/103, dusted with ammonium chloride; b same, schematic interpretation; c reconstruction; d same, schematic interpretation. Unbranched principal rays are highlighted in grey; graphical differentiation same as in Fig. 3 otherwise ◂ Fig. 8 †Oechsleria unterfeldensis gen. et. sp. nov., details of caudalfin supports. a SMNS 87457/103, dusted with ammonium chloride; b same, schematic interpretation. c HLMD-SMFF 18b, dusted with ammonium chloride; d same, schematic interpretation. Preural 1 and ural centrum 1 are highlighted in light grey, uroneural 1 and parhy-pural are highlighted in medium grey, and the unbranched principal caudal rays are highlighted in dark grey; the diagonally hatched area refers to artefacts in the resin plate. Graphical differentiation same as in Fig. 3 otherwise than six times in its maximum length. The cleithrum is well developed, and distally reaches the ventral body margin. The dorsal and ventral postcleithra are fused, the latter is posteroventrally inclined and seems to extend posteriorly to well beyond the pelvic girdle. The scapula is subquadrangular in shape and shows a comparatively large central foramen. The coracoid is slender, dorsoventrally compressed, and strongly curved. It is comparatively small and does not reach the ventral body margin. Four pectoral radials seem to be present, with the uppermost one being fused to the scapula. In SMNS 80738/81, a bunch of an estimated nine and possibly as many as 12 or 13 crushed pectoral-fin rays is visible, which more or less dorsally project from the scapula.
A reconstruction of the skull together with the shoulder girdle is provided in Fig. 5.

Pelvic girdle and pelvic fin (Fig. 3)
The basipterygium is broad, with a central axis in anterodorsal orientation. It has a median keel and a well-developed postpelvic process, similar to the one of Metavelifer multiradiatus that was figured by Olney et al. (1993: Fig. 4). Radials are missing, and the pelvic rays are poorly preserved. Therefore, it can neither be stated whether there is an autogenous lateral radial, nor whether the medial half of the medial pelvic-fin ray is fused to the medial radial.
There may have been six or seven pelvic rays in total and no spines. Squamation (Fig. 12) The body is covered with very thin cycloid scales. The bases of the dorsal and anal fin are covered with a scaly sheath as is typical in extant veliferids (Fig. 12a). The sizes of the scales are difficult to determine, since the marginal areas seemingly are not fully preserved in most of them. Some scales beneath the posterior part of the dorsal fin and close to the haemaxanal complex (sensu Monod 1968) have diameters of at least 1.5 mm. Others, in the dorsal part of the caudal peduncle, seem to be distinctly larger, with diameters from about 2 mm to more than 3.6 mm. What seems to be a growth cessation mark is visible in at least one of them (SMNS 87457/103, Fig. 12b). No radii have been found. Anteriorly, the squamation extends to the occipital and cheek regions. Posteriorly, it extends onto the proximal portion of the caudal rays. In SMNS 87457/103, one lateral line scale is preserved, indicating a course slightly below the midline of the upper body flank in the anterior part of the body (Fig. 12c).
In the middle part of the abdominal cavity, remnants of the original coloration are preserved in the form of small dark pigment grains that are distributed loosely over the scales and at some distance from each other. Somewhat more densely arranged pigment grains are preserved on some scales in the dorsal portion of the abdominal column, which indicates that the dorsal body flank may have been darker, as it is typical for many extant fish species (Fig. 12d).
An optimized reconstruction of the complete skeleton of †Oechsleria unterfeldensis is given in Fig. 13.

Assignment to Lampridiformes
On the one hand, only a few of the synapomorphies that were listed as supporting the monophyly of this order by Olney et al. (1993) and Davesne et al. (2014) can be assessed in the fossils under discussion herein. On the other hand, none of those that are verifiable also in these fossils contradict an assignment to this order. First of all, the anteriormost dorsal-fin pterygiophore inserts anterior to the first neural spine. Furthermore, there is a comparatively large ascending process of the premaxilla that is clearly longer than the alveolar one, although it is not clear whether this (together with the rostral cartilage) is accommodated by a vault or a cradle on the frontal. The first pectoral-fin radial is fused to the scapula, there are no supramaxillae, no epineurals on the postabdominal vertebrae (the presence of epineural-like elements above some postabdominal vertebrae in HLMD-SMFF 18b is considered a result of postmortem posterior displacement rather than a "real" morphological trait), no epipleurals, and the postcleithra are fused together, as in most lampridiforms.

Assignment to Veliferidae
In addition to these general lampridiform traits, and according to, e.g., Heemstra (1986), there are several character states that are typical of the members of the Veliferidae: the body is deep and laterally compressed; the dorsal and anal fins are extended and elevated, with numerous rays and scaly sheaths covering their bases. The caudal fin is forked, with 19 principal and 17 branched rays, and there is a typical veliferid composition of its endoskeleton with three epurals and third and four hypurals that are fused to the second ural centrum. In addition, there are less than 50 vertebrae (as in other bathysomous Lampridiformes), and no teeth are discernible in the oral jaws; a single supraneural (if present at all) that articulates in front of the anteriormost neural spine; unornamented dorsal-fin rays (without lateral spinules) that are bilaterally paired and have their halves fused proximally but separate and segmented distally, a dorsally orientated central axis of the pelvic bone, and a pelvic fin with seven rays and no spines. Untypical for veliferids, but not for Lampridiformes (e.g., Trachipterus, †Natgeosocus; see Olney et al. 1993;Bannikov 2014;Davesne et al. 2014;Keivany 2017), would be the presence of a pointed neural spine on the second preural centrum, instead of a broadened low crest. (Tables 1, 2) †Nardovelifer altipinnis. As has been described by Sorbini and Sorbini (1999), and as is evident from Tables 1 and 2, this species differs from †Oechsleria unterfeldensis in some body proportions and also in the meristics. These include a larger head, a slimmer body, and more posteriorly shifted insertions of the dorsal and anal fins. There is a higher number of vertebrae, a higher number of dorsal-fin spines, respectively rays, as well as fewer anal-fin spines and rays, respectively. In addition, there are considerable morphological deviations. These include a supraoccipital without an ornamentation of its anterodorsal lateral ridges, an ascending process of the premaxilla that is more than twice as long Fig. 11 †Oechsleria unterfeldensis gen. et. sp. nov., detail of anal fin and anal-fin supports. a HLMD-SMFF 18a, dusted with ammonium chloride; b same, schematic interpretation; c SMNS 87457/103, dusted with ammonium chloride; d same, schematic interpretation; e HLMD-SMFF 18b, dusted with ammonium chloride; f same, schematic interpretation. e and f are horizontally mirrored for ease of comparison; graphical differentiation same as in Fig. 3 as the alveolar one, a well-developed postmaxillary process in the upper jaw, a different caudal-fin endoskeleton with a shortened neural spine on the second preural centrum, third and fourth hypurals that are proximally fused to each other but not fused to the second ural centrum, two supraneurals that are located in front of the first neural spine, dorsal and anal fins with massive spines, short spines/rays in the anal fin, only two anal pterygiophores that articulate in front of the first haemal spine, and scales that are rather small. †Veronavelifer sorbinii. According to Bannikov (1991), and as Tables 1 and 2 show, there are differences in the body proportions, as well as in several skeletal details. There is, e.g., a shorter head, a slimmer body, a slightly posteriorly shifted insertion of the dorsal fin, as well as a distinctly posteriorly shifted insertion of the anal fin. There also are slightly more vertebrae (≥ 15 + 19), higher numbers of dorsal-and anal-fin rays, respectively spines, and a higher number of principal ventral caudal-fin rays. In addition, there are, e.g., a smooth supraoccipital crest that is not ornamented with anterodorsal lateral ridges, an ascending process of the premaxilla that is at least 2.4 times longer than the alveolar one, and a postmaxillary process in the upper jaw. The caudal fin endoskeleton is furthermore (most likely) characterized by a short (blunt) neural crest on the second preural centrum, proximally fused first and second hypurals, and completely fused third and fourth hypurals that seem to be also fused to the second ural centrum. There are two supraneurals in front of the first neural spine, and massive spines in the dorsal and anal fins. The pelvic bones are small, wedge-like, and arranged in an oblique position. †Wettonius angeloi. According to Carnevale and Bannikov (2018), this species differs from †Oechsleria unterfeldensis, e.g., by having a shorter head (HL 25.8% SL versus 32.8% SL), dorsal and anal fins that insert more anteriorly (PDL 33.1 SL versus 35.8% SL, PAL 44.3% SL versus 47.2% SL, respectively), and a larger caudal fin with a span of about 49.6% SL (versus 35.7% SL). In addition, there are slightly more vertebrae (31 versus 29-30) and dorsal-fin rays (35 versus 32), slightly fewer anal-fin rays (26 versus 29), and various other morphological deviations. Amongst others, the following characters and character states should be mentioned here as differentiating †Wettonius angeloi from †Oechsleria unterfeldenis: there is a long and broad ascending process of the premaxilla that is about 2.5 times longer than the alveolar one (much shorter in †Oechsleria unterfeldensis), a comparatively narrow opercle that is about 2.5 times (versus 1.7 times) deeper than wide, epineurals on only about seven (versus nine) abdominal vertebrae, a caudal-fin endoskeleton in which the fused first preural and ural centra together are clearly deeper than long (versus shorter), a neural spine on the second preural centrum that seems to be reduced to a broadened low crest (versus narrow and spine-like), first and second hypurals that seem to be proximally fused (versus separate), and third and fourth hypurals that are at least proximally fused and also fused to the second ural centrum (versus separate from each other), two supraneurals (versus one) that insert in front of the first Fig. 13 †Oechsleria unterfeldensis gen. et. sp. nov., graphically optimized reconstruction of complete skeleton. Scales are omitted, and, in contrast to the detailed drawings, the outlines of somewhat questionable elements are indicated as uninterrupted bold lines; diagonally hatched areas are insufficiently preserved in the fossils. The ornamentation of the caudal vertebrae is adopted from a mid-caudal centrum of SMNS 87457/103, that of the abdominal vertebrae is adopted from a typical abdominal centrum of the extant Velifer hypselopterus. For the dorsal-and anal-fin pterygiophores, a one-toone distribution over the interneural (respectively interhaemal) spaces is supposed for the anterior and middle portion of the fin, and a twoto-one distribution for the posteriormost one, as in Velifer hypselopterus neural spine; a simple (not compound versus compound) anteriormost dorsal-fin pterygiophore, three anal-fin pterygiophores (versus four) that articulate in front of the first haemal spine; an anal fin in which the second (versus fourth) ray is the longest, a pectoral girdle with a long coracoid that almost reaches the ventral tip of the cleithrum, four pectoral radials of which none is fused to the scapula, a postpelvic process that is shorter than the one of †Oechsleria unterfeldensis, and comparatively small (versus large) scales.

Comparison with extant veliferids
With reference to literature (e.g., Regan 1907a;Gregory 1933;Gosline 1961;Heemstra 1986;Davesne et al. 2014;Carnevale and Bannikov 2018), as well as to the available extant comparative materials, the following differences are to be stated (see also Tables 1, 2): Metavelifer multiradiatus differs from †Oechsleria unterfeldensis, e.g., by having a shorter head and a slightly deeper body; its dorsal and anal fins insert distinctly farther forward, but there is a more posteriorly located insertion of the pelvic fin. It furthermore has more vertebrae, fewer dorsal and ventral procurrent caudal-fin rays, and more dorsal-and analfin spines, respectively rays. The supraoccipital is without anterodorsal lateral ridges, and the ascending process of the premaxilla is very prominent, being almost five times longer than the alveolar one; there is a different count of vertebrae (16-17 + 18), the caudal-fin endoskeleton is characterized by fused hypaxial hypurals and a short neural crest on the second preural centrum; the anteriormost dorsal-fin pterygiophore seems to be simple (not a compound structure that probably resulted from the fusion of two formerly separated elements); there are massive spines in the dorsal and anal fins, and three anal-fin pterygiophores that articulate in front of the first haemal spine.  Carnevale and Bannikov (2018); those of †Nardovelifer are taken from Sorbini and Sorbini (1999); additional data for †Veronavelifer are from Bannikov (1991); further information is based on radiographs as well as on Froese and Pauly (2015) *Estimated measurements on the basis of the preserved parts of the body outline †Oechsleria unterfeldensis †Wettonius angeloi †Veronavelifer sorbinii

3
Velifer hypselopterus has, e.g., a shorter head, a somewhat deeper body, a dorsal fin that inserts farther forward, and an anal fin that inserts more posteriorly than that of †Oechsleria unterfeldensis; there are more vertebrae and dorsal-fin rays, respectively spines; in addition, there are fewer dorsal and ventral procurrent caudal-fin rays, and more pectoral-fin rays. It furthermore differs from the new genus and species, e.g., in having a supraoccipital crest with no anterodorsal lateral ridges, a well-developed ascending process of the premaxilla that is about 2.8 times longer than the alveolar one (Gregory 1933 : Fig. 173), a somewhat narrower opercle that is about 1.7 times deeper than wide (Gregory 1933 :  Fig. 173), a different count of vertebrae (16-17 + 18), two supraneurals in front of the first neural spine, third and fourth hypurals that are distally fused and also fused to the second ural centrum; spines in the dorsal fin, a comparatively long Dorsal procurrent caudal rays 7 5-6 5-6 Ventral procurrent caudal rays 7 4 4-5 Principal dorsal caudal rays 10 10 10 Bifurcated dorsal caudal rays 9 9 9 Principal ventral caudal rays 9 9 9 Bifurcated ventral caudal rays 8 8 8 coracoid that almost reaches the ventral body margin, and small scales.
The new genus and species appears to also differ from all fossil and extant Veliferidae in having an autogenous neural arch on the first abdominal vertebra and a spine-, respectively fulcral-like, scale in front of the anal fin.

Morphological particularities
Multiple anal-fin pterygiophores in front of the first haemal spine At a first glance, there are rather striking differences in the material under discussion here. In HLMD-SMFF 18b, up to seven such elements seem to precede the first haemal spine, while there are significantly fewer in SMNS 87457/103 (Fig. 11). Such variation would clearly exceed the range of veliferid intraspecific variability and would be indicative of the presence of a second taxon, an assumption that is highly improbable simply in view of the rarity of the respective extant forms. Principally, a loss of an anatomically consolidated complex such as this one, established by the anteriormost caudal vertebrae and the haemal spines in veliferids, is rather unlikely from a taxonomical point of view. Nevertheless, and as all other preserved structures and their arrangement are similar to those in the extant forms, this seemingly high number of anterior anal-fin pterygiophores in HLMD-SMFF 18b probably may be an artefact that resulted from the taphonomic loss of the first caudal vertebra and its haemal spine. If these were reconstructed in a "normal " manner, the number of anal-fin pterygiophores that are located in front of the first haemal spine would decrease to four, as it is (e.g.) usual for the extant Velifer hypselopterus and †Veronavelifer sorbinii.

Presence of a growth cessation mark
Unfortunately, such a structure seems to be preserved in only one of the scales of the materials described herein (SMNS 87457/103, Fig. 12b). According to the generally accepted interpretation of such structures in extant fishes (e.g., Bagenal and Tesch 1978), it would mean that this individual was more than one year old. This is somewhat incongruent with its comparatively small size. With a TL 74.5 mm and a SL of 53 mm, it clearly falls short of the lengths known from extant Veliferidae, as (e.g.) reported by Heemstra (1986): 280 mm TL for Metavelifer multiradiatus, and 400 mm TL for Velifer hypselopterus. Unfortunately, nothing has been reported as to their lengths at one year of age. All in all, the possibility cannot be excluded that this apparent annulus is an artefact resulting from a peripheral superimposition of two adjacent scales.

Other particularities and deviations
As already indicated in the descriptive section above, the different shapes of the opercle in SMNS 87457/103 and HLMD-SMFF 18b, the differences in the ornamentation of certain vertebrae (more ridge-like structures in SMNS 87457/103, more groove-like structures in HLMD-SMFF 18b), as well as the deviations in the caudal-fin endoskeleton (e.g., uroneurals that are closely allied with each other, insert almost at the same level, and their positioning closely above the vertebral column in SMNS 87457/103, versus uroneurals that insert one after the other and project dorso-posteriorly from the vertebral column in HLMD-SMFF 18b), all may be artefacts of preservation rather than being of taxonomical significance.

Palaeo-ecological and palaeogeographical implications
Unfortunately, almost no information exists on the ecologies of extant Veliferidae in the literature. According to Heemstra (1986) and Froese and Pauly (2015), and as already indicated in the Introduction, these are rare benthopelagic fishes of the Indo-Pacific and Indo-West Pacific. The Spinyfin velifer, Metavelifer multiradiatus (Regan, 1907a), has a predilection for temperate seas up to depths of 240 m. It is often found on the continental shelf and slope, and also on seamounts. The Sailfin velifer, Velifer hypselopterus Bleeker, 1879, also is demersal, but it appears to prefer somewhat more tropical waters and shallower depths, up to 110 m.
The general palaeo-environmental situation for the Grube Unterfeld deposits has been discussed with some controversy in the past (for a synopsis, see, e.g., Maxwell et al. 2016). Nevertheless, it is clear that all the presumed scenarios do not match the known ecologies of extant Veliferidae. Therefore, and when referring to the life habits of closely related extant forms (as is a frequently practised, especially in the interpretation of the batrymetrical preferences of fossil fish associations, e.g., Gaudant 1979)), their representation in the respective fish assemblage appears to have resulted from very particular circumstances. Maybe upwelling waters have played a role in that (Micklich and Hildebrandt 2010), or maybe occasional strong storms or other events caused tidal waves that washed them in from other water bodies (e.g., Schöggl and Micklich 2012;Eck 2018). Alternatively, the imperfect preservation of the fossils under discussion may indicate a longer post-mortem transportation. Even then it must be considered that the general depth of the Oligocene Upper Rhine Rift Valley during the deposition of the Bodenheim Formation (180-200 m according to Grimm et al. 2011) did not reach the maximum depth preference, at least not the one of Metavelifer multiradiatus, however.
As a matter of fact, and with reference to the life habits of the extant forms, also the palaeo-environmental situations (comparatively shallow waters) of the Monte Bolca and Nardo localities were likewise far from one that would promise finds of fossil Veliferidae (e.g., Sorbini 1978Sorbini , 1981Pappazoni et al. 2014;Friedman and Carnevale 2018). The respective records are at least better preserved than the ones from the Grube Unterfeld, though, and may not have been subject to similarly long post-mortem transportation. All in all, the best (and probably also most parsimonious) explanation for the extreme rarity of sailfin fish finds in the overall fossil record may indeed be mainly the difference between their (presumably) preferred habitats and the particular environmental and depositional conditions at the available fossil sites, rather than a change of their life habits since Palaeogene times. Even then it is difficult to understand, however, why such finds are as yet unknown from certain other localities, e.g., from the Eastern Paratethys, which have more benthic-bentbenthopelagic elements within their fish assemblages, and, therefore, should better represent the ecological preferences of extant veliferids. At least a few individual finds of other fossil Lampridiformes (e.g., certain Palaeocentrotidae and Turkmenidae) are known from some of them (Bannikov 2014).
As for the temperature preferences of the extant forms, it is rather clear that the new veliferid records support an assumption of their predilection for a warm climate. Similarly, when referring to their geographical distributions, they provide further evidence for an eastern (palaeo-mediterranean, respectively Paratethyan) seaway connection during that period in time.

Conclusions
The Grube Unterfeld specimens SMNS 87457/103, SMNS 88738/81 and HLMD-SMFF 18a, b can be assigned to the family Veliferidae within which they differ from all extant forms. Nevertheless, they also share, respectively unite, certain character states of some fossil and extant forms. As far as fossils are concerned, they seem to be closer to †Wettonius angeloi (similar morphology of the supraoccipital crest, absence of spines in the dorsal and anal fins) than to †Nardovelifer altipinnis and †Veronavelifer sorbinii, which both have, e.g., massive spines in their dorsal and anal fins. Compared with extant forms, they share more similarities with Velifer hypselopterus (e.g., compound first dorsal-fin pterygiophore, respectively very weak and particular dorsalfin spines, four pterygiophores in front of the first haemal spine) than with Metavelifer multiradiatus (simple first dorsal-fin pterygiophore, presence of massive spines in the dorsal and anal fins, three anal-fin pterygiophores in front of the first haemal spine). These are of course only superficial similarities that do not suffice to deduce a closer relationship between these taxa. The same applies to the comparatively poor state of preservation of the Grube Unterfeld specimens, which, in combination with the overall poor knowledge of the extant forms, does not allow a detailed phylogenetical analysis.
Palaeogeographically, they expand the known range of Veliferidae up into the Upper Rhine Rift Valley of the western Paratethys and also increase the known biodiversity of this family. Palaeo-ecologically, the representatives of the new genus and species are elements of the Grube Unterfeld fish fauna, that, with reference to the life habits of the extant forms (e.g., Heemstra 1986;Froese and Pauly 2015;Nelson et al. 2016), should be indicative of deeper water and also be suggestive for a warmer climate.