De très nombreuses missions géologiques européennes se sont succédé depuis la fin du XIX^e siècle au Spitzberg, révélant ainsi sa richesse et sa diversité en fossiles de vertébrés inférieurs. Cependant, aucun reste de dipneuste n'y avait jamais été signalé dans le Dévonien. Le matériel décrit dans ce travail a été récolté au cours de la mission française CAST (Cartographie géologique, paléontologie, stratigraphie et tectonique) dans les terrains d'âge Dévonien inférieur et moyen du Sud d'Andrée Land, au Spitzberg. Cette région, géologiquement et paléontologiquement mal connue en comparaison de la région nord d'Andrée Land, a livré les premiers restes de dipneustes dévoniens du Spitzberg.

The material described herein has been collected in 2002 during the French CAST expedition (Geological Cartography, Palaeontology, Stratigraphy and Tectonics) in the Lower and Middle Devonian outcrops, South of Andrée Land, Spitsbergen. This expedition, completed with a second one in 2003, has provided a detailed geological mapping and numerous palaeontological data on the Devonian sediments along a west–east transverse section of the southern Andrée Land (between the Ekman and Dickson fjords) and along the Nathorst and Hugin valleys, east of the Dickson fjord. The aim of this fieldwork was, amongst others, to establish a precise stratigraphic correlation between the southern Wood Bay Formation and the northern sequence. The Lower Devonian faunal assemblage of North Andrée Land is well documented, compared to that of South Andrée land, thanks to numerous European expeditions (e.g., 1898 Swedish expedition; 1906 to 1925 Norwegian expeditions, 1929 and 1939 English-Norwegian-Swedish expeditions; 1964 and 1969 French CNRS–MNHN expeditions). In spite of these palaeontological researches, and of many hundreds of metres of sequence, most of which yielding fishes (e.g., agnathans: [3], [4], [23], [29], [37], [41] and [42]; placoderms: [24], [25], [26], [27] and [35]; sarcopterygians: [17], [18], [19], [31] and [38]), no Devonian dipnoan remain had previously been found in Spitsbergen, whereas this group is usually common in the vertebrate assemblages of the classic Devonian localities of North America, China, Russia and Australia.

Most part of the dipnoan material comes from a large isolated block bearing scattered scales and some dermal bones, most of them broken and difficult to determine, except for an opercular (MNHN SVD 498, Fig. 1). The matrix is a slightly calcareous grey-greenish psammitic sandstone, showing numerous small calcareous-clayey pebbles and a layer of detritic gravels. The block also contains chopped plant fragments. The associated vertebrate fauna consists in various remains of Doryaspis sp. (Heterostraci, Pteraspidiformes, Protopteraspididae), Actinolepis sp. (Placodermi, Arthrodira, Actinolepidoidei), large Euleptaspis sp. (Placodermi, Arthrodira, Brachythoraci, Homostiidae), and Heimenia ensis (Sarcopterygii, Dipnomorpha, Porolepiformes, Porolepididae).

A small isolated pentagonal dermal element showing well-marked Westoll lines (MNHN SVD 499, Fig. 2) was also found during this expedition. Its matrix is a red-and-green psammitic sandstone and its associated vertebrate fauna consists of remains referred to Doryaspis sp., Euleptaspis sp., undetermined placoderms, and Heimenia ensis.

The large loose block with scattered dermal elements containing the dipnoan opercular was collected close to the bottom of Mount Toldstad, western coast of Dicksonfjord, on its eastern slope, at the altitude of 190 m; it has been impossible to find its original layer. However, according to the stratigraphic series of Mount Toldstad and to the associated vertebrate fauna, this dipnoan opercular most probably comes from the Stjørdalen faunal Division. Ørvig [38] claimed that “the Emsian–Eifelian boundary (here taken in the same meaning as in the Rhenish sequence) is situated somewhere in the Stjørdalen Division, but perhaps as high as in the Verdalen Member.” This dipnoan remain is thus considered here as Late Emsian to Early Eifelian in age [6].

The small isolated pentagonal element was found on the west side of Mount Kota at the altitude of 190 m; the block was found on a slope and, once again, its original position in the outcrop was impossible to recover. According to the stratigraphy of Mount Kota and of the vertebrate faunal assemblage, this specimen most probably comes from the Stjørdalen Division beds and is also considered here as Late Emsian to Early Eifelian in age.

Most of the dermal bones found in the large block are fragmentary and not determinable with certainty. The only complete bone is quite a large cosmine-covered opercular (Fig. 1), which certainly belongs to a sarcopterygian. Its quadrangular shape is very different from the elongated opercular of the porolepidid Porolepis [31]. Although the opercular of Heimenia, the second porolepidid previously known from the Devonian of Spitsbergen, is unknown, it is assumed that its shape would have been similar to that of Porolepis. This opercular is very similar in shape to that of all Devonian dipnoans, such as Tarachomylax [2 (fig. 12)], Speonesydrion [9 (fig. 27E and F)], Soederberghia [32 (fig. 3)], Dipterus [28 (fig. 8c)] and Pillararhynchus [1 (pl. 5, fig. 14)]. One of its margins, the only slightly concave one, shows a distinct overlapped area (Fig. 1, ov.a). The cosmine layer is missing on a large area of the bone, a condition supposed to be due to natural resorption and redeposition and not to wear. The resorbed areas show a thick smooth dentine layer and does not display tubercle or any ornamentation, to the contrary of dermal bones or scales of Tarachomylax [2 (fig. 5)], Speonesydrion [9 (fig. 27)], Iowadipterus [43 (fig. 9)], Dipnorhynchus [48 (figs. 86–89)], and Uranolophus [12 (fig. 43)]. The most remarkable feature of this opercular is the presence of some Westoll lines on its peripheral part (Fig. 1, w.l) a unique feature of dipnoans. These lines are very distinctive and follow the bone outline. They are clearly different from the cracks also present on the bone. Ørvig [39] and Thomson [47] considered these lines, where cosmine is never formed, as boundaries of separate growth episodes.

The inner surface of the bone cannot be observed and is difficult to clean with acid preparation, due to its numerous cracks. The presence or absence of a large sausage-shaped muscle scar high up on the inner anterior region of the opercular would have been interesting to check, since this scar seems to be present in Speonesydrion, Uranolophus and Chirodipterus [11].

The second specimen figured here is a small and thick pentagonal dermal element (Fig. 2). It presents five well-marked Westoll lines (Fig. 2, w.l), all more or less parallel to the unique convex margin of the bone. The distance between the Westoll lines increases with the distance from the convex margin. A similar Westoll lines pattern is present on the opercular of Speonesydrion [9 (fig. 27F)]. This new specimen shows three overlapped areas (Fig. 2, ov.a1–3). These areas have a short proximal section, but are well developed in a perpendicular direction of the bone; in fact they look more like processes than overlapped areas. One of these areas is much larger than the others (Fig. 2, ov.a1). It is developed from an angle of the bone, in a way that is very different from that of the dorsal articular process of a peg-and-socket articulation of scales. In spite of a roughly rhombic shape, this cosmine-bearing bone is not considered as a scale, but as an isolated dermal element of the head (skull, cheek or submandibular element). Unfortunately, this isolated bone cannot be more precisely determined due to the large number of small bones composing the dermal skull roof and of the important intraspecific variation in its pattern among the Dipnomorpha [20] and [21].

The external surface presents three depressions, each one of different size. These marks, due to the impression of some blunt objects, could be a result of the sedimentary compaction process. They have most probably been made by compression of detritic gravels after the death of the fish. This is probably the case since the matrix is a coarse intraformational conglomerate with mud pellets representative of a high-energy deposit.

The isolated pentagonal element and the square-shaped opercular, both cosmine-covered and bearing Westoll lines, are here considered as belonging to some indeterminate Dipnoi.

About 30 Devonian dipnoan genera are known. The majority of them are from Middle to Late Devonian localities. The oldest dipnoans are Pragian in age, disregarding Diabolepis [14] and [15], sometimes assigned to the Dipnoi, and some doubtful dipnoan remains.

Several Lower Devonian dipnoans are known (Table 1). It can be noticed that some isolated tooth plates of indeterminate Dipnoi have been described by Thanh and Janvier [45] and [46] from the Pragian of Vietnam and by Denison [22] from the Upper Pragian–Lower Emsian of Idaho, USA. A very doubtful isolated tooth plate from the Lochkovian–Pragian of Kotelny Island, Russia, has been described by Mark-Kurik [36] as a dipnoan remain [5].

The Lower and Middle Devonian sediments of Spitsbergen were deposited in a very large delta under a tropical or subtropical climate. The mouth of the palaeoriver is now situated in northern Spitsbergen (North of Andrée Land), the upstream being southward. Vertebrate fossil remains are abundant in the North of the island and scarcer in the South. Furthermore, the rare fossil material found in southern Andrée Land mainly consists of isolated scales or dermal elements, whereas the articulated or almost complete specimens found by previous field expeditions occur in the North of Andrée Land. It seems that the Lower and Middle Devonian vertebrates were essentially marine benthic animals living on the continental shelf. Some dislocated remains, or carcasses, may have been displaced somewhat upstream by strong rising tides. Thus, and although the dipnoan remains described herein were found in a relatively upstream channel palaeoenvironment, these new dipnoans from Spitsbergen could have been marine fishes, like 90% of the Lower and Middle Devonian dipnoans [11].

This ‘marine’ theory needs however to be better supported for the Spitsbergen data. The presence of dipnoan remains, in South Andrée Land (upstream) and the supposed absence of such remains in the North Andrée Land (downstream) may suggest, on the contrary, a freshwater mode of life for these fishes. In spite of Campbell and Barwick's attempt [11 (pp. 95–96)], an anadromous or catadromous model for the Devonian Dipnoi is impossible to establish. In any case, given the paucity of remains, the conclusion is that it is unfortunately not possible to define the life condition, marine or freshwater, of these new Devonian dipnoans from Spitsbergen.

This new dipnoan occurrence is nevertheless very interesting in a palaeobiogeographical point of view: it enlarges the Lower Devonian and Lower Eifelian dipnoan record of the Old Red Sandstone continent, to date only known by Tarachomylax from Severnaya Zemlya, Uranolophus and some indeterminate isolated tooth plates from USA, and maybe by Melanognathus from the Canadian Arctic.

The 2002 French CAST expedition has yielded some Dipnoi remains. The material consists of an opercular and an isolated pentagonal dermal element, both showing Westoll lines. This first occurrence of Dipnoan material in the Devonian of Spitsbergen increases the vertebrate palaeobiodiversity of this area of the Old Red Sandstone Continent. These indeterminate dipnoan remains are considered to be Late Emsian to Early Eifelian in age and are thus amongst the older dipnoan representatives. In spite of their upstream position, with regard to the palaeoenvironmental interpretation of the sediments, the question of the marine or freshwater environment of these dipnoans cannot be settled.