Monday, May 1, 2023

New Fossils of Coronodon 3: Implications for the early evolution and phylogeny of baleen whales

Welcome to part 3 of my blog post series on the toothed mysticete Coronodon - in part 1, I gave a bit of background on the history of research on Coronodon and other toothed mysticetes, as well as some of the new fossils of Coronodon havensteini. In part 2, I summarized the two new species, Coronodon planifrons and Coronodon newtonorum. In part 3, I dive into the broader evolutionary significance of Coronodon, whether or not it had baleen, its phylogenetic relationships, possible other relatives or members of the clade Coronodonidae from other ocean basins, and the early evolution of Neoceti. There's going to be a lot of different clade names in this - the most important one being Neoceti, which is the clade formed by baleen whales (Mysticeti) and toothed/echolocating whales (Odontoceti).

See Coronodon Part 1 here.

See Coronodon Part 2 here. 

And again, the paper can be viewed here.

Brief glossary:

Clade: a biologically 'real' group of organisms including a common ancestor and all of its descendants (monophyly means that a group of organisms is a clade, with one common ancestor)

Grade: a group of organisms excluding some ancestors or descendants; paraphyly means some members of a clade are excluded, polyphyly means that members of the group are distributed within several different clades and are unrelated (e.g. numerous ancestors are excluded)

Synapomorphy: some feature, typically an anatomical feature, that defines a particular clade

Sister taxon: a lineage that is the closest relative or branch to another

Character: some feature, usually an anatomical one, with various conditions that can be coded for in a cladistic analysis - some characters may end up being identified as synapomorphies

Operational Taxonomic Unit (OTU): The individual entry for a specimen or species (or occasionally group of species) in a cladistic data matrix

Matrix: All of the codings in a big spreadsheet, for all the characters and OTUs in a particular data set; the matrix is the data that is used to sort and identify the shortest, most parsimonious trees

Homoplasy: a character that evolves in parallel several times and may not be as informative as a derived character state that evolves once

Derived: a politically correct way of saying an 'advanced' character.

Plesiomorphic: the opposite of derived or synapomorphic - the 'primitive' character state

Odontoceti: 'toothed whales' or 'echolocating whales' - dolphins, sperm whales, beaked whales, and stem odontocetes like xenorophids, waipatiids, squalodontids, etc.

Mysticeti: all toothed and toothless 'baleen whales', regardless of possessing baleen.

Neoceti: the clade formed by the Odontoceti + Mysticeti. You can't be a neocete unless you're also either an odontocete or mysticete. 

Archaeocetes: all cetaceans outside of Neoceti, including pakicetids, ambulocetids, remingtonocetids, protocetids, basilosaurids, and probably Kekenodon.


Did Coronodon Possess Baleen?

In the original 2017 paper we mentioned a couple of palatal foramina in the supplementary description. If you recall from the first blog post I wrote on Coronodon, interpreting palatal foramina in baleen whale fossils used to mean that a mysticete likely had baleen: these are the nutrient foramina that supply blood vessels and nerves to the bed of epithelial tissue that the baleen grows from. All extant baleen whales have these foramina in the palate, along with long sulci (grooves). Some odontocetes have some tiny foramina, and a couple of archaeocetes have them on the premaxilla, but that's not exactly relevant since these structures in modern mysticetes are always on the maxilla - the primary bone of the rostrum.

Palatal foramina in the holotype specimen of Coronodon havensteini.

Upon closer examination, I noticed some additional foramina on the palate of Coronodon. I've only seen these in the Coronodon havensteini holotype, and they are curiously absent in the palates of juvenile specimens ChM PV 4745 and CCNHM 8722 - the latter of which is quite fractured, but actually has decent surficial preservation. Anyway, in the holotype, there are about nine foramina, six on the right and three on the left. All of these are greater than 1 millimeter in diameter; some are vertical, and none seem to be associated with the spongy bone frequently seen around the roots of modern odontocete and pinniped teeth probably driven by periodontal disease. Some of these, like foramina 1, 2, 6, and 9, all have sulci emanating from them. Others, like 3, 4, and 7, are vertical or nearly so.

In CT imaging, most of these foramina trend dorsally towards the tooth roots where the superior alveolar canal likely was - instead of the greater palatine canal. However, the scan is from a medical scanner so resolution is not fantastic, and the superior alveolar canal is damaged. Yet, we're more confident of a superior alveolar canal connection than the greater palatine canal. Why is this important? The superior alveolar canal is the canal in the maxilla that transmits all of the lateral palatal foramina and the blood vessels and nerves in extant mysticetes. This indicates that these structures are in fact homologous with extant baleen whales. 

The extensive, possibly rather disgusting gingiva of Coronodon. Is this the tissue that the foramina innervated? Illustration by yours truly.

So, does this mean Coronodon had baleen in addition to its teeth? Maybe, but we suspect not: these are quite tiny compared to skull size, being around the same absolute size as in the controversial Aetiocetus weltoni but in a skull twice as large - so proportionally, perhaps half the size - and there are fewer than in the toothed mysticete Aetiocetus weltoni. They are certainly much smaller (proportionally, and absolutely) and much fewer than in modern baleen whales. Could these instead feed thickened gingival tissue? We did suggest thickened gingiva may have been present in our 2017 paper. Marx et al. (2016) and Fordyce and Marx (2018) proposed thickened gingiva as an alternative to baleen - however, we note that it may not *actually be possible* to distinguish between the two hypotheses.

The Phylogenetic Position of Coronodon

Prior to our study, Coronodon had been included in a number of phylogenetic analyses as unpublished OTUs (operational taxonomic units: the individual branch coded as a discrete entity in a cladistic analysis). The first was Geisler and Sanders (2003), who coded a number of unnamed but phylogenetically informative Oligocene cetaceans into a relatively broad cladistic analysis aimed at Neoceti. This study coded both "Hoss", ChM PV 5720, and juvenile Coronodon havensteini specimen ChM PV 4745, into the matrix and recovered them as sister taxa and placed as the basal-most lineage within Mysticeti. Subsequent analysis using a similarly constructed matrix by Fitzgerald (2006) and (2010) for his studies of Australian mammalodontids Janjucetus and Mammalodon resulted in similar placement at the base of Neoceti. I produced a similar result in my analyses of mysticete relationships in my Ph.D. research on Eomysticetidae, with this "Charleston toothed mysticete clade" positioned at the base of Mysticeti (Boessenecker and Fordyce, 2015A, 2015B). 

 The phylogenetic analysis of mysticetes from Marx and Fordyce (2015).

A slightly different result was recovered in the analysis by Marx and Fordyce (2015), who found the North Pacific toothed mysticetes Aetiocetidae, and the Australian toothed mysticetes, Mammalodontidae, to form a sister taxon relationship and this Aetiocetidae + Mammalodontidae clade was positioned as the earliest branch within Mysticeti - followed by the Charleston toothed mysticete clade, Llanocetus, and then Eomysticetidae. Using the same matrix, Coronodon havensteini - recently named at this point - ended up in the basal-most lineage again in the phylogenetic analysis in the paper on Llanocetus by Fordyce and Marx (2018; note that the unpublished ChM PV 4745 and PV 5720 specimens were not included, there's a whole story there). In our 2017 paper reporting Coronodon, we achieved a similar result: Coronodon havensteini as the most basal lineage of mysticetes, but with Llanocetus as the last diverging toothed mysticete.

A more unusual solution was found by Lambert et al. (2017), who placed the newly named basilosaurid-like whale Mystacodon selenensis as the earliest diverging mysticete, followed by the Aetiocetidae + Mammalodontidae clade, then ChM PV 4745, then Eomysticetidae, and then ChM PV 5720 between Eomysticetidae and crown Mysticeti. In their 2019 followup monograph on Mystacodon, Muizon et al. (2019) found a far more typical result congruent with, for example, my Ph.D. matrix results: a basal Mammalodontidae, followed by the Charleston toothed mysticetes, then Llanocetus, Aetiocetidae, and Eomysticetidae.

 Phylogeny of toothed mysticetes and early Neoceti from the paper on Kekenodon onamata by Corrie and Fordyce (2022).

Much more recently, my Otago labmate Josh Corrie has published his redescription of the problematic archaeocete-like cetacean Kekenodon onamata last year - and quite frankly, I ought to write an entire blog post about that weird beast. Kekenodon has basilosaurid-like teeth along with a Neocete-like periotic that really resembles Coronodon of all things; the holotype specimen was collected from rocks correlative with the Kokoamu Greensand along the Waitaki River upstream from Kurow along the north Otago/south Canterbury border in NZ (Kurow was Fordyce's favorite ice cream pit stop after fieldwork, though depending upon the time, it also frequently turned into a fish 'n chips stop if it was a late field day). The holotype consists of a series of isolated teeth, partial periotic, bulla, frontal, and an atlas vertebra. Fordyce always considered Kekenodon, and more completely preserved specimen OU 22294 - to be transitional between archaeocetes and Neoceti, being somewhat more derived than Basilosauridae in many respects but basal to the odontocete-mysticete split. In the phylogenetic analysis published by Corrie and Fordyce (2022), there is a fairly shocking new result: Kekenodon falls outside Neoceti as predicted by RE Fordyce, but Coronodon, Mammalodontidae, Llanocetus, and Mystacodon ALL fall outside Neoceti as well - suggesting they're all archaeocetes! Only the Aetiocetidae plot out as toothed mysticetes. This result is quite provocative and suggests that there's quite a lot we may not yet understand - and historically have taken for granted - about the basal split between the odontocetes and mysticetes.

In order to approach the question of Coronodon's relationships, we greatly expanded my phylogenetic analysis from my Ph.D. research. I last published a version of this in Boessenecker and Fordyce (2017), my paper on Matapanui waihao, which was quickly adopted and added to by other authors (e.g. Peredo and Uhen, 2016; Peredo et al., 2018). I never stopped tinkering with the matrix - I added a bunch of new taxa into it and every summer would spend a week or so adding in newly named mysticetes, both stem, and crown. I also added more archaic odontocetes including Olympicetus, Ashleycetus, Agorophius, Ankylorhiza, Echovenator, and Xenorophus - odontocetes I am much more familiar with nowadays than during my Ph.D. We also added about 30 new characters, and with the addition of these odontocetes, Kekenodon, and newly described/studied mysticetes, we had an additional ~40 taxa in the analysis - for a total of 130 taxa coded for 392 characters. [I am tired y'all]. This is, to my knowledge, once again the largest analysis of mysticete relationships (both in terms of the number of taxa and number of morphological characters), just like my initial eomysticetid analysis a few years ago was, prior to other taxa being added to it.

Now, a few words for the uninitiated about cladistic analysis: this is the primary way in which we reconstruct evolutionary trees. For modern species this is done with DNA: each spot on the molecule has one of the four nucleotides (cytosine, guanine, adenine, thymine). The more positions sharing identical nucleotides in two different samples indicates they're more closely related. This usually results in many tens of thousands of 'characters' as we call them in morphological analyses. In morphological analyses, we use morphological (or anatomical) features, called characters - state 0 might be the primitive state, state 1 might be a derived state; multistate characters are also useful, with state 2, 3, and so on. An example might be "tooth count" and each state would be a range of tooth count numbers. Each time you go from one character state to another, that's called a 'step'. The computer program generates a large number of different tree shapes, and then sorts the trees based upon the number of steps - under the assumption that the fewest number of steps (character state changes) is most likely the one closest to the truth. I've just described the idea of parsimony. There are more complicated Bayesian analyses which are beyond my abilities to explain or execute, but this is fine for now. Essentially, we used nearly 400 different characters - some with two states, others multistate - to reconstruct evolutionary trees for about 130 different taxa. Lastly, there are at least two different cladistic camps in paleocetology. The first camp considers that only some characters are informative and that uninformative character data - whether they just show a bit of noise or fuzziness, or have some ecological signal that might override phylogenetic signal - should be excluded. The second camp believes that all cladistic analyses will be biased to a degree, and therefore we should attempt to minimize bias by including as many characters and taxa as possible. This is because if you are careful about what characters you pick and choose and which ones you exclude, you can steer the dataset towards delivering a preferred phylogenetic hypothesis, which is not great. "Cooking the books" has also absolutely happened in marine mammal paleontology, and I've witnessed it up close, so to speak. It's one reason why my Ph.D. matrix got so damned big in the first place: I wanted as objective a result as I could manage, and I included every known morphological character that had ever been used in the literature prior to my Ph.D. research. I've done a faithful job expanding the matrix, and the next task will be adding a host of new characters from Felix Marx's 2015 phylogeny, which we just didn't have time for with this study.

 Our enormous tree under equal weighting.

We ran our analysis in the program TNT, under equal weighting (every character is equal) and implied weighting (characters that are 'homoplastic' - evolving many times in parallel - are automatically downweighted by the computer). Each analysis gave us drastically different results. Under equal weighting (frequently the preferred method by many paleontologists, owing to guarded mistrust of the computer's ability to downweight characters and the 'black box' approach to it), the Llanocetidae (consisting of Llanocetus, Mystacodon, and a mandible from the Oligocene of NZ, ZMT 62) plot out at the base of Mysticeti; the Coronodonidae plots out as the next clade within Mysticeti, but in an unresolved position with Mammalodontidae. Various aetiocetids form the base of the next clade, though Aetiocetidae is rifted apart into a bunch of unresolved lineages. Aetiocetidae is a difficult family to code since few specimens have a rostrum, vertex, teeth, and earbones - most fossils have 2/4 preserved, so they tend not to stick together in analyses. The next clade are the Eomysticetidae and the rest of the toothless mysticetes. This tree is not terribly different from my Ph.D. results, the primary differences being the unresolved position of Coronodonidae and Mammalodontidae, addition of llanocetids, and the collapse of Aetiocetidae.

 Our enormous tree under implied weighting.

Under implied weighting, I'd normally hope for a bit more resolution on a relatively similar looking tree: maybe fixing the problems above. However, something very different happens under implied weighting (remember, this is the type of analysis that downweights characters that have evolved in parallel multiple times). Mystacodon and Coronodonidae are pulled outside Neoceti, with Coronodonidae being the sister taxon to the Odontoceti + Mysticeti clade (e.g. Neoceti) and Mystacodon and Kekenodon being successive sister 'stem' lineages to the Coronodonidae + Neoceti clade. Llanocetus forms a southern toothed mysticete clade with the Mammalodontidae, as the earliest diverging mysticete lineage, followed by the nearly monophyletic Aetiocetidae, and then the eomysticetids and rest of the true toothless mysticetes.

 Coronodon is much more similar to basilosaurid ancestors than the earliest diverging odontocetes, the Xenorophidae - which already have most of the hallmark features of Odontoceti.

Is Coronodon a Mysticete, or an Archaeocete?

One of the most serious problems affecting the early fossil record of Neoceti is that the earliest known fossil odontocetes have virtually all of the hallmark features of odontocetes: a large ascending process of the maxilla that overlies the frontal; premaxillary sac fossae; premaxillary foramina; deep antorbital notches, and a few others. This is certainly convenient for identifying fossil odontocetes, but it poses a larger headache: why are all of our early Oligocene odontocetes so derived? We don't really have that situation amongst mysticetes: we have, if anything, an embarassing number of transitional forms that are very, very archaeocete like (such as Coronodon and Mystacodon), some unusual forms that are slightly more derived (Llanocetus, Janjucetus, Mammalodon), and some flat-snouted taxa trending towards looking like a modern mysticete with teeth (Aetiocetidae). Why did mysticetes evolve so much more slowly? That is, after all, what is implied by these fossils: odontocetes evolved rapidly, achieved many of the key craniofacial adaptations for echolocation during a time we do not have fossils for (presumably in the late Eocene), and then radiated after.

Owing to this issue, Jonathan Geisler told me that every now and then, when he would run his matrix, the Charleston toothed mysticetes would pop out of Neoceti as a bit of a fluke - and he wondered if enough character data was amassed, and new specimens coded in - both of Coronodon and other toothed mysticetes - if the same issue might reoccur. Eventually it did, under implied weighting, anyway. So, what does this mean? Nothing, if you're in the camp that doesn't believe in implied weighting, or rather believes more strongly in results recovered under equal weights (for the record, I'm not in either camp).

Optimization of major characters that change at or near the base of Neoceti in our two analyses under equal weights (top cladogram) and implied weights (bottom cladogram). Black boxes show characters and states that support a particular clade under equal weighting, white boxes for implied weighting, and gray boxes for characters that support clades in both analyses. Many traditional characters supporting the monophyly of Neoceti may support a more inclusive clade if Coronodon happens to actually be an archaeocete!

Before we continue, let's review the synapomorphies of both Neoceti and Mysticeti. Some of the key features shared by virtually all Neoceti (with a couple exceptions) include 1) premaxilla contacts the frontal (in archaeocetes, the nasal and maxilla contact posteriorly, and the premaxilla terminates anterior to the frontal); 2) immobile or 'fixed' elbow joint, with flat facts on the distal humerus (all archaeocetes have a movable elbow joint); 3) nearly completely open and continuous mesorostral groove on the rostrum anterior to the bony nares (all archaeocetes have a firm medial suture between the premaxillae); 4) presence of an antorbital notch (no notch in archaeocetes); 5) a posterodorsally facing occipital shield with an apex (vertex) that is thrusted anteriorly to the level of the temporal fossa (in archaeocetes, the shield is vertical and posteriorly positioned); 6) three or more dorsal infraorbital foramina (basilosaurids typically have up to two foramina); 7) posterior process of periotic not exposed on lateral wall of skull - the 'amastoid' condition (basilosaurids have a long posterior process that is slightly exposed laterally); 8) monophyodont dentition (one set of teeth only; some basilosaurids replaced their teeth and were still diphyodont); 9) a ventral keel on the lumbar vertebrae, proposed recently by Davydenko et al. (2021; Basilosauridae have a rounded or flat ventral margin). There are other more poorly formulated synapomorphies that are not considered here because they either ignored many known odontocetes or mysticetes that lacked the feature (in other words, the proposed synapomorphies were present in only a subset of Neoceti).

We evaluated all of these, and a couple others, and found that owing to the placement of Coronodon outside of Neoceti in our implied weighting analysis, that many of these might actually define a group broader than Neoceti - which is a problem. Coronodon has all of these with the exception of two: 2), immobile elbow joint, is unknown in Coronodon, and 8) monophyodont dentition - we might need younger individuals to really tell. So, it's not because Coronodon LACKS these features - it's unknown, and we coded it as a '?' in our matrix. And honestly, that's not unusual: very, very few Oligocene cetaceans are coded for either character.

What about synapomorphies of Mysticeti (baleen whales)? Many proposed synapomorphies in older articles ignore already-established toothed mysticetes (e.g. Aetiocetidae) or have had to be chucked out thanks to the discovery of other toothed mysticetes more recently. Some of the solid synapomorphies proposed for Mysticeti include 1) antorbital process of the maxilla (a small ridge or flange of bone opposite the antorbital notch); 2) flattened rostrum that is <45 degree angle in cross section along at least 3/4 of the length of the maxilla; 3) wide basioccipital crests; 4) deep groove along premaxilla-maxilla suture; 5) triangular supraoccipital; 6) orbitotemporal crest extends anteriorly from parietals onto frontals; 7) swollen paroccipital process with a deep pit for the stylohyoid; 8) and an average of about 4.5-5 mesial cusps on the premolars. All of these features are present in Coronodon. As it happens, most of these synapomorphies have a shorter step length on the equal weighting tree with traditional relationships than on the unusual implied weighting tree. 

Owing to the more straightforward changes in these characters in the traditional tree, we cautiously suggested that Coronodon is probably within Mysticeti and Neoceti. We noted that 1) owing to low bootstrap support at the base of Neoceti and 2) owing to missing data in many early cetaceans at the archaeocete-neocete transition, the addition of just a couple of new fossils or specimens of existing taxa with additional character states might tip the balance one way or the other. To wrap this up - previous studies, including those published by me, have 1) either glossed over important details that are the exception to the rule and 2) certainly taken for granted whether or not some features are really synapomorphies of Mysticeti or not. We hope that these unusual results - like those by Corrie and Fordyce (2022) - result in more careful description of fossils and more careful character coding in future analyses. Our trees may look messier, but only because the truth is likely messier and more complicated than we've previously been able to appreciate.

Speciation and Recognition of Ancestry in Coronodon?

There are two fundamental 'styles' of evolutionary change at the most basic level: anagenesis and cladogenesis. Cladogenesis is the splitting of lineages - one species evolving into two; this is also called speciation. Anagenesis on the other hand is evolutionary change without speciation or the splitting of a lineage. Anagenetic evolution happens all the time, and should probably be viewed as the default mode of evolution outside of branching events. This is a bit different than gradualism v. punctuated equilibrium, which have to do more with how variable evolutionary rate is. Anagenesis remains a bit controversial, because many paleontologists do not 'believe' that ancestor-descendant relations can be identified in the fossil record. This attitude, in my opinion, is informed more by dogma than data. I won't bother going into the specifics, but the gist of it is that cladograms can only show relatedness - and not ancestor-descendant relationships.

Phylogeny of the Coronodonidae, under equal weighting (top) and implied weighting (bottom).

However, some cladograms actually *can* show ancestor-descendant relationships: the missing component is time, and there are certain cladistic tree shapes (topologies) that, if the individual lineages on the cladogram fit in the right time and place, could support ancestor-descendant relationships. In our new paper, we coded each specimen of Coronodon as a different OTU, for a total of six (four specimens of Coronodon havensteini and the Coronodon planifrons and Coronodon newtonorum type specimens) along with "Hoss", the unnamed coronodonid genus represented by ChM PV 5720. In both of our analyses, the two geochronologically younger species from the Chandler Bridge Formation - Coronodon planifrons and Coronodon newtonorum - were recovered as sister taxa. If the specimens looked identical, a sister taxon relationship might indicate (in the absence of other information) that they represented the same species. We know however that there are a number of features that distinguish these two species. The clade formed by these two species is not sister to the sample of Coronodon havensteini specimens, but is instead nested within it! In other words, some specimens of Coronodon havensteini are more similar to the younger species than to other specimens of Coronodon havensteini. Are these specimens not actually Coronodon havensteini? We're really talking about ChM PV 8722 - this specimen does not have any of the distinctive features of Coronodon planifrons or Coronodon newtonorum, and the frontals of this specimen most closely resemble the Coronodon havensteini holotype. Further, all four specimens are quite a bit older, and from the Ashley Formation, dating to 28-30 Ma, as opposed to the 23-24 Ma species from the Chandler Bridge Formation.  The evidence is a bit tenuous, but since we recovered this result under both methods, and the specimen ages line up, we interpreted this as evidence that the two late Oligocene species BOTH evolved from the single early Oligocene species Coronodon havensteini - so we seem to have evidence of ancestry, and a speciation event during the Oligocene. 

One of the isolated teeth of "Squalodon" gambierensis - admittedly a mystery. It doesn't have the same number of cusps, the root isthmus is shorter, and it is much smaller - but the enamel is similarly smooth and the remaining similarities are undeniable. What does the rest of this cetacean look like? Image from Pledge and Rothausen (1977).

On another level, it's interesting that we see a bit of diversity within the group locally - because we have ZERO evidence of Coronodon existing anywhere outside the Charleston Embayment. There are no teeth anywhere else on earth that are a good match. There are a few that somewhat resemble Coronodon but are much smaller - for example, "Squalodon" gambierensis from Australia and New Zealand. But these teeth are much smaller. Where might I predict actual teeth of Coronodon to show up for the first time outside Charleston? Two places: perhaps Onslow Beach in North Carolina, which has produced a number of xenorophids including the Albertocetus meffordorum holotype. Also a long shot, but perhaps Rupelian strata of Belgium - but very few cetacean specimens in general are known from these rocks, and if there's anything we know about Coronodon, is that it's quite rare here in Charleston.

Who else might belong to Coronodonidae?

Aside from aforementioned isolated teeth that resemble Coronodon but with unknown skull morphology, a couple of problematic Oligocene cetacean specimens are out there that have actually formed a clade with Coronodonidae - and *may* belong to the group, but we conservatively restricted the family to only include Coronodon spp. and its larger cousin "Hoss", known so far by ChM PV 5720. These two cetaceans are the recently named Borealodon osedax from the Pysht Formation of Washington, and Metasqualodon symmetricus, named in 1982 by Okazaki from the same locality as Yamatocetus.

 The holotype rostral fragment of Metasqualodon symmetricus, from Okazaki (1982). Top: ventral; middle: dorsal; bottom: lateral.

The utterly gorgeous teeth of the otherwise frustrating Metasqualodon symmetricus, from Okazaki (1982). Top: lingual. Bottom: labial.

Metasqualodon symmetricus is known from the right half of a rostrum from the Oligocene Ashiya Group of Japan. It has teeth that do not overlap like Coronodon, but have a similar number of cusps - and are remarkably symmetrical from the anterior (mesial) to posterior (distal) edges. The teeth are also similarly emergent from the jaw, and have a long isthmus but were probably double rooted given the sulcus in medial/lingual view. The rostrum looks similar  as well - likely triangular, with a deep groove between the premaxilla and maxilla; the maxilla is not flattened, but it is getting there - and the maxilla looks nearly identical to Coronodon in lateral view! In our analyses, Metasqualodon was one of the sister taxa to Coronodonidae, linked by two synapomorphies: having a thick edge of the maxilla and basal acessory cusps on the mesial side of the tooth that point mesially rather than apically. We will need better fossils of this poorly known species to evaluate whether or not it's a coronodonid or something else. Stay tuned =)

 A photo I took of the
Borealodon osedax holotype in January 2016, a few years before it was published. Specimen collected, acid prepared, and generously donated to the Smithsonian by our friend and colleague Jim Goedert.

Another taxon that is problematic, not because of its preservation but because of its penchant for phylogenetic 'flippancy' - it bounced around quite a bit during construction of this matrix. It is missing a lot of data, but has quite a few dorsal braincase features that can be coded, along with periotic, bulla, and a bunch of teeth. Sadly, when it was published, the lacrimal was mentioned nowhere in the description and Jim Goedert now fears the lacrimal to be lost. Regardless, this specimen looks about what you would predict the ancestor of all Aetiocetidae to look like, and also resembles Mammalodon and Janjucetus. As a matter of fact, at a conference presentation this was first reported as a northern hemisphere mammalodontid, and they stuck with the name Borealodon - even though the position was recovered as outside each clade. We consistently recovered Borealodon as the other sister taxon to Coronodonidae, but it shares really only one synapomorphy: possessing more than five accessory cusps on the cheek teeth.

Concluding Remarks

As expansive as this study is, it really only took about six months of furious writing to get done - but supported by five years of data collection on our part. We now have one of the best illustrated and described examples of a toothed mysticete, following earlier inspirations like the monographic descriptions of Mystacodon selenensis by Muizon et al. (2019) and Mammalodon colliveri by Fitzgerald (2010). We've reported new specimens of Coronodon havensteini, the first growth series for a toothed mysticete - illustrating that some features change ontogenetically. Great care must be taken when diagnosing new taxa that differences are not actually ontogenetic differences rather than taxonomic - no small problem, given that virtually all other toothed mysticetes are known from a single published specimen and little effort has been diverted towards assessing ontogenetic status. We reported two new species, Coronodon planifrons and Coronodon newtonorum, and identified ancestor-descendant relationships within the group. We also named the Coronodonidae to include some unnamed taxa from Charleston, like ChM PV 5720. Coronodon has some palatal foramina - but they're perhaps not large or numerous enough to indicate the presence of baleen. We confirmed that Coronodon has 12 rather than 11 mandibular teeth, demonstrating that it is one of the most plesiomorphic cetaceans to have evolved polydonty - and perhaps indicating a single origin for polydonty at the base of Neoceti. On that note, we found that perhaps our knowledge of the origin of Neoceti, and morphological character support for Neoceti - is not as well known or understood as previously assumed. Coronodon is probably a mysticete - but perhaps not. We've also identified a couple of unusual North Pacific toothed mysticetes that may be closely related to Coronodon, and possible members of the Coronodonidae.

What's next? Two big things for our team: first up, we barely touched on the feeding morphology of Coronodon. There's another paper in the works about this, with a looming deadline. In the distance, we will need to publish on and name "Hoss", the larger unnamed coronodonid represented by ChM PV 5720.


Davidow said...

The type species of Metasqualodon, M. harwoodi (Sanger, 1881), is known only from a single tooth unearthed from the Chattian-age Ettrick Formation of Wellington, southern Australia, and thus it's questionable whether Metasqualodon symmetricus is congeneric with M. harwoodi. If a re-assessment of Metasqualodon symmetricus based on comparisons with the figures of the holotype and paratypes of M. harwoodi in the paper by Pledge and Rothausen (1977) reveals a few morphological characters placing Metasqualodon symmetricus as distantly related to M. harwoodi, a new genus would be required for M. symmetricus.

As a side note, Fordyce (2004, 2006) regards "Squalodon" gambierensis as a probable relative of Kekenodon. With Ewan Fordyce having published his long-awaited redescription of Kekenodon last year, it'd be interesting to see if one of a few undescribed kekenodontid specimens is conspecific with "S." gambierensis.

Fordyce, R.E. 2004. The transition from Archaeoceti to Neoceti: Oligocene archaeocetes in the southwest Pacific. Journal of Vertebrate Paleontology 24 (Supplement to 3): 59A.

Fordyce, R.E. 2006. A southern perpective on cetacean evolution and zoogeography. pp. 755-778. In: J. Merrick, M. Archer, G. Hickey, and M. Lee (eds.), Evolution and Biogeography of Australasian Vertebrates. Auscipub, Oatlands.

Pledge, N.S., and Rothausen, K., 1977. Metasqualodon harwoodi (Sanger, 1881) - a redescription. Records of the South Australian Museum 17: 285-297.

Sanger, E.B., 1881. On a molar tooth of Zeuglodon from the Tertiary beds on the Murray River near Wellington, S.A. Proceedings of the Linnean Society of New South Wales 5:298-301

Petr said...

An absolutely fascinating read! Thank you!

Robert Boessenecker said...

@Davidow - all good points and accurate. I won't comment on "Squalodon" gambierensis other than to say that there is more research that awaits publication.

Agreed on Metasqualodon symmetricus not being congeneric with Metasqualodon harwoodi.