An exceptionally preserved Eocene shark and the rise of modern predator–prey interactions in the coral reef food web

Following extreme climatic warming events, Eocene Lagerstätten document aquatic and terrestrial vertebrate faunas surprisingly similar to modern counterparts. This transition in marine systems is best documented in the earliest teleost-dominated coral reef assemblage of Pesciara di Bolca, northern Italy, from near the end of the Eocene Climatic Optimum. Its rich fauna shows similarities with that of the modern Great Barrier Reef in niche exploitation by and morphological disparity among teleost primary consumers. However, such paleoecological understanding has not transcended trophic levels above primary consumers, particularly in carcharhiniform sharks. We report an exceptionally preserved fossil school shark (Galeorhinus cuvieri) from Pesciara di Bolca. In addition to the spectacular preservation of soft tissues, including brain, muscles, and claspers, this male juvenile shark has stomach contents clearly identifiable as a sphyraenid acanthomorph (barracuda). This association provides evidence that a predator–prey relationship between Galeorhinus and Sphyraena in the modern coral reefs has roots in the Eocene. A growth curve of the living species of Galeorhinus fitted to G. cuvieri suggests that all specimens of G. cuvieri from the lagoonal deposits of Bolca represent sexually and somatically immature juveniles. The modern trophic association between higher-degree consumers (Galeorhinus and Sphyraena) has a counterpart in the Eocene Bolca, just as Bolca and the Great Barrier Reef show parallels among teleost primary consumers. Given the age of Bolca, trophic networks among consumers observed in modern coral reefs arose by the exit from the Climatic Optimum. The biased representation of juveniles suggests that the Bolca Lagerstätte served as a nursery habitat for G. cuvieri. Ultraviolet photography may be useful in probing for exceptional soft tissue preservation before common acid preparation methods.


Introduction
Fossil record documents profound changes in both marine and terrestrial ecosystems across the Paleocene-Eocene Thermal Maxima (55.8 and 53.7 million years ago) [1][2][3][4]. These changes are preludes to the modern vertebrate faunas, with the increasing dominance of acanthomorph teleosts in marine systems and the emergence of artiodactyls, perissodactyls, and primates in terrestrial systems, among other radiation and extinction events [1,3]. It still remains an open question as to whether, how, and exactly what modern ecological traits and feeding interactions arose in such biodiversity hotspots as coral reefs during this time. These fundamental questions demand case studies of ecological traits in the fossil counterparts against those in components of living coral reef vertebrate fauna, but direct evidence that allows such one-to-one comparison is rare.
Despite the wealth of ecomorphological data on primary consumers, chondrichthyans from this latest phase of the Climatic Optimum have received little attention. Most chondrichthyans are secondary or tertiary consumers or apex predators, and those from the Bolca Lagerstätte should be no exception [5,22]. However, the overall morphology of chondrichthyans cannot be readily compared with that of teleosts using same anatomical landmarks in a single ecomorphological analysis, and direct evidence of trophic interactions has been lacking. Consequently, the teleost-centered paleoecological understanding of the Early Eocene vertebrate fauna has not readily transcended trophic levels above primary consumers. Several factors present a challenge to incorporating sharks into the paleoecological scheme, including documentation of head and body morphology, test of sexual dimorphism, and inference of trophic relationships. We solve these problems with the finely preserved specimens of the extinct carcharhiniform shark Galeorhinus cuvieri from Pesciara di Bolca Konservat-Lagerstätte. One of the specimens (MGGC 1976) receives a special focus for its exquisite preservation of soft tissues and for its stomach content.

Methods
The restoration project for damage caused by a 2012 earthquake allowed close examination of MGGC 1976 (slab and counterslab) using natural color and ultraviolet photography, stereomicroscopy, X-ray computed tomography, and SEM of extracted samples. The systematic use of UV light distinguished preserved tissues from cryptic reconstruction with fine grout and organic pigments in the 18th century. Furthermore, black light exposure revealed tissue-specific fluorescence, which delineated individual cartilages, muscles, brain, and visceral structures. We tested these interpretations under natural light, using differences in texture, topographical features, and colors.
We set three criteria for identification of the preserved elements: (1) clearly delineated edges under both visible and ultraviolet light, which are not obscured by more than 50 % of the outline; (2) area distinguished by texture, topography, and/or color enclosed within the outline; and (3) topographical and morphological consistency with corresponding elements in the living sharks ( Fig. 1). As G. galeus is a FAO-listed vulnerable species, we used males of the living carcharhiniforms Squalus acanthias and Mustelus mustelus for dissection.
Ultraviolet photography could improve screening for soft tissue preservation in specimens from Lagerstätten like Pesciara di Bolca. Although it is a common practice to prepare such specimens using acid solution [23], dissolution of matrix can lead to loss of preserved soft tissues or impressions.

Emended diagnosis
A carcharhiniform selachian with a unique combination of characters assembled from descriptions of living and fossil species [24,25]: mouth arcuate, with sides of lower jaw convex; pectoral fin greater in area than first dorsal fin; distance from snout tip to first dorsal origin much greater than interdorsal space; supraorbital crest moderately concave; teeth up to 5 mm tall; cusp bent toward the rear (from anterior files distally); upper anterolateral teeth low-crowned; mesial cutting edge much greater than distal cutting edge; distinct cusplets of decreasing size on distal heel, well-separated from cusp; thin root with a distinct furrow; basal edge weakly concave.   [25] and scaled to the same anteroposterior length. The slab was assembled from multiple parts using mortar glue (m1) and reconstructed using pigmented mortar (m2).   MGGC 1976 (the best-preserved specimen of G. cuvieri) lies on the right side of the body (main slab: upper; counterslab: lower), with the head twisted clockwise (Fig. 2). The pectoral fin has approximately twice the area of the first dorsal fin, and this ratio is greater than in the living species of Galeorhinus [25]. The caudal fin occupies a third of the entire body length, and its dorsal lobe has an accessory lobe. In comparison to the living species of Galeorhinus, the caudal fin is slender and the accessory lobe is small. The pelvic girdle extends beyond the pelvic fin posteriorly into a clasper, which indicates that MGGC 1976 is a male (Fig. 6b). MGGC 1976 (92 cm in total length) has a vertebral count of between 208 and 213 (114 to 118 in the trunk, 94-95 in the tail region). In comparison, MCSNV T 1124 ( Fig. 8d-f ) is a female of the same species (pelvic girdle clearly lacking claspers) and has a vertebral count between 200 and 204 (104 to 106 in the trunk, 96 to 98 in the tail). The vertebral count exceeding 200 in G. cuvieri is significantly higher than that in the living galeorhines (G. galeus: 136 centra, 83 in the trunk, 53 in the tail; G. japonicus: 160 centra, 106 in the trunk, 54 in the tail). Across the living selachians, vertebral counts of greater than 200 are rare. The exceptions are several carcharinid genera (Carcharhinus, Galeocerdo, Prionace, Triaenodon) and the leptochariid Leptocharias [25]. Dermal denticles are preserved in patches across the body (Fig. 7).

Chondrocranium and branchial skeleton
The tripartite rostrum of MGGC 1976 is relatively short among triakid sharks, not significantly longer than the orbital length (Fig. 3). The neurocranium is preserved with a clear outline on the left side of the head. The sigmoidal nasal cavity is entirely exposed on the right side and the sensory area for the ampullae of Lorenzini on the left side. Both right and left Meckel's cartilages can be delineated along the tooth series, meeting in roughly 70° (Fig. 3). The ceratohyals sit close to the proximal ends of the lower jaws. The palatoquadrates are preserved nearly in parallel to the Meckel's cartilages. On the left side of the head, a spiracle sits between the palatoquadrate and the hyomandibula. A full set of pharyngobranchials and epibranchials overlaps with at least three exposed ceratobranchials.

Dentition
The teeth of MGGC 1976 are small (3-4 mm crown height), cuspidate, and not bladelike (Figs. 4 and 5). Unlike the coeval Eogaleus [26], the crown height is never greater than the fore-aft basal length across the tooth series, even near the symphysis. From anterior files distally, the teeth have a cusp bent more posteriorly than dorsally so that the mesial cutting edge is greater than the distal edge. The cusp is upturned near the tip, which differs from other extinct species of Galeorhinus such as G. gomphorhiza, G. minutissimus, and G. ypresiensis and from other triakid genera such as Khouribgaleus, Palaeogaleus, and Triakis [24,25]. A similarly upturned outline is also present in G. duchaussoisi, G. goncalvesi, G. louisi, and Iago angustidens [24,27,28]. However, the teeth of G. cuvieri have dorsoventrally taller roots than the teeth of G. goncalvesi; have smaller cusplets relative to the bv, caudal vertebrae of S. bolcensis; cr, ceratotrichia; cs, claspers; i, intestine; ip, iliac process; ll, left liver; mp, metapterygium; pb, puboischiatic bar; pp, propterygium; rl, right liver; rp, radial pterigyophores; s, dermal denticle patches; st, stomach; sv, spiral valve; te, testis; vc, vertebral column Fig. 7 Additional information on the morphology of G. cuvieri, with emphasis on dermal denticles in the extinct and living species of Galeorhinus. a, b, X-ray radiograph of two vertebral centra sampled from a young female of G. cuvieri (MCSNV T 1124), showing four vertebral bands. G. cuvieri MGGC 1976 (c, e, g) and G. galeus (d, f, h). An interpretive drawing of MGGC 1976 shows areas of preserved dermal denticles sampled for scanning electron microscopy (SEM). c, d SEM micrographs of dermal denticles from the abdominal region in G. cuvieri (c) and G. galeus (d). e, f SEM micrographs of dermal denticles from the head region in G. cuvieri (e) and G. galeus (f). g, h SEM micrographs of dermal denticles from the pectoral fin in G. cuvieri (g) and G. galeus (h) cusp than in G. duchaussoisi and G. louisi; and are overall lower apicobasally than those of I. angustidens. Unlike G. duchaussoisi and Hypogaleus [25,28], the mesial and distal shoulders are at similar horizontal levels across the tooth series. From the anterior files distally, the distal heel develops distinct cusplets of decreasing sizes. The number and size of the cusplets vary from tooth to tooth (one in symphyseal to five in lateral; Figs. 4 and 5), but they are invariably oriented posterodorsally, irrespective of the orientation of the cusp. There are no more than five distal cusplets. The count of distal cusplets in G cuvieri is greater than in G. mesetaensis and G. minutissimus (less than three), but smaller than in G. duchaussoisi (up to six). The distal series of cusplets is separated from the cusp by a notch deeper than the distal cutting edge of the first cuspleta unique feature among triakids with the exceptions of Hypogaleus and Khouribgaleus [25,29]. G. duchaussoisi has a similarly deep notch in some of the anterior teeth, but these cusplets are larger relative to the cusp in this taxon than in G. cuvieri [28]. The teeth of G. duchaussoisi are also more asymmetric mesiodistally than those of G. cuvieri. In the anterior to anterolateral teeth of G. cuvieri, the mesial cutting edge has weakly developed denticles, but none of them is a distinct cusplet. This feature distinguishes G. cuvieri from other species of Galeorhinus such as G. ypressiensis [28]. The basal edge is weakly concave, and the basal furrow is as wide as the first distal cusplet.
These features distinguish the dentition of G. cuvieri from that of other triakids [24][25][26][27][28][29][30][31]. G. cuvieri differs from all other species of Galeorhinus in having the deep notch separating the cusp and the first distal cusplet, among other features that differentially sets G. cuvieri apart from each of the species of Galeorhinus in various combinations. The cusp is taller than at least three times the height of the first cusplet in G. cuvieri unlike Hemitriakis, Hypogaleus, Mustelus, Pachygaleus, and Triakis. The presence of multiple distinct distal cusplets in all but symphyseal teeth rules out Furgaleus, Iago, Paratriakis, Scylliogaleus, and Triakis, whereas the absence of parallel apicobasal ridges precludes Archaeotriakis, Mustelus, Palaeogaleus, Squatigaleus, Triakis, and Xystrogaleus.

Nervous system
Centered in the head of MGGC 1976, an area of intense UV reflection in contrast to the less reflective surrounding matrix indicates that the preserved structure is not simply an infilling of the endocranial cavity. Instead, it likely represents permineralization in the cerebrospinal space between the meninx and brain (M. I. Coates, pers. commn. 2015). Remarkable preservation allows identification of major external features in the central and  Table S1. Relative to G. galeus, G. cuvieri has a shorter rostrum, larger pectoral fins, taller second dorsal fin, shorter interdorsal distance, longer caudal fin, smaller accessory lobe, and smaller ventral lobe of caudal fin. Not to scale peripheral nervous systems (Fig. 3), which can be delineated with both highly reflective areas under UV light and texture/topography under natural light. The olfactory tracts (CN I) extend anteriorly in the approximate angle of 90°from each other. The right cerebral hemisphere sits above the base of the olfactory tract, although the left counterpart is obscured by the underlying teeth. These parts of the telencephalon are followed posteriorly by a pair of the optic lobes, and finally the cerebellum that sits on the medulla oblongata.
Within the left orbit, the eye and lens are clearly demarcated both in texture and outline.
Just anterior to the left preorbital process, an area with strongly UV-reflective dots may represent the chemosensory field of the ampullae of Lorenzini. On the right side of the head, the otic capsule is preserved with possible traces of semicircular canals. These potential semicircular canals are distinguished as narrow UVnonreflective loops within an intensely UV-reflective area. They appear to represent the proximal third of the anterior, posterior, and horizontal loops. Anterior to the otic capsule, an area filled with the non-UV reflective matrix represents the endolymphatic sac. Its potential counterpart on the left side sits on top of the medulla oblongata and cerebellum, presumably shifted as the chondrocranium was crushed.

Musculature
In MGGC 1976, several major muscles are visible both under natural and UV light on the left side of the head and in the branchial region (Fig. 3). The UV-reflective structure overlapping the left palatoquadrate and Meckel's cartilage represents the adductor mandibulae, a major jaw adductor that attaches to these two cartilages and occupies a large area of the head behind the postorbital process in lateral view. A similarly UV-reflective structure overlapping the hyomandibula posterior to the spiracle represents a second dorsal constrictor (levators hyomandibulae and hyoideus). Near the left series of the gill arches, two unclearly defined structures extend in parallel with the vertebral column. If these are muscles, the lateral element would represent the cucullaris, whereas the medial one would be the epaxial musculature.

Visceral tissues
The abdomen of MGGC 1976 is preserved in right lateral view on the main slab (Fig. 6). The stomach overlaps the large left liver, whereas the smaller right liver partly parallels the ventral margin of the stomach. The testis sits near the proximal end of the livers. The intestine follows the stomach in the sigmoidal twist on the left side of the body. Once at the midline, it parallels the ventral margin of the stomach posteriorly. The helical contents reveal that it was a spiral valve, as in the living selachians. Count of the spiral valves is six or more, slightly greater than that in the living species of Galeorhinus at 4 or 5 [25].
The pelvic girdle of MGGC 1976 has claspers. The stout, slightly anteriorly convex puboischiatic bar bridges the iliac process. The propterygium extends at approximately 80 degrees with respect to the puboischiatic bar. The metapterygium posteriorly extends into a clasper on both sides. This is not the case in MCSNV T 1124, which, therefore, is a female (Fig. 8d-f ).

Stomach content
The stomach content of MGGC 1976 consists of six articulated distal caudal vertebrae and dismembered caudal fin rays (Fig. 6). The amphicoelous centra are as long anteroposteriorly as tall dorsoventrally (approximately 7 and 8 mm each). In absolute size, the consumed prey had centra as large as those of the consumer (dorsoventral height of 8.2 mm for the centra below the first dorsal fin in MGGC 1976). The neural and haemal spines are semi-equal in length and oriented posterodorsally at approximately 40°. The spine length and the associated body outline show no evidence of tapering toward the caudal fin.
At this body size, candidates for the consumed fish are limited in the Bolca fauna. We considered scombrids and sphyraenids, both well represented in the locality [5]. The stomach content is not a scombrid, because neural and haemal spines do not decrease significantly in length from anterior to posterior positions along the vertebral column, because the spine orientation is at substantially lower angle around 40°, and because dorsal finlets are absent [32]. The vertebral centra are longer anteroposteriorly than tall dorsoventrally. These morphological features are consistent with the sphyraenid Sphyraena bolcensis [33,34] (Fig. 10). This association does not entirely rule out scavenging, but the intact preservation of an articulated skeleton in the stomach indicates that the preserved portion of S. bolcensis was ingested at or near death in a single large bite. Consistent with this, there is no evidence of erosion or weathering on the vertebrae of the prey.

Age and size estimates
Curiously, all six specimens of G. cuvieri fall into a range of similar body sizes from the total length of 54 to 92 cm (Figs. 2 and 8). Perhaps this small variation reflects similar ontogenetic ages among the individuals. We used the von Bertalanffy growth equation [35] to calculate age and mass estimates as routinely applied to modern chondrichthyans [36,37]: where t 1 is age of the individual, k is a growth coefficient (rate of change of length increment), L ∞ is the mean maximum body length for the population, L t is body length of the individual, and t 0 is hypothetical postnatal age extrapolated from a growth curve when length equals zero. We substituted for the parameters k, L ∞ , and t 0 with values taken from growth curves of multiple living carcharhiniform populations (Table 1). These calculations provide a range of numerical estimates. For example, a dataset for males of G. galeus from Australia [38] has the following life history parameters: where P ∞ is mean maximum weight for the population. Again, we use known values taken from a growth curve based on a dataset for males of G. galeus in Australia [38] in this example:  specimens of G. cuvieri ( Table 1). The data sets for different populations of G. galeus suggest the age of 3.9 to 5.8 years for the largest of the specimens (MGGC 1976 and MCSNV T 1124) and that of 2.2 to 3.1 years for the smallest (MGP-PD 8871C). This age estimate roughly agrees with the count of hypermineralized major vertebral bands of MCSNV T 1124 in X-ray radiograph (n = 4; Fig. 7a, b).
The biological implication of these estimates lies not in the absolute numerical values, but in the predicted relative age within each of the surrogate populations. When fitted against specimens of G. cuvieri, the von Bertalanffy growth curve for the Australian population of G. galeus suggests that all specimens of G. cuvieri with whole body preservation from Bolca, including MGGC 1976, represent sexually and somatically immature juveniles (Table 1; Table 1 Age estimates for specimens of Galeorhinus cuvieri using the von Bertalanffy functions for living relatives. All of the specimens are predicted to represent juveniles well beforeor in some cases nearsexual and somatic maturity The von Bertalanffy growth curves for small non-triakid carcharhiniforms do not fit readily to the specimens of G. cuvieri They either return unrealistic age estimates that are significantly older than the longevity of the living counterpart (x) or are unable to calculate the estimates because the specimens of G. cuvieri are larger than the mean maximum length (L ∞ ) of the original population (-). k = growth coefficient (rate of change in length increment); L ∞ = mean maximum length for the population; t 0 = hypothetical postnatal length (y intercept of the growth curve). S. = sex: C = male and female combined; F = female; M = male. Age at maturity is in years *Australian population; § New Zealand population; LF Length-frequency data; R1 Reader 1 (for vertebral bands); R2 Reader 2 (for vertebral bands) total body length at sexual maturity in the Australian population of G. galeus: males > 120 cm; females > 130 cm [38]). Although G. galeus has regional variations in growth rates, the body size at sexual maturity is largely consistent across the species. With the data set for the slow-growing New Zealand population of G. galeus [39], the age estimates are numerically different (MGGC 1976: approximately 5.8 years old). However, the specimens of G. cuvieri still fell into the range of juveniles (sexual maturity reached after 12 years of age in this population; Table 1). The total body length of 92 cm in MGGC 1976 would indicate age of 7-8 years in the Brazilian population of G. galeus, again well before sexual maturity (reached after 12 years of age at body length > 113 cm in males) [40]. Although numerical estimates for G. cuvieri vary geographically and from warm to cool waters, the inference of immaturity for all specimens of G. cuvieri remains consistent across different data sets for the living Galeorhinus.
To our knowledge, this study represents the first application of von Bertalanffy equation to fossil selachians. An important inference borne out in this exercise is probable sexual and somatic immaturity, not numerical values of the estimates. This is because of the assumptions underlying our fitting of the growth curves. These estimates assume that G. cuvieri had similar growth parameters to those of G. galeus. There is no direct evidence to support this assumption other than the congeneric status. However, the implication for sexually immature status would still be robust to violation of this assumption. When fitting the von Bertalanffy growth curves of other triakids and non-triakid carcharhiniforms, the specimens of G. cuvieri represent either sexually immature individuals or individuals near the sexual maturity in those respective taxa ( Table 1). The only exceptions are those in which the mean maximum size is substantially smaller than the specimens of G. cuvieri. In these small carcharhiniforms, the plateau of the growth curve is well below the sizes of the specimens of G. cuvieri so the age estimates are unrealistically high (denoted with 'x' or '-' in Table 1).
Hemitriakis, Mustelus, and Triakis may reach sexual maturity in the range of body size exhibited by the specimens of G. cuvieri (body length from 54 to 92 cm) among the living triakids (Additional file 1: Table S2). This is intriguing because triakids like Hemitriakis and Hypogaleus inhabit warmer waters than many of the sampled triakid populations (Table 1) [25]. Unfortunately, there is no robust data set for Hemitriakis and Hypogaleus to estimate von Bertalanffy growth curve. For this group of relatively small triakids, only the two largest specimens of G. cuvieri (MGGC 1976 and MCSNV T.1124) barely overlap the onset of sexual maturity in males of the documented populations of Mustelus and Triakis (Table 1). MCSNV T.1124 is a female (Fig. 8d-f) so this inference of sexual maturity does not apply to this specimen at the very least. In any case, none of these genera shares morphological characters with G. cuvieri that other species of Galeorhinus do not, both in the teeth and body proportions [24][25][26]. It is therefore difficult to consider them as a better analogue to estimate the relative age of G. cuvieri than the living Galeorhinus.

Taxonomy and sexual dimorphism
The alpha taxonomy of chondrichthyans from Pesciara di Bolca has a convoluted history. In brief, the six specimens referred to G. cuvieri (and to its junior synonyms) were distinguished from the coeval carcharhinid Eogaleus bolcensis on the basis of tooth and body morphology [24,26,41] despite some attempts for synonymization [42]. Subsequently, many tooth-based extinct species of Galeorhinus from other early Cenozoic localities were attributed to the tooth-based extinct carcharhinid Physogaleus, and G. cuvieri was also swept into that latter genus [22,28,43]. Finally, the fossil teeth classically assigned to Galeorhinus and Physodon were considered as female and male morphs of Physogaleus spp. [24,26]. This suspicion was raised because the type specimen of Physodon muelleri was re-identified as a male of Scoliodon laticaudus [44].
Our observations contradict the referral of G. cuvieri to Physogaleus and the implied sexual dimorphism in dentition in this species. G. cuvieri is best referred to the living genus Galeorhinus on the basis of a diagnostic combination of tooth and body characters. The living Galeorhinus does not exhibit distinguishable sexual dimorphism in tooth forms [30,38,45], and neither does G. cuvieri. This is borne out by the fact that a male (MGGC 1976) and a female (MCSNV T 1124) of G. cuvieri could be identified using the same diagnostic combination of tooth and body characters. However, the status of G. cuvieri does not determine whether or not other tooth-based species of Galeorhinus represent the carcharhinid Physogaleus [28] or whether or not the teeth of Physogaleus represent discrete sexual morphotypes [24,26].
The assignment of G. cuvieri to the living genus Galeorhinus reflects the observation that this taxon is morphologically closer to G. galeus than to any other living triakids. A suite of characters in G. cuvieriboth in the tooth morphology and body proportionsfollows that of the genus Galeorhinus (listed in Diagnosis) [25] and differs from the type G. galeus only in those diagnostic to the species (listed in Diagnosis; Figs. 4, 5 and 9). These characters diagnostic to G. cuvieri also differentially distinguish the taxon against other species of Galeorhinus (outlined in Dentition). The congeneric status with G.
galeus implies the existence of the genus for at least 50 million years. If G. glickmani from the Cenomanian of Russia also represents a congeneric species [28,31], the chronological range may even extend well into the Late Cretaceous times. With mosaic distribution of tooth characters among the living and fossil species of Galeorhinus, it would be difficult to distinguish G. cuvieri at the generic level to the exclusion of all other species of Galeorhinus. Optimally sorting the observed morphological variations at the generic level would require a comprehensive systematic reassessment of the fossil Galeorhinus spp. At this point, the conservative approach is to reflect apparent morphological proximity with G. galeus in the generic assignment of G. cuvieri.
The bolca as a possible nursery habitat for G. cuvieri The extant G. galeus inhabits warm temperate and tropical waters on continental shelves [25]. Juveniles tend to occupy inshore 'nursery' habitats such as protected bays or estuaries, whereas adults occur across the continental shelf. [30,38,[45][46][47][48][49][50] G. cuvieri may have had similar ecology to G. galeus. All six specimens of G. cuvieri with whole body preservation from Bolca likely represent sexually and somatically immature juveniles (discussed in Age and Size Estimates; Table 1). MGGC 1976 (length: 92 cm) is the largest, and five other coeval specimens range from 54 cm (BM B70) to 92 cm (MCSNV T 1124) in total length (Figs. 2 and 8; Additional file 1: Table S1). If all of the specimens of G. cuvieri are indeed sexually immature, the Pesciara di Bolca locality may have been the nursery habitat.
Consistent with the nursery interpretation, the Pesciara di Bolca assemblage deposited in a shallow lagoonal setting bordered by active fluvial and coastal systems and by coral reefs [8,9,51]. These environmental features agree well with the modern nursery areas (generally protected, nearshore/inshore habitats) for various coastal selachians, including triakids [52,53]. In such nearshore/inshore habitats, juvenile sharks tend to form a diffuse community of mesopredators feeding on similarly sized preys [54]. A nursery area may be identified on the basis of three criteria: (1) relative occurrence of the species in the habitat/ area versus outside; (2) duration of occupying the habitat/ area; and (3) repeated use of the habitat/area across years [52]. None of these criteria can be evaluated quantitatively for G. cuvieri with the current sample. Although the occurrence of likely juvenile individuals alone does not constitute direct support for the Bolca locality as a nursery for G. cuvieri, circumstantial evidence is consistent with that interpretation.

Trophic interactions in the Eocene coral reefs
The stomach content of MGGC 1976 indicates that G. cuvieri fed on S. bolcensis. The extant Galeorhinus frequently preys upon Sphyraena as a significant component of its diet that heavily relies on reef fish communities [25,30,38,45,46]. Sphyraena itself is a secondary to tertiary consumer in low-latitude coral reefs [47,55]. Therefore, the stomach content provides a rare glimpse of feeding relationships near top of the trophic network in this coral reef fish community near the exit of the Early Eocene Climatic Optimum [2,15].
The Galeorhinus-Sphyraena feeding interaction at higher trophic levels underscores modern features of the fish community of Bolca. The high functional diversity of herbivorous Bolca fishes is linked to post-Cretaceous increase in morphological disparity of acanthomorphs, and these are considered as important factors in shaping the Cenozoic coral reef communities [3,20]. Disparity analyses suggest that the colonization followed a classic niche-filling scenario [3,[18][19][20][21]56]. The invasion of herbivorous acanthomorphs into the coral reef habitats must have been accompanied by correlated shifts at higher trophic levels, and the Galeorhinus-Sphyraena association supports that prediction.
Coupled with the ecomorphological comparison of primary consumers [18][19][20]56], the Galeorhinus-Sphyraena association provides evidence for deep conservation of the food web structure in modern coral reef communities. The late Ypresian age for the Bolca locality suggests that this particular feeding interaction occurred under the shift of marine vertebrate fauna across the Paleocene-Eocene thermal maxima. The challenges to testing this interpretation are: (a) poor age constraints for both the predator and prey lineages; and (b) the paucity of pre-Ypresian marine Lagerstätten to link the modern features of the Bolca vertebrate fauna unambiguously to the climatic pattern. As for the Galeorhinus-Sphyraena association, the trophic relationship cannot be presumed earlier than the earliest occurrence of either predator or prey lineage. The uncertain identification notwithstanding, teeth referable to the Triakidae occur in the Cenomanian of Ukraine and, possibly, the Hauterivian of England [31,57]. It is difficult to constrain the origin of Galeorhinus or that of the Triakidae further. This is partly because the tooth morphology is highly heterogeneous among triakids [24,28] and partly because molecular evidence is equivocal about whether or not Galeorhinus is nested within the Triakidae [58][59][60]. As for Sphyraena, S. bolcensis provides the hard minimum calibration point for the split between sphyraenids and their sister groups [61]. So the trophic association is as deep into fossil record as the Sphyraenidae. This is consistent with the hypothetical link to the Climatic Optimum, but the causal relationship is non-falsifiable with the data available at hand.
At the faunal level, the paucity of pre-Ypresian Lagerstätten poses a challenge to associate modern features of the Bolca fauna with the exit of the Climatic