Thrissops formosus Agassiz, 1833.

Late Kimmeridgian; Kelheim, Bavaria, Germany.

(amended from Nybelin 1964; Bardack 1965; Alvarado-Ortega and Brito 2010): Elongate, slender or fusiform fishes known to reach a standard length of up to 0.9 m; head length included 4–6.5 times in standard length; mouth cleft directed upward; premaxillary teeth longer than those of maxillary and stoutly conical; triangular lower jaw; mandible without enlarged coronoid process; vertical and horizontal arms of preoperculum form right angle at posteroventral corner; preopercle with well-developed posteroventral process; tubules of preopercular sensory canal confined to horizontal arm of preoperculum; parietals not fused; dorsal fin posterior to anal fin origin; 13–15 dorsal pterygiophores; 23–31 anal pterygiophores; six elongated uroneurals; one urodermal.

In the late Kimmeridgian and early Tithonian of England, France, and Germany.

SNSB-BSPG AS VII 2 (Ebert 2021, fig. 6A).

Kelheim (most probably Kapfelberg), Bavaria, Germany.

Upper Kimmeridgian.

(for measurements of specimens and counts of features see Suppl. material 1: table S1). Maximum length 75 cm; 37–40 supraneurals; 57–61 vertebrae (without ural centra); 29–32 ribs anterior of anal fin; 28–31 (maximum at 30) anal pterygiophores; small teeth on all jaws (Figs 3, 18A, B); dorsally curved hook at the anteriormost tip of the dentary with two small, posteriorly directed teeth (Figs 3, 18A, B); curved maxilla (Figs 3, 18A, B, 20B).

BMMS (Eichstätt without number); CAM SM F.11220 (Kelheim); CM4702 (Eichstätt or Solnhofen); CM4083, 4091 (both Cerin); Coll. Tischlinger 88/91 (Eichstätt), 09/2, 92/2 (both Ettling); DMA (Painten without number); GPIT-PV-42046, 50597 (both Kelheim); JME-ETT 46, 47, 73, 74 (Fig. 2B), 75 (Fig. 2A), 76, 87a,b, 93, 103, 126, 139, 157, 209, 211, 245, 281, 283, 564, 869, 873, 879, 886, 887, 900a,b, 972, 1350, 1357, 1595, 1799, 1805a,b, 2076, 2079, 2166, 2171, 2550, 2750, 2872, 2941, 2942, 2966, 3108, 3109a,b, 3111, 3205, 3208, 3340, 3341, 3344, 3355, 3371, 3379, 3471, 3473, 3476, 3630, 3657, 3680, 3685, 3686 (Fig. 2C), 3909, 3914, 3917, 4094, 4108, 4315, 4379, 4413 (all Ettling); JME-SOS 2516 (Eichstätt), 4254 (Schernfeld), 7849 (Ettling); LF 1215 (Eichstätt), 2319 (Ettling); MB.f.1375 (Kelheim), 1590 (?Solnhofen), 9756, 15864, 15874, 18259 (all Kelheim); MHNL 20015136, 20015137, 20015185, 20015188, 20015760, 20150031, 20150198, 20150525, 20150558, 20150799, 20150913, 20272030 (all Cerin); MMG-SNSD BaJ2120 (Kelheim); MNHN SLN2 (Kelheim), MNHN CRN9, 73, 82 (all Cerin); NHMW 1852.II.61 (Kelheim); NHMUK PV OR 35013 (Kelheim), P.913, P.913B (both Kelheim), P.917 (Solnhofen), P.918 (Cerin), P.3683, P.3684 (both Kelheim); NMP UC10, 103, 104 (all three Kelheim), NMS 1879.30.5, 1882.32.4 (both Kelheim); NMWIN 801821 counterpart 801822 (Kelheim); NRM P00002924 (Solnhofen), PIMUZ A/I 0389 (Kelheim); SMNS-P-86701 (Eichstätt); SNSB-BSPG AS VII 2 (holotype), AS VII 175 (both Kelheim), SNSB-BSPG 1997 XVIII 1514, 1515 (both from Brunn); TM14842 (Cerin); UCBL-FSL 93382, 502300, 503200 (all Cerin); YPM VPPU.003282 (Eichstätt).

In the Upper Kimmeridgian and Lower Tithonian. Common in Ettling and Kelheim (both Solnhofen Archipelago, Bavaria, Germany) and Cerin (Ain, France); rare in the following localities of the Solnhofen Archipelago: Brunn, Eichstätt, Painten and Solnhofen (for localities and exact numbers see Fig. 1 and Suppl. material 1: table S1).

Due to space constraints, not all features of Th. formosus can be described in detail, as the new species described below have priority and features of Th. formosus have already been described before (see list of publications above). In this article, the features of Th. formosus are presented mainly in detailed photos and corresponding drawings of specimens from Ettling (for the cranium see Figs 3A, B, 4A–D, 18A, B, 20B and for the caudal fin and skeleton see Figs 5A–C, 21).

Thanks to the embedding of JME-ETT3371, it was possible for the first time to take a photo and a drawing of the skull of Th. formosus in dorsal view (Fig. 4C, D). Particularly interesting in this specimen is the parietal in dorsal view, as it seems that the parietal bones of both sides have fused together to form a median bone.

At the anteriormost tip of the dentary, all specimens of Thrissops formosus have a dorsally curved hook that carries two posteriorly directed teeth which have nearly the same size as the other dentary teeth (see Figs 3, 18A, B and comparison to the other species in the discussion section below).

The caudal fin is deeply forked, but in some very well-preserved specimens (JME-ETT126, 173, 283, 564, 887, 2750 (Fig. 21A), 2941, 3111 (Fig. 5C), 3340, 3473, 4108) an elongation of parts of the principal rays seven and eight is visible. The extension is in ray seven only on the ventral side and in ray eight on the dorsal side, so that together they form an inner caudal tip (Figs 2C, 5C).

The holotype of Thrissops formosus (SNSB-BSPG AS VII 2) and type specimen of the genus, which survived the bombing of WWII were stolen from a display case in Munich in 1975 and found again in 1979 in the collection of an invertebrate collector in Munich who claimed to have bought the piece at a flea market (see document in old catalogue of the BSPG).

SNSB-BSPG AS VII 178 (Ebert 2021, fig. 6C).

Kelheim (most probably Kapfelberg), Bavaria, Germany.

Upper Kimmeridgian.

Standard length (SL) to body depth (BD) specimens 26–29% (average 28%); Maximum length 43 cm; 59–60 vertebrae (without ural centra); 38–40 supraneurals; 30–31 abdominal ribs; 30–31 anal pterygiophores; lager, bigger and less numerous teeth than other Thrissops species (except for Th. cirinensis); dentary teeth of irregular size; ventral margin of maxilla more straightened than in the other Ichthyodectiformes; vertebral axis in front of the caudal fin first bends slightly downwards before turning into the upper caudal lobe.

Only known from the Solnhofen Archipelago, Bavaria, Germany, but rare in all localities (see Fig. 21 and Suppl. material 1: table S1).

CM 4030 (“Solnhofen”); Coll. Leich Bochum (Eichstätt without number); Coll. Tischlinger 01/26 (Öchselberg), 98/23 (Eichstätt); DMA (Painten without number); JME-ETT2639, 3465 (both Ettling); JME-SOS2525 (Solnhofen), 4253 (Wintershof, Eichstätt Basin); LF 1469 (Eichstätt); MB.f.15887 (Kelheim); MBH20253251, 20253252 (both Blumenberg, Eichstätt Basin); MMG-SNSD BaJ 2115 (Eichstätt); NHMUK PV P. 920, P.3678, P.3683a (all three Kelheim); NMP UC2280 (Solnhofen); NMS 1905.83.12 (Kelheim); ROMVP 947 (Solnhofen); SMNS-P-86700 (Zandt).

MNHN CRN 66 (Fig. 7), complete specimen in lateral view, 18.5 cm standard length (SL).

Cerin, France. The location of Cerin is marked with a question mark on the rock slab of the holotype. The rock is also somewhat unusual for the Plattenkalk of Cerin. However, this problem must be left to future research.

(amended from Nybelin 1964 and Taverne 1977). teeth in premaxilla and dentary larger than in other Thrissops species; 32 supraneurals; 27 anal pterygiophores; 54 vertebrae (without ural centra); 24 ribs.

Nybelin (1964) was somewhat cautious about calling this taxon a new species, since there was only a single specimen and that could be pathological. Taverne (1977) finally determined the species name Thrissops cirinensis.

(probably belonging to the same species): MHNL20150008, 20150311, 20150312, 20150313 (counterpart of 20150311), 20150401, 20150661, 20150690, 20150736, 20150831; UCBL-FSL P371, 502407, 502412.

All these specimens are poorly preserved and it is impossible to say for sure if they are specimens of Thrissops cirinensis. More detailed examinations under optimal UV light may help. These specimens are similar to the holotype of Th. cirinensis mainly in the following features: short length of the body (compared to other Ichthyodectiformes); the body shape with SL/BD 27–33%; the number of 30–31 supraneurals; 23–26 anal pterygiophores; ~49–51 vertebra and 24–25 abdominal ribs (for measurements of specimens see Suppl. material 1: table S1). In UCBL-FSL P371 some large teeth in the praemaxilla and in UCBL-FSL 502412 some larger teeth in the dentary, typical for Th. cirinensis, are visible.

JME-ETT3345 (Figs 8, 11, 13A, 18D), complete specimen in lateral view, 11.6 cm standard length (SL).

Ettling, Markt Pförring, Bavaria, Germany.

From the eigeltingense β horizon of the lower Tithonian (see Tischlinger and Schweigert 2020).

(all from Ettling). Coll. Tischlinger 20/4, 20/8; JME-ETT166 (Figs 9A, 12), 220 (Fig. 9B), 1360 (Fig. 10A, B), 4104a, b (Figs 9C, 13C, D).

The specific epithet in Thrissops ettlingensis refers to the village of Ettling, Markt Pförring, Bavaria, Germany where the specimens were found.

Thrissops reaching 15.5 cm total length (13 cm SL), with the following unique combination of characters (for measurements of specimens and counts of features see Suppl. material 1: table S1): standard length (SL) to body depth (BD) 26–33% (average 31%); 30–31 supraneurals; 49–50 vertebrae (without ural centra); 24–25 ribs anterior of anal fin; 23–25 anal pterygiophores; small teeth (Figs 11, 12, 18D); curved maxilla (Figs 11, 12, 18D, 20A); regular dentition in the lower jaw.

(for measurements of specimens see Suppl. material 1: table S1).

In Thrissops ettlingensis, the maximum body depth (MD) is 26–33% (average 31%) of its standard length (SL), with up to 15.5 cm in total length (13 cm SL). One juvenile specimen of 3.5 cm SL (JME-ETT1360) is also known from Ettling. The dorsal and the anal fins are in the posterior half of the body, with the dorsal fin origin slightly posterior to the anal fin origin, which is typical for Thrissops and Allothrissops (Nybelin 1964) (for differences to other species see Table 1).

Cranial morphology. Cranial morphology of Thrissops ettlingensis is best visible in the holotype (JME-ETT3345a, Fig. 11) and JME-ETT166 (Fig. 12), both showing the left side of the cranium. None of the cranial bones shows ornamentation or have serrated posterior borders.

Braincase and Ethmoid region. Most bones of the braincase are covered by other bones of the cranium, only the rostrodermethmoid and parts of the palatine, the ethmoid complex and the parasphenoid are visible. A small part of the palatine is visible at the anteriordorsal border of the maxilla ventral to the nasal pit. The lateral ethmoid separates the orbita posteriorly from the nasal capsule anteriorly. The ethmoid complex in Ichthyodectiformes is described in detail in Cavin et al. (2013). A small part of the parasphenoid is visible at the ventral border of the orbita. The rostrodermethmoid is an unpaired bone, at the anterior part of the cranium posterior of the premaxilla. Posteriorly it extends between the paired frontals.

Skull roof. The elongated frontals are the largest bones of the skull roof, with the posterior part twice as wide as the anterior part. The anteriormost part of the frontals is covered by the rostrodermethmoid, posteriorly the frontals overlap slightly the parietals. It is unclear if the parietal in Th. ettlingensis is an unpaired ossification median on the scull roof, because only one side of the cranium is visible. The visible parietal at the posteriordorsal border of the frontal, are nearly quadrangular, only about half the size of the extrascapular and probably as large as the pterotics. The pterotic is smaller anteriorly and gradually widens posteriorly, but the exact shape is not clearly visible due to the incomplete preservation of the specimens. There is one large extrascapular and one posttemporal on each side of the cranium. The nasal is a short, tubular, nearly rectangular bone at the posterior border of the nasal pit, with a canal running in its center.

Circumorbital series . Antorbitals anterior of the frontals are not recognizable in any of the specimens. There is a total of four thin infraorbitals (io) ventral and posterior of the orbit. The anteriormost two infraorbitals (io1, io2) are slender and anteriorly posteriorly elongated. The lacrimal (io1) extends from the nasal pit to the level of the first third of the orbita. The second infraorbital (subinfraorbital, io2) nearly went to the posterior border of the orbita. The third infraorbital (io3 = ventral postinfraorbital), is the largest infraorbital, it is nearly as long as it is deep. This bone is recognizable by postero- and postero-ventrally directed branches of the infraorbital canal. The fourth and last infraorbital (io4 = dorsal postinfraorbital) is best visible in the holotype JME-ETT3345a (Fig. 11). This bone is higher than long and lies on the posterior border of the orbit posterior to the sclerotic ring. A gap in which parts of the hyomandibula are visible separates the two posterior infraorbitals from the preoperculum. The infraorbital sensory canal is well visible in the holotype and JME-ETT166 (Fig. 11, 12). The autosphenotic is a somehow triangular bone at the posteroventral border of the orbit, but its exact borders are not clearly visible in any of the specimens. Suborbitals and supraorbitals are not recognisable in any of the specimens and probably generally absent. At least two large, broad bones of the sclerotic ring surround the orbita, but these fragile bones are broken and its borders are not everywhere clearly visible. The basal sclerotic bone is a thin, rounded bone in the posterior dorsal part of the orbita.

Opercular series . The preopercle is large with a long and slender dorsal part and a broad ventral part which extends anteriorly and posteriorly. It has slightly concave posterior margin; ventrally it has a well-developed posteroventral process. In the ventral part the preopercular canal is well visible. Some extensions of this canal bend ventrally and end in small openings on the surface.

The opercle is twice as high as it is wide (antero-posteriorly). Its dorsal border to the posttemporal is convexly curved, whereas its posterior border to the supracleithrum and cleithrum, its ventral border to the subopercle and the anterior border to the preopercle is nearly straight. The subopercle is small, with its anteriormost part with the anteriordorsal process covered by the preopercle. An interopercle is not visible, probably totally overlapped by the preoperculum.

Branchiostegal series and gular plate. There are 16 branchiostegal rays on the holotype (Fig. 11). The rays in the anterior half of the series being fine and hair-like. The posterior five rays are wider with the posteriormost being about three times as wide as the ray anterior to it. The anterior hair-like rays end in a point whereas the posterior rays have rounded posterior margin. Anterior to the branchiostegals, between the two dentaries there is a part of a small, thin bone visible, which is interpreted here as a small gular plate.

Jaws and suspensorium . The upper jaw consists of maxilla, premaxilla and two supramaxillae. The premaxilla at the anteroventral edge of the upper jaw is broad. On its convexly rounded anteroventral border there are eight to ten small teeth which are about twice as long and wide as the maxillary teeth (best visible in the holotype, Fig. 11). The posteriodorsal part of the premaxillae is covered by the maxilla and the rostrodermethmoid.

As in all Ichthyodectiformes the maxilla is long. In Thrissops ettlingensis at least as long as the dentary but not reaching the posterior border of the orbita. Its posterior border to the angular is rounded, covering the anterior part of the quadrate. The anteriormost part of the maxilla, which is covered by the premaxilla and the rostrodermethmoid, bents ventrally, whereas the visible anterior part is nearly straight. The posterior half of the maxilla bends continuously dorsally. Only one row of teeth with 51 small, pointy teeth is visible on the ventral margin of the maxilla of the holotype (Fig. 11). Almost all teeth are of the same length only the posteriormost are slightly smaller.

There are two supramaxillae on both sides of the cranium which run along more than half of the posteriodorsal margin of the maxilla. The anterior supramaxilla is slender, half as large as the posterior one and tapers towards the front. The posterior supramaxilla is as long but twice as wide as the anterior one. The posterior supramaxilla has an anteriodorsal process that ends in a point and covers the posterior half of the dorsal margin of the anterior supramaxilla. The maximum height of the posterior supramaxilla, is a little higher the maxilla at its highest point in the centre of the maxilla.

The mandible is moderately long, reaching the posterior edge of the orbita. There are 22 small, pointed teeth along the dorsal border of the dentary visible in the holotype (Fig. 11). The dentary teeth are slightly larger than the corresponding maxillary teeth. They are situated in one row and are of approximately the same size all along the whole length of the dentary. The mandibular sensory canal runs from anterior to posterior in the middle of the bone in a small groove. A border between the dentary and the angular is clearly visible at the dorsalmost part of the mandible. Dorsally the angular is covered by the posteriormost part of the maxilla.

The quadrate is large, but the articulation to the mandible is covered by the angular. The slender, elongated symplecticum branches off from the posteroventral part of the quadrate and extends posteriordorsally between the quadrate and the preoperculum.

Pectoral girdle . The large posttemporal borders the operculum dorsally. Along its ventralmost part, which is twice as broad as the dorsal part, the posttemporal sensory canal is well visible. This canal has at least four branches posteriorly, which ends in small pores. The supracleithrum, which lies on the posteriodorsal edge of the operculum, and covers some of the anteriormost vertebrae, is at most half as large as the posttemporal. In its dorsalmost part, the extension of the posttemporal canal runs to the lateral line. The cleithrum is, as usual, the largest bone in the shoulder girdle, but a large part of the cleithrum is covered by the operculum and suboperculum anteriorly. However, its approximate shape can be estimated through these transparent bones (Fig. 11A). As in all Ichthyodectiformes the coracoid is a large but thin bone ventral to the cleithrum.

Axial skeleton . The axial skeleton is clearly visible in nearly all specimens (Figs 8, 9), apart from the juvenile specimen JME-ETT1360 where most vertebra centra are not developed yet (Fig. 10). Only embryonal ring centra are visible on some parts of the vertebral axis of this juvenile specimen (JME-ETT1360); two behind the cranium and about 14 between the dorsal and anal fin. Embryonal ring centra are not fully ossified und without contact to each other. Interestingly they appear in black colour in normal-light (Fig. 10A) and UV-light (Fig. 10B), whereas all other bones are brown in normal light (Fig. 10A).

The vertebral column in adult specimens consists of 51 or 52 total centra, including 26 or 27 abdominal centra, 23 or 24 caudal centra including the parhypural, and two ural centra. The vertebrae are best visible in areas where the scales are absent. The centra in the anterior half of the body (best visible in JME-ETT4014a) have small anteriorly posteriorly aligned ridges and a few small pores. In the posterior half of the body (best visible in JME-ETT3345a) only the ridges are present, with the middle one or two ridges being slightly more pronounced. The anterior vertebrae are deeper than long, from about the 36^th to the 48^th vertebra they have about the same length as hight. All parapophyses, haemal arches and neural arches are autogenous from the centra. 24 or 25 long, paired pleural ribs cover the middle abdominal cavity. The anteriormost four vertebrae, which are covered by the operculum and the subcleithrum are without ribs.

The ribs are bow shaped, slightly curved anteriorly, only the posteriormost pair, which is the smallest, is straight.

There are 26 short paired neural spines in the abdominal part, with the two posteriormost, which are connected to the dorsal pterygiophores, being the longest. The epineural processes, which are connected to the ventral bases of the neural spines, are elongated, thread-like bones. Their anterior half runs dorsal to the neural arches, parallel to the body axis, but then bends dorsally to the tips of the neural spines. However, the posteriormost two or three epineural processes, ventral to the dorsal pterygiophors, hardly bend dorsally and therefore do not reach the ends of the neural spines. The epineural processes are as long as 8–9 centra. Epipleurals or epicentrals are absent. Small, unknown, ovoid structures are visible in JME-ETT3345a and JME-ETT4104a, dorsal to the vertebrae 10–16, between the epineurals, which have two third the length of a vertebra.

In the preural region there are 18 unpaired neural spines with the anteriormost, connected to the dorsal pterygiophores being the longest. All unpaired neural spines have an anterior projection near the base to the neural arch which is most prominent in the anterior part dorsal to the anal fin.

Additionally, there are 22 or 23 unpaired haemal spines between the paired abdominal ribs and the parhypural.

Dorsal to the paired neural spines, between the cranium and the dorsal pterygiophores there are 30 or 31 supraneurals. The anteriormost three have broad bases which are partly covered by the posttemporal. The following four also have slightly widened bases but are less pronounced as the neural spines anterior to them. The penultimate supraneural is dorsally in contact with the anteriormost dorsal pterygiophore and the posteriormost supraneural is situated in JME-ETT4104a between dorsal pterygiophores two and three. The ventralmost part of the supraneurals insert between the dorsalmost parts of the neural spines. The supraneurals has nearly the same length as the neural spines in the same area, apart from the posteriormost, which is only half as long as the neural spines ventral to it.

Pectoral fins . The shape of the pectoral fin is best preserved in JME-ETT220 (Fig. 9B). The anterior, unsegmented part of both pectoral fins in JME-ETT220 is visible from both sides of the body, with 13 rays each. The pectoral fin is larger and longer than the pelvic fin, with the anteriormost four rays being the largest and longest. The unsegmented proximal parts of the rays are long, approximately two third as long as the complete ray. In the juvenile specimen (JME-ETT1360) at least three pectoral radials are visible anterior of the pectoral fin (Fig. 10).

Pelvic fins . The pelvic fin is short, with approximately half the length of the pectoral fin. In JME-ETT220, the anterior part of both pelvic fins is visible from both sides of the body, with about nine rays each (Fig. 9B). Segmentation and branching are not well visible in any of the specimens. The pelvic bone is slender along the whole length; only the anteriormost and the posteriormost parts are slightly widened. It has the length of four vertebrae dorsal to it, but it is slightly shorter than the pelvic fin itself.

Dorsal fin. The dorsal fin is located far back on the trunk, posterior to the anal fin origin (Figs 8–10). The dorsal fin is slightly higher than long, forming a triangle. There are 14 (in one case possibly 13) dorsal pterygiophores supporting about the same number of distally segmented dorsal fin rays, plus anterior to the segmented rays three unsegmented procurrent rays. Each pterygiophore is composed of an elongated proximal radial and a middle radial. The middle radials are small and hardly visible. The first dorsal proximal pterygiophore is bifurcated at the dorsal base with a long anteriorventral directed process that corresponds to the posterior pterygiophores in position and length and a slightly shorter anterior process which is directed in the body axis. A thin bone membrane stretches between these two extensions. The anterior dorsal radials are elongated and deeply overlap the neural spines. The posterior dorsal radials extend into the interneural spine spaces ventral to them. The first segmented ray is unbranched, posteriorly the rays branch at least once (in adult specimens).

Anal fin. As in all Ichthyodectiformes, the anal fin is anterior-posteriorly elongated but not as long as in other Ichthyodectiformes (Figs 8–10). 22–24 segmented anal fin rays are supported by 23–25 anal pterygiophores (Table 1). The first segmented ray is unbranched, posteriorly all rays are segmented and branched. The first segmented ray is the longest, then the length decreases rapidly up to the 10^th ray. Behind the 10^th ray the ray length is very short and decreases only slightly. The posteriormost anal ray is an exception; it is slightly widened and has several branches. The posteriormost branch is thin and elongated, so that this ray becomes twice as long as the rays immediately anterior to it (best visible in JME-ETT4104a and already hinted at the juvenile JME-ETT1360). Whether this is a characteristic of sexual dimorphism can only be decided when more well-preserved specimens are available. Cavin et al. (2013) described a similar long filament on the IchthyodectiformesEubiodectes libanicus, but here for the dorsal fin. Anterior to the first segmented ray, there are three or four unsegmented procurrent rays. The first anal pterygiophore is very elongated and rests along the anterior margin of the first haemal spine. All anal pterygiophores situated posterior to this first pterygiophore are long, extending into the interhaemal spine spaces dorsal to them. The anal pterygiophores supporting the posterior half of the fin are longer than the fin rays connected to them. Middle radials are small and mostly covered by the bases of the fin rays in the anterior part of the fin. In the posterior part of the fin the middle radials are visible in JME-ETT4104a which are rectangular in shape.

Caudal fin and skeleton (Fig. 13A, B). The caudal fin is deeply forked with two triangular lobes of the same length. The caudal formula (see Alvarado-Ortega 2024 p. 20) is iv+I+9—8+I+v. Four dorsal procurrent rays are visible in JME-ETT3345a in JME-ETT4104a the number of dorsal procurrents seems to be higher, but these are probably disarticulated lepidotrichia from both sides of the body. The dorsalmost procurrent in JME-ETT3345a is the shortest, which is unsegmented. The following procurrents are segmented but unbranched, with the number of segments from one up to at least 13, increasing with length of the rays.

Ventrally, we have five ventral procurrents in JME-ETT4104a and apparently six in JME-ETT3345a where the smallest can also be a part of the procurrent of the other side of the body. As in all modern Teleostei (since Leptolepis cyprinoides of the Lower Jurassic) there are 19 principal rays with the dorsal and ventralmost being the largest rays, which are unbranched but multiple segmented (Arratia 2008). The rays between these two rays are branched up to six times. The segmentation of the innermost principal rays and the procurrent rays are straight, whereas the segmentation of the principal rays close to the leading edges (dorsal and ventral) is step-like or Z-shaped (comparable to Thrissops formosus Ebert and Kölbl-Ebert 2010a, fig. 3). The proximal base of the ventralmost dorsal ray (ray 10) bears a small dorsal process, while the proximal bases of the dorsal rays five to nine are bent ventrally. The bases of the innermost two rays (rays 10, 11) are broadened. The unsegmented ray bases of the median rays are entirely covered by scales, whereas the unsegmented bases of the marginal rays are covered by scales only in their anterior two thirds.

There is one dorsal and one ventral caudal scute (best visible in JME-ETT4104a).

The skeleton supporting the caudal fin rays is composed of six vertebrae with two ural centra being involved in supporting the fin rays. The first ural centrum consists of the fused centra U1+U2 on which hypurals H1 and H2 are attached posteroventrally. The posteriormost ural centrum is the smallest centum nearly completely covered by the uroneurals and the process of dorsal hypurals.

There are at least seven hypurals (H1–7) visible in JME-ETT3354a (Fig. 13A) and eight hypurals (H1–8) in JME-ETT4104a (Fig. 13B). The ventralmost three hypurals have well-developed articular heads which rest on U1–U3, and are developed as a thickening along its anteroventral edge. The heads of these hypurals are so broad that they are in contact with each other and enclose parts of the corresponding ural. The anteriormost two hypurals (H1, H2) attach the fused ural centra 1+2 (U1+2). Whereby in the holotype (JME-ETT3345a) the anterior bases of these two hypurals are separated with two separate heads (Fig. 13A), in JME-ETT4104a the anterior bases of H1 and H2 have grown together to form a joint head (Fig. 13B). The anteriormost two hypurals project posteroventrally, whereas the hypurals H3–8 joint the ural centrum (U3) and project posterodorsally, so there is a narrow and deep caudal diastema between the H2 and H3. The dorsalmost hypurals (H5–8) are nearly completely covered by uroneurals and fin rays. In addition, there is a very thin layer of scales on all elements of the caudal skeleton, so that the boundaries of these small bones can hardly be seen.

All haemal spines, the parhypural and the hypurals are clearly separated from the centra. Hypurals H1 and H2, the parhypural and the posteriormost three haemal spines are close together with almost no gap between them.

The parhypural is located at the preural centrum one, with the associated neural spine being significantly shortened.

All hypurals, the parhypural and the posteriormost three haemal spines are wider than the haemal spines more anteriorly. The hypural H1 is particularly wide and becomes twice as wide posteriorly where the anterior bases of six lepidotrichia rest.

There are six elongated uroneurals covering the dorsal surface of the urals and preurals PU1–4. These six uroneurals have approximately the same diameter, with the middle ones (U3, U4) being the longest. The anteriormost uroneural extends with its anterior end to preural centrum PU4.

One short urodermal lies on the surface of the first principal ray in JME-ETT3345a or slightly anterior to it on the posterior procurrent rays in JME-ETT4104a. The urodermal is the only element in the area of the caudal skeleton that is not covered by scales.

Dorsal to the uroneurals there are two elongated epurals which are dorsally covered by procurrent rays.

On preural centrum PU1, only a part of the neural arch is visible. A neural spine is absent. PU2 has a long neural spine, as have all vertebrae anterior to it.

Squamation . The cycloid scales of the body are thin, deeper than long and comparably large. The number of scales in the body axes is not countable exactly, but it corresponds approximately the number of vertebrae. The highest number of scales in a single transverse scale row from dorsal to ventral is about 15 between skull and dorsal fin, whereas there are only ten scales per transverse scale row in the area between the dorsal, anal and caudal fin. The anteroventral body scales are smooth, whereas most body scales have up to nine thin, elongated furrows that originate from an anterior point of a scale and radiate posteriorly, most prominently between the dorsal and anal fins (Fig. 8). Each scale of the body has a dark spot in the center, which is interpreted as “melanophores (chromatophores containing the dark pigment melanin)” (Tischlinger 1998: 1) and was probably a color pattern visible along the body axis with 10–15 lines with dots (in Thrissops ettlingensis best visible in JME-ETT3345a (Fig. 8) and JME-ETT220 (Fig. 9B). This color pattern of Ettling fishes was already described for Thrissops formosus by Tischlinger (1998) and Ebert et al. (2015).

Sensory canals and lateral line. The supraorbital sensory canal runs along the lateral margin of the frontal. The parietal branch of the supraorbital sensory canal is well developed and clearly visible in the holotype (Fig. 11), ending in the anteroventral part of the parietal. The preopercular sensory canal is with nine or ten ventral branches very well visible in the horizontal limb of the preoperculum in the holotype, whereas the sensory canal in the dorsal limb is less visible. The mandibular sensory canal runs anterior-posteriorly within a groove nearly within the middle of the dentary. It is clearly visible along its entire length in the transparent bone, and in some places pores open to the surface. The infraorbital sensory canal runs alongside the posterior and ventral orbital edges within the infraorbital bones. Branches, probably present in infraorbital 3 are hardly visible. A connection between the infraorbital and supraorbital canals is not visible.

The main lateral line along the body comes from the parietal, runs through the extrascapular, posttemporal, and the subcleithrum and is then visible as an indentation in the scales directly above the vertebrae on to the caudal fin (Figs 8, 11).

There is one single, quite large prey fish of Orthogonikleithrus hoelli in each of the stomachs of two specimens of Th. ettlingensis. In JME-ETT166 this single prey fish is well visible (Fig. 9A; Ebert and Kölbl-Ebert 2013, fig. 4b). In JME-ETT3345 the prey fish is already half digested, but here too there appears to be only one single prey fish in the stomach (see comparison with Th. formosus in discussion section below). In addition, the intestines in both fishes are filled with almost completely digested fish remains.

MJML K2295.

Kimmeridge, Isle of Purbeck, Dorset, England.

Upper Kimmeridgian.

Kimmeridge Clay, Dorset, England.

(for comparison of taxa see Table 1; for measurements of specimens and counts of features see Suppl. material 1: table S1). Maximum length 75.5 cm standard length (~90 cm total length); ~55–57 vertebrae (without ural centra); ~25–28 ribs anterior to anal fin; 29–30 anal pterygiophores; dentary with anterior hook armed with one large fang-like, posteriorly directed tooth; dentary teeth of irregular size, with the most anterior tooth and the teeth in the middle of the dentition being twice the size of the other dentary teeth (Figs 15A–D, 19A, B, D); curved maxilla with small teeth all in the same size (Figs 15A–D, 19A, B, 20C).

The specific epithet in Thrissops kimmeridgensis refers to the village of Kimmeridge, Dorset, England where most of the specimens were found.

MJML K26, 276, 306, 379, 452, 454, 518A, 518B, 525, 531C, 535, 547, 574, 575, 944, 949, 969, 998, 1001, 1002, 1015, 1069, 1085, 1129, 1142, 1155, 1159, 1167, 1173, 1194, 1196, 1230, 1282, 1283, 1290, 1301, 1304, 1313, 1316, 1338, 1373, 1378, 1379, 1395, 1407, 1419, 1439, 1440, 1456, 1506, 1520, 1521, 1534, 1550, 1590, 1653, 1669, 1672, 1676, 1677, 1690, 1714, 1745, 1786, 1792, 1804, 1839, 1925, 1934, 1946, 1981, 2003, 2022, 2063, 2079, 2119, 2138, 2167, 2294, 2333, 2399, 2491, 2700, 2770; NHMUK PV P.922, P.3686, P.3686a, P.54596, P.54597, P.54598, P.54599, NHMUK PV OR 40336, 40720.

The features of Th. kimmeridgensis are very similar to the other Upper Jurassic Thrissops species described above (for the most distinguishing features see Table 1). The cranial features of Th. kimmeridgensis are described and figured in detail in Cavin et al. (2013), especially the ethmoid region.

Here, I add two drawings of the cranium of Th. kimmeridgensis; one of the holotype (MJML K2295; Fig. 15A, B), the other of the slightly disarticulated cranium of MJML K1456 (Fig. 15C, D). Additionally, I figure the jaws and dentition of three specimens (MJML K1129, Fig. 19A; MJML K2119, Fig. 19B; MJML K1313, Fig. 19D), and the shape of the maxilla (Fig. 20C). In the following, I only describe features characteristic for Th. kimmeridgensisor which are not visible in the other above-described species.

The irregular size of the dentary teeth, the most striking feature to distinguish Th. kimmeridgensis from Th. formosus and Th. ettlingensis, was already recognised in Schaeffer and Petterson (1984, fig. 27J) and Cavin et al. (2013, fig. 33C) (see discussion in chapter discussion of features below).

The anterior ceratohyal is visible in some of the partially disarticulated craniums of Th. kimmeridgensis (MJML K1173, 1205, 1283, 1313, 1456, 1786, 1946, 2119). It is a rectangular bone, nearly two times longer than high, and has an ovoid foramen near the center (Fig. 15C, D). In MJML K1456 a series of at least 13 elongated and dorsally bent branchiostegal rays are articulated proximally to the lateral surfaces of the anterior ceratohyal (Fig. 15C, D).

The posterior ceratohyal or parts of it are visible in MJML K1173, 1313, 1456 (Fig. 15C, D), and 1946 (best visible in MJML K1173, 1946). It is a nearly reniform bone, which is broader than the anterior ceratohyal. At least the posteriormost five branchiostegals, which are twice as broad as the branchiostegals anterior to it, are articulated to the lateral surface of this bone.

Remains of the gill arch are visible in two specimens represented by rod-like elements of different length (MJML K1313, Fig. 19C; MJML K1407) probably epibranchials, hypobranchials and ceratobranchials). In MJML 1407 the ceratobranchials bear gill filaments.

Of Th. kimmeridgensis only few are preserved as complete fish (Fig. 14), most of the specimens listed under additional material are preserved as isolated crania or caudal fins. It seems reasonable to assume that these are the remains left by larger predators, which are well known from the same sites in the Kimmeridge Clay (for example crocodiles, ichthyosaurs, plesiosaurs, pliosaurs and sharks; Martill and Etches 2000).

The following Thrissops specimens are probably independent species. However, since only a few specimens from these localities are known and mostly in poor preservation, they are still awaiting revision.

Thrissops cf. formosus (late Kimmeridgian, Wattendorf, Bavaria Germany): NKMB (Wattendorf specimen, see Mäuser 2015, fig. 1014).

Thrissops ? curtus Woodward, 1919 (Tithonian, Purbeck and Portland, Dorset, England): NHMUK PV P.417a (holotype of Thrissops molossus Woodward, 1919; Fig. 16D), P.8381, P.10612 (holotype; Fig. 16A).

Thrissops portlandicus Woodward, 1895 (Tithonian, Isle of Portland, Dorset, England): NHMUK PV P.5538a (holotype; Figs 16B, C).

Thrissops sp. (early Tithonian, Creys, France): OSUG (UJF-ID. 16053*).

Thrissops sp. (early Tithonian, Daiting, Bavaria, Germany): coll. Tischlinger 76/72, 88/91; SNSB-BSPG 1964 XXIII 508.

Thrissops sp. (Tithonian, Purbeck, Dorset, England): NHMUK PV P.12643.

Allothrissops salmoneus (Blainville, 1818b)

The Upper Jurassic genus Allothrissops Nybelin, 1964, is problematic and needs to be revised. The existing literature has probably caused more confusion than clarity, because some important authors on this taxon (Blainville 1818a,b; Agassiz 1833–43 and Nybelin 1964) described fish from different species under the name of the type species Allothrissops salmoneus (Blainville, 1818b).

The holotype of the type species Allothrissops salmoneus (Blainville, 1818b) which is figured in Knorr (1755, pl. 31, fig. 1) is lost. Unfortunately, the illustration in Knorr (1755) does not show any diagnostic feature on species level. The specimen was most likely from Zandt (Zandt Basin, Solnhofen Archipelago, for localities see Fig. 1) due to the Liesegang rings typical for the Plattenkalk of Zandt depicted in Knorr (1755). The neotype (NHMUK PV OR 37078), proposed by Nybelin (1964), however, came from the Solnhofen Basin of the Solnhofen Archipelago (Fig. 1).

Blainville (1818a, p. 331) first names this species “Clupea elongata”. But since the species Clupea elongata Sesueur, 1818 (later recognized as synonymous with Clupea harengus Linneus, 1758) was described in the same year, Blainville (1818b, p. 27) renamed his species Clupea salmonea.

Nybelin (1964) split the genus Allothrissops from Thrissops Agassiz, 1833 with the species Allothrissops salmoneus, A. mesogaster both from Southern Germany and A. regleyi from Cerin, France.

However, the species Thrissops salmoneus described by Agassiz (1833) is not identical to the species Allothrissops salmoneus described by Nybelin (1964). All Th. salmoneus specimens, which Nybelin (1964) mentioned in his investigations originate from Eichstätt or Solnhofen, whereas those figured in the unpublished drawings intended for Agassiz’s work LDGSL/614/2/167, LDGSL/614/2/168 come from Kelheim (except for a specimen in Prague LDGSL/614/2/166 (NMP Uc64, missing today, see Ebert et al. 2022, fig. 18) which comes from Solnhofen, but cannot be determined more precisely due to its poor preservation. The unpublished drawings to Agassiz (1833–43) work are today in the Geological Society of London’s collection and online available (refs: LDGSL/613–616).

Evidence for the different interpretations of the species A. salmoneus may also be that Agassiz (1843, p. 128) says Allothrissops mesogaster is more elongated than A. salmoneus (“Thrissops mesogaster … une espèce très-voisine du Th. salmoneus, mais un peu plus allongée…”), whereas Nybelin (1964) described it the other way around.

Agassiz (1834) erected the nominal species Thrissops mesogaster with a brief description indicating its distribution in the Solnhofen limestones but without reference or illustration to a particular specimen. The first reviewer of this taxon, Nybelin (1964) refer the species to the genus Allothrissops and fixed the specimen NHMUK PV P.3679 as its lectotype.

Further specimens probably belonging to the type series of Allothrissops mesogaster are recognised in Ebert et al. (2022).