Cones, sticks and croissants: Early cephalopod evolution

Research on the earliest fossil cephalopods has been somewhat neglected in the past. In our study, we conducted the first large-scale phylogenetic analysis of Cambrian and Ordovician cephalopods, clarifying previously disputed relationships and providing a framework for systematic classification.
Published in Ecology & Evolution
Cones, sticks and croissants: Early cephalopod evolution

Read the paper

BioMed Central
BioMed Central BioMed Central

Early cephalopod evolution clarified through Bayesian phylogenetic inference - BMC Biology

Background Despite the excellent fossil record of cephalopods, their early evolution is poorly understood. Different, partly incompatible phylogenetic hypotheses have been proposed in the past, which reflected individual author’s opinions on the importance of certain characters but were not based on thorough cladistic analyses. At the same time, methods of phylogenetic inference have undergone substantial improvements. For fossil datasets, which typically only include morphological data, Bayesian inference and in particular the introduction of the fossilized birth-death model have opened new possibilities. Nevertheless, many tree topologies recovered from these new methods reflect large uncertainties, which have led to discussions on how to best summarize the information contained in the posterior set of trees. Results We present a large, newly compiled morphological character matrix of Cambrian and Ordovician cephalopods to conduct a comprehensive phylogenetic analysis and resolve existing controversies. Our results recover three major monophyletic groups, which correspond to the previously recognized Endoceratoidea, Multiceratoidea, and Orthoceratoidea, though comprising slightly different taxa. In addition, many Cambrian and Early Ordovician representatives of the Ellesmerocerida and Plectronocerida were recovered near the root. The Ellesmerocerida is para- and polyphyletic, with some of its members recovered among the Multiceratoidea and early Endoceratoidea. These relationships are robust against modifications of the dataset. While our trees initially seem to reflect large uncertainties, these are mainly a consequence of the way clade support is measured. We show that clade posterior probabilities and tree similarity metrics often underestimate congruence between trees, especially if wildcard taxa are involved. Conclusions Our results provide important insights into the earliest evolution of cephalopods and clarify evolutionary pathways. We provide a classification scheme that is based on a robust phylogenetic analysis. Moreover, we provide some general insights on the application of Bayesian phylogenetic inference on morphological datasets. We support earlier findings that quartet similarity metrics should be preferred over the Robinson-Foulds distance when higher-level phylogenetic relationships are of interest and propose that using a posteriori pruned maximum clade credibility trees help in assessing support for phylogenetic relationships among a set of relevant taxa, because they provide clade support values that better reflect the phylogenetic signal.

Most people know cephalopods such as squid, octopus and cuttlefish from various books, documentaries, movies, aquaria, snorkelling or perhaps even as food. They are particularly famous for their cognitive abilities compared to other invertebrates – recently, a cephalopod passed the so-called Stanford marshmallow experiment, an intelligence test originally designed for children. Almost equally famous are the fossils of cephalopods, where ammonites and belemnites represent probably some of the most iconic fossils of all time: everybody who went fossil collecting a few times has probably seen at least couple of them. It might thus be surprising that the earliest fossil cephalopods are still poorly understood – and despite the great amount of available material, research on them has been somewhat neglected. In our new study published in BMC Biology, we aim to change that a bit by providing a fresh perspective on the early evolution of cephalopods using quantitative methods. Hopefully this will inspire more research in the future!

Maybe the neglectance of research on early cephalopods was caused by the fossils looking relatively boring on first glance, especially when compared to the sometimes astonishingly beautiful ammonites. Rousseau H. Flower, one of the most prolific researchers on early cephalopods in the past century who single-handedly named more than 400 species, once wrote about endocerids (a group with very long slender straight shells) that “a collection of them seems about as fascinating as a collection of telegraph poles”. As with many things, the true beauty lies hidden beneath the surface. Cutting and polishing fossils of early cephalopods reveals an astonishing diversity of forms in their internal shell structures that essentially served as a buoyancy device. This is also what makes them so fascinating to me personally, because they show so many different morphologies that have no modern analogues.

Fig. 1. Examples of cephalopod fossils from the Ordovician (between 485 and 444 million years ago), with a human hand as scale. The top row demonstrates the size range in these animals (note that complete shells of some endocerids are estimated to have reached lengths of up to 6 m!), while the bottom row shows the diversity of internal structures. Top left: Charactoceras triagnulum. Top right: Microstomiceras holmi. Bottom left: Protocyclendoceras sp. Bottom right: Ordosoceras sphaeriforme. Photos: C. Klug, A. Pohle.

One important feature to distinguish between different fossil species of externally shelled cephalopods is the siphuncle. This is a tube that runs through their chambered shell and pumps out water to replace it with gas, essentially allowing the animal to become a little submarine, as seen in the only living cephalopod with an external shell, Nautilus. In Nautilus, the siphuncle has a relatively simple shape, as it basically consists of a thin straight tube. However, early cephalopods exhibit a great variety of siphuncle shapes, from thin to very broad and everything between siphuncles with very strongly expanded segments to concave segments and with various strange calcified structures that grew inside the siphuncle or even within the chambers (endosiphuncular and cameral deposits). Add to this the variety in external shell shape that basically ranges from long sticks to cones, croissants, drops, spirals and some others, and you get a large spectrum of different combinations of morphological structures (Fig. 1). And of course, these morphological characteristics had a very direct impact on the lives of these animals, as they directly controlled their swimming capabilities, e.g., the efficiency of buoyancy adjustments, manoeuvrability, swimming speed, etc. This means that there were probably a lot of different lifestyles and ecological roles among early cephalopods. It is hard to imagine how they might have looked like, but some of them may seem really strange to us (Fig. 2). 

Fig. 2. Cartoonish life reconstructions of early cephalopods. Note that colours and soft part anatomy are entirely speculative. Sizes not to scale. Drawing by E. Friesenbichler & A. Pohle.

The big question is here: how did these diverse forms evolve and how are all these groups related to each other? In the past, there have been multiple attempts to solve this, and different evolutionary scenarios have been proposed. The problem was that different researchers focussed on only few, but different characters to reconstruct their evolutionary trees and thus reached different, often contradicting results. In our study, we took a different approach by collecting a large amount of morphological data and analysing them quantitatively with state-of-the-art methods, i.e., Bayesian inference and the so-called Fossilized Birth-Death model. According to our results, early cephalopods diverged quite early into three major groups (Fig. 3): The Orthoceratoidea (mostly slender straight shells with calcified deposits within their chambers and siphuncles), the Multiceratoidea (very disparate shell shapes and thus hard to define, but generally ventrally enlarged muscle attachments and empty siphuncles) and the Endoceratoidea (sometimes very large straight to slightly curved shells, with characteristic conical deposits called endocones within their siphuncles).

Fig. 3. Simplified phylogeny of early cephalopods. Dashed lines represent paraphyletic groups. The presumable ancestors of coleoids (squid, octopus and cuttlefish) are found within the Orthoceratoidea, while it is still debated whether the ancestors of modern Nautilus are to be found within the Orthoceratoidea or Mutliceratoidea. Note that soft part anatomies are entirely speculative. Drawings by E. Friesenbichler, from Pohle et al. (2022).

In our paper, we go into detail about each of these groups and the methods. Here, I want to focus a bit more on the real heart of the paper: the morphological character matrix. Building a character matrix from the ground up was a challenging and long process that I started in 2017. Very roughly, my approach was as follows:

  1. Intensive literature research, reading through countless papers, some of them in Russian or Chinese (luckily, modern translation tools are available and work relatively well for this purpose).
  2. Defining characters and character states.
  3. Looking at specimens in Museum collections and the published literature to score characters and measure proportions.
  4. Realising that I need additional characters or change existing characters to capture some of the variation, thus revisiting all the previously scored species.
  5. Repeat 3. and 4. many (!) times.
  6. Check all species again, asking co-authors for their opinions, repeat 3. and 4. again where necessary, final revisions of characters.

As you can imagine, this was a very long process, but also very instructive at the same time and I was repeatedly perplexed by the strangeness of some of these extinct animals. During this entire process, I stumbled upon multiple cases where definitions of characters were problematic (perhaps no wonder, as they were not defined with phylogenetic analysis in mind) or where different names were applied for the same thing. Our paper is thus accompanied by an extensive supplementary material, which discusses every single character in detail and will hopefully prove helpful in future research.

I was lucky to be supported by an awesome, international team of co-authors (Fig. 4) from seven countries (Switzerland, Finland, Germany, United Kingdom, Czech Republic, Argentina and China), which helped a lot in above step 6, but also whenever I had questions or in preparation or revision of the manuscript, so a huge thanks goes to Björn, Rachel, Andy, Dave, Martina, Marcela, Xiang and Christian!

Fig. 4. Part of the author's team in Prague in 2018, with an original painting depicting life reconstructions of early cephalopods by the famous Czech palaeoartist Zdeněk Burian. Back row from left to right: Dr. Alexander Pohle, Prof. Christian Klug, Dr. Martina Aubrechtová, Dr. David Evans. Front row from left to right: Andy King, Vojtěch Turek, Martin Košťak. Photo: C. Klug.

Interested in the full article? Be sure to check it out here:

Early cephalopod evolution clarified through Bayesian phylogenetic inference

Please sign in or register for FREE

If you are a registered user on Research Communities by Springer Nature, please sign in