Introduction

Cephalopod molluscs have been one of the dominant and highly diversified groups of marine invertebrates since the late Cambrian to the present (House 1988). This is confirmed by the abundant fossil record of chambered shells in externally shelled (ectocochleate) taxa such as Nautiloidea and Ammonoidea. In contrast, the fossil record of Coleoidea, a group that is largely characterised by an internally shelled (endocochleate) body plan, is much poorer than that of the ectocochleate cephalopods, other than the extinct taxa with an aragonitic phragmocone and calcitic rostrum (Belemnitida, Aulacocerida, and Phragmoteuthidida). This is because a calcified chambered shell has been much reduced to a chitinous gladius or completely lost in most of the remaining group during their evolutionary history (Donovan 1977; Doyle et al. 1994; Fuchs 2006a; Kröger et al. 2011). For this reason, most previous studies on extinct coleoids lacking a chambered shell relied on extremely well-preserved fossils with soft tissue remains that were recovered from the Konservat- Lagerstätten of the late Palaeozoic to Cretaceous ages (e.g., Naef 1922; Donovan 1977; Donovan and Toll 1988; Fischer and Riou 1982a, b, 2002; Engeser and Reitner 1983, 1985, 1986; Engeser 1990a; Fuchs 2006a, b; Fuchs et al. 2009, 2013b; Fuchs and Larson 2011a, b).

The jaw apparatus, which is the primary feeding organ of cephalopods, appears to compensate for the relatively poor fossil record of Coleoidea because it has a preservation potential as a fossil even in taxa without a calcified chambered shell (Harzhauser 1999; Tanabe et al. 2006, 2008; Tanabe and Hikida 2010) and has been extensively used in taxonomical studies of modern fauna (Naef 1923; Clarke 1962, 1986; Clarke and Maddock 1988; Kubodera 2005). In modern cephalopods, the jaw apparatus consists of upper and lower elements (synonymous with beaks and mandibles) that are connected to each other by surrounding jaw muscles and housed within a globular-shaped muscular organ called the buccal mass in the proximal portion of the digestive tract (Nixon 1988; Tanabe and Fukuda 1999). The anterior portion of the upper and lower jaws is projected anteriorly to bite and cut prey, and chewed pieces of prey are conveyed to the oesophagus with the aid of the radula, which is accommodated in the space surrounded by the jaws (see Tanabe and Fukuda 1999: fig. 19.1). The jaw apparatus of modern coleoid cephalopods is made of a darkly tinted, organic hard tissue composed of a chitin-protein complex (Saunders et al. 1978; Hunt and Nixon 1981; Gupta et al. 2008), whereas in nautilids (Nautilus and Allonautilus), calcareous tips developed on the chitinous lamellae of upper and lower jaws (Saunders et al. 1978; Lowenstam et al. 1984). Fossilized chitinous and calcareous remains of cephalopod jaws have been discovered from the Devonian and younger marine deposits (Tanabe and Fukuda 1999; Tanabe et al. in press). They usually occur solitarily but are occasionally found in situ within the body chambers of ammonoids (e.g., Lehmann 1976, 1980, 1990; Tanabe and Landman 2002; Kruta et al. 2011; Tanabe et al. 2013) and nautilids (e.g., Dietl and Schweigert 1999; Klug 2001). Flattened jaws have also been preserved in situ in the buccal portion of exceptionally well-preserved late Palaeozoic and Mesozoic coleoid fossils retaining soft tissue remains (Saunders and Richardson 1979; Klug et al. 2005; Fuchs 2006a, b; Mapes et al. 2010; Fuchs and Larson 2011a, b). In addition to these flattened in situ jaws, unflattened isolated jaws referable to Coleoidea have recently been found from the Upper Cretaceous forearc basin deposits of Hokkaido, Japan (Yezo Group), and Vancouver Island, Canada (Nanaimo Group). Their comparative morphological examination with the jaws of modern coleoids allowed us to distinguish the following new genera and species (Tanabe et al. 2006, 2008; Tanabe and Hikida 2010; Tanabe 2012): Yezoteuthis giganteus of the order Teuthida, Nanaimoteuthis jeletzkyi and N. yokotai of the order Vampyromorpha, and Paleocirroteuthis haggarti and P. pacifica of the order Cirroctopoda.

Fig. 1.

Map of Haboro area, northwest Hokkaido, Japan, showing the locations of the coleoid lower jaws examined.

In the search for additional fossil material, we discovered two well-preserved coleoid lower jaws that we refer to here as another new species of Nanaimoteuthis and a new genus and new species of Teuthida from the Upper Cretaceous strata of Hokkaido. In this article, we describe the gross morphological and morphometric features of the jaws of these new taxa by comparing with those of modern coleoids and discuss the phylogenetic implications of the coleoid fauna from the Late Cretaceous of North Pacific.

Institutional abbreviations. —AMNH, American Museum of Natural History, New York, USA; CDM, Courtenary and District Museum and Palaeontology Centre, Courtenary, Canada; KMNH, Kitakyushu Museum of Natural History and Human History, Kitakyushu, Japan; NSM, National Museum of Nature and Science, Tsukuba, Japan; UMUT, University Museum of the University of Tokyo, Tokyo, Japan.

Other abbreviations. —HL, hood length of the lower jaw; ML, mantle length; RW, relative warp.

Material and recognition of jaw type

Two large cephalopod jaws that were recovered from Upper Cretaceous strata in the Haboro area of northwestern Hokkaido were used in this study. Both jaws were found individually in calcareous nodules; one (KMNH IvP 902,001) was collected from an outcrop exposed in the Detofutamata River before construction of the Haboro-futamata Dam (200 m north-northeast of the dam site; 44°18′19″N, 141°56′40″; Fig. 1: Locality 1), and the other (KMNH IvP 902,002) was recovered as float in the Sankebetsu River, about 1.2 km east-northeast of the Haboro Dam site (44°20′56″N, 141°56′35″; Fig. 1: Locality 2). KMNH IvP 902,001 came from the uppermost part of the Haborogawa Formation (Takashima et al. 2004), which was correlated to the Member Ui-j of the Nagareya Formation and dated the lower Campanian by Okamoto et al. (2003). The nodule that yielded KMNH IvP 902,002 was presumably derived from the upper part of the Haborogawa Formation (Takashima et al. 2004), which was correlated to the Member Uf-g of the Upper Haborogawa Formation and dated the upper Santonian by Moriya and Hirano (2001).

Since the two cephalopod jaws were preserved in calcareous nodules, their three-dimensional architecture is fa- vourably preserved. The two jaws consist of outer and inner lamellae that are jointed together in the anterior portion, forming a sharply pointed tip. The outer lamella is gently arched and extended posteriorly, while the inner lamella is partly enclosed by the outer lamella and projected posteroventrally. These features are commonly observed in the lower jaws of modern and fossil coleoids (Clarke 1986; Tanabe and Fukuda 1999; Neige and Dommergues 2002; Kubodera 2005; Tanabe 2012), indicating that the two jaws examined are identified as lower jaws of coleoids. Both lamellae are wholly composed of a black homogeneous material without any trace of calcification. We interpret that the lamellae were originally made of a chitin-protein complex as in those of modern cephalopod jaws (Saunders et al. 1978; Hunt and Nixon 1981; Gupta et al. 2008).

One of the two jaw specimens, KMNH IvP 902,001, is identified as a new species of Nanaimoteuthis Tanabe in Tanabe et al., 2008, of the order Vampyromorpha because it shares common generic features with other species of the same genus (see Diagnosis described in Systematic palaeontology). The other jaw specimen, KMNH IvP 902,002, seemingly exhibits mosaic features of the lower jaws of modern coleoids, especially sepiids and teuthids. To determine its higher taxonomic relationship, we performed morphometric analysis for the latter specimen with the lower jaws of 94 modern and 1 fossil coleoid species (listed in SOM 1: Supplementary Online Material available at  http://app.pan.pl/SOM/app60-Tanabe_etal_SOM.pdf), each represented by a single specimen. We followed Doyle et al. (1994) for higher systematics of coleoid cephalopods and Clarke (1962, 1986) and Clarke and Maddock (1988) for descriptive terms and measurements of a coleoid lower jaw (Fig. 2).

Morphometric analysis

Data source and methods

Geometric morphometrics is useful for detecting slight differences in shape among taxa with great within-group variation because it retains full geometric information on the spatial configuration of landmarks throughout analysis. In the present study, two-dimensional geometric morphometrics was used to assess among-species variation in the form of the lower jaw in extant and fossil coleoid species and to compare the specimen KMNH IvP 902,002 with Recent taxa.

Landmark and outline data were collected from photographs by Clarke (1986) and Kubodera (2005) and from photographs taken by us for the specimens housed at UMUT, AMNH, and CDM. Some of the latter specimens have been provided in Tanabe et al. (2008) and Tanabe (2012). Specimens were photographed from the right or left side using a digital camera, and the image was saved as a jpeg file. The left-side images were used for morphometric analysis, if available, and the right-side images were reflected to obtain reversed images prior to analysis. Eight landmarks were fixed on the surface of the lateral-side image of each jaw (Fig. 3A). Neige and Dommergues (2002) defined 10 landmarks on the lateral side of the lower jaw, but two were omitted in the present study because they were poorly defined in some specimens. Coordinates of all landmarks were collected on a computer image using the digitizing program DIGITIZER (Ubukata 2004). Each segment of the outline curve between neighbouring landmarks was equally spaced to define comparable points, or semi-landmarks, along the curve (Fig. 3A). Analysis using semi-landmarks allows us to simultaneously compare an outline curve with a landmark configuration (Bookstein 1997). The outline curves were digitized using the same computer program by locating a series of coordinate points along the curves, and sets of coordinate points were interpolated so that each segment was represented by a fixed number of equally spaced semi-landmarks (Ubukata 2003; Ubukata et al. 2010). A total of 89 semi-landmarks were used in addition to the eight fixed landmarks.

Fig. 2.

Diagrammatic drawing of a coleoid lower jaw. Left lateral (A) and frontal (B) views. 1, rostral length; 2, hood length; 3, crest length; 4, wing length; 5, width of outer lamella. After Clarke (1962, 1986), see Clarke and Maddock (1988) for terms and measurements.

Fig. 3.

Eight landmarks and 89 semi-landmarks collected on a coleoid lower jaw (A, B). 1, the rostral tip; 2, the base of the rostrum; 3, the posterior end of the wing; 4, the posterior intersection of the outer and inner lamellae; 5, the posterior end of the inner lamella; 6, the anteroventral end of the crest; 7, the anterior intersection of the outer and inner lamellae; 8, the anteroventral end of the hood. B. A square grid showing shape differences between two individuals as deformation from B1 to B2 drawn by use of a thin-plate spline.

Fig. 4.

All coleoid lower jaws registered by the generalized least squares superimposition. A. A full set of landmarks (large symbols) and semi-landmarks (small symbols). B. A subset of landmarks and semi-landmarks put along the outline of the outer lamella.

Centroid size (Bookstein 1991; Slice et al. 1996) was calculated for each specimen and was used for representing specimen size. A configuration of landmarks and semi-landmarks was translated, rotated, and scaled using generalized least squares (Rohlf and Slice 1990), allowing semi-landmarks to slide to minimize the bending energy of deformation (Bookstein 1997). The mean shape of all specimens was then calculated as the set of mean coordinates of landmarks and semi-landmarks of the aligned specimens by least squares superimposition. The shape of a specimen can be characterised by the deformation of its landmark configuration from the mean shape. An interpolation technique applied for coordinates of landmarks and semi-landmarks using the thin-plate spline function defines a deformation grid (Fig. 3B) for data of two shapes (Bookstein 1989, 1991; Rolf 1996). The thin-plate spline deformation from the mean shape of each specimen can be decomposed into a series of components, called partial warps (Bookstein 1989, 1991). Scores of the partial warps for each specimen (partial warp scores) were then used as variables in principal component analysis (relative warp analysis) (Bookstein 1991; Rohlf 1993). Relative warps, which correspond to principal components, were estimated in each specimen to examine the variation in shape of the lower jaw.

The lower jaw consists of outer and inner lamellae, and the length and orientation of the inner lamella seem to be highly variable among the species examined. Generalized least squares superimposition adopted here is sensitive to large displacements at one or several landmarks (Zelditch et al. 2004). The “Pinocchio Effect”, a great change concentrated at one landmark with little at others, may considerably affect analytical results in the present case.Therefore, shape of the outer lamella, as well as gross jaw form, was also compared among species of different taxa. In this option, a subset of six landmarks and 64 semi-landmarks along the outline of the outer lamella was used for the analyses.

The morphometric analysis was performed using the computer program TPSRELW version 1.49 (Rohlf 2010). For sliding semi-landmarks, the default setting was used. In computing the partial warps, α = 0 (Rohlf 1993) and the estimates of the uniform components by Rolf and Bookstein (2003) were included as an option of the TPSRELW program.

Results

Procrustes analysis.—A set of the lateral views of coleoid lower jaws were registered using Procrustes superimposition to visualize variation in shape. Procrustes analyses for the full set and a subset of landmarks and semi-landmarks both led to a similar result (Fig. 4). Teuthids generally have an outer lamella with a wide jaw angle and a great lower rostral length; the shape of the outer lamella shows remarkable interspecific variation with species of other orders. Interspecific shape variation appears to be greater in the inner lamella than in the outer lamella. Octopods tend to possess a longer crest than most teuthids (Fig. 4A). In KMNH IvP 902,002, the jaw angle is acute (ca. 80°), and the rostral length is relatively as large as the rostral lengths of modern teuthids. However, unlike typical teuthids, its shoulder is distinctly projected dorsally, and its ventral margin of the outer lamella is almost straight (Fig. 4B).

Relative warp analysis for the full data set.—The first relative warp (RW1) accounts for 46.48% of the total variance, and the first five relative warps together account for approximately 85% (SOM 2). RW1 mainly represents overall stretching or compression in the dorsoventral direction (Fig. 5A). As the score proceeds negatively along the RW1 axis, the outline of the lower jaw is compressed dorsoventrally, with a wider jaw angle and greater lower rostral length. In contrast, a jaw form accommodated in the positive RW1 area of the morphospace is characterised by dorsoventral elongation with a reduced rostrum. The second, and smaller relative warp (RW2) defines local shape variation. In the positive direction of RW2, the posterior ends of the wing are stretched posteriorly, but the outer lamella reduces its anterior, forming a dogleg shape (Fig. 5A). In the negative direction of RW2, the wing is largely reduced, while the outer lamella is greatly enlarged. The third relative warp (RW3) seems to correlate with the proportion of the inner lamella to the outer lamella as well as with the proportion of wing size to rostrum size. A specimen with a negative RW3 score tends to possess a short lateral wall of the inner lamella, covered mostly by the hood of the outer lamella and with a large rostrum, while a specimen with a positive RW3 score has a developed inner lamella with a flat cap of the outer lamella (Fig. 5B). The fourth relative warp (RW4) is closely related to the length of the lateral wall (Fig. 5B). As RW4 score decreases, the lateral wall is elongated posteroventrally, and the outer lamella is anteroposteriorly compressed. A positive perturbation along the RW4 axis reduces the lateral wall and elongates the wing.

Fig. 5.

Plots of the relative warp scores computed for the full set (A, B) or subset (C, D) of landmarks and semi-landmarks and the relative warps displayed as thin-plate splines. The lower and left grids show deformations from the mean shape (undeformed grid) corresponding to perturbations associated with RW1 and RW2, respectively (A, C) or RW3 and RW4, respectively (B, D).

RW1 fairly separates higher taxa of coleoids. Most teuthids (> 85%) have RW1 scores lower than 0.02, while all sepiids examined, excluding Sepia longipes, have RW1 scores higher than 0.02 (Fig. 5A). Most species of Octopoda, Cirroctopoda, and Vampyromorpha, all of which are included in the superorder Octobrachia, occupy the larger RW1 re- gion than the species of Teuthida and Sepiida of the superorder Decabrachia (>0.12). Compared to RW1, RW2 seems useless for higher taxonomy. Stauroteuthis sp., an octopod species, has by far the lowest RW2 scores of any coleoid species. Except for this species, species of Vampyromorpha and Cirroctopoda have extreme RW2 values. In RW1-RW2 morphospace, KMNH IvP 902,002 is located near the region predominantly occupied by decabrachian species (sepiids, octopods, and teuthids) and has a higher RW1 score (0.0885) than most teuthid species (Fig. 5A).

In RW3–RW4 morphospace, all the decabrachian species occupy the central region and are surrounded by octobrachian species, most of which have forms that are quite different from the mean shape (Fig. 5B). This result indicates high variability in the lower jaw morphology of octopuses. KMNH IvP 902,002 is distributed at the periphery of the decabrachian species and has a lower RW3 value than any sepiid species.

Relative warp analysis for subset data on the outer lamella.—RW1 accounts for 44.63% of the total variance, and the first five relative warps together account for approximately 88% (SOM 2). As in the result for the full data set, RW1 also represents stretching or compression in the dorsoventral direction and relative size of the rostrum, although the positive direction of the RW1 axis is reversed in this case (Fig. 5C). RW2 for the subset data corresponds to RW3 for the full data set, i.e., the proportion of wing size to rostrum size. RW3 for the subset data is characterised by compression of the anterior hood and posterior wing (Fig. 5C). The anterior part of the hood is slightly compressed when the RW3 score is positive, whereas the negative perturbation of RW3 reduces the posterior part of the wing. The individual characterised by a positive RW4 score tends to have a well-developed shoulder and a straight ventral margin with great lower rostral length.

Most octobrachian species examined have lower RW1 (less than -0.1) scores than the decabrachian species (Fig. 5C). KMNH IvP 902,002 is accommodated near the origin of the RW1–RW2 morphospace, which is occupied by teuthids as well as by sepiids. This specimen has one of the highest RW2 scores among the decabrachian species. Neither RW3 nor RW4 separates higher taxa of coleoids (Fig. 5D).

Our two-dimensional geometric morphometric suggests that the higher taxonomic relationship of KMNH IvP 902,002 is assigned to the Decabrachia because its morphometric features are more similar to extant decabrachian species than to modern and fossil octobrachian species, especially in the overall shape of the outer lamella. The other characteristic feature observed in KMNH IvP 902,002 is the development of a distinct fold on the lateral wall (see Systematic palaeontology). In modern decabrachian coleoids, this feature is recognised in many genera of Teuthida, such as Lepidoteuthis (Octopoteuthidae), Histioteuthis (Histioteuthidae), and Cycloteuthis (Cycloteuthidae) (Clarke 1986; Kubodera 2005), but does not occur in the lower jaws of Sepiida, Sepiolida, and Spirulida (Clarke and Maddock 1988: fig. 2). For these reasons, we assign the higher taxonomic position of KMNH IvP 902,002 to Teuthida.

Systematic palaeontology

Fig. 6.

Octobrachian coleoid Nanaimoteuthis hikidai sp. nov. KMNH IvP 902,001, holotype, a lower jaw; early Campanian, Late Cretaceous, Haboro area, Hokkaido, Japan. Right lateral (A), left lateral (C), dorsal (D), and frontal (E) views, and right lateral view after removing part of outer lamella (B).

Fig. 7.

Decabrachian coleoid Haboroteuthis poseidon gen. et sp. nov. KMNH IvP 902,002, holotype, a lower jaw; late Santonian, Late Cretaceous, Haboro area, Hokkaido, Japan. Frontal (A), left lateral (B), dorsal (C), and right lateral (D) views.

Fig. 8.

Phylogenetic relationships of fossil and modern coleoid major taxa complied from the data by Kröger et al. (2011: fig. 5) and Fuchs et al. (2013: fig. 12). Fossil records of Cretaceous decabrachian and octobrachian coleoids from the North Pacific region are given in black circles. 1, Haboroteuthis (this study); 2, Yezoteuthis (Tanabe et al. 2006); 3, Paleocirroteuthis (Tanabe et al. 2008); 4, Nanaimoteuthis (Tanabe et al. 2008; Tanabe and Hikida 2010; this study). Age after Cohen et al. (2012).

Discussion

Modern coleoid cephalopods are currently classified into two superorders, Decabrachia Boettger, 1952 (10 arms, fourth pair modified as tentacles), and Octobrachia Fiorini, 1981 (= Vampyropoda Boletzky, 1992) (10 arms, second pair modified or lost) (Doyle et al. 1994). The superorder Decabrachia includes four orders (Spirulida, Sepiida, Sepiolida, and Teuthida), and the superorder Octobrachia includes three orders (Octopoda, Cirroctopoda, and Vampyromorpha) (Doyle et al. 1994: table 1). Molecular clock analysis of modern coleoids using different genes, models, and software suggested a Palaeozoic origin for the coleoid crown (Strugnell et al. 2006; Bergmann et al. 2006; Warnke et al. 2011; Kröger et al. 2011). Although divergence time estimates of the coleoid crown do not match among researchers (453 ± 60 Ma, Ordovician/ Silurian boundary interval, Warnke et al. 2011; 415 ± 24 Ma, early Palaeozoic, Bergmann et al. 2006; 276 ± 75 Ma, early to middle Permian, Kröger et al. 2011), they are congruent at several points with the available fossil record (see Kröger et al. 2011: fig. 5). For these reasons, the origin and phylogenetic relationships of individual modern coleoid orders are still controversial.

Well-preserved coleoid fossils with squid-like compressed soft tissue remains (“fossil teuthids”) are known to occur episodically from the Konservat-Lagerstätten of the late Palaeozoic to Cretaceous ages (Naef 1922; Fischer and Riou 1982a, b, 2002; Fuchs 2006a, b; among others). These “fossil teuthids” sometimes exhibit a mosaic distribution of characteristic features of modern Decabrachia and Octobrachia (see discussion by Vecchione et al. 1999); hence, they have been mostly placed in Decabrachia and Octobrachia (Naef 1922; Donovan 1977; Doyle et al. 1994; Young et al. 1998; Vecchione et al. 1999) or in Octobrachia (Bandel and Leich 1986; Engeser and Bandel 1988; Engeser 1990a, b; Haas 2002; Fuchs 2006a).This discrepancy among previous researchers contributes greatly to the different evaluations of the arm structures and gladii preserved as fossils. Concerning the origin of Decabrachia, Fuchs et al. (2013a) suggested that the Cretaceous belemnoid-like phragmocone-bearing coleoid Longibelus, which has a marginal siphuncle, stripelike dorsal attachment scars, and a poorly developed layer of tabular nacre, represents a sister taxon of the decabrachian clade comprising modern Spirulida, Sepiida, and Teuthida. Meanwhile, Tanabe et al. (2006) described an extremely large coleoid upper jaw from the Campanian (Upper Cretaceous) of Hokkaido as a new genus and species (Yezoteuthis giganteus) and assigned its higher taxonomic position as Teuthida based on cladistic analysis of several characters with modern cephalopods.

This study and others (Tanabe et al. 2006, 2008; Tanabe and Hikida 2010; Tanabe 2012) on the jaws from the Upper Cretaceous of Hokkaido and Vancouver Island (Canada) fill the gap in the relatively poor fossil record of the mainly soft-bodied coleoids (Fig. 8).

Conclusions

On the basis of the comparison of gross morphology and morphometric data of the lower jaws of modern and fossil coleoids, two extremely large lower jaws that were recovered from the Upper Cretaceous strata in Haboro area, Hokkaido, Japan, were assigned to the following new taxa: Nanaimoteuthis hikidai sp. nov. of the order Vampyromorpha (superorder Octobrachia) and Haboroteuthis poseidon gen. et sp. nov. of the order Teuthida (superorder Decabrachia). Because of the unusually large lower jaws, these new taxa appear to be comparable in body size to modern giant squids (Architeuthis spp.) and the Humboldt squid (Dosidicus gigas). This and other discoveries of large jaws referable to octobrachian and decabrachian coleoids from the Upper Cretaceous strata of the North Pacific fill the gap in the relatively poor fossil record of mainly soft-bodied coleoids and demonstrate that large non-belemnoid coleoids existed in this bioprovince during that time.