The family diagnosis is based on a unique combination of characters (uniquely derived features among teleosteomorphs are identified with an asterisk [*]): Small fishes about 55 mm maximum length, with naked body, and highly modified protractile upper and lower jaws giving the anterior part of the head a characteristic profile [*]. The body shape is torpedo-like, with a head about 50% deeper than the caudal peduncle [*]. T-shaped mesethmoid with strong lateral processes. Two pairs of nasal bones [*]. Absence of supramaxillae [*]. Absence of dentition [*]. Preopercle L-shaped. Interopercle small triangle-like. Vertebral column with persistent notochord in older forms; chordacentral vertebral column in younger. Vertebral caudal region diplospondylous, with small interdorsal and interventral elements. Ossified ribs absent. Short, stout epineural processes associated to the abdominal neural arches. Large and curved pelvic plates. First dorsal fin proximal radial enlarged and plate-like, resulting from fusion of three or more radials and supporting four or more dorsal rays [*]. Enlarged last dorsal proximal radial supporting several dorsal rays [*]. First anal fin proximal radial basally expanded and very elongate and dorso-anteriorly bent, acting as post-coelomic bone [*]. Last anal fin proximal radial highly modified, expanded, and plate-like, supporting three or more lepidothrichia [*]. No fringing fulcra associated with paired, dorsal, or anal fins. Homocercal caudal fin with both lobes deeply forked. Body lobe of the caudal fin completely reduced. Ural region with five or six broad and short hypurals. Diastema hypural absent or very narrow. Caudal fin with dorsal and ventral scutes; well-developed epaxial and hypaxial basal fulcra; short series of epaxial and hypaxial fringing fulcra reaching about half length of first and last principal rays. Accessory fulcra present in hypaxial caudal lobe. Procurrent rays only present in the hypaxial lobe of caudal fin. Eighteen to 21 principal caudal rays. A few large scales around urogenital opening [*].

One genus and four species known, Marcopoloichthys ani, M. andreetti, M. faccii, and M. furreri sp. nov.

Eurasian distribution, including Southern China (Yunnan and Guizhou Provinces), Northern Italy (Lombardy and Friuli), and eastern Switzerland (Canton Graubünden). Another undescribed species is present in the Middle Triassic of southern Switzerland, in Monte San Giorgio, Canton Ticino; T. Bürgin, pers. comm., 2022.

From Anisian (Middle Triassic) to Norian (Late Triassic).

Same as family diagnosis.

The species diagnosis is based on a unique combination of characters: The largest marcopoloichthyid reaching ca 55 mm maximum length. Skull roof covered with small and rounded oval tubercles and a few ridges of ganoine. Premaxilla and maxilla with slightly expanded articular region, spatulate-like and with crenulated anterior margin. Dentary ornamented with strong ridges and deep grooves; anterior margin covered with well-developed tubercles of different shapes. Short vertebral column with 33 to 35 vertebral segments, the first five fused into one element, the supradorsal carrier. With about nine supradorsal bones; the first five expanded distally, followed by sigmoid-shaped supradorsals; last supradorsal bones placed in front of the plate-like first compound dorsal proximal radial. Abdominal and first caudal neural arches with stout epineural processes reaching the next posterior neural arch. Dorsal fin support with first expanded proximal radial a massive squarish plate formed by fusion of four proximal radials. Last anal proximal radial with long and distally expanded region supporting several lepidotrichia. Five hypurals; no hypural diastema present. Ten or 11 epaxial basal fulcra. Short series of epaxial fringing fulcra. Twenty or 21 principal caudal rays with straight segmentation. One to three short hypaxial procurrent rays; accessory hypaxial fulcra present. About 12 hypaxial basal fulcra. No urodermals present. With three or four large, ovoid scales associated with the urogenital region.

The species name, furreri, honors Dr. Heinz Furrer who has dedicated most of his distinguished professional career to Triassic fossils of Switzerland, especially those of the Prosanto Formation.

PIMUZ A/I 2886, an almost complete specimen, very well preserved (Fig. 2A) with a bent abdominal vertebral region; it was collected in Gletscher Ducan, Davos, in the Canton of Graubünden, Switzerland on July 30, 2003. Upper Prosanto Fm., Early Ladinian, Middle Triassic.

PIMUZ A/I 1194, TL about 23 mm; poorly preserved. PIMUZ A/I 1924, almost complete specimen, but it appears longitudinally compressed; poorly preserved. PIMUZ A/I 1958 almost complete, very well-preserved specimen from the same locality as holotype. PIMUZ A/1 2886, complete specimen: TL ca 50 mm. PIMUZ A/I 2888, incomplete, disarticulated specimen with well-preserved disarticulated pectoral girdle and fin. PIMUZ A/I 2889, incomplete specimen; anterior part of body, some abdominal vertebrae, pectoral and pelvic fins poorly preserved. PIMUZ a/1 2890, incomplete specimen missing part of head, paired fins, anal fin and posterior part of caudal fin. PIMUZ A/I 3209, an almost complete specimen of about 54.5 mm maximum length, with nicely preserved head and caudal fin. BNM 201166, almost complete specimen of ca 45 mm total length. See Table 1 for more information concerning specific specimens.

Gletscher Ducan, Davos, in the Canton Graubünden, Switzerland. Upper Prosanto Fm., Early Ladinian, Middle Triassic. See Table 1 for information on localities and ages of specimens studied.

General description. The fish is ca 55 mm total length, slightly torpedo-like form (Fig. 2), with the head about three times deeper than the caudal peduncle. The dorsal fin insertion placed near to or at the midpoint of standard length (51–53% of SL). The pelvic fin insertion is placed at the same level of the dorsal fin insertion, but in one specimen is placed anteriorly (48–54%). The anal fin insertion is closer to the insertion of the pelvic fins than to the caudal fin (60–68% of SL); consequently, the fish has a long peduncle. The head is proportionally large, about 33 to 38% of standard length, and its aspect is very different when the mouth is closed compared to open. When the mouth is closed, most of the dorsal profile of the head looks gently rounded, decreasing in depth anteriorly (Fig. 3A; PIMUZ A/I 1958). When the fish is preserved in “feeding mode”, the mouth is extended anteriorly, as well as the bones supporting the lower jaw, giving the head a characteristic profile (Figs 2C, 4; PIMUZ A/I 3209). The orbit is moderately large, about 28 to 39% of head length, and the preorbital region is moderately short, ca 25% of head length (specimens with closed mouth). The pectoral fins have a low position, closer to the ventral margin of the body than to the middle region of the flank (Figs 2, 3). The caudal fin is homocercal with both lobes almost the same size and with its posterior margin deeply forked. All exposed surfaces of cranial bones are ornamented with tubercles and longitudinal ridges covered with a thin layer of ganoine. The lateral surface of fin rays and fulcra is covered with a thin layer of ganoine. The body is naked, except for a few large scales (or scutes?) around the urogenital region, probably one in front of the dorsal fin, and dorsal and ventral scutes in the caudal fin.

Skull roof and braincase. Although the skull roof is preserved in several specimens, it is almost impossible to trace each bone, because sutures are not visible due to fusion (Figs 4, 5). As the preservation permits, the bones of the skull roof apparently have smooth surfaces; however, under magnification, the bony surfaces may be densely ornamented with small, round or oval tubercles and short, longitudinal ridges (Fig. 5B). The ornamentation is covered by a thin layer of ganoine.

The parietal [= frontal] region is about 2.5–3 times longer than the postparietal [= parietal] region, and the limit between dermopterotic and postparietal cannot be traced (Figs 4, 5A, B). Consequently, it is assumed here that the postparietal and dermopterotic are fused to each other. This possibility is supported by the skull roof of the paratypes PIMUZ A/I 2887 and PIMUZ A/I 3209 (Figs 4, 5). According to available information, no specimen illustrates a complete fusion involving left and right sides of the skull roof, but each side is independent from its antimere.

From posteriad to rostrad, the skull roof is formed by the broadly and latero-ventrally expanded dermopterotic fused with the postparietal (postparietal + dermopterotic), which are densely covered with small tubercles (Fig. 5A, B). Apparently left and right bones are contacting each other through a straight suture (= sutura harmonica). It is unclear whether the parietal branch of the supraorbital canal extends into the compound bone, or the anterior middle pit-line is the one placed from the anterior margin to almost the half of the bone almost reaching the middle pit-line (Fig. 5B). The middle pit-line, as well as the anterior pit-line, are placed in conspicuous grooves. A posterior pit-line has not been observed. It is unclear if the supraorbital canal was covered by thin bone that collapsed after death and burial. The trajectory of the otic canal is not evident in the available specimens. The latero-ventral region of the postparietal + dermopterotic together with the autosphenotic are the main elements that articulate with the hyomandibula. Posterior to the postparietal + dermopterotic is a narrow, triangular bone that it is interpreted as an extrascapular (Fig. 5B). Although the extrascapular is incomplete in the available material, it appears to be in contact, or at least becomes closer, to its antimere medially. The postero-median region of the skull roof is not preserved in any specimen, hence it is unknown whether a supraoccipital bone was present.

The short lateral process of the autosphenotic is well-ossified, but its dorso-lateral walls are not well preserved (Fig. 4). The autosphenotic seems to be fused with the postparietal + dermopterotic region posteriorly in the holotype (Fig. 2A), whereas it is not preserved in PIMUZ A/I 2887 (Fig. 5), which raises the possibility that the autosphenotic is not fused to any of its surrounding bones in this specimen.

The parietal [= frontal] is the longest bone of the skull roof, about twice the length of the postparietal + dermopterotic, and it ends just short of the postero-dorsal corner of the orbit; anteriorly it ends near the antero-dorsal corner of the orbit. Due to conditions of preservation, the interparietal [= frontal] and postparietal [= parietal] sutures are not discernable in most specimens, except for PIMUZ A/I 2887 that shows straight sutural borders (Fig. 5A, B). Because of the parietal preservation, it looks like a fontanelle was partially separating both left and right bones medially. The trajectory of the supraorbital canal is partially visible in PIMUZ A/I 2887 (Fig. 5), and no lateral sensory tubules or pores are observed. It is unclear whether the sensory canal was placed in a groove or whether the groove was covered by thin bone that has collapsed. The bones described above are part of the immovable region of the skull roof. In contrast, the so-called snout is formed by bones that are loosely articulated and changed position during the suction feeding process.

The anterior movable region of the skull roof includes, from posteriad to rostrad, an extra bone identified here as a posterior nasal or additional nasal, a nasal bone, and the mesethmoid (Figs 4A, A, 5A, B, 6, and 7). The first two are paired, whereas the latter is an unpaired bone. The additional nasal is a somewhat ovoid-shaped bone with slightly irregular anterior and posterior margins loosely articulated with the parietal posteriorly and the nasal bone anteriorly; when the fish is not feeding, this bone is placed downward, forming a kind of anterior margin to the parietal bone, and because of its position, it can be confused with the lateral ethmoid. This additional nasal is preceded by the nasal that is almost as long as the additional nasal in PIMUZ A/I 2887 (Figs 5, 6), but is about twice the length in the holotype (Fig. 6) and is almost rectangular-shaped. Unfortunately, its lateral margins are damaged in most specimens. Both nasals seem to be loosely articulated medially. The supraorbital sensory canal is positioned almost in the mid-region of the additional nasal and nasal bones. Part of the surface of the nasal bone is covered by rounded tubercles in the holotype (Fig. 6). Forming the tip of the snout is a T-shaped median bone, the mesethmoid (Figs 4–7), with strongly ossified lateral processes, as well as a strongly ossified and elongate posteromedian process. There is no evidence of a rostral commissure. The holotype, PIMUZ A/I 2886, has an outstanding element preserved, which by comparison with some living atherinomorphs and cyprinodontiforms with suction feeding mechanisms, is interpreted as the rostral cartilage (Fig. 6). The rostral cartilage can be a continuous element extending in front of the parietal to the mesethmoid anteriorly. It can be perforated or not. In this case, only one ovoid foramen is observed, and because of this, I interpret that the rostral cartilage was broader, and its right side is incompletely preserved.

Morphologically, the anterior tip of the skull roof looks very different when the mouth is not open (e.g., Figs 5–7) compared to open (Figs 4, 7). When the mouth is closed, the anterior articular margin of the parietals together with the additional nasals produce a marked curved, downward region where the additional nasals lie. When the upper jaw is protracted, the profile of the anterior part of the head changes with the additional nasals, nasals, and mesethmoid placed almost in a straight line in front of the anterior margin of the parietal bones and the well-ossified lateral ethmoids. Since the mentioned bones are loosely connected, it is assumed here that the bones involved in the suction mechanism were kept in their position by the aid of ligaments and the rostral cartilage, but due to their soft structure, they were lost after death and burial.

The orbitosphenoid is not preserved in most specimens, but apparently both eyes are separated by an incomplete interorbital septum as shown by specimen PIMUZ A/I 3209 (Fig. 4). The lateral ethmoid is well-ossified and slightly bent, but its preservation does not allow a proper description.

The antero-middle region of the parasphenoid is visible in one of the fishes (Fig. 4), permitting its partial description. The parasphenoid is narrow anteriorly, it expands slightly posteriorly, and part of its ascendant process is poorly preserved just posterior to the orbital region. There are no teeth associated with the ventral surface of the bone or scattered below the parasphenoid. The parasphenoid joins anteriorly a small, narrow, triangular-shaped, and unpaired vomer (Fig. 4). No teeth are associated with the vomer either.

Orbit and circumorbital series. The fish has a moderately large orbit (Figs 2, 3A, 7), ranging from 28 to 39% of head length. When the fish was not feeding, the orbit was almost rounded, but when the fish was in feeding action, and the mouth protracted anteriorly, the orbit became oval-shaped.

The series of circumorbital bones is incomplete; supraorbital bones are absent dorsally, as well as an antorbital that seems to be missing at the antero-dorsal margin of the orbit in most specimens. Since I do not feel confident about the presence of an antorbital in this fish, I consider its presence uncertain. The infraorbital bones are thin and fragile and destroyed in most specimens. They are partially preserved in the paratype PIMUZ A/I 1958 (Fig. 2B); however, their delicate preservation makes their description difficult. Their total number is probably five plus a small dermosphenotic (Fig. 7). It is unclear whether the small flat bone placed between the dermosphenotic and anterior margin of the preopercle is a suborbital or part of the most dorsal infraorbital, but it is interpreted here as a suborbital.

Infraorbital 1 is the largest bone of the series, somewhat rectangular-shaped and with some broad sensory tubules that are difficult to count (Fig. 2B; PIMUZ A/I 1958); infraorbital 1 is incompletely preserved in PIMUZ A/I 2887, and the main infraorbital canal seems to be placed in a groove, but this could be misleading, since a groove is not observed in PIMUZ A/I 1958. Infraorbital 2 is an elongate, narrow bone, bearing a groove for the infraorbital canal (or the outer wall of the sensory tube is broken away). Infraorbital 3, at the posteroventral corner of the orbit, is slightly enlarged, reaching the anterior margin of the preopercle and its circumorbital margin, as well as that of the dorsal most infraorbital(s); it is well-ossified, and its few sensory tubules seem to be positioned in grooves. Infraorbital 4 is square-shaped, with its margin heavily ossified and with at least one sensory tubule. If a fifth infraorbital is present, it should be mainly represented by the thickened orbital margin. A description of the dermosphenotic is not possible because of poor preservation. The possible suborbital is a narrow, squarish bone dorsally and triangular-shaped ventrally, but this also could be the flat laminar surface of infraorbital 5. Unfortunately, there is not another specimen preserving the infraorbital series, so these uncertainties cannot be clarified with the available material.

In most specimens there are no orbitosphenoid or sclerotic bones preserved, and the orbital space looks “clean”. It is uncertain whether this condition is the result of the preparation of this area, but one specimen (PIMUZ A/I 3209; Fig. 4) shows remnants of bones preserved. Due the flatness of the bones, it is unclear if these can be considered sclerotic bones or parts of a broken orbitosphenoid. In another specimen (Fig. 3A), there is one elongate bone at the anterior part of the orbit, giving the impression of the presence of an enlarged, slightly concave anterior sclerotic bone, but a possible posterior sclerotic is not preserved.

Upper jaw. Premaxilla and maxilla form the upper jaw. A supramaxilla has not been observed in any specimen, and it is assumed here to be absent. Both bones lack teeth, and their ventral margin is smooth. The premaxilla is about half of the length of the maxilla, and when the mouth is closed, the premaxilla is placed ventral to the ventral border of the maxilla, but when the mouth is open, both premaxillae project anteriorly in a very distinct position (compare Figs 2A and 3A with 2C and 4; and Fig. 7A with 7B).

The premaxilla (Figs 4, 5C) is a slightly bent bone, with its proximal end slimmer than the main section of the bone, which expands gently distally, ending in a straight margin. The slightly curved proximal region of the bone (Fig. 5C) lacks an ascendant process or any other process and is slightly spatulate, with a few short interdigitating ridges separated from each other by short grooves, giving this region a characteristic surface.

The maxilla (Figs 4, 5C) is an elongate bone, ending below the posterior half of the orbit and about the level of the articulation of the quadrate-lower jaw when the mouth is closed. It is narrower in its anterior half and slightly expanded at its anterior tip, with similar interdigitations as in the anterior tip of the premaxilla; in contrast, the maxilla expands gently posteriorly, keeping an elongate, straight aspect in its middle region, and then expands posteriorly, ending in a slightly triangular or rounded tip. A supramaxillary process is absent on the dorsal margin of the bone. The ventral margin is almost straight. The surface of the maxilla is covered with longitudinal bony ridges, which in some specimens retain remnants of ganoine. When the mouth is open, the maxilla is displaced anteriorly (compare Fig. 2A with Fig. 4 and Fig. 7A with 7B).

Lower jaw. The jaw (Figs 4, 8) is massive, relatively short, deep, and somehow triangular-shaped, with the quadrate-mandibular articulation placed below the posterior half of the orbit when the mouth is closed and displaced anteriorly, below the anterior half of the orbit, when the mouth is open (compare Figs 2A, 7A, 8 and 4, 7B). The jaw is formed laterally by three bones: dentary (= dentalosplenial or dentosplenial), angular, and surangular. Medially, an ossification interpreted here as a coronoid bone is present (Fig. 8). Since the medial view reveals only one bone posteriorly, it is assumed here that the angular, articular and retroarticular are fused into an angulo+articulo+retroarticular (Fig. 8). The lower jaw is toothless, and no evidence of sockets for teeth has been observed in any specimen.

The sutures between angular, surangular and dentary reveal that the dentary forms most of the jaw (Figs 7, 8). From a narrow but thick mandibular symphysis, the dentary expands abruptly dorso-posteriad, producing a massive and high coronoid process that is thicker and strongly ossified at its antero-dorsal region. The latter has a large contribution of the surangular. The antero-ventral portion of the dentary projects anteriorly and ventrally in a kind of flap or broad process (Figs 6A, 8) that commonly is broken, but it is well-preserved in A/I 3209 (Fig. 4). The postero-ventral process of the dentary narrows posteriorly and extends ventrally, almost reaching the posterior corner of the angular. A notch is absent in the ascending margin of the dentary. The surangular is an elongate bone, suturing ventrally with the dorsal region of the angular portion of the angulo + articulo + retroarticular and the postero-dorsal region of the dentary. The postarticular process is short.

The mandibular sensory canal is placed near the ventral margin of the jaw, and its trajectory is marked by a conspicuous ornamentation that has preserved remnants of ganoine. Sensory pores have not been observed in the postero-ventral region of the angulo + articulo + retroarticular, so it is assumed that the mandibular canal exits medially.

The lateral surface of the lower jaw of certain specimens presents a curious ornamentation at its antero-dorsal region of the dentary along the oral margin (Fig. 8). The ornamentation consists of well-developed, massive protuberances of various sizes and shapes that make the oral margin uneven. In other specimens such ornaments are missing (Figs 3A, 4, 6). It is unclear if these differences in ornamentation are sexual dimorphism, a hypothesis that should be tested when more specimens become available. The surface of the postero-ventral process of the dentary presents marked longitudinal ridges in most specimens; the deep ridges are also observed in the medial view of the jaw. Both the external protuberances and ridges are partially covered with a thin layer of ganoine.

The medial view of the lower jaw (Fig. 8) is somehow concave, with a deep triangular depression at the anterior confluence of the dentary and the angular portions. In some jaws, this region gives the impression of the presence of a space between bones. In front of the depression/space, a rectangular, well-ossified bone is positioned. Because of its position, I interpret this bone as a coronoid devoid of teeth.

Palatoquadrate, suspensorium, hyoid arch, and urohyal. Most of these elements are partially hidden by other bones or are destroyed so that the description is restricted to a few of them.

The regions where the metapterygoid, entopterygoid and ectopterygoid would be placed are damaged in most specimens, but a section of a bone that is interpreted here as the ectopterygoid is preserved in PIMUZ A/I 3209, anterior to the anterior margin of the quadrate (Fig. 4). Because of the size of the preserved areas, it is assumed here that the metapterygoid, as well as the entopterygoid, was a narrow bone. Another long, thin, and narrow bone anteriorly placed to the ectopterygoid is interpreted here as a palatine. Under the present conditions of preservation, it is impossible to clarify whether this is a dermal (dermopalatine) or a chondral bone (= autopalatine).

The quadrate is hidden by the anterior arm or ramus of the preopercle and the posterior region of the maxilla when the mouth is closed (Fig. 3A) and is partially exposed when the mouth is open, because the lower jaw displaces anteriorly (Figs 4, 7). The main body of the quadrate (Fig. 8) is slightly triangular close to its articular condyle with the lower jaw. The articular condyle is strong and slightly laterally projected to articulate with the lower jaw. The posterior margin of the quadrate, which shifts to a horizontal position in continuation with the jaw, when the mouth is open, is massive, and together with the symplectic, which lies ventrally to the quadrate, provide a strong support for the lower jaw. The quadrate seems to continue posteriorly in a flat, almost rectangular process in the holotype, whereas the process ends in a sharp tip in PIMUZ A/I 3209 (Fig. 4). The complete length of the symplectic is unknown, because the bone is covered by the anterior margin of the preopercle or is broken, but considering its position and that of the hyomandibula, it is assumed here that it was a long bone. A quadratojugal has not been observed, and it is interpreted as absent.

The hyomandibula is incompletely preserved in all specimens, but in some its contour is visible throughout the preopercle. In specimen PIMUZ A/I 3209, it appears as a long, columnar bone that is inclined ventro-anteriorly when the mouth is open, and together with the long symplectic gives support to the jaw; the hyomandibula is placed in an almost straight line when the mouth is closed. Its dorsal region articulating with the cranium is broader and well-ossified and continues ventrally as a well-ossified shaft; it is unclear whether an anterior membranous flange is present or not. The dorsal articular region of the hyomandibula (Fig. 4) apparently has only one elongate articular condyle with the latero-ventral articular facets of the dermopterotic and autosphenotic regions laterally. Nothing can be said about the opercular process. Considering the length of the jaw and the position of the quadrate-mandibular articulation, the symplectic is assumed to be a long and strong bone that is partially exposed in PIMUZ A/I 3209 and PIMUZ A/I 2841; an alternative possibility is the presence of an elongate cartilaginous articular region filling the space between the ventral margin of the hyomandibula and the dorso-posterior margin of the symplectic.

The lower part of the hyoid arch preserves a posterior ceratohyal (Fig. 4C, D) that is almost as long as the anterior ceratohyal, which is an almost rectangular bone, lacking a foramen or a notch close to its smooth, dorsal margin. Only one massive, squarish hypohyal articulating with the anterior margin of the anterior ceratohyal (Fig. 4C, D) is present. A urohyal has not been observed in any specimen, and it is assumed here to be absent.

Opercular and branchiostegal series, and gular plate. Although the preopercle is an element associated with the suspensorium, it is included here to describe the opercular series together. The preopercle (Figs 2C, 3A, 4, 8) is a large and L-shaped bone, which is slightly expanded postero-ventrad. The dorsal lobe is slightly longer than the ventral one when the mouth is closed (PIMUZ A/I 1958); however, when the mouth is open, the angle of the preopercle changes, and both arms are about the same length (Figs 4, 7B). Its dorsal arm is about 57% longer than the ventral one, almost reaching the ventro-lateral margin of the dermopterotic region. When the mouth is closed, both arms form an almost right angle, whereas the angle increases to over 100 degrees when the mouth is open, as a result of the anterior extension of the mouth and the action of assumed ligaments joining the posterior margin of the lower jaw and the anterior margin of the anterior arm of the preopercle and interopercle. The preopercle has a gentle flange just anterior to the confluence of both arms where a curvature of the preopercular canal is present. A notch at the posterior margin of the bone is absent. The preopercular canal (Fig. 8) apparently only bears the main preopercular canal, because no tubules are conspicuous at is dorsal arm. A few tubules (Figs 3A, 4, 7, 8) fill the preopercular ventral arm; more precise information is not available because of incomplete preservation of the available preopercles. The sensory tubules are delicate, simple and narrow, and open irregularly near to or at the ventral margin of the bone.

The opercle (Figs 3A, 4, 8) is not very well preserved in the available specimens, but still, it is possible to observe that is the largest element of the series, slightly deeper than broad, and slightly narrower at its dorsal margin, whereas the ventral margin is slightly broader. Dorsally, the opercle reaches the latero-ventral margin of the dermopterotic region, the extrascapular and the posttemporal, and posteriorly, the supracleithrum and cleithrum. Its dorsal and anterior margins are almost straight, whereas the posterior margin in gently curved, and the ventral margin is markedly oblique. Anteriorly, the margin of the opercle is thickened and joins the dorsal limb of the preopercle, whereas it joins the subopercle postero-ventrally and the interopercle antero-ventrally. The opercular surface is irregularly covered with short ridges and rounded and oval tubercles. The subopercle (Figs 3A, 4, 8) is large, as broad as the opercle, and slightly shorter. The general aspect of the bone is not easy to describe, because it is gently curved ventrally in some and markedly rounded in others. Information on the size of the antero-dorsal process is not possible based on the available specimens. A small interopercle (Figs 4, 8) is partially covered by the postero-ventral margin of the preopercle so that its complete shape and size remains unknown.

Branchiostegal rays are not preserved, except for one specimen (holotype PIMUZ A/I 2886) with two narrow and spine-like posterior branchiostegals associated with the posterior ceratohyal. The absence of branchiostegals or their low number in one specimen could simply indicate that the fish has very few that are usually not preserved. Only one short and rounded branchiostegal ray was mentioned and illustrated for Marcopoloichthys ani by Tintori et al. (2007: p. 16, fig. 3). A gular plate has not been observed, and it is assumed here that it is absent.

Vertebral column, intermuscular bones, and ribs. The information on the whole vertebral column is incomplete, because most specimens provide partial or no information. An almost complete vertebral column is preserved in several specimens, including the holotype (PIMUZ A/I 2886; Fig. 2A) and paratypes (BNM 201166 and PIMUZ A/I 19568; Figs 2B, 3A), while the caudal region is well-preserved in PIMUZ A/I 2890 and several other specimens.

The vertebral column is aspondylous (see Arratia et al. 2001 for different types of the vertebral column), with well-developed arcocentral elements forming the centra, but the notochord remains persistent and functional in adults. There are about 33 to 35 vertebral segments, including those of the hypurals. About 13 to 18 are abdominal, monospondylous vertebral segments, whereas the caudal region is diplospondylous, with very small interdorsal and interventral arcocentral elements alternating with the well-developed basidorsal and basiventral arcocentral elements. Because of their small sizes, many of the interdorsal and interventral elements have not been preserved. No remains of centra are present in the ural region.

The first five neural arches and spines are fused into one special, previously unreported element that is preserved in the holotype PIMUZ A/1 2886, as well as in PIMUZ A/1 1958 (Figs 2, 3, 9). This compound bone is named here “supradorsal carrier” and is formed by the lateral, fused expansions of the neural arches and hemispines of the first abdominal vertebrae, forming two lateral wings (Fig. 9). Five supraneurals are in a median position between the two lateral wings of the supradorsal carrier. I expect that this special structure is a synapomorphy of marcopoloichthyids, a character that should be checked in other species when better-preserved material becomes available.

There are about 13 or 14 parapophyses (Figs 3, 9), the first ones covered by the opercle and the dorsal bones of the pectoral girdle. The parapophyses are comparatively large for the size of the fish, and they are well-ossified; they are squarish in shape and each bear a small cavity close to its ventral margin. No ribs are preserved in the available material, and they were not reported or illustrated in Marcopoloichthys ani (Tintori et al. 2007: fig. 4) either; thus, it is accepted here that marcopoloichthyids do not have ossified ribs.

The neural arches of the abdominal vertebrae (Figs 3, 9) are slightly expanded, and the halves of each arch, plus their elongate neural spines, are unfused medially. The lateral wall of each neural arch projects in a stout and short epineural process (= epineural bone; see Arratia 1997 or 1999 on the terminology) emerging at the postero-lateral margin of the arch. They are easily broken because of their position and structure.

The neural arches of the first caudal vertebrae (Figs 2, 3A, 9) are slighter broader than those of the abdominal vertebrae, and each has an epineural process until the third or fourth vertebra posterior to the last anal pterygiophore. The neural and haemal spines of the caudal region are narrow, except for those of the preural centra (see below). The neural and haemal spines are moderately inclined toward the body axis in the precaudal region, increasing their inclination caudally (Figs 10, 11). The first haemal spines (Figs 10, 11) are short, not extending between the anal pterygiophores or just reaching them. The neural and haemal spines of the mid and caudal regions are ossified, showing an internal core of cartilage where the bones are broken.

The series of supraneural bones is commonly not preserved, distorted, or covered by other structures. The series is formed by nine bones in the paratype (PIMUZ A/I 1958), with the first five associated with the supradorsal carrier. These anterior supradorsals are slightly ovoidal and expanded, especially supraneural 3, whereas supraneural 4 and 5 are partially fused proximally. The subsequent supradorsals are slightly sigmoid-shaped. The series extends up to the expanded, plate-like, compound first dorsal proximal radial, and it does not extend between the most anterior proximal radials as in Marcopoloichthys ani (Tintori et al. 2007: fig. 4).

The epineural processes of the neural arches (Figs 9, 11) extend along the abdominal region, ending posterior to the last dorsal pterygiophore. The broad and well-ossified epineural processes are short, extending laterally on the neural arch of the next vertebral segment. Epipleural bones are absent.

Pectoral girdle and fins. The pectoral girdle includes dermal and chondral bones. The dermal bones are the posttemporal (linking the girdle with the cranium), supracleithrum, cleithrum, and postcleithra. It is unclear whether a clavicle was present, but see below. The chondral bones are the scapula, coracoid, and proximal and distal radials. The posttemporal is incompletely preserved in the available material (Figs 4, 5B). Apparently, it is a relatively small and narrow bone, placed laterally to the extrascapular; it is unclear whether a dorsal process for articulating with the cranium is present. The main lateral line is not observed.

The supracleithrum (Figs 4, 7) is incompletely preserved or covered by the opercle, but it seems to be an elongate bone. The trajectory of the lateral line is not observed. The sigmoidal-shaped cleithrum (Figs 4, 5A, 7–9) is a heavily ossified bone, with a moderately long dorsal limb and markedly developed, expanded and curved ventral limb, which is partially broken in the available material, making identification of its complete area difficult. The cleithrum is slightly expanded at its postero-dorsal corner and becomes narrower at its dorsal region. The anterior surface of the cleithrum is covered by a long and broad serrated appendage that is almost completely preserved in the paratypes PIMUZ A/I 2841 and 2887 (Figs 5A, 8). The external surface of the cleithrum in PIMUZ A/I 2888 (Fig. 10) is abraded so that the serrated appendage is not preserved. A broad clavicle in front of the antero-ventral region of the cleithrum is observed in specimen BNM 201166.

Three postcleithra are present (Figs 4, 7). Postcleithrum 1, the uppermost element of the series, is elongated, with a slightly rounded posterior margin. Dorsally, it articulates with the supracleithrum and anteriorly with the upper part of the cleithrum and ventrally with postcleithrum 2. Postcleithrum 2 is slightly narrower than postcleithrum 1 and is curved postero-distally. Postcleithrum 3 is a splint-like bone. By comparison with other teleosteomorphs, it is assumed here that the three bones were not externally placed, but they were covered by the body hypaxial musculature.

The scapula and coracoid (Fig. 10) are incompletely preserved in the available material, and they are not informative. Four proximal radials are observed in the paratype PIMUZ A/I 2888 (Fig. 10), with the first two being larger than the third and fourth proximal radials, which are square-shaped. At least three small distal radials are preserved between the broken proximal region of some pectoral rays.

The pectoral fin (Figs 2, 3, 5A) is positioned near the ventral margin of the body. The total number of pectoral rays is unknown, because commonly the fins are incomplete, but 15 rays are preserved in the right fin in PIMUZ A/I 2841, and most fins in other specimens have ca 12 rays preserved. All rays have very long bases, are scarcely branched, and segmented distally; a few last rays, closer to the body, are smaller than the lateral ones. The first pectoral ray, which is exposed in PIMUZ A/I 2888 (with the pectoral girdle and fin displaced), merits a description. The first ray is a massive ray formed by the fusion of three rays at least (Fig. 10). These rays are fused at their bases, being separated distally. This first compound ray is slightly expanded and thicker at its proximal portion where the propterygium is fused with its base.

Pelvic girdles and fins. The pelvic girdles are partially exposed in several specimens (Figs 2A, B, 3A, B, 11). A large, elongate plate-like basipterygium (or pelvic plate) is slightly curved medially, with its lateral margin more strongly ossified than the rest of the plate. The posterior part of the basipterygium is slightly broader than the anterior margin and presents a short postero-medial process. The number of rays per fin are difficult to count, due to preservation. Ten or 11 rays are present in the holotype; eight of them are thicker and longer than the two or three medial rays. In contrast, nine long pelvic rays are preserved in each fin in specimen PIMUZ A/I 2888. Eleven rays were mentioned for Marcopoloichthys ani, but the number of rays remains unknown for M. andreetti and M. faccii (Tintori et al. 2007). The pelvic rays of M. furreri sp. nov. have long bases, are distally segmented, and apparently branched only once. This information is collected from the holotype, with one ray distally exposed (Fig. 2A). In other specimens, the distal parts of the fin rays are disarticulated or overlapping so that they are not informative (Fig. 11). Because of the position of the articular region of each ray, it is unknown whether proximal radials were present.

Dorsal fin and radials. The dorsal fin (Figs 2, 3, 11, 12) is commonly not well preserved with its rays partially displaced or damaged so that a precise total number of dorsal fin rays cannot be provided, but considering that the paratype PIMUZ A/I 2841 has 15 rays preserved, including a short, thin one segmented anteriorly, this could indicate that the fin has ca 15 rays.

Commonly, the dorsal pterygiophores preserved the proximal radials, however in the holotype, some of the anterior middle and distal radials are also preserved (Fig. 12). The series of proximal radials presents distinct features characterizing marcopoloichthyids, for instance, the modifications in the first and last proximal radials. In M. furreri sp. nov., the first proximal radial can be plate-like and square, but in others, the proximal radials are incompletely fused so that the elements forming this complex structure can be counted (Fig. 11). There are six intermediate proximal radials followed by one modified last radial bearing an undetermined number of rays in PIMUZ A/I 2841 and holotype (Figs 11, 12). This last proximal radial has an expanded distal articular region that projects ventrally in a narrow, markedly curved process. The complex plate-like first proximal radial in Marcopoloichthys ani is ax-shaped, whereas it is pear-shaped in M. andreetti (Tintori et al. 2007); in addition, M. ani has an ax-shaped proximal radial and nine to 10 proximal radials posterior to the first, which is a higher number than in Marcopoloichthys furreri sp. nov.

Anal fin and radials. The anal fin and its pterygiophores are not well preserved in the available material, and because of this, a description is difficult, and a total count of fin rays is not available. Additionally, there is variation in the number and amount of fusion of the proximal radials. The most complete series of proximal anal radials, or the most informative, is that present in the holotype (Fig. 12). In this specimen, the first anal proximal radial is a compound element resulting from the incomplete fusion of two proximal radials. This first element curves antero-dorsally giving the radial a characteristic shape, reminiscent of the postcoelomic bone of pycnodontiforms (Tintori et al. 2007). The first anal proximal radial is followed by a second, long, narrow radial, that is followed by a third element that results from the partial fusion of two proximal radials which are broken at their bases. Behind this element is one simple proximal radial that is followed by the last radial. The last radial is an elongate element bearing a narrow, thin anterior process that extends dorsally between the distal tips of the haemal spines and has a broad distal portion for articulation with several lepidotrichia (Fig. 12). In total, the anal series of proximal radials in the holotype included five separate elements. In the paratype PIMUZ A/I 2841, only three proximal radials are preserved, and the first and last are not preserved.

Caudal fin and endoskeleton. The caudal fin and endoskeleton are preserved in several specimens, but the dorsal elements of the ural region are poorly or not preserved at all. The homocercal caudal fin (Figs 2, 3) is deeply forked, with few short middle principal rays compared to the long first and last leading marginal ray that frame the segmented and branched principal rays. Many rays preserve a thin layer of ganoine.

One or two preural vertebrae support the most anterior basal fulcra. The preural vertebrae, as well as the ural ones, are supported by a functional notochord. Consequently, except by the arcocentra, no centra are formed, and the region is monospondylous, in contrast to diplospondylous vertebral segments in anterior and mid-caudal vertebrae (Figs 2, 13, 14). The two preural segments (corresponding to preural centra 1 and 2) are characterized by the presence of well-developed ventral arcocentra with broad and flat haemal spines, which distally support the last principal rays, one procurrent ray, and the series of hypaxial basal fulcra (Figs 13, 14). Dorsally, the neural arches or arcocentra of these two vertebrae are well-developed, and their neural spines are broad and of similar length. The neural spines of the last caudal and preural vertebrae are inclined posteriorly, closer to the body axis, and they do not support the most anterior basal fulcra.

The preservation of the neural spines of preural vertebrae 1–5 suggests they have a central core of cartilage surrounded by a thin, perichondral ossification. In the vertebrae that are completely preserved, an anterior process at the base of neural spines 1–5 is apparently absent. The haemal spines of preural centra 1–3 are moderately broad, but narrower than their respective neural spines. The haemal spine of preural vertebra 4 and more anterior ones are narrower. The haemal spines of the most preural vertebrae are perichondrally ossified thinly. The haemal spines of preural vertebrae 1–3 (Fig. 13) bear a short and narrow anterior process dorsally, at their limit with the expanded ventral arcocentra. A complete neural arch or dorsal arcocentrum, with a well-developed spine, is present on preural centrum 1. A hypurapophysis on the lateral wall of the ventral arcocentrum or haemal arch of preural centrum1 is absent.

Posterior to the neural spine of preural centrum 1, a series of slightly modified chondral neural elements is positioned (Fig. 13). In most specimens, this region is damaged or badly preserved, except for the holotype, which is illustrated in Fig. 13. The first two are elongate laminar elements resembling neural spines, and lacking the ural arcocentra; a third broad, laminar element, also lacking an arcocentrum follows. There is a fourth small, plate-like element posteroventral to the third, which extends caudally between the bases of the epaxial basal fulcra, but it is unclear if this could be a broken section of the enlarged third bone. Because of their position as part of the ural region and the lack of their ural neural arches or arcocentra, these bones are considered here as uroneurals “of a special kind”. They are different from the uroneural-like elements present in pachycormiforms or some present in aspidorhynchiforms and Eurycormus, which are modifications of spines of the preural region. They also differ in shape from the uroneurals of Leptolepis coryphaenoides plus more advanced teleosts (see Discussion below). Certainly, these elements in Marcopoloichthys furreri sp. nov., because of their position and shape, increase the stiffness of the tail during locomotion, which is a function of the uroneurals.

No epurals are present in the holotype, and there is no space left for them between the distal tips of the enlarged uroneurals and the bases of the epaxial basal fulcra.

Five hypurals (Figs 13, 14) are present, all of them close together so that a diastema between hypurals 2 and 3 is absent. Hypurals 1–4 are slightly expanded at their proximal regions and seem to have preserved part of the ventral arcocentrum. Hypurals 1 and 2 are the longest elements of the series, and hypural 3 is the broadest. A small element is positioned between the distal portions of hypural 2 and 3, and it is interpreted here as an interhemal. Hypural 5 is the smallest of the series of hypurals. Hypurals 1 and 2 (Fig. 13) are weakly supporting the thin bases of part of the hypaxial basal fulcra, the procurrent ray, and the lowest principal rays. Several thin and narrow bases of the principal rays articulate directly with one hypural without producing a special angle.

There are ten or eleven epaxial basal fulcra, which are followed by 10 or 11 fringing fulcra and only reach to the mid-region of the dorsal margin of the first unsegmented principal ray. There are 20 or 21 principal rays that are segmented and branched distally, and their bases are narrow. The articulation between segments of the principal rays is straight. Ventrally, the basal fulcra are usually incompletely preserved so that a total count cannot be given, but the holotype presents 12 hypaxial basal fulcra. There are one or two short procurrent rays that are followed by a short series of hypaxial fringing fulcra; however, PIMUZ A/I 3209 has a third short procurrent ray (Fig. 14). In addition, accessory fulcra are present between the principal rays and the hypaxial basal fulcra (Figs 13, 14). The external surface of the different kind of fulcra and rays is covered by a thin layer of ganoine.

One elongate and slightly oval dorsal scute and a slightly shorter ventral scute (Figs 13, 14) precede the epaxial and hypaxial lobes, respectively. No urodermals have been observed in the available material.

Scales. The body is devoid of scales, with the exception of two to four large oval scales (Figs 2, 3, 10) placed around or close to the urogenital region and a possible elongate one in front of the dorsal fin in one specimen (Fig. 9).