Abstract:  Three previously unknown octopods are described from Upper Cenomanian limestones of the Hâqel and Hâdjoula localities (Lebanon). Keuppia levante gen. nov., sp. nov., Keuppia hyperbolaris gen. nov,. sp. nov. and Styletoctopus annae gen. nov, . sp. nov. are regarded as the earliest representatives of the Octopoda (= Incirrata). This assumption is mainly based on their medially isolated bipartite gladius vestige. As can be inferred from growth increments, Keuppia gen. nov. can be distinguished from the genus Palaeoctopus by blades that grow forwards along their longitudinal axis. The gladius vestige of Keuppia hyperbolaris sp. nov. differs from that of Keuppia levante sp. nov. in having a more heterogeneous course of growth lines. Based on a pair of widely separated stylets, which closely resemble the rods of modern octopods, Styletoctopus annae gen. nov., sp. nov. is assigned to the Recent family Octopodidae. Peculiar encrustations, which are situated in close association with the gladius vestiges of Keuppia levante sp. nov., Keuppia hyperbolaris sp. nov., and Styletoctopus annae sp. nov. are interpreted as basal fin cartilages. The gladius vestige morphology of Keuppia hyperbolaris sp. nov. and Keuppia levante sp. nov. opens the possibility that both the Octopda and the Cirroctopoda originated from loligosepiid vampyropods instead of teudopseid. The surprising existence of a stylet-like gladius vestige in Styletoctopus annae sp. nov. suggests that the octopod clade branched off much earlier than previously believed. Octopod apomorphies such as the development of stylets, loss of fins and cirri must have been occurred before the Cenomanian.

For a long time, Palaeoctopus newboldi Woodward, 1896 from Santonian limestones at Sâhel Aalma (Lebanon) was the only known pre-Cenozoic coleoid cephalopod that can be seen as an unambiguous stem-lineage representative of the Octobrachia Fioroni, 1981. Thanks to the extraordinary soft-part preservation in the Lebanon limestones, we have a comparatively precise knowledge about the morphology of this ancient octopus. Palaeoctopus newboldi had a spherical mantle sac, a head-mantle fusion, eight equal arms armed with suckers, an ink sac, a medially isolated shell vestige, and a pair of (sub-) terminal fins. Particularly the bipartite shell vestige suggests that Palaeoctopus already belongs to the octopod stem-lineage, for the sister taxon of the Octopoda, the Cirroctopoda, is characterized by an unpaired clasp-like shell vestige (Engeser 1988; Haas 2002; Bizikov 2004).

However, increased research efforts recently delivered further evidence on fossil Octobrachia. From the Campanian of Canada, Fuchs et al. (2007a) presented the first record of an unpaired, saddle-shaped shell vestige that might have belonged to a cirroctopod. From the Santonian–Campanian of Canada and Japan, Tanabe et al. (2008) reported furthermore the existence of at least four different jaw morphotypes. Two of them (Paleocirroteuthis haggarti Tanabe et al., 2008 and Paleocirroteuthis pacifica Tanabe et al., 2008) have been interpreted as being of cirroctopod type, one of octopod type, and one of uncertain octobrachiate type. Finally, Fuchs et al. (in press) described the second species of Palaeoctopus, the Turonian Palaeoctopus pelagicus from limestones at Vallecillo (Mexico), on the basis of a medially isolated shell vestige.

The well known Upper Cenomanian limestones at Hâqel and Hâdjoula (Lebanon) have recently yielded five specimens that can be reliably placed within the Octopoda. It is the aim of the present work to describe these exceptionally well-preserved specimens and to discuss their morphology in the context of phylogeny and evolution.

Geological setting

The examined specimens come from the sub-lithographical limestones of Hâqel and Hâdjoula, in north-west Lebanon. These localities are about 15 km apart, 45 km away from Beirut and 15 km away from the coastal city of Jbail (Text-fig. 1).

Figure TEXT‐FIG. 1..

 Topographic map of north-western Lebanon with the outcrop area in the upper right hand corner.

The debate about a precise stratigraphical age of the exposed limestones in the Hâqel and Hâdjoula quarries is long. Geological studies on the Cretaceous outcrops in Lebanon started with Botta (1833), who investigated the lithology of the sediments in the Hâqel area. Lewis (1878) and Fraas (1878) postulated a Turonian age for both Hâqel and Hâdjoula. Patterson (1967) and Hückel (1970) later recognized a Cenomanian age of these localities. Patterson (1967) estimated a Mid-Cenomanian age for both Hâqel and Hâdjoula on the basis of fish fauna, whereas Hückel (1970) placed the beds at Hâqel as Early Cenomanian based on the record of the foraminifer Orbitolina concava concava and the index ammonite Mantelliceras mantelli already reported by Zumoffen (1926). Biostratigraphical studies of Dubertret (1959, 1966) and Saint-Marc (1974) determined the sequences at Hâqel and Hâdjoula to be Early Cenomanian. Hemleben (1977), on the other hand, again dated them as Late Cenomanian in view of the assemblage of the planktonic foraminifera Praeglobotruncana stephani, Rotalipora cushmani and Rotalipora greenhornensis. More recently, Wippich and Lehmann (2004) confirmed a Late Cenomanian age for both Hâqel and Hâdjoula owing to the presence of the ammonite Allocrioceras cf. annulatum, which is a member of the lower Upper Cenomanian Sciponoceras gracile Zone in the Western Interior of USA and the Metoicoceras geslinianum Zone of the international standard.

Lithologically, the limestones from Hâqel are described as hard, fine-grained, well-bedded and laminated, often characterized by a rich fossiliferous content and by a yellowish colour that sometimes may become greyish (Hückel 1970, 1974; Hemleben 1977; Capetta 1980). Some levels of the Hâqel outcrop are particularly rich in flint nodules. The same authors described the limestones from Hâdjoula as a more compact, soft and laminated rock, characterized by a lighter yellow or grey-yellow colour and without flint nodules.

During the Cenomanian, Lebanon and the whole Arabian Peninsula were part of the African platform in the northern part of the Gondwana super-continent (Philip et al. 1993). In Cenozoic times, the opening of the Red Sea interrupted the connection between the Arabian Peninsula and Africa. As the Late Cenomanian Al-Namoura outcrops (Dalla Vecchia et al. 2002; Dalla Vecchia and Chiappe 2002), Hâqel and Hâdjoula were probably deposited in small and shallow Tethyan basins with a reduced water circulation, in a carbonate platform setting that covered much more of the Arabian craton between Albian and Turonian. In this scenario, mild oscillations of the relative sea level produced an exceptional sandwich of shallow water carbonate facies (Ferry et al. 2007). The described palaeoenviroment produced favourable conditions of preservation of the palaeofauna comprising traces of terrestrial plants together with a rich ichtyofauna, decapod and brachyuran crustaceans, annelid polychates, nematodes, ammonoid and coleoid cephalopods, tylachocephalan arthropods and others (Bracchi and Alessandrello 2005; Fuchs 2006a).

Material and methods

Four of the studied specimens are housed in the palaeontological collection of the Museo Civico di Storia Naturale di Milano: those from Hâqel are labelled MSNM i26321 and MSNM i26323, those from Hâdjoula MSNM i26320 and MSNM i26742. The fifth specimen belongs to the Musée national d’histoire naturelle de Luxembourg and is labelled MnhnL-CRE045.

All the specimens are compressed and flattened on a two-dimensional surface of yellow-greyish limestone. Although a mineralogical analysis has not been applied, the shell vestige of MSNM i26320 and MSNM i26321 is preserved in a brownish material very similar to that of other Mesozoic gladii (Doguzhaeva and Mutvei 2003; Fuchs 2006a; Fuchs et al. 2007b, c). This observation indicates that the original composition was chitinous. Both MSNM i26323 and MSNM i26742 preserve the shell vestige too, but as an imprint only.

Soft parts of all the specimens (mantle, head, arms, ink sac, and muscle fibers) are preserved as imprints or they were mineralized post-mortem in white-yellowish apatite (calciumphosphate). Fuchs (2006a) described an analogous preservation for vampyropod coleoids from Hâqel and Hâdjoula. For more information about the mineralization of soft-tissues see Allison and Briggs (1991), Wilby (1993), and Kear et al. (1995).

The studied material was investigated by means of a binocular microscope and was compared with both fossil and living specimens. Ultraviolet light has been used to enhance phosphatized tissues.

Systematic Palaeontology

The higher-level taxonomy follows Boletzky (1992, 1999), in which Vampyropoda Boletzky, 1992 include the Octopoda Leach, 1818, the Cirroctopoda Young, 1989 and the Vampyromorpha Robson, 1929, with their only living representative Vampyroteuthis infernalis Chun, 1903.

As will be elaborated in the discussion, the term ‘gladius vestige’ (synonym: shell vestige, fin support, muscular support) is used here to emphasize, its morphogenetic origin, the gladius.

Institutional abbreviations.  MnhnL, Musée national d’histoire naturelle, Luxembourg; MNHN, Musée National d’Histoire Naturelle, Paris; MSNM, Museo Civico di Storia Naturale, Milano; UANL-FCT, Universidad Autonoma de Nuevo Leon, Faculdad de Ciencias de la Tierra, Linares, Mexico.

Subclass COLEOIDEA Bather, 1888
Superorder VAMPYROPODA Boletzky, 1992
Order OCTOBRACHIA Fioroni, 1981
Suborder OCTOPODA Leach, 1818

Type genus. Palaeoctopus Woodward, 1896a, from the Santonian of Sâhel Aalma, Lebanon.

Included genera. Palaeoctopus Woodward, 1896a and Keuppia gen. nov.

Genus KEUPPIA gen. nov.

Derivation of name.  The name is in honour of Prof. Dr. Helmut Keupp (Berlin) and his extraordinary contributions to palaeontology.

Type species. Keuppia levante sp. nov.

Diagnosis.  Gladius vestige with a primordial shell situated at the posterior end.

Remarks.  In Palaeoctopus newboldi (Woodward, 1896b) and Palaeoctopus pelagicus Fuchs et al. (2008) the primordial shell is situated in the centre of the gladius vestige, i.e. growth occurred concentrically around a nucleus. In Keuppia gen. nov., in contrast, the vestige grew along its longitudinal axis.

Occurrence.  Known only from the Upper Cenomanian of Hâdjoula (Lebanon).

Keuppia levante sp. nov.
Text-figures 2, 3, 4D

Figure TEXT‐FIG. 2..

Keuppia levante sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon). A, holotype, MSNM i26320a, ×0.5. B, sketch of the holotype.

Figure TEXT‐FIG. 3..

Keuppia levante sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon). A, holotype (MSNM i26320a+b), photographs taken with UV light of part and counterpart have been put on top of each other to demonstrate the extraordinary preservation of musculature, ×0.5. B, close-up of A to show the right ventrolateral (right hand) and dorsolateral (left hand) arm. The inner surface of the ventrolateral arm bears two longitudinal rows of circular suckers, ×1.5. C, close-up of A to show the anterior part of the mantle sac. The series of lappet-like staining in the lateral mantle are interpreted as outer gill lamellae, ×1.5. D, close-up of A to show the posterior part of the mantle sac and the bipartite gladius vestige.

Figure TEXT‐FIG. 4..

 Comparative morphology of palaeoctopodid gladius vestiges (right column models of the vestiges). A, Palaeoctopus newboldi (Woodward 1896b) from the Santonian of Sâhel Aalma (Lebanon), MNHN B18834a, ×1.5. B, Palaeoctopus pelagicus Fuchs et al., 2008 from the Turonian of Vallecillo (Mexico), UANL-FCT-VCI/150A, the associated blade is missing why the orientation is questionable, ×1.0. C, Keuppia hyperbolaris sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon), holotype (MnhnL-CRE045); two asymptotes delimiting the blade into three fields are indicated by dashed lines, ×4. D, Keuppia levante sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon), holotype (MSNM i26320a), ×1.5. E, Styletoctopus alessandrellianus sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâqel (Lebanon), holotype (MSNM i26323), ×2.

Derivation of name.  From the latin levante (= sunrise, orient).

Holotype.  MSNM i26320a+b.

Type locality.  Hâdjoula (Lebanon).

Type horizon. Metoicoceras geslinianum Zone (international standard; corresponds to Sciponoceras gracile Zone in Western Interior); Lower Upper Cenomanian.


Gladius vestige.  In the posterior extremity of the mantle, the holotype includes a completely preserved gladius vestige (Text-fig. 2). Its morphology shows that the animal was embedded ventrally. The gladius vestige consists of a pair of subtriangular blades (Text-fig. 4D). The apexes of both halves are close to each other. A median connection can be excluded. Together, the two blades form a U-shaped complex with the vertex turned towards the posterior end of the mantle. The outer margins of the two blades precisely follow the rearmost mantle outline. Each blade has a maximum length of 18 mm and a maximum width of 10 mm.

A dense ornamentation of transverse ridges that start from the posterior apex is evident on the dorsal surface of each blade (Text-fig. 4D). The ridges are oriented perpendicular to the longitudinal axis and most probably represent growth increments. Growth lines are pronounced at the inner margins and become indistinct towards the outer margins. A distinct longitudinal rib is present along the inner margins. Furthermore, two more indistinct longitudinal ribs are intersecting the transversal growth lines in the median part. Posterior apexes show a slight inward curvature. Anterior margins of the blades are almost straight, corresponding to the homogenous growth lines.

However, anterior limits are difficult to determine as a peculiar structure covers the anterior outer portion of both blades (Text-fig. 4D). The more or less circular encrustation of 7 mm diameter exhibits a regular radial ornamentation. The centre of the rays is located in its lower outer parts and elevated capulus-like above the surface of the gladius vestige. In the centre, the encrustation is clearly thickened. In the periphery, it appears very thin, as growth lines of the gladius vestige are perceivable.

Mantle outline.  The mantle outline is easily visible thanks to the preservation of mantle musculature, which largely appears as whitish francolite (Text-fig. 3A). The dorsoventrally compressed mantle is 112 mm in maximum length and 63 mm in maximum width (ratio width/length = 0.56). It clearly gives evidence of a sac-like body outline. It is partially possible to distinguish between dorsal and ventral parts of the mantle musculature (Text-figs 2B, 3C).

Head/nuchal region.  Although the preserved musculature shows an obviously narrow nuchal region, imprints indicate that the head/nuchal region was much wider (Text-figs 2, 3A). The head appears therefore not clearly demarcated and assumption of a head-mantle fusion is probable. The muscles preserved in this region might probably represent the nuchal muscles (anterior mantle adductors). On both sides outside these muscles, the position of the eye capsules are conceivable (Text-fig. 2). The buccal mass, which is located more anteriorly close to the arm bases, is visible as a distinct depression (elevation in the counterslab). Beaks, radula teeth or statoliths are not preserved.

Arm crown.  Eight exceptionally well-recorded arms are visible (Text-figs 2, 3A). They are obviously long (approximately 202 mm). The ratio arm length/total body length is 0.64. Arms seem to be of equal length and show no evidence of modification. At their bases, arms are 10 mm in thickness. As can be seen in figure 3B, suckers are evident under UV light. They are arranged in two parallel rows (i.e. they do not seem to alternate). Their circular diameter is gradually decreasing from the arm base (5 mm) towards the tip. The position of each arm can even be inferred from the suckers and the dorsal aspect of the specimen (Text-figs 2B, 3A). Evidence of an interbrachial web is not discernible. In the light of this extraordinary preservation, we can ascertain that cirri seem to be absent.

Ink sac.  An ink sac is present in the anterior third of the body (Text-fig. 2). A short S-shaped furrow, which is anteriorly connected to the ink sac, is interpreted as the ink duct. More anteriorly from the ink sac, ink has leaked.

Gills.  Two series of comparatively short, lappet-like yellowish staining are barely discernible on both sides of the lateral mantle (Text-figs 2, 3C). These structures might be relicts of gills (since they did not phosphatize when held under UV light, they are unlikely to be remnants of the mantle musculature). The 6–7 lappets can be interpreted as the outer branches of the primary efferent vessel.

Fins.  Remnants of the fins are not recognizable.

Differential diagnosis. Keuppia levante sp. nov. can be clearly distinguished from Palaeoctous newboldi (Woodward, 1896b) and Palaeoctopus pelagicus Fuchs et al. (2008) by the gladius vestige (Text-fig. 4). In general, the shape is boomerang-like in P. newboldi (Text-fig. 4A), patella-like in P. pelagicus (Text-fig. 4B), and subtriangular in K. levante sp. nov. (Text-fig. 4D). Both P. newboldi and P. pelagicus possess a reinforcement along the longitudinal axis, which is absent in K. levante sp. nov. Furthermore, K. levante sp. nov. has a more obtuse posterior end compared to the nipple-like extensions in P. newboldi and P. pelagicus. Finally, the most important difference concerns growth patterns. The gladius vestige of K. levante sp. nov. grows linear in a longitudinal direction, whereas P. newboldi and P. pelagicus show concentric growth increments around a nucleus in the centre of the gladius vestiges.

Keuppia hyperbolaris sp. nov.
Text-figures 4C, 5

Figure TEXT‐FIG. 5..

Keuppia hyperbolaris sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon). A, holotype (CRE045), ×0.5. B, sketch of the holotype. C, close-up of A showing the head-nuchal region. D, close-up of A showing imprints left half of the gladius vestige and the fin cartilage.

Derivation of name.  From the hyperbolaric course of growth increments on the gladius vestige.

Holotype.  MnhnL-CRE045.

Type locality.  Hâdjoula (Lebanon).

Type horizon. Metoicoceras geslinianum Zone (international standard; corresponds to Sciponoceras gracile Zone in Western Interior); Lower Upper Cenomanian.


Gladius vestige.  In the posterior extremity of the mantle, the holotype offers well preserved dorsal imprints of a bipartite gladius vestige (Text-figs 4C, 5D). It consists of two semicircular blades that together form a V-shaped complex. Each blade measures 18 mm in length. The two posterior apexes are very close to each other. We cannot exclude that both blades have been weakly connected and/or articulated (Text-fig. 5D). The rounded outer margins of each blade precisely follow the rearmost mantle outline. The inner margins are more or less straight. Dense transverse growth increments are oriented perpendicular to the longitudinal axis and incipient from the posterior apex. The growth lines become more indistinct from inside to outside. According to their heterogeneous course, three different fields are notable: an inner field with weakly concave (parabolar) growth lines, a narrow central field typified by a convex (hyperbolar) striation, and an outer field that is characterized by oblique growth lines. Hence, the inner field slightly protrudes from the central and the outer field. Longitudinal reinforcements intersecting the growth lines are absent.

Similar to Keuppia levante sp. nov., an oval structure with a radial striation covers the broad anterior part of each blade (Text-figs 4C, 5D). The maximum radius of the encrustation is arranged in antero-posterior direction and measures 9 mm. The centre of the ray-like striation of this possibly capulus-shaped encrustation is situated in its outer posterior part. In parts, the periphery appears very thin, as growth lines of the gladius vestige are conceivable.

Mantle outline.  The dorsoventrally embedded mantle is 97 mm in maximum length and 47 mm in maximum width (width/length ratio = 0.48). Mantle musculature is only partially preserved.

Head/nuchal region.  The head/nuchal region appears wide compared to the mantle width (Text-fig. 5A–C). Therefore, the head seems to be fused with the dorsal mantle. The buccal mass is located more anteriorly close to the arm bases (Text-fig. 5C). Between the presumed anterior mantle edge and the buccal mass, a pair of longitudinal muscles (nuchal muscles?) is visible. Beaks, radula teeth or statoliths are not visible.

Arm crown.  Seven arms have a preserved length of 202 mm, i.e. they take 63 per cent of the total body length. At their bases, arms are 10 mm in thickness. Clear evidence of arm modification is missing. Suckers are not discernible. An interbrachial web is not visible, but was possibly present, as can be judged from the conjoined arms.

Ink sac.  An ink sac is present in the centre of the mantle (Text-fig. 5A).

Gills.  Lappet-like red staining is weakly visible on the left lateral mantle (Text-fig. 5A). The 6–7 lappets are interpreted as remnants of the outer gill blades.

Fins.  Evidence of fins is not recognizable.

Differential diagnosis.  The gladius vestiges of K. hyperbolaris sp. nov. and K. levante sp. nov. show some faint differences (Fig. C–D). In K. hyperbolaris sp. nov., each blade appears semicircular, whereas in K. levante sp. nov., the latter is more subtriangular. Furthermore, K. hyperbolaris sp. nov. exhibits, in contrast to K. levante sp. nov., a clear ‘hyperbolar zone’ that subdivides each blade of the gladius vestige in three different fields.

Keuppia sp.
Text-figure 6

Figure TEXT‐FIG. 6..

Keuppia sp. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon). A, specimen MSNM i26742, ×0.5. B, close-up of A showing the imprints of the gladius vestige and the presumed fin cartilage.

Type locality.  Hâdjoula (Lebanon).

Type horizon. Metoicoceras geslinianum Zone (international standard; corresponds to Sciponoceras gracile Zone in Western Interior); Lower Upper Cenomanian.

Material.  One specimen (MSNM i26742).

Description.  The preservation of the dorsoventrally flattened specimen is unlike the K. levante sp. nov. and K. hyperbolaris sp. nov. (Text-fig. 6). Nevertheless, the specimen preserves the complete mantle (100 mm) and an arm crown (210 mm). In the most posterior mantle, imprints of at least one half of the originally bipartite gladius vestige as well as the associated encrustation are weakly visible (Text-fig. 6B). Except the possession of an ink sac, no additional details (fins, number of arms suckers, gills, etc.) are recognizable. Even observations under UV light have been without success.

Remarks.  Owing to an imperfect preservation (particularly of the gladius vestige), a differential diagnosis whether the specimen belongs to Keuppia levante sp. nov., Keuppia hyperbolaris or a third species is impossible.

Family OCTOPODIDAE d’Orbigny, 1840
Genus STYLETOCTOPUS gen. nov.

Derivation of name.  From the stylet-like gladius vestige.

Type species. Styletoctopus annae sp. nov., from the Upper Cenomanian of Hâqel (Lebanon).

Diagnosis.  Gladius vestige developed as a pair of widely separated and noticeably narrow stylets.

Styletoctopus annae sp. nov.
Text-figure 7

Figure TEXT‐FIG. 7..

Styletoctopus annae sp. nov. from the Upper Cenomanian (Metoicoceras geslinianum Zone) of Hâqel (Lebanon). A, specimen MSNM i26323, ×1.5. B, close-up of A showing the imprints of the stylets situated in the lateral mantle sac. C, close-up of A showing the star-like remnants of the arms. D, close-up of A photographed with UV light exemplifying the arm musculature.

Derivation of name.  The name is dedicated to Anna Alessandrello, for her guidance to one of the authors (G. B.) during the years of scientific studies.

Holotype.  MSNM i26323.

Type locality.  Hâqel (Lebanon).

Type horizon. Metoicoceras geslinianum Zone (international standard; corresponds to Sciponoceras gracile Zone in Western Interior); Lower Upper Cenomanian.


The holotype comprises a complete body including mantle and arms (Text-fig. 7A).

Gladius vestige.  It exhibits a pair of stylet-like imprints in the lateral mantle (Text-fig. 7B). The stylets (rods) have a maximum length of 6 mm and curve towards the inside. Similar to Recent Enteroctopus dofleini, an anterior and a posterior shoulder can be distinguished. The anterior shoulder left distinct pointed imprints, whereas the posterior shoulder is hardly perceptible. Both are delimited by a swelling (bend) which is situated in the posterior third of the total stylet length. It seems that the swelling is oriented towards the inside.

Mantle outline.  Imprints show that the mantle is dorsoventrally embedded. It measures approximately 20 mm in length and 9 mm in width. Mantle musculature is hardly preserved. The ink sac is preserved anterior to the stylets.

Head/nuchal region.  The area between the arms and the mantle is slightly constricted, but this is possibly due to incomplete preparation.

Arm crown.  According to the well-preserved arm crown, the specimen is seen in ventral aspect (Text-fig. 7C–D). Eight comparatively short arms are arranged in a circle, indicating that the dead body landed in an oral-down position. The ventral arm pair is the shortest (5 mm); the dorsal pair the longest (8 mm). An interbrachial web is not evident. Suckers are not observable.

Fins.  Fins are not visible.

Differential diagnosis. Styletoctopus annae sp. nov. considerably differs from other palaeoctopodids by its strongly reduced gladius vestige. The gladius vestige of Styletoctopus annae sp. nov. is situated in the lateral mantle and stylet-like (Text-figs 4E, 7B).

Styletoctopus aff. annae
Text-figure 8

Figure TEXT‐FIG. 8..

Styletoctopus aff. annae from the Upper Cenomanian. from the Cenomanian (Metoicoceras geslinianum Zone) of Hâdjoula (Lebanon). A–B, holotype (MSNM i26321a+b), ×4.0. C–D, close-up of A–B showing details of the style-like gladius vestige and the associated fin cartilage.

Material.  One specimen (MSNM i26321a+b, positive and negative).

Type locality.  Hâdjoula (Lebanon).

Type horizon. Metoicoceras geslinianum Zone (international standard; corresponds to Sciponoceras gracile Zone in Western Interior); Lower Upper Cenomanian.


Specimen MSNM i26321 is small and poorly preserved (Text-fig. 8A–B). It includes dorsoventral imprints of a mantle as well as an incomplete arm crown.

Gladius vestige.  In the posterior lateral mantle, the specimen preserves a pair of stylets, whose posterior end appears thickened (Text-fig. 7C–D). They have a length of approximately 2 mm. Close to their anterior end, two circles of 2 mm diameter each are recognizable. In the positive, the circles appear as white spots, usually indicating phosphatized soft tissues. It is therefore possible to interpret them as fins. However, on the other hand, it is also likely that the circles correspond to the oval to circular encrustations found in association with the gladius vestiges of Keuppia levante sp. nov. and K. hyperbolaris sp. nov.

Mantle outline.  The sac-like mantle has a length of 15 mm and a width of 12 mm. The ink sac is located in the centre of the mantle.

Head/nuchal region.  Evidence about the head/nuchal region is scarce, but the head seems not to be demarcated.

Arm crown.  The 4–5 arms have a preserved length of maximum 9 mm. Suckers are not visible.

Fins.  Evidence of fins is absent.

Remarks.  Owing to its small size and inadequate preservation, specimen MSNM i26321 is tentatively identified as Styletoctopus aff. annae. In terms of the general habitus, specimen MSNM i26321 might be conspecific with Styletoctopus annae sp. nov. The main difference between specimen MSNM i26321 and the holotype of Styletoctopus annae sp. nov., specimen MSNM i26323, concerns the morphology of the stylets. In specimen MSNM i26323, the stylets are subdivided into an anterior and a posterior shoulder. Specimen MSNM i26321, in contrast, seems to lack the posterior shoulder. However, in the light of the poor preservation of the posterior shoulder in specimen MSNM i26323, the posterior shoulder may not recorded in specimen MSNM i26321. Vice versa, the same can be assumed for the absence of the associated circles in Styletoctopus annae sp. nov.


About the origin of Cirroctopoda and Octopoda

So far, only two pre-Cretaceous specimens with a preserved body outline have been attributed to the Octobrachia. Pohlsepia mazonensis Kluessendorf and Doyle, 2000 from the Late Carboniferous of Mazon Creek (USA) was classified as a cirroctopod. Proteroctopus ribeti Fischer and Riou, 1982 from the Middle Jurassic (Callovian) of La Voulte-sur-Rhône (France) was assigned to the Octopoda. Classification of Pohlsepia mazonensis is solely based on a single, poorly preserved specimen with a sac-like body outline without evidence of hard parts. Proteroctopus ribeti is occasionally regarded as an octopod, because it seems to lack a gladius vestige as well as cirri (Fischer and Riou 1982; Kluessendorf and Doyle 2000). However, the absence of a gladius vestige is most probably a diagenetic artefact, since gladii are never preserved in La Voulte.

Preservation of cirri is furthermore an extremely rare event. For these reasons, Fuchs et al. (2008) followed Engeser (1988), Haas (2002), and Bizikov (2004) and doubted the systematic position of Pohlsepia mazonensis and Proteroctopus ribeti. Fuchs et al. (2008) therefore considered Palaeoctopus pelagicus from the Turonian of Vallecillo (Mexico) to be the oldest record of an octopod. Consequently, Upper Cenomanian Keuppia levante sp. nov., Keuppia hyperbolaris sp. nov. and Styletoctopus annae sp. nov. represent now the oldest known records of octopods (Text-fig. 9).

Figure TEXT‐FIG. 9..

 Phylogeny of the Octobrachia based on the shell complex. As can be inferred from the gladius growth increments, the shell sac separates and emigrates forwards. Both the time of divergence of Octopoda and Cirroctopoda and the stem group of the Octobrachia is still uncertain.

Fuchs et al. (2008) assumed the rise of Octopoda and Cirroctopoda to have occurred between the Toarcian (180 Ma) and the early Turonian (93 Ma). The present Cenomanian record slightly reduces this estimated time span to 180–95 Ma.

However, the advanced gladius reduction found in Styletoctopus is surprising. Whereas the blade-like gladius vestiges typified by Keuppia and Palaeoctopus permit a relatively late separation of the Octopoda and Cirroctopoda to be assumed in mid-Cretaceous times, the widely isolated and strongly reduced lateral stylets of Styletoctopus rather suggest an earlier separation in Early Cretaceous or even Jurassic times.

About the morphogenetic origin of the octobrachiate gladius vestige

With respect to the obscure morphogenetic origin of the octopod shell vestiges, Naef (1921, p. 695) suggested that it represents a remnant of a gladius, the stiff, but flexible chitinous pen within the dorsal mantle of many coleoid groups. Jeletzky (1966, p. 88), too, assumed that the bipartite shell rudiment of Palaeoctopus newboldi derived from a gladius. He considered it reminiscent of the posterior conus-like median field. Voight (1997, p. 320), in contrast, stated that the octopod stylets embedded in the lateral mantle musculature ‘…evolved de novo, independent of the plesiomorphic gladius’. Haas (2002) and supported later by Bizikov (2004) presented the idea that the clasp- and stylet-like shell vestiges of Recent Cirroctopoda and Octopoda evolved through gradual reduction from a teudopseid gladius, a gladius with a posteriorly wide (large apical angle) and an anteriorly narrow median field (Text-fig. 9). In the light of their recently described Palaeoctopus pelagicus, Fuchs et al. (2008) followed Haas (2002) and Bizikov (2004) in considering a wide gladius similar to Trachyteuthis or Teudopsis to be an ideal prerequisite to open and reduce the median field in a longitudinal direction to obtain the body plasticity typical of octobrachians. This assumption would imply that the shell rudiments represent mainly the gladius’ lateral fields (wings). Indeed, e.g. the genus Trachyteuthis shows some characters typical for vampyropods: 1) eight arms with an elongated dorsal arm pair (Fuchs 2006b; Bandel and Leich 1986), 2) interbrachial web (Bandel and Leich 1986), 3) two pairs of fins (Donovan 1995; Donovan et al. 2003; Fuchs et al. 2007b), and 4) beaks of octopod type (Klug et al. 2005). Because of this character complex, Fuchs (2006a, b) and Fuchs et al. (2007b, c, 2008) regarded the Teudopseina as a stem-group of the Octobrachia.

However, with respect to the shell morphology found in Keuppia levante sp. nov. and particularly Keuppia hyperbolaris sp. nov., an alternative origin is conceivable (Text-fig. 9). If one imagines the paired blades of Keuppia levante sp. nov. and Keuppia hyperbolaris sp. nov. (and Palaeoctopus pelagicus, too) as a united structure, the gladius vestiges bear resemblance to a loligosepiid gladius (Text-fig. 9; more about loligosepiid gladii see Fuchs and Weis (2008)). The anterior gladius rim is straight, the posterior gently curved, and the lateral rims evenly rounded. Furthermore, Keuppia hyperbolaris sp. nov. exhibits a hyperbolar zone that subdivides each blade into an inner (median) and an outer (lateral) field. Especially the slightly protruding inner field is typical for loligosepiid gladii. On this account, the octobrachian gladius vestiges might have also evolved from the Loligosepiina, a Jurassic group of vampyropods. Geopeltis simplex, for instance, is characterized by a gladius that is considerably wider than in other loligosepiid genera (Loligosepia, Parabelopeltis), i.e. the median field opens at a very wide angle (approximately 45 degrees). In terms of this, the gladius of Geopeltis bears characteristics comparable to Teudopsis and Trachyteuthis.

In all Recent Octopoda, stellate ganglia are widely separated. Bizikov (2004, p. 79) therefore assumed ancestors with a wide anterior median field. In this respect, the ‘loligosepiid pathway’ appears more plausible than the ‘teudopseid pathway’, because the anterior gladius margin is pointed in teudopseids (Text-fig. 9).

The ‘loligosepiid pathway’ would greatly affect the phylogeny recently proposed by Fuchs (2006b) and Fuchs and Weis (2008). According to their phylogeny, loligosepiids belong to the vampyromorph lineage. This proposal must be erroneous taking into account that the octobrachian blades originated from a gladius of loligosepiid type. In this case, loligosepiids would rather represent the root-stock (stem group) of both Octobrachia and Vampyromorpha (Doyle et al. 1994, fig. 1).

About the occurrence of fins and fin cartilage in early octopods

Octopod embryos show Anlagen of fins but they disappear during the development. The loss of fins is therefore clearly secondary and widely accepted as an apomorphy of the Octopoda (Young and Vecchione 1996; Voight 1997; Haas 2002). In Santonian Palaeoctopus newboldi, soft-part preservation shows clear evidence of fins. Although Keuppia levante sp. nov. and Keuppia hyperbolaris sp. nov. do not preserve fins, the well-developed gladius vestiges support the assumption that early octopods still possessed fins.

Modern cirroctopods retained a pair of strong fins. The fin musculature is supported by axial cartilage, the fin cartilage (Bizikov 2004; Collins and Villanueva 2006). It tightly attaches the gladius vestige through the shell sac. According to Bizikov (2004, p. 19), the basal cartilage is more dense and stiff than the rest of the cartilage. The paired circular encrustations found respectively in close association with the gladius vestiges of Keuppia levante sp. nov. and Keuppia hyperbolaris sp. nov. possibly represent remnants of the basal fin cartilage. If so, the fins might have been able to execute powerful upward and downward swimming strokes. This, in turn, clearly induces a pelagic or bentho-pelagic life style, but certainly not a strictly benthic. A pelagic or bentho-pelagic mode of life was also suggested for Palaeoctopus pelagicus (Fuchs et al. 2008).

Is the absence of basal cartilage in Palaeoctopus pelagicus and P. newboldi diagenetic or real? If it is real, one might argue that members of the genus Keuppia have been better swimmers and that the genus Palaeoctopus represents an intermediate stage on the shift from a pelagic or bentho-pelagic to a strictly benthic life.

The presence of fins in Palaeoctopus newboldi is well supported. In Palaeoctopus pelagicus they are still unknown. Independent from the question, whether the genus Palaeoctopus had well-developed or reduced fins, the interpretation of distinct circular structures described in Styletoctopus aff. annae is highly problematic (Text-fig. 8). Concerning their phosphatised composition, one tends to assume the existence of a pair of small globular fins. On the other hand, one would clearly expect no fins with respect to their strongly reduced gladius vestige. Unfortunately, the preservation of both specimens of Styletoctopus annae sp. nov. is not good enough to make further conclusions about the presence of fins and thus on the mode of life. The point of time when the Octopoda have lost their fins must therefore remain uncertain.

About the evolution of the shell sac

A separated shell sac, the gladius secreting epithelium, characterizes modern octopods. As Haas (2002, p. 346) mentioned, the medially isolated gladius vestiges demonstrate that the separation of the shell sac has already been realized in palaeoctopodids. Growth increments on the gladius vestiges of Keuppia gen. nov. and Palaeoctopus provide further evidence on the evolutionary development of the octopod shell sac (Text-fig. 9).

In the Turonian–Santonian genus Palaeoctopus, the blades grow concentrically around nuclei. This kind of growth pattern indicates that the primordial shells and thus the early shell sacs are clearly distant and capable of growing posteriorly.

In the slightly older genus Keuppia gen. nov., on the other hand, initial growth increments indicate shell sacs that are very close to each other and able to grow only anteriorly. Compared to Palaeoctopus, the primordial shells (and thus the early shell sacs) of Keuppia levante sp. nov. and Keuppia hyperbolaris sp. nov. were situated very close in the most posterior part of the mantle.

The lateral position of the stylets in Styletoctopus gen. nov. already demonstrates an evolutionary state of the shell sac that is comparable with Recent octopods. Hence, the position of the shell sac as typified by Palaeoctopus represents a transitional state between Keuppia gen. nov. and Styletoctopus gen. nov.

About the loss of cirri in the Octopoda

The arms of both Recent cirroctopods and Vampyroteuthis are equipped with cirri. The absence of cirri is therefore widely regarded as an apomorphy of the Octopoda (Voight 1997, p. 319; Haas 2002, p. 345). Haas (2002) believed that the holotype of Palaeoctopus newboldi exhibits cirri, but evidence of cirri is very sparse in the holotype, as Engeser (1988, p. 84) correctly stated. Furthermore, in additional specimens of Palaeoctopus newboldi from the Musée d’histoire naturelle in Paris, we could not find them. Also in the present older material, there are definitely no signs of cirri. We are aware of the poor preservational potential of cirri; nevertheless, the fossil record clearly suggests that the loss of cirri in the octopod lineage most probably occurred prior to the Cenomanian, just like the separation of the shell sac, the development of stylets (and possibly the reduction of the fins).


Octopods and cirroctopods have been divers and widely distributed groups during the Late Cretaceous (Tethys: Palaeoctopus newboldi, Keuppia hyperbolaris sp. nov., Keuppia levante sp. nov., Styletoctopus annae sp. nov.; Atlantic: Palaeoctopus pelagicus; North Pacific: Paleocirroteuthis haggarti, Paleocirroteuthis pacifica). Keuppia hyperbolaris sp. nov., Keuppia levante sp. nov. and Styletoctopus annae sp. nov. from the Cenomanian are considered the oldest unambiguous members of the Octopoda.

The gladius vestige morphology of Keuppia hyperbolaris sp. nov. and Keuppia levante sp. nov. establishes the possibility that the Octobrachia originated from loligosepiid vampyropods; instead from teudopseids. The existence of basal fin cartilage in these taxa let assume a pelagic or bentho-pelagic life.

Styletoctopus annae sp. nov. shows that the Recent family Octopodidae already existed in Mesozoic times. Particularly, the strongly reduced gladius vestige morphology is very similar to modern benthic octopod genera Enteroctopus, Bentoctopus and Eledone, indicating an Early Cretaceous or even Jurassic origin of the Octopoda.

Terminal (or nearly terminal) position of the gladius in the Palaeoctopodidae differ significantly from clearly subterminal position of stylets in the Octopodidae, indicating that the gladius of the former functioned in a different way.


Acknowledgements.  Thanks to the alertness of Giorgio Terruzzi (Milano) and Albert Abi-saad (Versailles), we came across this beautiful material. We are also grateful to Alain Faber and Georges Bechet (both Luxembourg), who enable to purchase the Luxembourg specimen. We thank Anne Beck (Berlin) for her careful proofreading. Additionally, we thank Larisa Doguzhaeva (Moscow), Helmut Keupp (Berlin) and Vyacheslav Bizikov (Moscow) for reviewing the manuscript, for helpful discussions and valuable ideas.