Material.

Nine articulated skeletons, MC-P-2011-01-TF1 to -TF9.

Provenance.

Corner of Avenue Philippe Solari and Chemin du Pin, city centre Aix-en-Provence (Bouches-du-Rhône, France).

Formation.

Aix-en-Provence Formation, member Calcaires et Marnes des stations d’essence.

Age.

Latest Oligocene.

Geographical and stratigraphical range.

†Lepidocottus aries is additionally known from Oligocene strata near Martigues, Bouches-du-Rhône, France [57].

General description.

The size ranges between 70 and 96 mm total length and 51 to 81 mm standard length (SL). The body is subcylindric, with a relatively long caudal peduncle (22–26% of SL, Table 1); body depth is about one-quarter (23–28%) of SL (Table 1, Figs. 3, 4). The body is covered with ctenoid scales (Fig. 3B4), but cycloid scales occur in the predorsal region (Fig. 3B3). The lateral line is absent.

The head is large and robust; its length is included about three times in the standard length (Table 1). The orbit is of medium size, subdorsal, and slightly elliptic; the eye diameter is approximately one sixth of the head length (Fig. 3B1). The mouth is terminal, the gape oblique, with the posterior end slightly in front of the orbit (Figs. 3B1–B2).

Neurocranium.

The ethmoid region is short and bears a short mesethmoid; other ethmoid bones and the nasal bone are not recognizable. The vomer is short and rounded and apparently bears no teeth. The frontal bones display an elongate supraorbital and a widened postorbital section and show prominent crests of the supra- and postorbital sensory canals (Figs. 4B–C1). A small infraorbital (io2 or io3) is present (Fig. 4B). In the otic region appear round epiotic bones, in four specimens with otoliths in situ (Figs. 3A1, 4C–D). Parietal bones are absent. The parasphenoid is visible in the lower third of the orbit (Fig. 3B2).

Otoliths.

The shape of the saccular otolith (termed otolith in the following) is rectangular, the lateral (outer) side slightly convex and the medial (inner) side almost flat. The dorsal otolith margin displays a fine crenulation and a prominent posterodorsal projection. The posterior otolith margin is first concave (below the posterodorsal projection), then it runs slightly oblique in posterior-ventral direction and meets the ventral margin with an angle of about 70–80°. The ventral otolith margin is crenulated and faintly bent; it bears a small praeventral projection. The anterior otolith margin is slightly concave; its junction with the dorsal margin is angular or faintly rounded.

The medial side of the otolith presents a sulcus that shows the “shoe-sole-like” shape that is present in most gobioid otoliths (Fig. 3A4). The sulcus covers approximately 85% of the median otolith length and is anteriorly extended, i.e. positioned closer to the anterior than to the posterior otolith margin. A distinct, but thin line borders the tip of the ostium; a small prae-ostial area is visible between the ostium tip and the anterior otolith margin. A slightly thickened crista superior is present above the middle and posterior part of the sulcus, while a thin crista inferior appears along the entire sulcus. Above the sulcus appears the elongate, moderately incised dorsal area. The ventral line is running along the ventral rim of the otolith and ascending posteriorly.

Note that the otolith preserved in situ in specimen MC-P-2011-01-TF1 (Fig. 3A4) displays an ostium that seems to be widely opened to the anterior margin, but this is an artifact produced by the poor preservation of this otolith; the thin suture bordering the ostium tip is destroyed due to corrosion.

Branchiocranium.

The general structure of the upper jaw is well exposed in several specimens. The premaxilla is bent in the anterior section and bears a prominent processus articularis (Figs. 3A1–2, 4A) and an almost rectangular-shaped, large posterior processus (Fig. 3B2); the processus ascendens is longer than the processus articularis (visible in specimen NMW 1910-V-12). The oral edge of the premaxilla has three rows of irregularly arranged alveoles and conical teeth of different sizes, i.e. small (0.08–0.17 mm), medium-sized (0.25 mm) or rather large (0.4 mm) (Fig. 3B5); the largest teeth insert mostly along the outer margin of the premaxilla. The maxilla is bent anteriorly and slightly expanded posteriorly, with a prominent articular head (Fig. 4A); it is toothless.

The lower jaw is robust, but not preserved completely. The dentary is associated with a wedge-shaped angulo-articular (Fig. 3A3), the oral edge of the dentary displays two or three rows of irregularly arranged alveoles and conical teeth of different sizes; the largest teeth reach 0.55 mm (the total length of this specimen is 90 mm).

The palatine is elongate, robust, and L-shaped, i.e. it has a prominent anterior process (Fig. 4A). The quadrate is three-dimensionally preserved in several of the specimens. It is triangular, with a well-developed posterior process and a large articular head. Posteriorly, the quadrate is articulated with the symplectic that is rod-shaped anteriorly and V-shaped posteriorly (Fig. 3A1), anteriorly it is associated with the elongate ectopterygoid and entopterygoid (Fig. 3B2). A gap between the symplectic and the preopercle is visible in the comparative material (NMW 1910-V-12, HLMD 1910-V-2543).

The hyomandibula is large, but usually damaged. The further hyoid region includes a large ceratohyal, which can be subdivided into an anterior ceratohyal and a triangular-shaped posterior ceratohyal ( = epihyal). Two narrow branchiostegal rays (rays 1–2) and four robust branchiostegal rays (rays 3–6) articulate with the anterior ceratohyal, between rays 1–2 and rays 3–6 is a distinct gap (Fig. 3B6).

The opercular bones are large, but mostly not well preserved. The subopercle is semi-circular (Fig. 4D). The crescent-shaped preopercle (best visible in specimen NMW 1910-V-12) has a lower and upper arm of almost equal length, and the opercle is rounded-triangular with the tip pointing to the ventral margin, a thickening is present at its anterior and posterior margins (visible in specimen HLMD 1910-V-2543).

Many conical pharyngeal teeth ranging in size between 0.05 and 0.35 mm are present in the branchial region of most specimens. Slender ceratobranchials and numerous, very small, spiny gill rakers are also visible (Fig. 4D).

Vertebral column.

It comprises 25–26 (10+15–16) vertebrae (v), which have a delicate net-like structure (Fig. 4C2). The anteriormost vertebral centra (v1 to v4/v5) are less elongate and less constricted in the middle than the subsequent ones.

The neural spines of v1 to v3 are expanded and triangular. The other neural spines are of almost equal length and elongate (Figs. 3A1, B1), with the exception of the short neural spine of the second preural centrum (PU2; Fig. 5A3). In addition, the neural spines of PU3–5 are more oblique than the preceding ones; they appear at the posterior end of the respective centrum.

The haemal spines are usually as long as the neural spines, with the exception of the haemal spine of the first caudal vertebra (v11), which is slightly shortened (Fig. 5A1), and the haemal spine of PU2, which is enlarged (Fig. 5C).

All abdominal vertebrae have long and prominent parapophyses, in some specimens preserved in connection with the ribs. Seven to eight rib pairs are recognizable, the last two rib pairs are slightly shorter than the preceding ones (Fig. 5A1). The ribs are long, have a strong and thickened proximal portion, and a pointed distal tip; long epipleurals are also present (Fig. 4D). There are no supraneurals.

Pectoral girdle.

The cleithrum is long, slender and reveals a broad to triangular widening in its ventral portion (visible in specimen NMW 1910-V-12). The supracleithrum is rather robust (Fig. 3A1). The endoskeleton of the pectoral fin consists of four well-developed hour-glass radials (R) with ovate-shaped gapes in between (Fig. 3B7). A scapula and coracoid are not recognizable.

Paired fins.

The pectoral fins, which are inserted in the lower third of the flank, comprise 14–16 rays. The pelvic fins are inserted just under or slightly behind the pectoral fins and are probably separated (Fig. 5A2). Each pelvic fin includes one spine (visible in specimen NMW 1910-V-12) and five rays. The endoskeleton of the pelvic fin consists of an elongate, slightly triangular basipterygium (Fig. 5A2).

Dorsal fins.

The first dorsal fin (D1) has six unbranched and unsegmented, medially paired rays, increasing in length from ray 1 to 3 and then slightly descending; the last ray follows with a short gap the preceding ones. Every ray is supported by an elongate pterygiophore, distally ending in a slightly concave depression (for the articulation of the ray) (Fig. 5B). The first pterygiophore inserts behind the neural spine of vertebra 4 and the last pterygiophore is situated behind the neural spine of vertebra 7; the D1 pterygiophore formula is 4(22110) (Figs. 5A1, B).

The second dorsal fin (D2) is inserted slightly in front of the insertion of the anal fin. The first ray may perhaps be a spine, and then follow 10 segmented and branched rays (Fig. 3B1). The first ray (or spine) is supported by two pterygiophores, which both insert in the interneural space behind the vertebra 9 (Figs. 5A1, B). Every pterygiophore supporting the rays is associated to a neural spine.

Anal fin.

The anal fin consists of a very small spine and 8 to 9 rays. The spine and the first ray are supported by a single pterygiophore. The first three or four pterygiophores insert in front of the haemal spine of the second caudal vertebra (v12), they are slightly longer than the following ones (Fig. 3B1).

Caudal fin.

The caudal endoskeleton bears two large, triangular hypural plates (HY1+2, HY3+4) and an additional small hypural plate in the dorsal part (HY5), which is separated by a thin suture from HY3+4 (Fig. 5C). A short and slender parhypural is present and is closely associated with HY1+2; its proximal region is reduced and does not reach the terminal centrum (Fig. 5C).

Two epurals (EP) are present (Figs. 5A3, C). The anterior one is characterized by a longitudinal median rib and pointed proximally; its proximal end is close to the terminal centrum. The posterior epural is slightly shorter than the anterior epural and also shows a longitudinal rib.

The number of principal caudal fin rays that are segmented and branched is 13. The caudal fin formula is 7/6 (Fig. 5C). In addition to the segmented and branched caudal fin rays, three long and ten short unbranched rays are present dorsally, and six rather long and an undetermined number of short rays appear ventrally (Fig. 5C).

The uppermost segmented and branched caudal fin ray is supported by HY5, the next six rays are supported by HY3+4, the next five rays are supported by HY1+2, and the parhypural supports the lowermost segmented and branched ray (Fig. 5C). In addition, the epurals and the widened neural spine of PU3 contribute to the caudal endoskeleton by supporting the dorsal unbranched rays (Fig. 5C). The expanded haemal spine of PU2 supports the first (longest) ventral unbranched ray (Fig. 5C).

Body scales.

Ctenoid scales are present all over the body except in front of the first dorsal fin. They do not differ much in size and shape in the dorsal and ventral body parts (Figs. 3A1, B1) and display regularly arranged radii and tiny ctenii (Fig. 3B4). Specimen MC-P-2011-01-TF2 shows well preserved rows of scales with an average width of 0.9 mm for the not-imbricated part of the scale (Fig. 3B4); thus, the number of scales along the lateral series ( = longitudinal scale row, see Fig. 2) can be estimated as being about 50.

Predorsal scales.

Cycloid scales are present in the predorsal region (Figs. 3A1, B1). They show distinct radii, numerous fine circuli, and are thinner and relatively higher than the ctenoid scales. In specimen MC-P-2011-01-TF2, the width of the not-imbricated part of a predorsal scale is about 0.5 mm and the predorsal segment that is covered with this type of scales is about 12.8 mm. Therefore the number of predorsal scales can be calculated to have been about 25.

Stomach content.

Specimen MC-P-2011-01-TF2 displays well-preserved content of the stomach and gut (Fig. 3B1), which consists of densely packed gastropod shells.

Osteology.

The separation of the Butidae and Eleotridae is supported by molecular data [13], but is less clear if anatomical data is taken into consideration [5], [7], [12]. Specializations both of the jaw musculature and caudal cartilage are the known synapomorphies used to characterize the Eleotridae [7]. However, muscles and cartilages are not preserved in fossil gobioids. Moreover, no synapomorphy is presently available to classify the Butidae; rather, this family is currently identified as lacking the apomorphies that define the Eleotridae [7].

The caudal endoskeleton, usually a significant source of diagnostic characters in fish taxonomy, may display apomorphies that help in correctly identifying butid or eleotrid genera [7], [51], [65], but it does not provide any diagnostic feature that can be interpreted as a synapomorphy at the family level [7]. The presence of two epurals and one parhypural, as in †Lepidocottus aries, occurs in several butid and eleotrid taxa [7]. The number of branched and segmented caudal rays is 15 in Butidae, with the exception of Kribia, which has 11–13 branched and segmented caudal rays [7]. Eleotridae usually possess 13 branched and segmented caudal rays (precisely as in Kribia and †L. aries).

Characters of the cranium of †Lepidocottus aries are the presence of an infraorbital bone (Fig. 4B) and irregular ridges along the orbital margins of the frontal bones (Fig. 4C1). Considering the Butidae and Eleotridae, the only taxa with which †L. aries shares an infraorbital bone are Bostrychus and Oxyeleotris (both belong to the Butidae), and the only taxon with which †L. aries shares the irregular ridges on the frontals is Butis (data from [52]). It is possible that these two characters represent a synapomorphy of the family Butidae (vs. Eleotridae) because both characters are absent in the Eleotridae (at least what can be said based on the available data). If this is correct, then the presence of these characters in †L. aries is indicative of affinities with the Butidae, and the absence of these features in certain butids represents a secondary loss. However, a more complete data set on these characters in extant Eleotridae is necessary to rule out the possibility that the infraorbital bone and irregular ridges along the orbital margins of the frontals represent homoplasies.

An additional interesting character is the pterygiophore formula of the first dorsal fin; it is 4(2211) in †Lepidocottus aries. Both Butidae and Eleotridae include genera in which the first pterygiophore inserts behind vertebra 4 (see [7] and here Table 3), like in †L. aries, and thus the position of the first pterygiophore is not significant at family level. However, the pterygiophore arrangement (2211), as seen in †L. aries, is equal to that of most Butidae, with a few exceptions in Kribia and Parviparma (see [7] and here Table 3). In contrast, Eleotris and several other Eleotridae display a (1221) pterygiophore arrangement (see [7], [9] and here Table 3). We agree with previous authors in that the phylogenetic significance of the pterygiophore formula is difficult to grasp, however, the (2211) arrangement of the pterygiophores appears more common in the Butidae than in the Eleotridae and hints to an assignment of †L. aries to the Butidae.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Meristic characters of †Lepidocottus aries (this study) and selected extant butid and eleotrid species.

https://doi.org/10.1371/journal.pone.0064117.t003

Meristics.

The composition of the unpaired and paired fins, the number of longitudinal and transverse scales, as well as the number of vertebrae are not appropriate to separate between Butidae and Eleotridae. Examples are the counts of I8 for the second dorsal fin, I9 for the anal fin, 9 for the number of transverse scales, and 10+15–16 for the precaudal/caudal vertebrae that all are present both in butid and eleotrid species (see Table 3). Moreover, it appears that the only meristic characters that are diagnostic for an individual genus include the number (or ranges) of predorsal and longitudinal scales (Table 3). Adding support to this suggestion comes from the “keys” for the genera of Butidae and Eleotridae [52], in which the number (or ranges) of these scales is used as a key character in the identification of genera. As a result, the number of predorsal or longitudinal scales represents an autapomorphy for a given genus of the Butidae and Eleotridae, but does not discriminate between the families.

Scales.

The scales of extant butids and eleotrids usually are ctenoid, but several exceptions exist. For example, Eleotris possesses small cycloid scales in the predorsal region, at the pectoral-fin base, at the abdomen, and in one or two rows along the borders of the median fins, whereas large ctenoid scales cover the sides of its trunk [66]. The eleotrid Philypnodon grandiceps (Krefft) also shows a combination of both scale types along the body, whereas Erotelis has exclusively small cycloid body scales [9], [67]. Among the Butidae, Pogoneleotris displays ctenoid plus auxiliary cycloid body scales, Kribia produces cycloid predorsal scales, but large and ctenoid body scales (with the exception of a few cycloid scales that may appear on the border of the second dorsal fin), while Bostrychus is the only butid genus that is covered exclusively by cycloid scales [52], [53]. As a result, the presence or absence of cycloid or ctenoid scales cannot be used for a general discrimination between Butidae and Eleotridae.

However, Pezold and Cage [66] have shown that the occurrence of many small cycloid body scales (termed as secondary cycloid scales in the following) can be interpreted as a derived character in Erotelis, while the appearance of large ctenoid scales is regarded as conservative. Accordingly we hypothesize that, during evolution, secondary cycloid scales successively replaced ctenoid scales in both Butidae and Eleotridae. If this hypothesis is accurate, then the many small cycloid scales of Bostrychus would be more derived than the scale pattern of any other butid genus. This is consistent with the position of Bostrychus on the molecular tree presented by Thacker [13]. As a result, the presence of secondary cycloid scales may represent a tool for determining relationships within the Butidae and Eleotridae.

Otoliths.

The morphological characters of the saccular otolith (termed otolith in the following) hitherto have received little attention in phylogenetic studies. Otolith formation involves hormonally-regulated calcium carbonate deposition within an organic framework, but otoliths are not part of the skeleton and have evolved independently [68]. Their significance in phylogenetic analyses remains to be fully explored; however, it is long since known that otolith morphology can be characteristic not only at the species level, but also at genus and family level ([29], [69] and many others).

We have compared the otoliths of †Lepidocottus aries with the otoliths of those extant butids and eleotrids of which data on otolith morphology are available (Figs. 6, 7). Clearly, the rectangular otolith shape and the anteriorly extended sulcus (with the ostium tip very close to the anterior otolith margin) of †L. aries are also present in otoliths of several Butidae (Figs. 6B, 7; see also [30]: Pl. 4, [65]: Figs. 11, 12). In contrast, otoliths of Eleotridae do not show these characters; they are usually quadratic or higher than long (exceptions occur in Eleotris) and their sulcus consistently is centered (Fig. 6C, see also [65]).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Otoliths of selected extant Butidae (refigured from [65]).

A. Butis butis (Hamilton), Sumatra, coll. Schwarzhans, leg. ZMH. B. Ophiocara porocephala (Valenciennes), Manus Island, Bismarck Archipel, ZMUC P.781771–78. C. Oxyeleotris lineolata (Steindachner), northern Australia, coll. Schwarzhans, leg. WAM. D. Bostrychus sinensis Lacepède, China, coll. Schwarzhans, leg. ZMH.

https://doi.org/10.1371/journal.pone.0064117.g007

Among the Butidae, the otolith of †L. aries shares the rectangular shape and the anteriorly extended sulcus with Bostrychus africanus (Steindachner), B. strigogenys Nichols, B. sinensis Lacepède (Fig. 7D), Kribia kribensis (Boulenger) (Fig. 6B), Ophiocara porocephala (Valenciennes) (Fig. 7B), and Oxyeleotris lineolata (Steindachner) (Fig. 7C). The otolith of Butis butis (Hamilton) is different from the former taxa in that it is rectangular to trapezoid in shape and possesses an almost centered sulcus (Fig. 7A). Otoliths of Oxyeleotris species other than O. lineolata are rectangular in shape, but their sulcus is centered (figured in [65]: Figs. 12L–N), which might suggest that their assignment to the Butidae may deserve further investigation.

The differences between the otoliths of the Eleotridae, and those of the Butidae and †Lepidocottus aries raise the question as to whether the characters otolith shape (rectangular in most Butidae and †L. aries vs. quadratic or higher than long in most Eleotridae) and position of sulcus (anteriorly extended in most Butidae and †L. aries vs. centered in Eleotridae) have phylogenetic implications. The different otolith shapes cannot be explained in terms of function, but "it is possible to argue from parsimony and suggest that the complex shapes are biologically meaningful” ([70]: 502). The biological meaning of a sulcus that is anteriorly extended is also difficult to interpret because the relationship between otoliths and inner ear function in teleost fishes are incompletely understood to date [70]–[73]. Perhaps otoliths with an anteriorly extended sulcus (as seen in †L. aries and the butids Bostrychus, Kribia, Ophiocara, Oxyeleotris lineolata) have the advantage of a broader range of sensitivity (such as improved hearing) in comparison to otoliths with a centered sulcus. If this is correct, then the anteriorly extended sulcus represents an apomorphic character for most butid lineages and †L. aries. This hypothesis is supported by a study on the oldest record of gobioid otoliths from the Lower Eocene of India [27]. In this paper, two new otolith-based fish species are described and interpreted as belonging to the Gobiidae. One of these is characterized by triangular otoliths, in which the ostium is wider than the cauda (see [27], Fig. 2a–f). These otoliths resemble those seen in present-day Odontobutidae (here Figs. 6E–F), and thus may represent an ancient member of this family, rather than a species of the Gobiidae. The otoliths of the second species, however, are nearly quadratic and have a centered sulcus (see [27], Fig. 2g–o). They resemble the otoliths of certain extant eleotrid and gobiid species, but clearly differ from the otoliths of extant Butidae and †L. aries. As a result, the otolith-based fossil record of Gobioidei from the Lower Eocene of India adds support to the hypothesis that the “quadratic shape” and “centered sulcus” are plesiomorphic otolith characters in an ancient lineage of gobioid fishes. Accordingly, the “anteriorly extended sulcus” of several butids and †L. aries would represent an apomorphic character. We therefore conclude that †L. aries most likely belongs to the Butidae (Fig. 8).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Hypothetical phylogenetic scenario of †Lepidocottus within the Butidae.

For genus apomorphies we used the number of scales in the longitudinal row (LR) according to Larson and Murdy [52].

https://doi.org/10.1371/journal.pone.0064117.g008

On the other hand, the most recent molecular phylogenetic hypothesis [13] interprets the Eleotridae as the sister of a clade containing the Butidae as sister of the Gobiidae and Gobionellidae. This raises the question as to whether †L. aries may represent a stem taxon to the [Eleotridae+[Butidae+[Gobiidae+Gobionellidae]]]. However, the fossil record does not provide support for this hypothesis. Fossil otoliths indicate that the Gobiidae were present in the Middle Eocene of India [74]. The oldest record of Gobiidae from Europe consists of a species of Pomatoschistus Gill from the Lower Oligocene of the southern Upper Rhinegraben [75]. Additional evidence of Pomatoschistus in the Oligocene of Europe comes from isolated otoliths from the Western Paratethys [76]–[77]. As a consequence, the split leading to the Gobiidae had probably occurred already in the Eocene, and the Oligocene-Early Miocene taxon †Lepidocottus cannot be regarded as a stem taxon to the [Eleotridae+[Butidae+[Gobiidae+Gobionellidae]]]. Based on the otolith data, osteological characters and the fossil record of the Gobioidei, we conclude that †Lepidocottus is closely related to the extant Butidae and represents the first skeleton-based record of this group.