The Cephalopoda are a class of molluscs first known from the Late Cambrian; among modern taxa, cephalopods include the shelled Nautilidae and the Coleoidea (squids, cuttlefish, octopods) that lack an external shell. Traditionally, the cephalopods have been divided between three subclasses, the Nautiloidea, Ammonoidea and Coleoidea, but the Nautiloidea in the broad sense are paraphyletic with regard to the other two groups.

The earliest cephalopods are known from the Upper Cambrian and had straight to slightly curved shells with tightly packed septa; these stem representatives may not yet have adopted the nektonic lifestyles of crown cephalopods. An alternative structure for cephalopod evolution was proposed by Smith & Caron (2010), who interpreted the Early to Middle Cambrian Nectocarididae as early cephalopods and suggested that the first cephalopods may have lacked a shell (see below for comments on why this may be unlikely).

Open query: what are cephalopod shells for?
Published 29 October 2008
Cross-sectioned reconstruction of the Late Cambrian cephalopod Plectronoceras, from Palaeos.com.

Palaeozoic cephalopods have been occupying my mind lately. I am currently working my way through the Treatise of Invertebrate Paleontology volume on “nautiloids” (the lumping of all non-ammonoid, non-coleoid cephalopods under the name of “nautiloid” is a really unfortunate example of paraphyletic lumping, which is why I’m insisting on the quotation marks). For those unfamiliar with them, the Treatise of Invertebrate Paleontology (or simply the Treatise to its friends) is a series of volumes cataloguing the known genera of fossil invertebrates, and each volume also includes a series of introductory essays describing the anatomy, palaeoecology, etc. of the group it covers. I’ve commented before that there is something incredibly purgatorial about a Treatise volume. It’s a long, painful, torturous slog that hammers you both mentally and physically*, but when you finally come out the other end there can be no doubt that you’ve done something really worth achieving.

*Anyone who doubts that a book is able to challenge you physically has not had to carry a bag holding the entire 1000 pages plus in three volumes of the Treatise section on crinoids, nor been assualted by a falling copy of the 1100 pages of Roewer’s Die Weberknechte der Erde.

The question that is currently residing in my mind also, in a roundabout sort of way, relates to the recent publication of the fossil avialian Epidexipteryx. Typically for a Nature article, the revolutionary nature of that fossil has been more than a little exaggerated, but it has publicised the growing consensus among palaeontologists that feathers were not originally used for flight when they first evolved. It is more likely that they were originally used for insulation, and only later became used for other functions such as display and flight*. Gould & Vrba (1982) actually coined a term for this phenomenon, ‘exaptation’, which was meant to refer to cases where a feature that originally evolved for one function was co-opted for use in another function, as opposed to ‘adaptation’, when a feature originally evolved directly for its current function. The term ‘exaptation’ in itself never really caught on, because almost all ‘adaptations’ are in some way ‘exaptations’. However, the verb form ‘exapted’ remains useful when describing examples of such a process.

*I’m going to speculate a little and suggest that the use of feathers for display may have even been a necessary prerequisite for their use for flight, because the planar surface required for flight may have been less likely to be selected for insulation than for maintaining a stereotyped form for display. This may be why flightless maniraptorans such as Caudipteryx possessed planar tail and arm fans.

Cross-section of the Late Cambrian cephalopod Yanheceras, showing the closely-spaced septa in this 12 mm shell, from Palaeos.com.

What is the connection between cephalopods and the origin of feathers? As note before, fossil cephalopods can be distinguished from all other molluscs by their unique shell structure, with the shell divided into a series of internal chambers. In most shelled cephalopods, the chambers would have been/are mostly hollow and filled with gas whose volume can be adjusted to control shell buoyancy. But how and why did this unique structure evolve in the first place? Contrary to first impressions, the chambers were probably not used as floats when they first appeared. The really early cephalopods, such as plectronoceratids and ellesmeroceratids, were small subconical shells, generally only a few centimetres in length. Despite their small size, their shells were divided into relatively large numbers of chambers, with the dividing septa packed close to each other. Even if the chambers were filled with gas (which they may not have been—it is currently unknown just when in cephalopod evolution the chambers became gas-filled), it is unlikely that the volume of the chambers would have been enough to lift the weight of the shell. Plectronoceratids, etc., were almost certainly benthic (Furnish & Glenister 1964). Because these early forms are mostly outside the cephalopod crown group, there is currently no way of knowing whether they shared any of the soft-body features associated with modern cephalopods, such as tentacles and the siphon, or whether they still had a more primitive, superficially snail-like (though untorted) morphology.

Holland (1987), if I understand correctly, suggests that the septate shell, either by increasing relative buoyancy (even if it did not make the animal actually buoyant) or by changing the distribution of the shell’s weight, may have made cephalopod locomotion more energy-efficient, allowing greater mobility. Does this seem like a likely explanation? Can any of you think of an alternative? And how would you suggest we test any likely explanations?

How to be straight
Published 5 November 2008

Externally-shelled cephalopods in the Palaeozoic showed a far greater diversity of basic morphologies than their Mesozoic and Caenozoic successors—coiled gyrocones, long straight orthocones, short fat brevicones. By the beginning of the Mesozoic, almost all cephalopod shells were planar coils. A few orthoconic orthocerids lingered into the Triassic (and some ammonoid families did later experiment with different arrangements), but, overall, the coil was king.

The primary reason for this was probably buoyancy management. The cephalopod shell, with its inbuilt flotation chambers, is a marvellous thing indeed, and was doubtless a crucial factor in allowing some cephalopods to become the biggest animals in the Palaeozoic. An exogastric (i.e. away from the venter) coil brings the centre of buoyancy more or less directly above the animal. Straight-shelled forms, of which there were many during the Palaeozoic, faced more of a challenge in this regard. Simply extending and enlarging the shell would have increased the potential buoyancy, but with the animal’s buoyancy shifted towards the back end and its mass centred towards the living chamber at the front, orthoconic cephalopods with simple shells would have ended up floating permanently head-downwards with their arses sticking towards the sky—a rather inconvenient position for doing anything much. Some alternative approach was required to allow the shell to remain horizontal.

Diagram of cameral deposits from Kevin Bylund.

One approach that was used by a number of Palaeozoic cephalopods, such as orthocerids, was the formation of cameral deposits. Cameral deposits were mineralised layers coating the insides of the chambers (Teichert 1964a). They became progressively thicker as they approached the apex of the shell, thus counter-balancing the weight of the living chamber at the front. They were also generally thicker ventrally than dorsally, to keep the animal upright.

If we were to assume that Palaeozoic cephalopod anatomy was just like that of a modern Nautilus (a completely unwarranted assumption, but one that has been made all too often), explaining cameral deposits poses a major dilemma. In Nautilus, the only part of the soft anatomy extending behind the living chamber is the siphuncle, a backward extension of the mantle. The siphuncle is a narrow cord running (more or less) through the centre of the chambers. Otherwise, the chambers are devoid of tissue, and the internal walls are bare (Stenzel 1964). If orthocerids and such had the same arrangement, then the cameral deposits would have had to have been laid down in each successive living chamber before that chamber was closed off by the development of a new septa and forward contraction of the mantle. Though such an arrangement has indeed been suggested in the past, Teichert (1964a) pointed out that it was probably impossible. In many orthocones, the cameral deposits are so well-developed that the most apical chambers are entirely or nearly entirely filled by them. If they had been laid down before the formation of the next chamber, there would have been no room left in the shell for the animal itself! Also, when cameral deposits growing from opposing walls of the chamber meet in the middle, they are generally divided by a thin line, a pseudoseptum. It seems more likely that orthocerids and many other Palaeozoic cephalopods differed from modern Nautilus in possessing a “cameral mantle”, a further extension of the mantle that lined the inner walls of the chambers*. While a cameral mantle may have been an ancestral feature for cephalopods (in light of the presence of cameral deposits in a number of phylogenetically disparate lineages, though some authors, e.g. Kolebaba, 2002, have recognised an order Pallioceratida defined by the presence of cameral mantle), it has not been preserved in any living cephalopod.

*A third alternative, suggested by some authors such as Mutvei (2002), is that the cameral deposits were not laid down during the lifetime of the animal at all, but instead represent post-mortem deposits formed by minerals precipitating from water penetrating the chambers. If so, they would be completely irrelevant to the animal’s lifestyle. I agree with Teichert (1964a) that this seems unlikely considering the even, regular arrangement of the deposits.

Cross-section of the breviconic endocerid Cassinoceras, with the endocones visible in the lower part of the shell. Image from Palaeos.com.

An alternative solution to cameral deposits was employed by groups such as endocerids. Endocerids (which include the largest of all orthocones) had very large siphuncles, sometimes occupying nearly half the diameter of the shell (in another example of how Palaeozoic cephalopods may have differed in anatomy from modern cephalopods, Teichert [1964a] suggested that the siphuncular space in such forms may have been large enough that not only the mantle but also some of the visceral mass probably extended back past the living chamber). Instead of forming cameral deposits, endocerids weighted the apex of the shell by mineralising the siphuncle itself. The siphuncular space became filled with endocones, conical mineral layers stacked one into the next like a series of waffle cones (Teichert 1964b). A hollow tube running through the centre of the endocones probably contained the living tissue. Because the siphuncle was such a sturdy structure, it is not uncommon for endocerids to be preserved as isolated pieces of siphuncle, with no trace of the more delicate external shell. Some structurally very distinctive groups, such as the Allotrioceratidae with two stacks of endocones pressed into the siphuncle alongside each other, are only known from such fragments of siphuncle (Teichert 1964b), and what the remainder of the animal looked like is a complete mystery.

I do have to end this post on something of a complaint. The Endocerida, as recognised by Teichert (1964b), contains an assortment of families united primarily (as far as I can tell) by the presence of endocones. However, elsewhere in the same volume, Teichert (1964a) refers to the presence of endocones in some members of at least two other cephalopod orders, the Discosorida and Orthocerida. At least one of the families included by Teichert (1964b) in the Endocerida, the Narthecoceratidae (then known only from isolated siphuncles), has been transferred to the Orthocerida after the discovery of more complete specimens (Frey 1981). So it would appear that all cephalopods with endocones are endocerids—except for when they are not. The stench of potential polyphyly hangs heavy in the air…

Nectocaris: largely irrelevant to cephalopods?
Published 27 May 2010
Nectocaris pteryx as reconstructed by Marianne Collins in Smith & Caron (2010).

Today’s issue of Nature sees the publication of a paper presenting a radical reinterpretation of the Middle Cambrian nektonic animal Nectocaris pteryx (Smith & Caron 2010). Previously only known from a single specimen, Smith & Caron increase the hypodigm of Nectocaris by a whopping 91 specimens, an absolutely mindblowing advance. Unfortunately (and, I’m sad to say, not uncommonly for a Nature paper), the authors then take this amazing discovery and use it to make some decidedly unwarranted inferences.

Smith & Caron reconstruct Nectocaris as a small squid-like animal with two anterior tentacles, broad lateral fins and a ventral cylindrical funnel close to the head. Based on the similarity of the funnel to the siphon of living cephalopods, the authors infer a relationship between Nectocaris and cephalopods and suggest that the former is representative of the ancestral morphology of the latter. One problem with that – Nectocaris doesn’t have a shell and cephalopods have always been assumed to have evolved from shelled ancestors like other mollusc classes. Smith & Caron suggest that this assumption is incorrect and that each of the living mollusc classes acquired shells independently.

This is the representation given by Smith & Caron (2010) of molluscan evolution and the known fossil record of each of the classes:

Smith & Caron (2010): “Arrows indicate the crown groups of 1, molluscs; 2, conchifera; 3, cephalopods. Stars represent the earliest record of mineralization in each lineage (after ref. 23). Clade divergence times (dotted lines) are unconstrained. Early branches follow previous phylogeny (after ref. 20).”

Simple, straightforward and very misleading. The diagram only shows the living classes of mollusc but omits all lineages not directly relatable to one or another of the recent taxa—a category that includes most Cambrian molluscs, including many that are directly relevant to cephalopod ancestry. The phylogenetic positions of Tryblidiida (including modern ‘monoplacophorans’) and Polyplacophora (chitons) as sister group or serial* sister groups to other molluscs, together with features of putative stem molluscs such as Wiwaxia and their possible nearest living relatives the annelids, suggest that serially-repeated structures were part of the ancestral ground plan for molluscs. The absence of indications of serial structures in many Cambrian ‘monoplacophorans’ such as helcionelloids suggests that they were (at least) part of the clade including bivalves, gastropods and cephalopods, and the fossil record for helcionelloids extends back to the very earliest Cambrian (Runnegar & Jell 1976). The supposed absence of an early fossil record for scaphopods overlooks good support for a derivation of scaphopods from the Rostroconchia, another Palaeozoic mollusc group (Peel 2006) which may take the scaphopod lineage back to the early Cambrian. Smith & Caron dismiss the possibility that Nectocaris may have secondarily lost an ancestral shell by claiming that it is too early in the fossil record and lacks likely predecessors; however, shells have been lost on a large number of occasions in molluscan history; shelled molluscs appeared in the fossil record some twenty million years or so before the earliest known nectocarids; and the relative rarity and simplicity of early molluscan fossils (early molluscs were generally small and fairly delicate) means that it is quite possible that a direct nectocarid ancestor may not have been preserved, nor is there any guarantee that it would be recognised as such if it had.

*No pun intended.

The earliest known stem cephalopods (from the Late Cambrian) possessed shells with large numbers of very tightly packed septa and were unlikely to have been very buoyant. Their generally short conical shape would have been ill-suited for jet-propelled swimming as in modern cephalopods and they were most likely benthic. As other molluscan classes were also ancestrally benthic, it seems unparsimonious that the actively swimming Nectocaris represents the ancestral cephalopod lifestyle.

If Nectocaris is a stem cephalopod (which essentially depends on how strong the siphon is as a supporting apomorphy), then the most likely scenario is that its shell loss and squid-like form is an independent convergence on modern shell-less cephalopods rather than representing the ancestral form for cephalopods as a whole. Nectocaris would not be an ancestor, but a highly specialised side branch of its own.

So nice when people agree with you
Published 14 January 2011
Nectocaris pteryx

When I wrote the above section, I was not convinced by Smith & Caron’s (2010) interpretation of Nectocaris as an early cephalopod. Nor am I the only one: a paper recently released by Mazurek & Zatoń (2011) comes to much the same conclusions. They point out that Smith & Caron’s proposed model of cephalopod evolution conflicts strongly with what we previously knew from the cephalopod fossil record (and that’s no small amount—there are few groups of organisms whose fossil record has been as intensely studied as cephalopods), and that most of the characters supposedly shared between Nectocaris and cephalopods are in fact only shared between Nectocaris and coleoids (modern octopods and squid) that did not appear in the fossil record until during the Mesozoic, considerably later than the Cambrian Nectocaris. Smith & Caron (2010) suggested that the absence of a shell in Nectocaris indicated that the cephalopod shell had been evolved independently to that of other molluscs, but Mazurek & Zatoń point out that the only known cephalopods to completely lack any trace of a shell are octopods, and that octopods are secondarily shell-less is indicated, not only by their phylogenetic position, but also by the presence of a remnant shell in Cretaceous stem-octopods.

What I find particularly interesting about Mazurek & Zatoń’s paper, however, is how much it brings up the same points already raised by commenters here. The primary feature cited by Smith & Caron (2010) as connecting Nectocaris with cephalopods, the presence of a funnel, is contradicted by the apparent difference in functional structure between a cephalopod siphon and Nectocaris‘ ‘funnel’, as noted by Adam Yates. Aydin Örstan commented on the absence of a radula. Just goes to show that I’ve got some pretty clever readers here. Mind you, I’m not happy with everything in Mazurek & Zatoń’s paper. They make the argument that Nectocaris could not have evolved from a shelled and radula-possessing ancestor because it was ‘too early’, but the known fossil record of molluscs pre-dates Nectocaris by about twenty million years, more than long enough for shell loss to potentially occur. The absence of a beak in Nectocaris is also of doubtful significance as this is a cephalopod autapomorphy.

I am also not swayed by Mazurek & Zatoń’s (provisional) alternative placement for Nectocaris as a dinocarid. Though it is tempting to compare the funnel of Nectocaris to the proboscis of dinocarids such as Opabinia, and dinocarids have the advantage over coleoid cephalopods of being coeval with Nectocaris, no dinocarid has cephalic tentacles like Nectocaris. Also, some of the figures in Smith & Caron (2010) appear as if the pharynx might pass through the funnel; in Opabinia, the proboscis was a separate structure in front of the mouth. I can think of two groups of animals for which cephalic tentacles are definitely known: annelids and molluscs (presuming that the tentacles of Nectocaris are not a unique autapomorphy of its own). The lack of obvious segmentation makes it unlikely that Nectocaris is an annelid. That leaves us with mollusc. While Smith & Caron’s identification of a pinhole camera eye in Nectocaris could still connect it to cephalopods, I believe that this is outweighed by the arrangement of the gut. The presence of an apparent through-gut, opening with a terminal anus, was identified by Chen et al. (2005) in Vetustovermis (now recognised as a synonym of Nectocaris)*. Cephalopods, however, have a U-shaped gut, opening in the mantle cavity not too far from the head. This is related to the 90° shift that the cephalopod body plan has gone through during its evolution: the apparent “front-back” axis of a squid actually represents the “top-bottom” axis of other molluscs. A similar U-shaped gut is present in scaphopods, the probable living sister group of cephalopods, so a molluscan Nectocaris would have to sit outside the scaphopod-cephalopod clade. As far as is known, cephalic tentacles are a possible synapomorphy of the clade uniting cephalopods, scaphopods and gastropods**, so a molluscan Nectocaris would probably have to be either a stem representative of this clade, or just possibly a stem gastropod.

*My apologies to the commenter somewhere whose identity I’ve forgotten who brought my attention to this point.

**Cephalic tentacles are definitely absent in polyplacophorans and tryblidiids***. They are also absent in bivalves, but bivalves don’t have a head to have cephalic tentacles on in the first place, so the absence of tentacles is probably best treated as ambiguous for bivalves.

***Or whatever Neopilina and its ilk are going by these days.


Chen, J-Y., D.-Y. Huang & D. J. Bottjer. 2005. An Early Cambrian problematic fossil: Vetustovermis and its possible affinities. Proceedings of the Royal Society of London B 272: 2003–2007.

Frey, R. C. 1981. Narthecoceras (Cephalopoda) from the Upper Ordovician (Richmondian) of southwest Ohio. Journal of Paleontology 55 (6): 1217–1224.

Furnish, W. M., & B. F. Glenister. 1964. Paleoecology. In: Moore, R. C. (ed.) Treatise on Invertebrate Paleontology pt K. Mollusca 3 pp. K114–K124. The Geological Society of America and The University of Kansas Press.

Gould, S. J., & E. S. Vrba. 1982. Exaptation; a missing term in the science of form. Paleobiology 8 (1): 4–15.

Holland, C. H. 1987. The nautiloid cephalopods: a strange success. Journal of the Geological Society of London 144 (1): 1–15.

Kolebaba, I. 2002. A contribution to the theory of the cameral mantle in some Silurian Nautiloidea (Mollusca, Cephalopoda). Bulletin of the Czech Geological Survey 77 (3): 183–186.

Mazurek, D., & M. Zatoń. 2011. Is Nectocaris pteryx a cephalopod? Lethaia 44: 2–4.

Mutvei, H. 2002. Connecting ring structure and its significance for classification of the orthoceratid cephalopods. Acta Palaeontologica Polonica 47 (1): 157–168.

Peel, J. S. 2006. Scaphopodization in Palaeozoic molluscs. Palaeontology 49 (6): 1357–1364.

Runnegar, B., & P. A. Jell. 1976. Australian Middle Cambrian molluscs and their bearing on early molluscan evolution. Alcheringa 1 (2): 109–138.

Smith, M. R., & J.-B. Caron. 2010. Primitive soft-bodied cephalopods from the Cambrian. Nature 465: 469–472.

Stenzel, H. B. 1964. Living nautilus. In: Moore, R. C. (ed.) Treatise on Invertebrate Paleontology pt K. Mollusca 3. Cephalopoda—General Features—Endoceratoidea—Actinoceratoidea—Nautiloidea—Bactritoidea pp. K59–K93. The Geological Society of America and The University of Kansas Press.

Teichert, C. 1964a. Morphology of hard parts. In: Moore, R. C. (ed.) Treatise on Invertebrate Paleontology pt K. Mollusca 3. Cephalopoda—General Features—Endoceratoidea—Actinoceratoidea—Nautiloidea—Bactritoidea pp. K13–K53. The Geological Society of America and The University of Kansas Press.

Teichert, C. 1964b. Endoceratoidea. In: Moore, R. C. (ed.) Treatise on Invertebrate Paleontology pt K. Mollusca 3. Cephalopoda—General Features—Endoceratoidea—Actinoceratoidea—Nautiloidea—Bactritoidea pp. K160–K189. The Geological Society of America and The University of Kansas Press.

Leave a comment

Your email address will not be published. Required fields are marked *