2. A short history of true and mythical discoveries of vertebrates in the Early Cretaceous of Libya

According to several authors (Robaux et al. Reference Robaux, Choubert, Flandrin and Drosdoff1942; Lapparent, Reference de Lapparent1951, Reference de Lapparent1960; Ouaja et al. Reference Ouaja, Ferry, Barale and Srarfi2002; Srarfi et al. Reference Srarfi, Ouaja, Buffetaut, Cuny, Barale, Ferry and Fara2004; D. Srarfi, unpub. Ph.D. thesis, Univ. Claude Bernard, Lyon, 2006), Early Cretaceous continental vertebrates were first recognized in Libya by Italian geologists, who described fossil fishes in the deposits of the ‘Continental Intercalaire’. This mention of ‘Italian authors’ can be traced originally to Robaux et al. (Reference Robaux, Choubert, Flandrin and Drosdoff1942), and was then cited by later authors. In this paper (the notice of a hydrogeological map of southern Tunisia; original pamphlet undated and perhaps effectively published in 1942 or 1943), Robaux et al. mention the recent discovery of a ‘rich fish fauna made in Tripolitania and recently described by Italian authors which leaves no doubt about the Wealden age of these deposits; these fishes are as follows:’ (translation from the French by the authors) but the text stops here without any list following the colon. In a long erratum kept at the Société géologique de France in Paris, it is noted that this publication was made difficult by the ‘1943 Tunisian events’ which prevented the authors from correcting the proofs. The so-called events are related to some episodes of the Second World War in North Africa, when German and Italian troops invaded the French protectorate of Tunisia at the end of 1942, before they had to leave the country in 1943 following the battle of El Alamein. Their defeat led to the departure of Italians from Libya and the closure of their geological institutions in Africa. However, we could not find a single original reference in the Annali del Museo Libico di Storia Naturale about these purported discoveries of fishes, and Desio (Reference Desio1950) in his bibliography of the geology of Libya does not mention Early Cretaceous (‘Wealden’) fossil fishes found by Italian geologists. It thus seems likely that Robaux et al. would have heard about discoveries of fossil fishes in Tripolitania by Italian colleagues, which were never published and soon forgotten following the collapse of Italian colonies in North Africa. An alternative possibility is that Robaux et al. mistakenly mentioned fishes when referring to Sassi's recent (Reference Sassi1942) report of Wealden plants in the Early Cretaceous of Libya. Although a forgotten publication in some obscure journal cannot be excluded, we suspect that one of our hypotheses is correct. Lapparent was thus the first author to report Early Cretaceous vertebrates from Libya in 1960; he mentioned several sites discovered by him along the Dahar escarpment in South Tunisia and Tripolitania (northwestern Libya) during a field trip in 1951, on the tracks of the French geologist Pervinquières who was the first, in Reference Pervinquière1912, to describe fossil vertebrates from the same horizons in southern Tunisia. In Libya, Lapparent mentions several localities in Nalut, Giado, Iefren and Garian. The Giado locality yielded a theropod tooth and a bone referred by him to Elaphrosaurus iguidiensis. Other sites provided fishes and crocodiles only and were not described in detail by Lapparent. Later, Burollet & Manderscheid (Reference Burollet and Manderscheid1965) mentioned teeth and jaws of crocodiles from the Cabao Formation. Hammuda mentioned in 1969 a crocodile tooth found at Wadi al-Mezaida near Giado. The Libyan geologist El-Zouki (Reference El-Zouki, Salem and Busrewil1980) later figured various teeth and scales referred to fishes, turtles and crocodiles as well as a dinosaur vertebra from the Cabao Formation in a sand quarry at Jannawan, near Giado. Although El-Zouki refers this material to Spinosaurus (?), the published plate (a vertebral centrum in ventral view) does not allow any precise assessment. More recently, Smith & Dalla Vecchia (Reference Smith and Dalla Vecchia2006) described a possible abelisaurid tooth collected by an Italian amateur palaeontologist near Nalut (see also Dalla Vecchia, Reference Dalla Vecchia1995 for a preliminary description of this tooth). The same locality has yielded abundant material of the hybodont shark Priohybodus described by Duffin in 2001. This material has been found, according to Duffin (Reference Duffin2001) and Smith & Dalla Vecchia (Reference Smith and Dalla Vecchia2006), in the Kiklah Formation, which overlies the Cabao Formation, but our own investigations in this area suggest that this material comes rather from the Cabao Formation. Smith et al. (Reference Smith, Tshakreen, Rasmussen and Lamanna2006) and Smith & Lamanna (Reference Smith and Lamanna2009), in two abstracts, briefly reported more discoveries in the Cabao and Kiklah formations, including abelisauroid bones, a possible titanosauriform, along with shark, turtle, lungfish and crocodyliform remains.

3. Geology

Several sand quarries located along the northern border of the Jabal Nafusah have yielded vertebrate remains. The material described here was found in three quarries: Tendamirah, Jawsh and Jannawan (Fig. 1). These are all opened in the Cabao Formation, a continental to shallow marine unit (estuarine with tidal influences on depositional settings: cf. Rubino et al. Reference Rubino, Lafont, Métais, Blanpied, Ferry, Ouaja and Sbeta2009) first described by Burollet in 1963 (Burollet, Reference Burollet1963a,Reference Burolletb), near the village of Giado. At this locality, it directly overlies the marine Late Jurassic Shakshuk Formation and underlies the Kiklah Formation (Fig. 2), a possible lateral equivalent of the Oum ed Diab and Chenini members of the Aïn Guettar Formation of southern Tunisia considered as Late Aptian to Albian in age (see Ouaja et al. Reference Ouaja, Ferry, Barale and Srarfi2002). Since that time, this stratigraphic succession has been adopted by all authors working in Djebal Nefusah (Hammuda, Reference Hammuda1969; El-Zouki, Reference El-Zouki, Salem and Busrewil1980; Fatmi, Eliagoubi & Hammuda, Reference Fatmi, Eliagoubi, Hammuda, Salem and Busrewil1980), but eight sedimentological sections logged between the Giado locality and the Tunisian border clearly demonstrate that an intermediate sandy and shaly interval appears and develops between the Shakshuk and Cabao formations in this area (Fig. 3). This series, up to 120 m thick is topped by a distinct member of lagoonal green shales and dolomites that clearly corresponds to the Douiret shale, first described in southern Tunisia by Busson in 1967 and ranged more recently in the upper Douiret Formation by Ouaja et al. (Reference Ouaja, Ferry, Barale and Srarfi2002). This additional interval intercalated between the Shakshuk and the Cabao formations is herein mentioned for the first time in Libya and should be considered as the lateral equivalent of the Tunisian Boulouha and Douiret formations (see Dehibat section, modified from Ouaja et al. Reference Ouaja, Ferry, Barale and Srarfi2002 in Fig. 3). The overlying Cabao Formation would itself be contemporaneous with the sandstones overlying the Douiret shales in Tunisia and appearing south of Touil el Hira (‘Barremian sandstones’ of Busson, Reference Busson1967) and later attributed to the Chenini Formation by Bouaziz et al. (Reference Bouaziz, Buffetaut, Ghanmi, Jaeger, Martin, Mazin and Tong1988) and Ouaja et al. (Reference Ouaja, Ferry, Barale and Srarfi2002). We suspect that only the uppermost part of this sandy interval should be considered as the Aptian–Albian Chenini Formation (see Fig. 3).

Figure 1. Location map.

Figure 2. Composite stratigraphical section of the Jabal Nafusah, NW Libya.

Figure 3. Lithostratigraphical correlation between Jannawan, Wazzin (Tunisian–Libyan border) and Dehibat (SE Tunisia) sections. Dehibat section is from Ouaja et al. Reference Ouaja, Ferry, Barale and Srarfi2002 (modified). Horizontal aspect not to scale. Distance between Dehibat and Jannawan: 120 km.

Tunisian geologists refer the Douiret Formation to the Aptian or Barremo-Aptian. This age assignment is based on its geometric position above Hauterivian to Barremo-Bedoulian levels of the Boulouha Formation (the latter being dated by the occurrence of the Late Hauterivian to Early Aptian (Bedoulian) rhynchonellid Loriolithyris russillensis: cf. Peybernès et al. Reference Peybernès, Vila, Souquet, Charrière, Youssef, Zarbout and Calzada1996) and below the Chenini member of the Upper Aptian to Albian Aïn Guettar Formation. The Douiret Formation is thus generally considered as Barremo-Aptian in age by Tunisian geologists (Ouaja et al. Reference Ouaja, Ferry, Barale and Srarfi2002; Srarfi et al. Reference Srarfi, Ouaja, Buffetaut, Cuny, Barale, Ferry and Fara2004). However, the brachiopod L. russilensis has also been described from older Berriasian strata in Algeria (Benest, Gaspard & Gali, Reference Benest, Gaspard and Ghali1996), which would rather suggest an age somewhere between the Berriasian and the Early Aptian for the Boulouha and the Douiret formations and the overlying lowermost sandstone interval in Tunisia, equivalent of the Cabao Formation. The fossil content of the Douiret Formation can help to constrain its age: the absence of angiosperms in its flora (Barale & Ouaja, Reference Barale and Ouaja2002) would also suggest an age older than Aptian. The microvertebrates from the Douiret Formation are known from the Late Jurassic to the Aptian, the youngest record of several taxa being in the Hauterivian or Barremian (G. Cuny, pers. comm.): the latest ascertained record of Egertonodus is in the Barremian (Rees & Underwood, Reference Rees and Underwood2008); Caturus has been found in the Barremian of Spain (Buscalioni et al. Reference Buscalioni, Fregenal, Bravo, Poyato-Ariza, Sanchíz, Báez, Cambra Moo, Martín Closas, Evans and Marugán Lobón2008) but not later, and the youngest record of Gyrodus is in the Hauterivian of Germany (Kriwet & Schmitz, Reference Kriwet and Schmitz2005). This faunal assemblage thus strongly suggests a pre-Aptian age for the Douiret Formation; the macrofauna does not yield useful stratigraphical information. In all likelihood, a Hauterivian to Barremian age is plausible for the Douiret Formation, although an older Berriasian or Valanginian age cannot be completely excluded.

4. Vertebrates from the Cabao Formation

The material described in this paper belongs to the collections of the Geology Department at the University Al Fateh in Tripoli, Libya (abbreviation: UT).

4.c. Dinosaurs

Only Saurischian dinosaurs have been found in the Cabao Formation. No ornithopod teeth have been found so far during our field work, although it would be premature to conclude that they are absent in the Cabao Formation given the limited size of our sampling (surface collecting). It should be noted that one isolated iguanodont tooth has been mentioned in the Douiret Formation of Tunisia by Srarfi et al. (Reference Srarfi, Ouaja, Buffetaut, Cuny, Barale, Ferry and Fara2004). Smith & Dalla Vecchia (Reference Smith and Dalla Vecchia2006) referred an isolated tooth from Nalut to the family Abelisauridae on the basis of a morphometrical analysis.

Two teeth from the Jawsh quarry share several very peculiar characters. The best preserved specimen (UT-JAW2; Fig. 4) is a 27 mm high recurved tooth slightly compressed labio-lingually with a granular enamel. The tooth is fluted and its mesial and distal carinae show extremely small serrations (7–8/mm). Their morphology is very close to that of the teeth of the spinosaurid Baryonyx from the Barremian of England (Charig & Milner, Reference Charig and Milner1997), although they seem to be broader in relation to the crown height than is typical for Baryonyx. We refer them to the sub-family Baryonychinae. Baryonychine spinosaurids have been described from the El Rhaz Formation in Niger, Suchomimus tenerensis Sereno et al. Reference Sereno, Beck, Dutheil, Gado, Larsson, Lyon, Marcot, Rauhut, Sadleir, Sidor, Varricchio, Wilson and Wilson1998 being probably a junior synonym of the extremely incomplete Cristatusaurus lapparenti Taquet & Russell, Reference Taquet and Russell1998; according to Milner (Reference Milner, Hurtado and Fernandez-Baldor2003), however, the spinosaur from Niger is indistinguishable from Baryonyx walkeri. Although the authors of these two papers did not fully illustrate the teeth of these dinosaurs, they mention that they bear extremely small serrations. The presence of serrations on the teeth is a good diagnostic character to separate baryonychine spinosaurids which have such serrations from spinosaurine spinosaurids (that is, Spinosaurus aegyptiacus, Irritator challengeri), the teeth of which have no serrations. In Africa, Spinosaurinae occur in the youngest formations of the ‘Continental Intercalaire’, that is, in Albian and Cenomanian deposits from Egypt, Algeria, Tunisia and Morocco, whereas older ante-Albian spinosaurids, which are known from Niger and Cameroon (J. D. Congleton, unpub. Master's thesis, Southern Methodist Univ. 1990), all belong to the Baryonichinae, suggesting a replacement during the Late Early Cretaceous (Buffetaut & Ouaja, Reference Buffetaut and Ouaja2002). Buffetaut (Reference Buffetaut, Mazin, Pouech, Hantzpergue and Lacombe2008) has mentioned possible spinosaurid teeth with larger serrations from the Late Jurassic of Tanzania.

Figure 4. UT-JAW2; tooth of cf. Baryonyx (Jawsh Quarry, Cabao Formation, Libya) in (a) labial, (b) distal, (c) lingual and (d) mesial views. (e) Enlargement of the mesial carina. Scale bars: 10 mm (a–d) and 2 mm (e).

A single tooth from the Tendamirah Quarry (UT-TEN15, Fig. 5) can be referred to a sauropod dinosaur. It is a large (preserved height: 74 mm) spatulate tooth with a wrinkled enamel, the crown of which is broken at mid-height. There is a distinct neck between the root and the crown. The labial face is convex and the lingual face is flat with two coalescing central ridges delimiting a mesial and a rostral concavity. The rostral and mesial edges of the tooth are parallel and vertical. This tooth (which belongs to the spoon-like tooth morphotype of Calvo, Reference Calvo1994) is clearly different from the teeth of Brachiosauridae, which show no constriction between the crown and the root, and is also different from the various morphotypes illustrated by Srarfi (D. Srarfi, unpub. Ph.D. thesis, Univ. Claude Bernard, Lyon, 2006) from the Chenini member in Tunisia. This constriction is reminiscent of the Turiasauria (cf. Royo-Torres, Cobos & Alcala, Reference Royo-Tores, Cobos and Alcala2006) but the crown does not show the typical heart-shape of this poorly known clade. This tooth has the closest affinities with those of Camarasaurus from the Late Jurassic of the United States, as well as with teeth from the Tiouraren Formation of Niger initially described by Lapparent (Reference de Lapparent1960) under the name of Rebbachisaurus tamesnensis and later referred to Jobaria tiguidensis by Sereno et al. (Reference Sereno, Beck, Dutheil, Larsson, Lyon, Moussa, Sadleir, Sidor, Varricchio, Wilson and Wilson1999), although the figured teeth of Jobaria are markedly different (see next Section). We provisionally refer the tooth from Tendamirah to the family Camarasauridae (?Camarasauridae indet.). Although Rich, Molnar & Rich (Reference Rich, Molnar and Rich1983) mentioned camarasaurid teeth from the Late Jurassic–Early Cretaceous Kirkwood Formation in South Africa, their figures do not show any constriction at the base of the crown and their material rather suggests a brachiosaurid sauropod.

Figure 5. UT-TEN15; tooth of a ?camarasaurid sauropod, (Tendamirah Quarry, Cabao Formation, Libya) in (a) mesial, (b) labial, (c) lingual and (d) distal views; scale bar 20 mm.

5. Biostratigraphical implications of the vertebrate assemblage

The ubiquitous abundance of the hybodont Priohybodus arambourgi may be interesting for biostratigraphical purposes as the genus is only known from Late Jurassic levels outside Libya and Tunisia (that is, in Yemen, Uruguay, Somalia and Ethiopia). Its last occurrence indeed was supposed to be in the overlying Kiklah Formation (see Duffin, Reference Duffin2001), but our field work at the site which has yielded the material studied by Duffin, as mentioned above, shows that it was actually found in the Cabao Formation, not in the Kiklah Formation. It is extremely abundant in the Cabao Formation but conspicuously absent in the Kiklah Formation, which is believed to be of Albian age (Novovic, Reference Novovic1977; Tekbali, Reference Tekbali2000). It can be noted that this genus is also the most frequent hybodont in the Tunisian Douiret Formation, whereas a much more diversified shark fauna is known from the Chenini and Oum ed Diab members of Late Aptian to Albian age (Cuny et al. Reference Cuny, Ouaja, Srarfi, Schmitz, Buffetaut and Benton2004); even though this may be linked to environmental differences as Priohybodus is considered a freshwater form, it is interesting to note that it is absent from the Tunisian formations of supposed Aptian–Albian age and that its last well-dated occurrence is in the Douiret Formation (Cuny et al. Reference Cuny, Ouaja, Srarfi, Schmitz, Buffetaut and Benton2004), that is, in the Berriasian to Early Aptian of Tunisia. The record of Priohybodus would thus instead suggest an age close to the beginning of the Cretaceous for the Douiret and Cabao formations.

The presence of a baryonichine spinosaurid also suggests an ante-Albian age, as African baryonichines are so far known only from the supposedly Aptian El Rhaz Formation of Niger and the Koum Formation of Cameroon, the vertebrate assemblage of which seems very close to the El Rhaz fauna (Taquet, Reference Taquet1984; J. D. Congleton, unpub. Master's thesis, Southern Methodist Univ. 1990); European baryonichines are known from the Hauterivian to the Early Aptian in England, Portugal and Spain (Milner, Reference Milner, Hurtado and Fernandez-Baldor2003; Buffetaut, Reference Buffetaut2007; Canudo et al. Reference Canudo, Gasulla, Gómez-Fernández, Ortega, Sanz and Yagüe2008). The rich Albian and Cenomanian deposits of North Africa have so far yielded only spinosaurine spinosaurids (Cavin et al. Reference Cavin, Tong, Boudad, Meister, Piuz, Tabouelle, Aarab, Amiot, Buffetaut, Dyke, Hua and Le Loeuff2010; Buffetaut & Ouaja, Reference Buffetaut and Ouaja2002) and spinosaurids are so far unknown from the European Albian localities. The precise age of the El Rhaz Formation is not so clear: it was considered as Barremian to Late Albian by Taquet (Reference Taquet1976), who suggested an Aptian age, an opinion largely followed by subsequent authors, although Taquet's suggestion was based on the referral to the Aptian by Busson (Reference Busson1967) of an isolated ornithopod tooth found in Tunisia by Lapparent. It has since been demonstrated that this iguanodontid tooth (found in the Oum ed Diab member of the Ain Guettar Formation, a few metres below the Cenomanian marine deposits) was more likely Albian, after the discovery of ammonites in the vicinity (D. Srarfi, unpub. Ph.D. thesis, Univ. Claude Bernard, Lyon, 2006). However, iguanodontid teeth are also known in the older Douiret Formation (Srarfi et al. Reference Srarfi, Ouaja, Buffetaut, Cuny, Barale, Ferry and Fara2004). Thus Taquet's main argument for an Aptian age seems weaker and his ‘Barremian to Albian’ interval for the El Rhaz Formation is likely, although an older age is not impossible as the lower part of the formation is not stratigraphically constrained.

The presence of a sauropod spatulate tooth in Libya would also suggest an age older than Aptian for the Cabao Formation. Although the best known camarasaurid genera (Camarasaurus and Haplocanthosaurus) are restricted to the Late Jurassic of the United States, isolated bones and teeth from Europe are slightly younger (Berriasian of Western France: cf. Le Loeuff, Buffetaut & Merser, Reference Le Loeuff, Buffetaut and Merser1996; Colin et al. Reference Colin, El Albani, Fursich, Martin-Closas, Mazin and Billon-Bruyat2004). The youngest possible camarasaurids known so far are Hauterivian to Early Barremian in age in Europe (locality of El Pelejon in Galve, Spain: cf. Sanz et al. Reference Sanz, Buscalioni, Casanovas and Santafé1987); incomplete spatulate teeth occur in the Tiouraren Formation of Niger, which underlies the El Rhaz Formation and the age of which has been strongly disputed by Rauhut & Lopez-Arbarello (Reference Rauhut and López-Arbarello2009), who, on the basis of the evolutionary grade of the dinosaurs, suggest a Late Middle Jurassic age for this unit. El Rhaz sauropods include the rebbachisaurid Nigersaurus taqueti and an undescribed titanosaur (Taquet, Reference Taquet1976; Sereno et al. Reference Sereno, Beck, Dutheil, Larsson, Lyon, Moussa, Sadleir, Sidor, Varricchio, Wilson and Wilson1999), but spoon-like teeth have been reported only in the older Tiouraren Formation (see Lapparent, Reference de Lapparent1960). Sereno et al. (Reference Sereno, Beck, Dutheil, Larsson, Lyon, Moussa, Sadleir, Sidor, Varricchio, Wilson and Wilson1999) have described the sauropod Jobaria in this formation; the single illustrated tooth shows serrations and is markedly different from the Libyan tooth and also from the spoon-like teeth illustrated by Lapparent; this suggests that the spoon-like teeth from Niger may not belong to Jobaria but to a second sauropod, possibly a camarasaurid. The absence in the Cabao Formation of ornithopod teeth which are extremely abundant in the El Rhaz Formation (but absent in the Tiouraren Formation) can hardly be taken into consideration given the small size of our sampling.