Palaeoecology of Voulteryon parvulus (Eucrustacea, Polychelida) from the Middle Jurassic of La Voulte-sur-Rhône Fossil-Lagerstätte (France)

Exceptional and extremely rare preservation of soft parts, eyes, or syn-vivo associations provide crucial palaeoecological information on fossil-rich deposits. Here we present exceptionally preserved specimens of the polychelidan lobster Voulteryon parvulus, from the Jurassic of La Voulte-sur-Rhône Fossil-Lagerstätte, France, bearing eyes with hexagonal and square facets, ovaries, and a unique association with epibiont thecideoid brachiopods, giving insights onto the palaeoenvironment of this Lagerstätte. The eyes, mostly covered in hexagonal facets are interpreted as either apposition eyes (poorly adapted to low-light environment) or, less likely, as refractive or parabolic superposition eyes (compatible with dysphotic palaeoenvironments). The interpretation that V. parvulus had apposition eyes suggests an allochthonous, shallow water origin. However, the presence of thecideoid brachiopod ectosymbionts on its carapace, usually associated to dim-light paleoenvironments and/or rock crevices, suggests that V. parvulus lived in a dim-light setting. This would support the less parsimonious interpretation that V. parvulus had superposition eyes. If we accept the hypothesis that V. parvulus had apposition eyes, since the La Voulte palaeoenvironment is considered deep water and had a soft substrate, V. parvulus could have moved into the La Voulte Lagerstätte setting. If this is the case, La Voulte biota would record a combination of multiple palaeoenvironments.

epibionts. The holotype displays on its carapace eight complete valves, shelly remains, or attachment scars of brachiopods (Fig. 4A). The most complete brachiopod, exhibiting a well-preserved ventral valve, is attached on the right side of the lobster near the median line, immediately posterior to the cervical groove (Fig. 4B). Three other epibiotic structures may correspond to the discoid attachment scars of dorsal valves. They occur near the ocular incision (Fig. 4A) anterior (Fig. 4C) and posterior (Fig. 4D) to the postcervical groove, on the left side of the specimen. Two other scars, located posterior to the branchiocardiac groove on the anterior right side of the carapace (Fig. 4E) and near the posterior margin (Fig. 4F), evoke shell-shaped surface swellings that would correspond to highly abraded ancient valves. These remains range from 180 to 500 µm in diameter.
The brachiopods attached to the specimen of Voulteryon parvulus have a characteristic sub-circular to sub-triangular outline (Fig. 4B,C), a feature typical of thecideid brachiopods 15,16 . Unfortunately, the diagnostic characters used in the systematics of these brachiopods are generally located on their dorsal (=brachial) valves and not ventral (=pedicle) ones, which are preserved in the specimens here discussed. In addition, the diameters of the ventral valves (or their traces - Fig. 4D-F) suggest that these brachiopods were juveniles. Due to their preservation and development stage, precise determination of specimens is impossible. Indeed, according to Baker 17 , early juveniles are not only less common than adults in the fossil record, but they are also more difficult to identify [16][17][18] . Nevertheless, the right antero-lateral portion of the best available specimen (Fig. 4B) preserves the exposed edge, which is typical of the ventral valves of early juvenile thecideoid shell 16 . Based on such criteria we consider these specimens as Rioultina-like forms. These thecideoid brachiopods were abundant during the Jurassic, with a range spanning from late Bathonian to late Oxfordian 18,19 . (G) hexagonal array with imprints of ommatidial lenses (SEM); (H) quadrate array with imprints of ommatial lenses (SEM); (I) more abrupt transition between hexagonal (black asterisks) and quadrate (white asterisk) arrays with imprints of ommatidial lenses (SEM). Abbreviations: ala, anterolateral angle; b 1 , hepatic groove; bc, branchial carina; bi, hepatic incision; c, cervical groove; cse, short expansion of carapace; d, gastro-orbital groove; e 1 e, cervical groove; ei, cervical incision; fm, frontal margin; o, eye; P1, pereiopod 1 (=thoracopod 4); pc, postcervical carina; pla, posterolateral angle; pr, postrostral carina; s1-s6, pleonite 1-6.   www.nature.com/scientificreports www.nature.com/scientificreports/ with the carapace length. A small phylogenetic effect is possible, since eryonids tend to have smaller eyes than coleiids, the other main polychelidan clade 5 .
Interpretation of ommatidia packing and facet shape. Hexagonal facets. The hexagonal facets packed in hexagonal lattice of V. parvulus are typical of apposition eyes, refractive and parabolic superposition eyes.
If we consider the apposition hypothesis: Eyes with apposition optics correspond to the simplest type of compound eyes, are a plesiomorphic condition for eucrustaceans, and are present in fossil and extant decapod larval stages 13,20,21 . Adult decapods in most shrimp, lobster, galatheoid anomuran, and ancient brachyuran groups, however, share a unique reflecting superposition visual system coupled to square-shaped ommatidial cornea, not seen in other marine eucrustaceans [22][23][24][25] . Noticeably, a number of decapod lineages have independently retained the larval apposition eyes 20,26 , or have evolved refractive or parabolic superposition eyes while retaining a hexagonal packing 27 . Previously, Rogeryon oppeli (Woodward, 1866) was the only polychelidan considered to have hexagonal facets (most likely apposition eyes) retained due to paedomorphic, more precisely neotenic evolution 5 . Voulteryon parvulus is probably also a paedomorphic form 28 , so it is more parsimonious to consider that its hexagonally-packed ommatidia correspond to larval apposition optics retained in adulthood as in R. oppeli. In contrast to R. oppeli, however, the adults are smaller in V. parvulus, and either progenesis or a combination of progenesis and neoteny may have occurred in this case. Another process might explain the presence of apposition eyes; in small scarabeoidea beetle, the eye is sometimes so small (less than 0.6 mm in diameter) that superposition optics cannot be accommodated 29 . For V. parvulus, such an interpretation is unlikely because: (1) a small specimen of Palaeopentacheles roettenbacheri (Fig. 3) has smaller eyes which are probably of the reflective superposition type 5 yet are smaller than those of V. parvulus and (2) V. parvulus has eyes that are more than 50% bigger than the biggest eyes of the aforementioned scarabeoidea.
It is also possible, yet probably less parsimonious, that V. parvulus, and perhaps also Rogeryon oppeli 5 , had a type of superposition optics that did not require square facets: refractive or parabolic superposition optics. From the fossil alone we cannot distinguish between these two options. However, while we can propose a simple evolutionary scenario explaining the presence of apposition optics in an adult by heterochronous retention of the larval www.nature.com/scientificreports www.nature.com/scientificreports/ apposition eyes, having refracting or parabolic superposition optics, although likely, is less parsimonious than the retention of apposition optics.
Square facets. The eyes of the holotype (and likely the paratype) show small patches of square-facetted ommatidia (Fig. 1E,H). These facets with rectilinear packing are strongly reminiscent to those seen in decapods with reflective superposition eyes 26 (although this shape can sometimes occur in parabolic and refracting superposition eyes 27,30 ). These 'mirror' eyes are well suited for dim-light conditions, and are commonly seen across nocturnal or deep-water decapod taxa 31 .  . Original data for V. parvulus from the present study; for polychelidans, from Audo et al. 5 ; for phyllosoma, from Tanaka et al. 34 ; for Dollocaris ingens, from Vannier et al. 10 .
www.nature.com/scientificreports www.nature.com/scientificreports/ Association of square and hexagonal facets. The combination of two types of structured arrays is unusual: in polychelidans, most documented arrays are square (9 species), with only one other example of hexagonal arrays 5 . Even within decapods, eyes combining hexagonal and square facets are extremely rare. They have only been reported in a handful of taxa: (1) the post-zoeal stage with all pleonal appendages of the hydrothermal vent caridean shrimp Rimicaris exoculata Williams & Rona, 1986 32 ; (2) the adult benthesicymid shrimp 30 ; (3) isolated decapod fossil eyes, supposedly of a phyllosoma larva from the Santana Group 33 ; and (4) at least one extinct and three extant brachyuran species 34 (Luque, personal observation). Yet, it is unclear whether such a combination of facet shape and packing implies that two discrete and functional visual systems are present (which may need different underlying optical neuropiles), or if it may occur as the outcome of the packing, size, and position of marginal facets being accreted to the eye.
Another fossil with an apparent combination of facet types is the polychelidan lobster Hellerocaris falloti (Van Straelen, 1923), also from La Voulte (Fig. 3). In H. falloti, most of the facets are square and packed orthogonally as in reflecting superposition eyes, although some facets are hexagonal or irregular (Fig. 3C). However, the non-square facets in H. falloti may just be a local packing artifact rather than a true regionalization of the eye. Contrary to H. falloti, in V. parvulus hexagonal facets are the dominant type, and may correspond to larval apposition optics retained in later instars, with the discrete squarish facets likely representing either a) an artefact of the packing, or b) a transition towards the adult reflecting superposition eye type seen across most of polychelidans from La Voulte.
In general, the number of ommatidia increases with growth, and the ommatidia may be inserted from a dorsal accretion zone, which points to a gradual development of the adult eye from the larval eye 13,35 . In the case of mantis shrimps, the adult optics of the eye arise on the ocular peduncle separated from the larval optics of the eye, and pushe it aside as they develop 35,36 . In the case of the gradual development of the eye, it seems that ommatidia optics are "pre-adapted" to their adult forms: light is reflected or refracted as in superposition optics, but the lack of clear zone and adaptations of the rhabdom force the eye to function as an apposition eye. This scenario is similar to that proposed by Tanaka et al. 33 to explain the occurrence of a patch of square facets on the anterior part of isolated fossil eyes.
Another interesting possibility is presented by benthesicymid shrimps, where the visual surface can be composed of more than one third of square facets, the rest being hexagonal. It has been proposed that these square facets are an evolutionary relict of ancestral reflective superposition optics 30 . Such a case would not be too surprising in the case of V. parvulus since the presence of reflective superposition optics is the ancestral condition in polychelidan lobsters is the presence of reflective superposition optics.

Visual capacities of the eyes in a deep-water environment. Voulteryon parvulus differs from most
other polychelidans and glypheidans from La Voulte, which are very likely to possess reflective superposition eyes 5 (see also Charbonnier et al. 37 , Fig. 337). Besides, due to its small size, the corneal lenses cannot compensate the absence of superposition optics by their size (larger ommatidial lenses can collect more light, as in deep-sea isopods or anomura, which can have large ommatidial lenses 25,38 -but also consider that small ommatidial lenses in V. parvulus are expected, since size of ommatidial lenses is mostly proportional to the size of the animal 5 ). The pooling of ommatidia signal as it occurs in deep-water hyperiid amphipods 39 also seems unlikely, since hyperiids have bilobed eyes (i.e., one lobe with a higher resolution and another with higher sensitivity) absent in V. parvulus or other polychelidan lobsters. Yet, we cannot exclude other processes that would increase dramatically the sensitivity of the eye such as temporal summation (photons are collected over longer periods of time, at the expense of temporal resolution) or neural resolution (adjacent rhabdoms combine part of their signal by neural connection 40 , herein impossible to study).
From the available data, we infer that the eyes of V. parvulus were likely poorly adapted to dim-light environments, as expected from eyes of the apposition type (note however that some possible mechanisms to increase sensitivity such as tapetum or rhabdom adaptations cannot be observed). Yet, whether the small square-facetted area in the eyes of V. parvulus functioned as superposition optics or not is unlikely, since superposition optics necessitate numerous ommatidia to collect the light optimally. Contrary to most other polychelidans and glypheidans in La Voulte, and similarly to Rogeryon oppeli, V. parvulus might have been better suited for bright-light conditions. This is surprising in a disphotic palaeoenvironment as it has been proposed by Charbonnier et al. 9,41 for La Voulte. In fact, the size of corneal facets and apposition optics of V. parvulus are similar to that of the co-occurring thylacocephalan crustacean Dollocaris ingens Van Straelen, 1923, which was considered to be adapted to well-illuminated environments.
We also acknowledge that it is possible that V. parvulus had a type of superposition optics that does not require square facets. i.e., refractive superposition, parabolic superposition. As above explained, this hypothesis is slightly less parsimonious, but it would however explain the presence of V. parvulus in a supposedly deep-water palaeoenvironment.
ontogeny. Despite its small size (holotype carapace maximum length = 10.0 mm, paratype carapace maximum length = 9.1 mm), Voulteryon parvulus was at a sexually mature size as attested by the development of the ovaries in the holotype 28 , and thus considered as paedomorphic. The development of the visual surface in V. parvulus herein documented reinforces the interpretation that V. parvulus was paedomorphic. If we consider that its ommatidial array is ontogenetically intermediate between larval and adult eye, it is surprising to observe such a late replacement of apposition optics. Even small juvenile specimens of Palaeopentacheles roettenbacheri (Münster, 1838) and Hellerocaris falloti have fully developed reflective superposition eyes 5 . The apposition optics of the eye of V. parvulus might therefore represent a case of neoteny, just as in Rogeryon oppeli, with the added minute size of specimens. It is possible that despite being sexually mature, V. parvulus had not finished its growth. Indeed, other eucrustaceans continue to grow after their sexual maturity 42 . www.nature.com/scientificreports www.nature.com/scientificreports/ epibionts. Age of the epibionts. Contrary to the well documented size-frequency patterns in fossil and modern brachiopods [43][44][45] , data reporting precise size/age ratio are far less available, especially for very young shell stages. Indeed, very small brachiopods are rarely recovered as fossils 46 and modern early instars are difficult to survey because of their small size. They often display transparent valves, and have a propensity for cryptic settlement 47 .
The age of the epibiotic brachiopods on V. parvulus (Fig. 4) can be assessed from measurements obtained from communities of modern benthic and subtidal temperate species (Scotland and New Zealand, respectively), and on one Antarctic species whose growth occurs much slower than that of temperate ones. Extrapolating data obtained on post-larval terebratulids 48 , we infer that the largest fossil brachiopods associated to V. parvulus attached and grew to their size in a timeframe of more than a month (500 µm reached in 43 days).
The exquisite preservation of the polychelidan host with appendages and delicate structures such as the eyes, all preserved in connection but above the internal organs fossilized in 3D 28 , implies a very short post-mortem exposure of its carcass. The inferred rapid burial of the host body (or displacement of the carcass to an environment unfavourable to many organisms) contradicts any post-mortem attachment/growth of the brachiopods that would have lasted more than a month. Thus, the brachiopods most likely colonized V. parvulus while alive, further supported by the location of the thecideoids on the crustacean, all attached to the carapace and none on the more articulated segments of the pleon (Fig. 4). This selective distribution evokes a biological elimination of the brachiopods. Indeed, attachment of the young epibionts to the pleonites should have been limited by the constant movements along the pleonal articulations.
Rarity of the association. Due to their sessile habit, brachiopods may be expected to live onto many other hard-bodied organisms. However, cases of fossil brachiopods identified as ancient ectosymbionts are few. Most examples known so far correspond to Cambrian associations from the Burgess shale-type deposits and Mesozoic sponge-reef foulings. Indeed, symbiotic lingulate brachiopods are known to have colonized the frond of the Cambrian algae-like organisms Malongitubus 49 and the spicules of sponges (Paterinata and Kutorginata) 50 . Nisusiid brachiopods are known to have been living in commensal of the motile Wiwaxia corrugata Walcott, 1911, attached to the dorsal spine of these Cambrian possible molluscs 12 . They are also reported as ectosymbionts of other lophophorates, like other brachiopods from the Cambrian of China 51 and Canada 52 or the Silurian of England 53 ; but also possibly attached to the disk of the middle Cambrian eldonioid Rotadiscus guizhouensis Zhao & Zhu, 1994, whose systematic nature and life habits remain poorly understood (semi-motile?) 54 . Spiriferid brachiopods attached on the proximal part of echinoderms spicules from the Carboniferous of Texas were also identified as likely commensals of their hosts 55 . From the Jurassic, some studies attempted to demonstrate the syn-vivo nature of some large scale foulings-like on sclerosponges 19,56 , bivalves 57 or other brachiopods (lacunosellids) 19 .
Evidences of clear post-mortem associations have also been identified; they involve a "stem-group" of brachiopod on the carcasses of the Cambrian arthropod Sidneyia inexpectans Walcott, 1911 58 but also much more recent thecideoids on the inner mould of Paleocene crabs 59 .
Consequently, confident reports of fossil symbiotic brachiopods on motile organisms were hitherto restricted to the colonization of Wiwaxiidae (possible molluscs), and to typical Burgess-like environments and fauna. The association of Jurassic thecideoid brachiopods and a polychelidan lobster here described corresponds to novel association between brachiopods and a motile benthic host (see Table 1 in the Supplementary Information). The colonisation of such a substrate may have provided various benefits to the thecideoids: (1) obtaining an attachment on a hard substratum; (2) avoiding being silted by sedimentary supplies; (3) access to an increased food supply for such filter-feeders provided by the locomotion of their host; (4) obtaining a favoured dispersion of the larvae, and thus a wider range of potential substrates to colonise 60 .
In the Ravin des Mines, where the Lagerstätte was deposited and V. parvulus was discovered, the substrate was soft (see Geological context in the Supplementary Information). Therefore, at this place, thecideoids were more likely to attach to other organisms. On the other hand, the Ravin du Chénier, a fossil locality without exceptional preservation, contains a rather rich sessile bathyal fauna involving a diversity of hexactinellid sponges and crinoids, ranging from the disphotic to aphotic portion of the continental slope, suggesting the presence of a hard substrate 8,41 . Surprisingly, in the latter, epibionts including thecideoids are rare 41 , and this fouling competition must have been inferior to that occurring in circalittoral environments. Given that the brachiopods presence should not have hindered the host fitness in any significant way, a commensal and facultative interaction between these thecideoids and V. parvulus can be inferred. This relation displays no counterpart reported in modern environment. That uniqueness could result of the under-examination of modern marine arthropods for their epibionts, like observed for other taxa (e.g. foraminiferans and bivalves 60,61 , but also of the reduced diversity of extant brachiopods. Palaeobiology of the epibiont. Thecideoid brachiopods are small cementing articulated cavity-dwelling forms. In modern environments, they occupy dim-light environments of cryptic habitats (=cryptobionts or coelobites 62 -sensu Kobluk 63 ), and the most probably colonised similar habitats in the past 44,[64][65][66][67] ). More precisely, in modern environments, they are typical for shallow water crevices, submarine caves, undersides of sponges and/or corals or other marine invertebrates and are widely distributed in tropical and subtropical seas 64,68-70 at a depth ranging from a few meters to about 150 m 68,[71][72][73][74][75][76] ). In the fossil record, thecideoid brachiopods often used lower surfaces of sponges as hard substrate 19,56,77 , coral-reef cavities 78 , empty shells on the sea floor 79 , and small cryptic spaces resulting from the disintegration of hardgrounds 62,[80][81][82][83][84] . They also incrusted surfaces of Jurassic brachiopods (Oxfordian lacunosellids 19 ), bivalves (Kimmeridgian oysters Actinostreon 85 ; Callovian Ctenostreon 57 ), sclerosponges (Oxfordian Neuropora 19 ), oncoids (Bajocian 86 and Bathonian 87 ) or bored cobbles 83,88 . Exceptionally, some extant thecideoid brachiopods live in deep-water environments at depths between 400 and 1000 m 76 . www.nature.com/scientificreports www.nature.com/scientificreports/ The cavity-dwelling strategies of cryptic brachiopods have been used earlier, not only in Mesozoic times (e.g. Triassic Thecospira [89][90][91] ), but probably also in the Palaeozoic (e.g., Silurian Leptaenoidea and Liljevallia 92 , Devonian Davidsonia 93 ), which have close phylogenetic relationships with thecideoid brachiopods 94 . Therefore, these thecideoid epibionts on the carapace of V. parvulus are a rare example of using mobile benthic animals as hard substratum for their colonisation. This contrasts with the standard occurrences of these brachiopods on sessile benthic invertebrates or pieces of rocks. Additionally, their photophobic character indicates at least dark palaeoenvironments during incrustations and can be used to infer bathymetric conditions of their host, deeper than photic zone of the sea floor.

Conclusion
The exceptional preservation of Voulteryon parvulus from the Middle Jurassic of La Voulte, France, including the presence of compound eyes bearing facets (this work), a unique example of a syn-vivo association between epibiont brachiopods and crustaceans (this work), and ovaries 28 , gives new insights on the palaeoecology of this polychelidan lobster and the La Voulte Lagerstätte.
The epibiont brachiopods of V. parvulus generally inhabit dim-light environments, which is congruent with previous hypothesis by Charbonnier et al. 41 and Charbonnier 8 of deep-water, low-lit conditions during the deposition of the La Voulte Lagerstätte. Surprisingly, the most parsimonious interpretation of the surface of the eyes in V. parvulus suggests limited capacity for vision in low-light conditions. These two apparently opposed results are difficult to explain with certainty. One hypothesis is that La Voulte Lagerstätte was not as deep as what seems to be indicated by the geology and sponges in the nearby Chénier ravine (hypotheses of Charbonnier et al. 41 and Charbonnier 8 ). This would agree with Vannier et al. 10 , who suggested more lit conditions based on the presence of the large-eye predatory arthropod Dollocaris ingens. Yet, the presence of thecideoid brachiopods on V. parvulus strongly indicates dim-light conditions: either due to the depth in La Voulte or to the microenvironment of crevices in a shallower environment.
Voulteryon parvulus was a mobile organism, and the geology in the vicinity of La Voulte is dominated by numerous faults that probably created an irregular bathymetry at the time of deposition of the La Voulte Lagerstätte. We therefore propose two additional hypotheses that could explain all observations (Fig. 6): 1. Nature of the eyes: Adults of V. parvulus had refractive or parabolic superposition eyes, so they were capable of living in a dim-light environment. The only problem of this hypothesis is that it requires an independent evolution of parabolic or refractive superposition eyes from either reflective superposition or apposition eyes. 2. Displacement or natural migration: V. parvulus may have lived in a more well-illuminated environment (as suggested by its eyes); which probably would have been different from the one in La Voulte. Since thecideoid brachiopods prefer crevices or dim light, it is possible that V. parvulus used to hide in rock crevices; V. parvulus was then either flushed to the deposition environment of La Voulte, or, more probably, La Voulte could have been a reproductive ground for V. parvulus 28 . This last suggestion seems plausible but  (1) www.nature.com/scientificreports www.nature.com/scientificreports/ opens the question: did V. parvulus spend enough time in the depth of La Voulte for the fixation and growth of the epibionts? Unfortunately, too little is known about the reproductive behaviour of extant polychelidans 95,96 to compare it with their fossil relatives.
Transportation of the thylacocephalan Dollocaris ingens, facilitated by its nektobenthic lifestyle 10 , and the irregular palaeobathymetry in the La Voulte area, could also explain the apparent mismatch between geological data and the ecology of some arthropods as suggested by their exceptionally preserved visual systems. La Voulte biota clearly combines the influence of multiple palaeoenvironments, as expected from the complex local geology.

Material and Methods
Material. We study the only two known specimens of Voulteryon parvulus (holotype: MNHN-F.A50708; paratype: MNHN.F.A29151), from the La Voulte-sur-Rhône Lagerstätte, housed in the palaeontology collection of the Muséum national d'Histoire naturelle (MNHN.F specimens). Both specimens are three-dimensionally preserved inside sideritic nodules and show very fine anatomical structures 1,28 . The holotype consists only of the part, and the paratype consists of both part and counterpart. Imagery techniques. Global views of the holotype and paratype of Voulteryon parvulus were done using a cross-polarized light setup 97,98 ) coupled to image stacking. Colour detailed views of the epibionts and eyes on the holotype were obtained with a 24 × 36 mm digital SLR equipped with a x10 microscope objective. The holotype was micrographed using a Hitachi Analytical Table Top Scanning Electron Microscope. The paratype was micrographed using a Tescan SEM (VEGA II LSU) linked to an X-ray detector SD3 (Bruker) (Direction des Collections, MNHN, Paris).
Additionally, we used the virtual slices and the reconstruction of the ventral surface of the holotype by Jauvion et al. 28 , reconstructed using the VG-Studio MAX 2.2 (© Volume Graphics) software on X-Ray Tomography (XTM, CT-Scan) data (voxel size = 9.55 µm, 2007 virtual slices), acquired at the AST-RX platform (USM 2700, MNHN) on a v/tome/x 240 L tomograph (GE Sensing & Inspection Technologies Phoenix X/ray) and equipped with a microfocus 240 kV/320 W tube delivering a current/voltage of 485 mA/95 kV.

Data Availability
All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Information.