The original slab was discovered during the 1970s among a batch of large slabs (Dalles de Provence) sold by quarry-workers for paving. Later, it was sold by the first owner to Mr Guy, who graciously loaned the slab to the Muséum national d’histoire naturelle (MNHN) in Paris, in order to cast and study it. The original slab is at the moment in a small French museum, the Préhistorama de Rousson, in the Department of Gard (catalogue number: MPR1). Unfortunately, some 40 years after its discovery, it is impossible to find out where the slab comes from with certainty. Nevertheless, the main quarry of Canjuers was exploited during the 1970s for the Dalles de Provence and the former owners of the slab stated to Mr Guy that it came from this site. There is no reason to doubt this testimony. Moreover, the sedimentological features of the slab fit perfectly with other slabs stored at the MNHN.

The original slab preserved the negative imprint of a trackway composed of 18 individual tracks (Fig. 2 and Fig. 3). It was cast at the MNHN in order to obtain the counterpart of the slab and study the positive imprint of the trackway (Fig. 4).

The trackmaker had been first considered a pterosaur based on the main imprints. But, given the presence of intermediate tracks, the second hypothesis establishing a limulid as the trackmaker was more congruent with our observations. In the past, limulid tracks have already been mistaken for bird or pterosaur tracks (Lockley et al., 2008 and Wellnhofer, 1991). Moreover, the trackway from Canjuers is very similar to one recently described from Cerin (Gaillard, 2011).

The Canjuers Lagerstätte is located in the military camp of Canjuers, in southeastern France, in the Haute-Provence region, more precisely in the Department of Var, near the village of Comps-sur-Artuby. The Canjuers lithographic limestones were first exploited in the late 1960s by several families of quarry-workers to produce ornamental stones called “Dalles de Provence” (see above). The commercial exploitation continued until the early 1970s when Canjuers was declared a military camp. Fortunately, access to the site by the scientific community was still granted. The first scientific studies of the locality date back to the 1970s (Ginsburg and Mennessier, 1970). Until the early 1980s, J. Fabre led the excavations at Canjuers (Fabre et al., 1982). During the 1980s, the Ghirardi family sold their collection of fossils, including the small dinosaur Compsognathus, to the MNHN. From 1983 to 1993, paleontologists from the Paris Museum excavated the site exclusively during the summer; these excavations resulted in the discovery of new fossils (de Broin, 1994 and Poyato-Ariza and Wenz, 1994). From the mid-nineties to 2004, paleontologists cancelled field trips to the Canjuers Lagerstätte. In 2010 and 2011, two geological field seasons permitted a fuller understanding of the geology, sedimentology, and paleoenvironmental history of the site (Charbonnier et al., unpublished data).

The ten quarries of Canjuers that have yielded fossils are located in a single area called “Les Bessons”, which covers a surface of more than 15 ha. The outcrop is composed of three distinct lithological units. There are, from the base to the top, lithographic limestones, bioclastic limestones and sublithographic limestones. Only the basal unit, characterized by alternations of thin lithographic limestone beds, corresponds to the Canjuers Konservat-Lagerstätte (Charbonnier et al., unpublished data). The composition of the lithographic beds indicates that the different units were deposited during different hydrodynamic cycles within the lagoon. Indeed, the lagoon's water was evaporating or refilling depending on the connection with the open sea. This connection was based on environmental conditions. Sometimes a complete rupture of this link caused the drying of the lagoon. Hence, the exceptional preservation of the fossilized organisms was favored, above all if microbial mats were laid down (Atrops, 1994). The organisms found in the lithographic limestones are either from marine or continental environments.

The first observations were done by the naked eye in order to observe the main characteristics of the trackway such as the alignment and the disposition of the tracks. Then we used the methodologies proposed for locomotion traces of arthropods (Gaillard, 2011 and Trewin, 1994) in order to carry out precise measurements on the whole trackway. We first evaluated the external (EW) and internal (IW) widths of the trackway, which were measured between each pair of main tracks and correspond to the maximum and minimum width of the trackway. We estimated the length of each series (S) of imprints. One series is composed of a main imprint and the three following intermediate imprints found on the same side of the trackway. The stride was also measured; it corresponds to the repeat distance (RD), it means the length between two sets of imprints. The width (w) and length (l) of each main imprint were evaluated. Concerning the intermediate tracks, we only estimated the length. Finally, the value of the angle (A) to mid-line was measured (Fig. 4). All the measurements were made twice using a ruler graduated in millimeters. In order to check some of them, we used the software Mesurim (Madre, 1998–2011) on digital pictures of the slab.

The preserved part of the trackway is symmetrical, straight and 39 cm long. It represents two complete sets of imprints and the beginning of a third one. The two first sets show two well-preserved main imprints and two or three intermediate imprints. The third set is located at the end of the slab and only preserves main imprints (intermediate are not visible); the one on the right is damaged at the extremity. No median imprint was observed on the trackway, which means that the limulid's telson was not in contact with the substrate during its motion (Fig. 3). The mean of the external width of the trackway is 175 mm and the mean of the internal width is 121 mm. The average series length is 180 mm (SL1 corresponds to the distance between PL1 and LL1-III, SL2 is from PL2 to LL2-III, the same for the right side). The mean stride is 178 mm. The angle to the mid-line was measured in the four cases; its mean is 13.75° (Table 1).

The main imprints of the trackway are the most external ones. In each set, two of them are symmetrically opposed: they correspond to the imprint left by the sixth pair of legs. These tracks are triangular-shaped: their anterior part is tetradactyl whereas the posterior part is narrower and bifid. They are the result of the contact between the substrate and the extremity of the sixth pair of legs, which leads to the flattening of their five individual pieces: four pieces spreading on the anterior part of the extremity and the fifth, which is bifid at the end, restoring the balance on the posterior part. The six imprints show equivalent proportions. On average, imprints made by the left leg VI (PL) are 28 mm long and 25.7 mm wide. Those of the right leg VI (PR) are 31 mm long and their width is 25.5 mm (Table 2).

These vary in number: two or three are in each set. Among them, two are fine, elongate and single: they are the imprints of appendages IV and V. The other type is slightly thicker and double: it is the mark of the third appendages, the first walking legs. The marks of appendages IV and V are inserted between two main imprints, and more internal than the latter. They are straight and slightly oblique because their anterior part points to the median axis of the trackway. Tracks of appendages III are located next to the main imprints, on their inner side. Two imprints do not appear on the trackway: LR1 – III and LR2 – III. Appendage III does not always contact the substrate, probably because of their small size or because we are dealing with undertracks (Goldring and Seilacher, 1971). The latter hypothesis is nevertheless unlikely because the preservation of imprints of the other appendages is good. On average, intermediate imprints are 15 mm long on the left and 14.6 mm long on the right of the trackway (Table 2).

Limulidae belong to Xiphosura, a taxon that includes both extinct and extant families (Dunlop, 2012). Recent forms are marine and their body plan consists of two parts: the cephalothorax or prosoma and the abdomen or opisthosoma (Fig. 5a). The latter ends in a long, peaked, tail called the telson. The upper surface of the cephalothorax is covered by a solid, horseshoe-shaped carapace and its underside bears six pairs of uniramous appendages (Dunlop, 2012). At the front of the animal, the first pair of appendages (I), the chelicers, is the smallest. They are used to grasp food and are followed by a pair of pedipalps (II) and four pairs of walking appendages, or legs (III to VI). Within Chelicerata, pedipalps are often transformed, yet within the limulids they are morphologically identical to the legs although smaller. The three first pairs of legs (III to V) are real locomotory appendages, bifid at their ends (Fig. 5a). The last pair (VI) is also used in locomotion but is transformed into more complex and longer legs, adapted to push the animal forward and is often called the pushers. The last segment of the pusher is not bifid anymore but is formed of five individual pieces (Fig. 5b).

Pusher appendages are only known within Xiphosura (Davis et al., 2007 and Eiseman and Charney, 2010). Thus, within Chelicerata, Limulidae is the only group able to leave a trackway such as the one we are studying. Pycnogonida and Arachnida also leave numerous tracks but none of them looks like the main imprints here (Davis et al., 2007).

The morphology of a complete limulid trackway is now well established based on observations of fossil and recent taxa (Caster, 1944 and Gaillard, 2011). It is composed of two main lateral imprints, corresponding to legs VI (Fig. 5c). There are also eight intermediate imprints, four on each side of the trackway (legs II to V). Chelicers are too short to leave a trace on the substrate. The telson can leave, in some cases, an axial median imprint. The prosoma only leaves a trace if the animal is ploughing. The studied trackway does not exhibit any telson or prosoma imprint. It is only composed of the imprints of legs III to VI. We observed that the legs VI left relatively well-preserved imprints compared to others: it confirms that the pushers play an important role in the locomotion of limulids. Conversely, the tracks of the other appendages are shorter and more superficial, some legs III imprints are even missing; this is the proof they have a minor role in the limulid locomotion. However, the visible imprints left by legs III show a bifid part (Fig. 2, Fig. 3 and Fig. 4); it probably means that when these appendages did contact the substrate, their entire extremity was used. All the tracks we have described as “oblique” show a low angle (on average 13.75°, Table 1) toward the central axis of the trackway. The regularity of the position of the tracks is consistent with a regular gait of the limulid. The principal stride (177.7 mm on average, Table 1) combined with the linearity of the trackway reveal a high speed. We used the characterization of limulid trackways defined by Gaillard (2011) to reconstruct the locomotory behavior of the Canjuers specimen. All the characteristics we have pointed out above, from the tracks to the set size, including the angle of the tracks regarding the median axis, demonstrate that the limulid was walking.

To infer the size of the limulid that made the tracks, we used a model peculiar to limulids (Malz, 1964) based on the following formula: I = W × 1.5, where I is the width of the prosoma of the trackmaker and W the external width of the trackway. In the Canjuers specimen, W equals 175 mm (Table 1), giving a width of the prosoma of 262 mm. In extant limulid species, the total length is 2 to 2.2 times the width of the prosoma (Sekiguri, 1988). It suggests a length ranging from 525 to 577 mm for the Canjuers specimen.

The limulid from Canjuers is slightly larger than the largest extant species, Tachypleus gigas Müller 1785, which has a 250 mm wide prosoma (Vijayakumar et al., 2000). Among extinct species, only the limulids from Solnhofen and Cerin seem to be larger (Frickhinger, 1994 and Gaillard, 2011). Indeed, most of the fossil limulids are small, many only a few centimeters long (Babcock et al., 2000, Lomax and Racay, 2012 and Pickett, 1984). The most imposing limulid trackway known is from Solnhofen and was described by Frickhinger (1994): it is 330 mm wide but remains an exception, because few invertebrate ichnofossils reach such a spectacular size.

The use of locomotor behavior is a recurrent problem in invertebrate ichnotaxonomy. Bromley (2004) stressed that, within fossil traces, “the same individual or species can produce different structures corresponding to different behavior patterns”. On the one hand, some authors hold that the only valuable ichnotaxonomic basis is morphological criteria (Bertling, 2007) and reject using variations in the behavior of the producer to synonymize ichnotaxa. On the other hand, other authors acknowledge that limulid trackways are among the most variable, depending on how they behave (Gaillard, 2011 and Seilacher, 2007). Thus, several limulid ichnospecies have been described based on this criterion (for instance the different ichnospecies attributed to Kouphichnium such as Kouphichnium variabilis Linck, 1949 and Kouphichnium rossendalensis Hardy, 1970). Here we describe traces rather close to those studied by Gaillard (2011) and we decided to consider behavior as valuable for taxonomic assessment.

The limulid ichnogenus Kouphichnium is quite common during the Late Jurassic. It was described by Nopsca (1923) and is based on the type ichnospecies Ichnites lithographicus Oppel, 1862 from Solnhofen. In 1964, Malz coined the specific name Kouphichnium walchi to the tracks of a dying limulid (mortichnia), also from Solnhofen. Later, Hasiotis (2004) introduced resting and locomotion traces from the Morrison Formation, which he referred to Kouphichnium isp. Finally, K. lithographicum tracks were found in the Nusplingen lithographic limestone (Schweigert and Dietl, 2002) and, more recently, Gaillard (2011) assigned this ichnospecies to a straight trackway from Cerin.

The Canjuers trackway cannot be referred to K. walchi (Malz, 1964) because the latter is a mortichnia, which really differs from the tracks described above. However, our trackway is quite similar to the one described in Cerin (Gaillard, 2011) and can be referred to the ichnospecies K. lithographicum. The only difference between both trackways is the size of the imprints and the width of the trackway, which could be the result of a different ontogenetic stage of the animal.

Because a single taxon can produce different kinds of trackways, the identification of a trackmaker is never obvious. However, two species are known from the Upper Jurassic fossil record. The first one, Crenatolimulus sp. n., is from Owadów-Brzezinki in Poland (Kin et al., 2013). The second one, Mesolimulus walchii (Desmarest, 1817), was found at the end of a mortichnia in Solnhofen and was also reported in Cerin (Gall et al., 1996). Both Kouphichnium and M. walchii are recorded in Solnhofen and Cerin. If M. walchii is responsible for the mortichnia K. walchi, we can assume that the same species is very likely the trackmaker of a part or the entire ichnogenus Kouphichnium, at least in the Upper Jurassic of western Europe. We tentatively refer the trackmaker of Canjuers to M. walchii, although we are aware that K. walchi, being part of a parataxonomy, is related to body fossils as well as tracks. We also acknowledge that the trackmaker could also be another species, not described yet. Indeed, extant limulid species are geographically isolated but have the same ecology, morphology (Shuster, 1982) and serology (Shuster, 1962). This implies that they could have coexisted during the Late Jurassic. The discovery of limulid specimens in Canjuers would reset our hypothesis.

Extant limulids are mostly marine, live close to the shore (Dunlop, 2012) and are euryhaline (Barthel, 1974). However, some fossil limulids from the Paleozoic and Triassic are considered freshwater because they were found in continental freshwater deposits (Babcock et al., 2000, Häntzschel, 1975 and Pickett, 1984). It was established that during the Late Jurassic Canjuers was a lagoon sometimes connected to the open sea via channels cutting across reef barriers and reef patches (Atrops, 1994 and Charbonnier et al., submitted for publication). This connection was not stable, so the site had to withstand temperature and salinity variations. Hence, limulids living in the Canjuers lagoonal environment were marine, just like the extant ones. The size of the trackmaker is consistent with this conclusion: the Canjuers limulid is indeed very large while freshwater species are said to be smaller than marine ones (Babcock et al., 2000).

Extant limulids mainly feed on bivalves and polychaete worms (Botton et al., 2003), two organisms found in the lithographic limestones of Canjuers (Charbonnier et al., unpublished data), worms being the tracemakers of Tubularina lithographica (Gaillard et al., 1994). Considering this point of view, Late Jurassic limulids could have fed in the lagoon. Nowadays, marine turtles (like Caretta caretta Linnaeus, 1758) are one of the principal predators of limulids (Keinath, 2003). Plesiochelydae have been found in the Canjuers deposit (de Broin, 1994). Their presence could have exerted a pressure on limulids, turning the lagoon into a hostile environment for the latter. The discovery of organisms from various habitats in Canjuers (Charbonnier et al., submitted for publication and Fabre et al., 1982) allows us to hypothesize that limulids were maybe transient in the lagoon. Storms could be at the origin of an accidental, probably temporary, introduction of many animals in the lagoon (Bourseau et al., 1991, Roman, 1994 and Wellnhofer, 2009). Another possibility is that limulids came into the Canjuers lagoon in order to lay their eggs on nearby beaches as extant species do. Those two hypotheses would explain why we study here the only one trackway known from Canjuers and why no body fossils of this group have yet been found there.