Cephalopod development and evolution


People are always arguing about whether primitive apes could have evolved into men, but that one seems obvious to me: of course they did! The resemblances are simply too close, so that questioning it always seems silly. One interesting and more difficult question is how oysters could be related to squid; one’s a flat, sessile blob with a hard shell, and the other is a jet-propelled active predator with eyes and tentacles. Any family resemblance is almost completely lost in their long and divergent evolutionary history (although I do notice some unity of flavor among the various molluscs, which makes me wonder if gustatory sampling hasn’t received its proper due as a biochemical assay in evaluating phylogeny.)

One way to puzzle out anatomical relationships and make phylogenetic inferences is to study the embryology of the animals. Early development is often fairly well conserved, and the various parts and organization are simpler; I would argue that what’s important in the evolution of complex organisms anyway is the process of multicellular assembly, and it’s the rules of construction that we have to determine to identify pathways of change. Now a recent paper by Shigeno et al. traces the development of Nautilus and works out how the body plan is established, and the evolutionary pattern becomes apparent.

Working out the development of a cephalopod is very hard work. They aren’t trivial to raise, requiring large quantities of filtered sea water and constant tending — imitating an open ocean environment in a tank in a lab isn’t easy. The Shigeno group has managed to raise 3 generations of Nautilus in the lab, which is an accomplishment in itself; they collected 1035 eggs over the course of five years, of which 81 reached the hatching stage. I think you can understand why most of us work with model systems rather than these difficult species. In my small zebrafish colony, I can get that many eggs in a week, and have well over a 90% hatching rate. Furthermore, it takes 8 months for a Nautilus egg to reach the hatching stage; that takes 2 days in a zebrafish. Ouch. The investigators have my sympathy. This is slow, difficult work.

The animals they were raising were Nautilus pompilius. If you need to be reminded of the differences between a nautiloid and a squid, here’s a handy reference diagram to their gross anatomy.

Comparative scheme of the body plans in nautiluses
and coleoids (e.g., squid), lateral view. The head complex is distinct from the visceral mass and mantle particularly in squid.
These schemes were described by the physiological orientation
defined by Hoyle (1886).

The affinities are clear. Both can be roughly divided into two body parts, a posterior visceral mass (that bag-like “head” of an octopus isn’t actually a head, it’s where it keeps its guts, and similarly the mass in these animals is within the shell or mantle) and an anterior head, with eyes and tentacles/arms and a collar and funnel. Nautiloids have a shell, simpler eyes, and more tentacles that lack suckers. These regions are set up early in development, and the purpose of this particular paper is to sort out what’s going on in just that head region.

Let’s get one confusion straightened out quickly. We usually think of the tentacle side as the front side, but embryologically, it’s the ventral side, and the visceral mass is dorsal. Just that swiveling around of perspective helps clarify the developmental process. In the picture below, you can see just how cute and adorable a baby nautilus is, but you can also see that the external morphology of the head complex is also the most complicated part of the animal.

Body plans viewed from the anterior, lateral, and posterior aspects, showing the large head complex with hood, eyes,
digital tentacles, and funnel. In contrast to Figure 1, these specimens are arranged by embryological orientation (see Fioroni,
1978), in which dorsal is towards the top of panel. Embryological dorsal 5 physiological posterior. Scale bars, 5 mm.

So now we roll back the clock and look at an earlier stage of development, at 3 months after fertilization. You have to imagine taking the animal above, putting one hand on top (dorsal), one hand on the bottom (ventral) and squishing it into a pancake-shaped disc. The picture in (a) below is looking down on the disc. The visceral mass, the mantle and shell field, is in the middle, and the more ventral head parts are now splayed out in concentric rings around the periphery.

A flat and less than 3-month-old embryo of Nautilus
without outer yolk sac, DAPI staining. The embryo
has a total of five lateral compartments, composed of nine bud-like structures, on each side (I-III, yellow; IV and V, red). (a)
dorsal, (b) lateral, and (c) posterior view. The cephalic compartment (eyes, hood, and mouth) is shown by green. apr, anterior
projection. Scale bars, 300 µm.

Now you might be able to see some of the similarities to other molluscs. Another point of interest most easily seen in (b) is that there are little buds that will eventually form the tentacles — and there are nine on each side, for a total of 18. The primitive number of tentacles in cephalopods is thought to be 10, and what we can see here is that embryonically, each tentacle (except one pair in Nautilus) are formed from a pair of buds that are thought to fuse later in development.

One of the common hallmarks of papers describing the development of a species is the establishment of a staging series. Because the number of specimens is so small in this case, though, the staging is understandably a bit rough, and there aren’t a lot of detailed steps described. The process of tentacle bud fusion, for instance, wasn’t seen, and has to be inferred from widely spaced samples. The photos below show embryos at 3, 4, and 6 months, at least, and can give you a sense of the changes going on. And (h) in particular is very pretty — a kind of short, stumpy version of the later nautiloid to emerge.

(click for larger image)

Embryos 3 months (a-c), 4 months (d-f), 6-months-old (g-i) of Nautilus pompilius without the outer yolk sac and shell,
DAPI staining (a-f). The cephalic compartment (the hood, mouth, and eyes) is shown by green. The green arrowheads indicate differentiated cephalic compartments. The compartments for hood (I-II) and digital tentacles (III-V) are indicated by yellow and red,
respectively. The whole or partial mantle was removed in all embryos to expose the visceral mass, collar, and funnel. These
embryos are arranged by the embryological orientation. Dorsal (a,d,g), lateral (b,e,h), and posterior views (c,f,i). Scale bars, 200 µm
(for a-c; d-f), 300 µm (for g-i).

Tentacle/arm development and evolution is confusing! The authors compared Nautilus with a coleoid cephalopod, Idiosepius paradoxus. Idiosepius is a little strange itself; it’s a highly specialized, tiny (less than a centimeter long) squid with reduced arms that at least is prolific and easily harvested, and does represent the coleoids in this study. Nautiloids add additional tentacles beyond the basal 10 that muddle up the issue enough, but we can still see a core similarity. Idiosepius is shown in (e-g), and they also have 9 buds on each side, grouped in pairs (except for one). The irregularities in the distribution suggest to me that when we someday get around to identifying the molecular/genetic patterning elements in the cephalopod, we aren’t going to find a simple pattern generator—I suspect we’re instead going to find hard-coded specific regulatory elements for each arm.

Development of digital tentacles in Nautilus pompilius or arms in Idiosepius paradoxus embryos. (a,b) Early
arrangement of compartment III (yellow) and IV 1 V (red) in 3-
month-old Nautilus embryos (a, no. 020401; b, no. 020412). The
two buds in compartment V fuse to one as a bipartite arm,
whereas bud IV is still single. The arm base-like projection of
thick tissue is present (an arrowhead in a). Asterisks indicate
newly differentiated tentacle buds. (c,d) Digital tentacles in the
3-month- (no. 020326) and 4-month-old (no. 020304b) embryos
viewed from the ventral aspect without outer yolk sac. Derivatives of the cephalic compartment, lateral compartments I-III,
and IV 1 V are indicated by green, yellow, and red, respectively.
The buccal tentacles (buc) appear in a 4-month-old embryo
(blue in d). A large projection is present toward the anterior
(see apr) to form a base of the medial side of the digital tentacles. Asterisks indicate small buds considered to be newly
formed tentacles. The digital tentacles arranged with similar
topological manner are indicated by dotted lines. apr, anterior
projection. The identity of compartment III is unclear. Hoechst
nuclear staining (a-d). (e-g) The scanning electron micrographs
show the early arrangement of arms and arm bases in I. paradoxus (e, stage 20, lateral view) compared with 3-month-old
embryos of N. pompilius (Fig. 3b). In Idiosepius (f, stage 21; g,
stage 24, posterior views), the two buds of a bipartite arm (I-III, yellow and V, red) fuse into one arm anlage except for IV
(red) in later stages. The arm bases begin to cover the eye
region to form a head-foot complex (arrowheads, red). The
developed cephalic bulges (ceb) are shown by closed dot lines
(in e, the head compartment is shown by green). fu, funnel.
Scale bars, 200 µm.

One other question people always ask — we’ve got mostly ten-armed squids, and eight-armed octopus. What happened? We don’t know. The paper briefly discusses the homologies between different species, but unfortunately the homologies in the octopods seem to be an open question, still, with different competing explanations. Obviously, what we need is more octopus embryology!

Comparative scheme and arm homologies in Nautilus and coleoid cephalopods. The plesiomorphic arrangement of
arms in coleoid cephalopods is considered to be 10 equal arms
on each side with the five bipartite arm condition (Roman
numerals) shown as a cephalopod phylotype composed of shared
characters. In Nautilus, tentacle anlagen IV and V have a
unique differentiation manner. Other species have a conserved
five arm formula (I-V) except lack or degeneration of arms
Octopoteuthis, asterisks indicate the
adult stage] and delay of differentiation timing in each arm
during the early ontogeny (arm IV in the embryos of Idiosepius
), arm III and V in the embryo and paralarvae of
Todarodes pacificus. In octopodiformes, there are some unsolved hypotheses
for arm homologies. In Octopus, the Arabic numerals indicate arm for-
mula adopted by each author.

At least when comparing nautiloids and coleoids, we aren’t completely lost. The similarities in the organization at that early pancake-like stage are easy to see, and are color coded in this diagram.

Note also how the arm buds are initially located posteriorly and the mouth anteriorly, like a more typical descendent of a bilaterian worm. Later in development, the arms migrate to wrap around the mouth, to produce the familiar central mouth surrounded by arms.

(click for larger image)

Comparative scheme in the early embryonic bodies between the
Nautilus and squid. Similar characters are represented by the same colors (e.g., the cephalic compartment and brain cords are
shown by orange and red, respectively). Bipartite arms are indicated by dotted lines.

Now to answer that question raised at the beginning of this article: what is the evolutionary relationship between the organization of a primitive gastropod and a cephalopod? The diagram below relates the parts of Patella, more familiarly known as a limpet, to Nautilus and Idiosepius, and also within the cephalopod group. What cephalopods did was modify the muscular gastropod foot into an array of tentacles, and then elaborate the set of organs above them (eyes, ganglia, funnel, etc.) into a head complex.

(click for larger image)

A suggested alternative scenario for the evolution of the cephalopod head complex by assembly of multiple molluscan
body parts. (a) Simplified schematic figures of conchiferan body-plans to show comparable and derived features. (Each color represents suggested homologous parts in a primitive form of the gastropod Patella, Nautilus, and a representative derived form of the coleoid Idiosepius. The homologies between gastropod “head” part
and cephalopod cephalic compartments are not certain (the orange color). The similar topographical arrangement of body plans is
emphasized for comparision. (b) Cephalopod body plans are characterized by elaboration of the head complex. The arm bases cover
the whole head part and funnel fused to them to originate a more integrated head-arm part in coleoids
(only the transient condition is represented in this figure). No such cover is identified in Nautilus embryo, although some arm-base-like structures develop during tentacle formation. Homology inferences are still controversial, but loss of arm II in octopodiformes (or vampyropods) composed of vampyroteuthids and octopods, might occur as seen in late embryos of a basal coleoid, vam-
pire squid. ab, arm base; olf, olfactory organ; rhi, rhi-

We can also sketch out the molluscan body plan and see the relationships in the phylotype of gastropods and cephalopods. The ironically amusing part is that what we’re calling the “head complex” of a squid is actually derived in part from the foot — it’s an amazing piece of of morphological juggling that actually makes a heck of a lot of sense from a developmental point of view.

(click for larger image)

A conchiferan and cephalopod phylotype is composed of shared morphological characters to show a drastic transition
from the benthic to nektonic forms. These figures indicate that the cephalopod tentacles and arms are derived from the ancestral
foot. Morphological novelty is seen in the appearance and elaboration of collar-funnel compartments as a part of the head complex
(black). The various organs including cephalic compartment (orange) and neural cords (red) are assembled to form an integrated
head complex along the anterior-posterior body axis.

This was an awesomely data-rich paper, and I haven’t even touched on some of the information on gene expression, so I’m just going to give you the very handy brief summary from the end of the paper.

  1. The tentacles/arms were derived from the foot
    region. Acquisition of a nektonic life from a
    benthic ancestor accelerated the loss of a creeping pedal sole and the development of tentacles
    from a freely mobile foot.

  2. In an ancestral cephalopod, the number of ten-
    tacles/arms was five pairs (or 10 pairs of bipartite
    arms). This means that the large number of tentacles in Nautilus results from secondary multiplication. Alternatively, the 10-arm condition of
    coleoids could be neotenous.

  3. The mouth surrounded by foot-derived tentacles/arms is unique among molluscs. This
    body plan was created by enwrapping the head
    part by epidermal tissues of pedal origin. During embryogenesis the pedal region shifts for-
    wards on the body surface, and eventually the
    “foot” is displaced anterior to the head.

  4. The rhinophores of Nautilus and olfactory
    organs of coleoids are presumably homologous,
    since they develop at similar posterior parts
    of the cephalic compartment as discussed earlier. Therefore, ancestral olfactory organs might
    have been present at an early stage of cephalopod evolution.

  5. An unfused hyponome as a primitive funnel
    might have arisen from the posterior part of the
    hood-collar compartment, which is possibly
    derived from an intermediate zone between the
    head-foot and visceral mass in the monoplacophoran ancestor. Alternatively, there
    is a possibility as suggested by Naef that
    a region of epipodial (dorsal) tentacles may differentiate into the hood-collar compartments.
    Then, with modifications for free-swimming
    behavior, the collar became distinct at the lateral region of the mantle and well-developed,
    including the funnel.

  6. Cephalopod brain masses centralized from the
    primitive tripartite neural-cord condition
    as seen in the embryonic nervous system of
    coleoids. Early cephalopods probably had a
    cord-like brain (not ganglia) as is found in the
    pedal cords of primitive gastropods.

  7. The optic lobes innervating cerebral eyes were
    derived from the cerebral cord, since this connection is found in the early embryos of Nautilus.

  8. The hood seems to be a secondarily-derived
    structure, convergent with the operculum of
    gastropods, which was coopted from two dorsal
    arm pairs together with ocular tissue and part
    of the collar/funnel complex.

  9. The ancestral function of transcription factor
    engrailed was conserved during the shell formation process, given that similar expression patterns were seen in Nautilus, Idiosepius, and
    other molluscan embryos. Further
    analysis is required; however the expression
    patterns may suggest a role for arms (pedal
    components), funnel, collar, and eyes in the evolution and development of molluscs.

The whole system is beautifully complicated, but what we see in this work is the power of developmental biology to illuminate the underlying, fundamental rules that define the evolution of organismal form.

Shigeno S, Sasaki T, Moritaki T, Kasugai T, Vecchione M, Agata K. (2007) Evolution of the cephalopod head complex by assembly of multiple molluscan body parts: Evidence from Nautilus embryonic development. J Morphol. [Epub ahead of print].


  1. #1 Torbjörn Larsson, OM
    July 30, 2007

    Awesome, even though the reasons for the gastronomic advantages of molluscs, if any, is still unexplained.

    Is that why tetrapods taste like chicken?

    Tetrapods outside most mammals, you mean?

    PZ sez: “Hold your molluscs at arm’s length, but your tetrapods closer – they aren’t as wet.” But he still uses the same platter, I bet.

    Say, why did tetrapods cross the beach?

  2. #2 Torbjörn Larsson, OM
    August 2, 2007


    “one can believe in god and accept evolution and reality and even be a scientist.”

    Absolutely. It is then one changes the science to pseudoscience, for example by inserting teleology, that it becomes problematic. I would also argue that a biologist that has two versions of evolution, one in the laboratory and one privately but publicized, is incompetent at what he does.

    From the biology universe, I would offer Francis Collins and Ken Miller as examples that support my position.

    Both of these are doing something else than you claimed earlier though. They are inserting teleology into a theory that doesn’t predict it from the current data.

    Don’t you think you are confusing your clearly stated position from earlier from what people actually do, and that you at times formulate the later? That would surely start a discussion on the irrationality and anti-scientific taint of the later position.

    Jim Roberts:

    I’ve just noticed that James Collins (Anonymous) posted the shorter version yesterday at Sandwalk

    And, IIRC, he also drove by Panda’s Thumb and posted the longer inconsistent one.

  3. #3 Torbjörn Larsson, OM
    August 4, 2007


    Since we’re already here, and looking back, we cannot possibly see anything other than a history of a universe which is suited to the emergence of human life.

    Hmm. That was the best formulation I’ve seen on how to avoid the usual mistake of confusing a priori probability with a posteriori outcome.

    And yes, Collins on morality is not agreeable with biologists who looks on altruism and kinship relations.

    I’ve just noticed that some people get triggered when they see that word capitalized (which I think is a bit pedantic).

    In my case it is also a reaction to english usage of capitalization – I’m used to it for names of persons and geography only, so it seems like a personalization and honorific. For example, not all christians believe in a personal god.

    But generally I think it is because individuals of different world views recognizes (or not) none, one, or several gods, all different from each other. So it seems funny and/or impolite to confuse them with each other, and to confuse them with the general concept.

    It must make analysis harder as well for those who use this convention. I don’t notice it any more, it is like those religious texts that can haphazardly be quoted in the middle of some discussions, they are just bypassed like getting used to the sounds of Tourette sufferers.

  4. #4 David Marjanovi?, OM
    October 28, 2007


    So 10 or 20 tentacles are the normal state for (…crown-group…) cephalopods… and Nautilus adds completely neomorphic tentacles… cool.

    James Collins above hasn’t noticed that molecular phylogenetics is neither done by comparing genome sizes nor by comparing gene numbers. It is done by selecting genes that evolve at appropriate speeds and looking for innovations in their sequences that are shared between species.