Nous décrivons ici le squelette d'un nouveau sauropode, incluant des restes crâniens, récolté dans une succession marno-calcaire (60 m d'épaisseur) intercalée par de gros bancs de grès, datée du Jurassique moyen [10,17] dans les monts des Ksour de l'Atlas saharien occidental (Haut Atlas algérien) (Fig. 1). D'autres restes de dinosaures sauropodes, ainsi que de théropodes de grande et moyenne taille, mais d'affinités incertaines pour le moment, ainsi que quelques troncs d'arbres non déterminés ont été découverts. Seul un fragment de bennettitale (Cycadeoidea) provenant du mont Larouya (légèrement plus récent que le mont Rouis El Djir) est bien préservé, mais ce végétal est commun à tout le Jurassique et ne peut donc pas nous fournir plus d'informations pour une datation plus précise du site.

Sauropod remains were documented from the Early and Middle Jurassic strata of North Africa by Moroccan sauropods through the descriptions of « Cetiosaurus » mogrebiensis [13], Atlasaurus imelakei [22] and more recently Tazoudasaurus naimi [2]. In Algeria, Lapparent and Lucas [15] report the discovery of about 10 middle caudal vertebrae of an indeterminate sauropod from the Middle Jurassic (Callovian) in the Rhar Rouban region (15 km southeast of Oujda, Algeria).

Recently, we collected parts of a sauropod skeleton, including cranial material, discovered in a detrital series about 60-m thick intercalated between sandstone bars from the Middle Jurassic [10] and [17] of the Occidental Saharan Atlas (Algerian High Atlas), in the Ksour Mountains (Fig. 1). Sauropod remains were found associated with other indeterminate sauropod bones and theropod teeth. A bennettitale (Cycadeoidea) fragment was preserved in the Larouya mount (slightly more recent than the Rouis El Djir Mount) near the site, but this vegetal is representative of the Jurassic flora and does not provide a better support to date the site.

Dinosauria Owen, 1842 [23]

Saurischia Seeley, 1888 [25]

Sauropodomorpha Huene, 1932 [11]

Sauropoda Marsh, 1878 [18]

Cetiosauridae Lydekker, 1888 [16]

‘Géant des Ksour’ Chebsaurus algeriensis gen. et sp. nov.

Etymology. The generic name is from the Arab word Cheb, which designates a teenager, in a friendly and tenderly way in the West Algerian region (this sauropod being a juvenile), plus sauros, Greek for lizard. Specific name is from Algeria, referring to the site country.

Holotype. D001–01 to 78 (Fig. 2), temporarily in Sonatrach CRD, Boumerdès, Algeria; partial skeleton and cranial material of a juvenile sauropod including a part of basioccipital, a surangular and teeth.

Locality and horizon. Rouis El Djir mount, Oulakak, Sfissifa daïra, Naama wilaya, Occidental Saharan Atlas, Algeria. The site is at mid-slope of the Rouis El Djir mount on its northern hillside, in a continental detrital series from the indeterminate Middle Jurassic that overlaps the clay-limestones series from the Early Jurassic [10], [17] and [21].

Diagnosis. Primitive eusauropod displaying the following autapomorphies: a slender and elongated surangular; teeth crown and root with tuberculated surface; short and robust basioccipital tubera which diverge with a V-shape (60°) and their facets are inclined at about 70° to the horizontal in caudal view (their elliptical surface is oriented with a craniodorsal long axis); the base of the transverse process postzygodiapophysial lamina (podl, [29]) of the is included within the cranial part of the postzygapophysis in lateral view; there is a small elliptical fossa caudodorsally to the lateral excavations (pseudopleurocoels) on one middle dorsal vertebra centrum.

Chebsaurus is a sauropod, because it possesses teeth with V-shaped wear facets, spatulate (D-shaped) teeth crown with a wrinkled enamel surface texture, opisthocoelous cervical centra, a triradiate proximal condyle of ulna with a deep radial fossa, a reduction of an olecranon process on ulna and the pedal ungual of the first digit is sickle-shaped. We can specify that it is an eusauropod by the block-shaped carpal bones and the spreading metatarsus configuration [19], [30] and [31].

It is not a neosauropod because the dorsal surfaces of the cervical parapophyses are excavated and they are not separated from the lateral excavation by a ridge [26] and because it possesses marginal tooth denticles on the anterior margin of crown [5], [20] and [30].

Chebsaurus was a small, ~ 8–9-m-long sauropod that was at a juvenile stage because vertebra centra are not fused with its neural arches. A piece of the basicranium (D001–63) corresponds to the junction between basioccipital and basisphenoid, at the level of basal tubera (basioccipital tubera [BoT] for the basioccipital part and basisphenoid tubera [BsT] for the basisphenoid part). In caudal view, BoT are short and robust; they diverge from one another at an angle of about 60° with a V-shape (30–35° in Rapetosaurus [8]; BoT are thicker and more bulbous in Shunosaurus [6 (Fig. 5B)]) and their facets are inclined at about 70° to the horizontal in caudal view. Their surface is elliptical and oriented with a craniodorsal long axis. In lateral view, the arching neck between the occipital condyle and the BoT is missing. Between BsT, a deep fossa is pierced by several small foramina.

The lower jaw is represented by a long surangular (D001–77), broken cranially, which exhibits a smooth, sigmoidal curvature in lateral view and in dorsal view. It tapers caudally. Unlike the high and robust surangular of Atlasaurus (Läng, pers. obs.) or Shunosaurus [6], this is very low, long and slender (as Euhelopus [32]). On about the middle third of the lateral surface of the surangular, there are two oval (anteroposteriorly elongated) neurovascular foramina side by side and the dorsal margin gets thicker with prominent medial and lateral edges. In Atlasaurus, these two foramina are situated on the caudal third of the lateral surface (Läng, pers. obs.).

The dentary crown has a horizontal section in D-shape (spatulate teeth D001–78). The enamel is wrinkled as in other sauropods [30] and [31], and its texture is composed of many tubercles (without orientation) that we observe again on the root surface. Lapparentosaurus shows the same texture on its teeth crown surface but its root is slightly smooth with fine longitudinal ridges (Läng, pers. obs.). Only the neck is smooth and is very tightened as in Lapparentosaurus. Occlusal pattern is V-shaped as in primitive sauropods [26], [30] and [31]. Teeth are all isolated so there is no possibility to know the crown orientation. As in Mamenchisaurus and Omeisaurus, denticles are present on the cranial margin of teeth crowns (on cranial and caudal margins in Tazoudasaurus; absent on both margins in Neosauropoda) [26] and [30].

The bone structure of the presacral vertebrae is solid. Parts of proatlas and axis as six incomplete cervical centra were found. All cervical centra are opisthocoelous as within all sauropods [19] and [30]. The high/diameter ratio of cranial cervicals centra is more than 1 (1.14 for D001–15) as Chinese eusauropods (Euhelopus, Mamenchisaurus, Shunosaurus, Omeisaurus) [26]. Dorsal surfaces of the cervical parapophyses are excavated, but they are not separated from the pleurocoel by a ridge (unlike neosauropods). Pleurocoels seem to be simple and not divided. No cervical neural arc was found. The caudal part of centra is always missing, except for the smaller vertebra.

Six anterior and middle dorsal centra (among two incomplete), notably cranial centra, are amphicoelous as in Lapparentosaurus but the cranial end is very slightly more convex and the caudal end is very shallowly more concave, as in middle dorsals in Patagosaurus. They possess deep excavations on their lateral surfaces as in all sauropods (except Shunosaurus, Mamenchisaurus, Lapparentosaurus, Malawisaurus and Dicraeosauridae) [26], but it is rather simple pit (pseudopleurocoel), which is not extensively ramified within the centrum as in Cetiosaurus oxoniensis from Oxford and ‘Cetiosaurus oxoniensis’ from Leicester and which lies on the cranial two-third of the lateral surface. In lateral view, on the caudodorsal part of one vertebra centrum (?6^th, D001–17), there is also a small rather deep elliptic fossa (with a large craniodorsal diameter of 25 mm and a short cranioventral diameter of 15 mm). This feature is not found within other sauropods so it is regarded as an autapomorphy in this study. Just one caudal part of a dorsal neural spine was found and there is no complete neural arc. In dorsal view, the spine is at once lengthened craniocaudally (as Lapparentosaurus and ‘Cetiosaurus oxoniensis’ from Leicester) and at once somewhat widened transversely (crosswise-like): very thin at the centre and very widened transversely with robust spinoprezygapophysial and spinopostzygapophysial laminae cranially and caudally, respectively. The spinopostzygapophysial lamina (spol) is single (as in Lapparentosaurus, Cetiosaurus, and Shunosaurus). Postzygapophyses have an ovoid plane surface with a craniocaudally long axis that faces ventrally and very slightly laterally. In lateral view, forwards, in the postzygapophysis, there is the base of the postzygodiapophysial lamina of transverse process. It is remarkable in this specimen that the base of this lamina is included within the postzygapophyses; in other specimens, it is well separated: cranial to the postzygapophysis. So this character state is regarded as an autapomorphy.

The only sacral centrum (D001–19) is very eroded.

Only three cranial middle caudal vertebrae were found (one complete, one incomplete and one very eroded). As Ferganasaurus, Lapparentosaurus and Cetiosaurus, they are amphicoelous and do not show lateral excavations. The length/height ratio of anterior caudal centra is about 0.63 (D001–20) as in many eusauropods (except in Malawisaurus and Titanosauridae without Opisthocoelicaudia) [26]. The suture with the neural canal lies on the cranial two-third of the centrum, as in many sauropods. The caudoventral margin bears chevron facets on D001–20 (others caudals centra are too less preserved to determine that).

The scapular blade (D001–52) has an elliptical-like section, with a thicker caudoventral border and a thinner craniodorsal border, as Lapparentosaurus and Patagosaurus. The caudoventral border presents a narrow furrow as in Lapparentosaurus.

The two heads of each humerus (D001–31 and 41) are available. They have the characteristic spoon shape of all sauropod (with a concave cranial and a convex caudal side). The proximal margin is rounded and the articular head is projected backward. In proximal view, the cranial margin is relatively straight (not sigmoid in a transverse section as in Lapparentosaurus or Ferganasaurus, where the lateral border is curved towards front to form the beginning of the deltopectoral crest). The right ulna (D001–32) is nearly complete; the left is reported by the cranial branch of its proximal end and by the distal end. The proximal end of the ulna is triradiate and forms a deep radial fossa as in most of sauropods [30], [31] and [33]. The ulnar olecranon process is low and reduced, at the level of the proximal articulation as in sauropods [30] and [31]. It is gracile as in Lapparentosaurus (midshaft width/length ratio about 10% in Chebsaurus and Lapparentosaurus). Only the proximal end of the right radius (D001–34) is available and presents a caudomedial beak. Carpal bones are block-shaped as in eusauropods [30] and [31] with a rounded palmar surface and a flattened dorsal surface as in Lapparentosaurus (both flattened surfaces in Shunosaurus, Diplodocus). No metacarpal bone (D001–25 to 28) is complete, only ends are preserved but their assemblage is semitubular as in many eusauropods [26] and [33]. A nonungual phalanx (D001–01) is broader transversely than long craniocaudally as in eusauropods. Manual unguals (D001–08 to 10) are sickle-shaped and much deeper dorsoventrally than broad transversely as in eusauropods, those of the digit I being the largest [30] and [31].

There are many flat pieces that should be part of ilium. The ischiatic peduncle (D001–37) is triangular in caudoventral view. Its cranioventral part, which participates to the acetabulum, is flat. The pubis (D001–36), represented by its proximal end, has a poorly developed ambiens process. The extremity of this process is rather rounded in both lateral and proximal views whereas in Lapparentosaurus it is more developed: rather tapered in proximal view and mildly prominent forward in lateral view.

Hindlimbs are represented by the distal end of a fibula (D001–40) that presents a distal condyle with a medial margin expanded transversely, more than twice the mid-height diameter as Barapasaurus, Mamenchisaurus and Brachiosaurus. The assemblage of metatarsus (D001–03 to 07) has a spreading configuration as in eusauropods [7], [12], [14], [18], [20] and [30]. The transverse axis of its proximal condyle of the Mt I is perpendicular to its long axis (as Rayososaurus, Patagosaurus and Barapasaurus). Mt I is short and very robust as in many eusauropods [26]. A nonungual phalanx (D001–02) is rectangular and more elongated craniocaudally than transversely. Pedal unguals (D001–29, 38, 39, 48, 60) are sickle-shaped and much deeper dorsoventrally than broad transversely as in eusauropods and those of the digit I is the largest [30] and [31].

Chebsaurus is the most complete sauropod skeleton found in Algeria. It allows us to fill out the Middle Jurassic sauropod register in an anatomical and systematical viewpoint.

Compared to the other well-preserved Early Jurassic sauropod skeletons like Vulcanodon (Vulcanodon Beds, Mashonaland North, Zimbabwe [24]), Tazoudasaurus (Toarcian continental detrital series, Toundoute, Morocco [2]) and Middle Jurassic sauropod skeletons, including Shunosaurus (Xiashaximiao Formation, Sichuan, China [9] and [34]), Ferganasaurus (Balabansai Formation, Fergana Valley, Kirghizia [1]), Patagosaurus (Cañadon Asfalto Formation, Chubut, Argentina [4]), and Atlasaurus (Wawmda, Central High Atlas, Morocco [22]), Chebsaurus is different on the basis of numerous anatomical characters. But it shows much more affinities with the Middle Jurassic taxa Lapparentosaurus madagascariensis (Isalo III Formation, Bathonian, Madagascar [3]), Cetiosaurus oxoniensis (Forest Marble Formation, Bathonian, Great Britain [28]) and ‘Cetiosaurus oxoniensis’ (Rutland Formation, Bajocian, Great Britain [27]): the neck of teeth rather very tightened, amphicoelous dorsal and caudal centra (the latter lack excavations on their lateral sides), a single spinopostzygapopysial lamina on dorsal neural spine, gracile ulna, carpal bones with a rounded ventral and a flattened dorsal surfaces.

Looking forward to discovering the remainder of the dinosaur, a more detailed description of this new specimen as well as a phylogenetic study will be necessary to test the monophyly of this Middle Jurassic group (Läng and Mahammed, in prep.). It might be a comeback to Bonaparte's hypothesis (basing on vertebral anatomy) [3] that proposed to form a Cetiosauridae monophyletic group composed of Lapparentosaurus, Volkheimeria, on the one hand, and Barapasaurus, Cetiosaurus and Patagosaurus, on the other hand.