Mesonychians part III: Andrewsarchus and the triisodontines

Is there any particular reason the fossil bits of the mammal tree seems so much more in flux than those of the dinosaurian one? Comparative lack of attention to fossil mammals?

REVISION OF THE PROBLEMATIC EARLY PALEOCENE GENUS OXYCLAENUS
(MAMMALIA: OXYCLAENIDAE) AND A NEW SPECIES OF CARCINODON
THOMAS E. WILLIAMSON*,1 and THOMAS D. CARR2
http://www.bioone.org/doi/pdf/10.1671/0272-4634%282007%2927%5B973:ROTPE…

A cladogram showing ANdrewsarchus as closer to Oxyclaenus.

"Loren Coleman thinks that the Sarkastodon in this picture looks much like a famous depiction of the Cheshire cat."

Looks rather like a mantichore to me.

Correct reference for "Impact of increased character sampling on the phylogeny of Cetartiodactyla (Mammalia): combined analysis including fossils" is
Cladistics Volume 24, Issue 4, Date: August 2008, Pages: 397-442 Maureen A. O'Leary, John Gatesy

Andreas: Mammals are all alike, more or less, so mammalogists are reduced to squabbling over minutiae.

Ancodonta's (anthracothere-like artiodactyls, monophyletic or not) taxonomy is a mess. Each article sugested a different cladogram. There's a bunch of anthracothere-like beasts, Raoellids, Anthracotheriids, Cebochoerids, Choeropotamids, and etc, whose relationships are obscenely dubious.

Phylogenomic grouping of hippos and whales (unfortunately there's no living anthracothere to provide DNA) makes me suspect that hippos has a long ghost-linneage in Pre-Miocene Africa, deriving from some African Eocene ancodont (there was Anthracotheres in Fayum, it's a fact), or from some Southeast or Indopakistanese stock.

That cladogram is surprising. Why are aardvarks there at the bottom?

@#5: Oh! Snap!
Actually, you could say it's the other way around: Dinosaurs are all alike, so they're easy to place in a few categories.

About monospecific higher taxa. I first saw this argument in a paper by Lynne Parenti, basically saying that placing a species in a monospecific genus does not increase information. I have described two monospecific genera, and have presented my case in the reviewed literature. I think there are two reasonable critera. First, the species is so unique that it does not fit comfortably into any present genus. I think placing it in a separate, well diagnosed genus, is informative. Secondly, if there is a taxonomic scheme where the sister group of the species is a genus, then it needs to have generic status as well. To do otherwise would just destroy the beauty of the scheme. I realize that the diagnosis of a monospecific genus and that of its one included species are identical. But what if one discovers another species which belongs in that genus?

That cladogram is surprising. Why are aardvarks there at the bottom?

Presumably they're the outgroup. Odd choice, but safe, right?

I wouldn't take the outgroup area too seriously, nor some of the polytomies (e.g., the hippopotamid one, which would make Andrewsarchus a hippopotamid!). These are probably just artifacts of character and taxon selection, to be refined later.

Cetartiodactyla phylogenies are so diverse, but the most supported by DNA studies, shows a clade clustering hippo+whales (Cetancodonta or Whippomorpha), then united to Ruminantia, then to Suina, last to Tylopoda. Since there is a chaotic crowd of anthracothere-like families, European endemic families, oreodonts, entelodonts, diacodexeids, dichobunids, and et-et-et cetera, we dont know how to fit extinct groups into the phylogenomic scheme.

I think there are two reasonable critera. First, the species is so unique that it does not fit comfortably into any present genus. I think placing it in a separate, well diagnosed genus, is informative. Secondly, if there is a taxonomic scheme where the sister group of the species is a genus, then it needs to have generic status as well.

But in both these cases, what you're doing is merely a side-effect of the current binomial system, with a heaping helping of subjectivity (and I say this as someone who has himself established a monotypic genus). It's not certain that anyone else would see the new species as "so unique".

I realize that the diagnosis of a monospecific genus and that of its one included species are identical. But what if one discovers another species which belongs in that genus?

But because "genus" is an arbitrary, subjective level, the only thing that determines if someone describes a new species of that genus is whether or not they decide to include a new species in that genus.

As regards the affinities of arctocyonids, I'm getting the impression that arctocyonids as a whole are likely to be polyphyletic. Protungulatum and Oxyprimus are "arctocyonids", and they're not even definitely placentals.

I was speculating recently about whether a "condylarth" morphology could be ancestral for placentals as a whole. Afrotheria and Cetferungulata/Laurasiatheria (depending on preferred phylogeny) are two placental supergroups that seem likely to have "condylarth"-type origins. Any thoughts?

Arctocyonids looked like racoons or civets. Omnivore generalists, gave rise to carnivore and herbivore linneages. I think it's very plausible that those we call "arctocyonids" were a bunch of polyphyletic basal eutherians. Some of them were basal Afrotherians, another ones basal Laurasiatherians, maybe basal xenarthrans or even basalmost placentals (or pre-placentals). I suspect some of the South American ungulates or even pantodonts could be linked to xenarthrans.

Well, they *are* an outgroup, being afrotherians and all. It does seem an odd choice of outgroup, though. And why isn't Hyracotherium a perissodactyl?

Secondly, if there is a taxonomic scheme where the sister group of the species is a genus, then it needs to have generic status as well. To do otherwise would just destroy the beauty of the scheme.

All subjective, as Chris said. In that case, why erect a monotypic genus rather than including the species in the genus you say is its sister group? This would do no damage at all to the scheme. Now, the time you might want to avoid damage to previous classifications by erecting a monotypic genus is the more complex case of a single species being sister to a clade composed of two or more genera. Placing the new species in an existing genus would require submerging all but one of the genera in the clade. And we have emotional attachments to our favorite genera.

I haven't had time to read the O'Leary & Gatesy article yet, but glancing at it I was relieved to see that one of their other cladograms (consensus of top 300 trees...) DID show Hyracotherium as a Perissodactyl. (Whatever they took Hyracotherium to be: the species assigned to the genus seem to be a miscellany of primitive Perissodactyls, some closer to Equidae, some to Palaeotheriidae, some... It has even been proposed that one species should be given the generic name... Eohippus!) ... Even that tree, however, put Phenacodus as outside a Perissodactyl + Cetartiodactyl clade!

There's one thing about this "Mesonychia might be close to Cetacea after all" idea. When the ankle of the quadrupedal Cetacean Artiocetus was published (Gingerich et al., "Science" vol. 293 (2001), pp. 2239ff.; cf. also the additional illustrations in the accompanying commentary by Rose, same issue pp. 2216f) -- this was the bit of morphological evidence that convinced lots of people that the molecular crowd might be right after all and whales might NOT be derived Mesonychians -- the key fact was that the primitive whale astragalus looked MUCH more like a standard Artiodactyl double-pulley than the Mesonychian one did: the Mesonychian astragalus (pictured in Rose) is MUCH more primitive.

So. Suppose the cladogram reproduced above (from O'Leary and Gatesy) is right, and Mesonychians and Cetaceans do form a clade excluding Ruminants, Suina and Camels. Then Mesonychians, with primitive ankles, have an "extant phylogenetic bracket" of classic Artiodactyls (+ whales), all of which have a double-pulley. I don't think a reversal is likely-seeming (functionally the standard Artiodactyl ankle seems like something in the "once you've tried it you'll never go back to Brand X" category), but the alternative is that the Artiodactyl ankle -- long held up as an example of undoubted synapomorphy uniting the clade -- was evolved independently several times.

Mesonychians may have been weird, but they seem to have even weirder implications!

Fossil mammals are based mainly on teeth, and complex teeth can tell you a lot about an animal's biology (size, ontogenetic age, historically adapted diet, actual diet based on wear, yada yada) but clearly fail to provide reliable phylogenetic characters in a lot of cases. Slightly better than snake vertebrae in that respect (the latter don't even tell you about diet). The necessity of precise occlusion (resp., articulation of verts) probably has a lot to do with the amount of homoplasy.

Also, I was going to say roughly what John Harshman said in response to Jim Thomerson. (I voted to refer 'Brachyaspis atriceps' to Hoplocephalus, but Scott Keogh thought that someone would be certain to split it at the earliest opportunity, so it might as well be us... and I guess there would be many such cases.) Of course a 'monotypic' named clade whose sister taxon comprises one or more genera may turn out to have (or have had) more than one species in it, and then any 'taxonomic efficiency' advantage of sinking it would vanish (among my thoughts on reading Kluge's 1993 Erycinae paper).

Andrewsarchus was clearly more unusual than I thought, however many of the skull features suggest to me that Andrewsarchus was not the mighty top predator it is put out to be.(walking with beasts lied!) The unusually long jaws suggest that Andrewsarchus needed to reach towards its prey. Other Mesonychids like Sarkastodon had short powerful jaws for attacking close contact prey of large sizes. Andrewsarchus's jaws give the impression of being able to withstand only verticle bite forces rather than the sideways struggles of large and powerful prey. Aware of this it seems to me that Andrewsarchus was a predator of small 30cm to 1m herbivores and predators close to ground level. As you may have noticed; the incisors and canines of Andrewsarchus are positioned right at the front of the jaw like some sort of pick. Therefore when hunting Andrewsarchus would use its low set eyes to scan the ground for prey. It also could have used the nostrils set at the very tip of its snout to trace any scent trails (however I am not very farmiliar with Andrewsarchus olfactory anatomy.). When a creature like a "Condylarth" for example is spotted Andrewsarchus would have pursued the prey and plucked it from the ground using its inscisor and canine teeth like a pair of tongs. Then the morsel was passed to the back of the mouth still alive where it would be crunched like a nut in the Enteledont like molars.

I hope you like my theory on Andrewsarchus feeding biology.
If you can please tell me what you think of my theory by replying through the comments (if you email me you will probably be sending it to my step dad). By the way, i'm 13.

Andrewsarchus crassum

A. crassus of course.

Loren Coleman thinks that the Sarkastodon in this picture looks much like a famous depiction of the Cheshire cat.

In any case, it looks very human...

Is there any particular reason the fossil bits of the mammal tree seems so much more in flux than those of the dinosaurian one? Comparative lack of attention to fossil mammals?

1) Yes. Most Paleogene mammals have never seen a phylogenetic analysis from the inside.
2) Careful: it's just Placentalia or a slightly larger group, not Mammalia as a whole.
3) Similarly, the tree of Neornithes is very much in flux, even (for much of it) if only molecular data are considered.

That cladogram is surprising. Why are aardvarks there at the bottom?

As part of the outgroup, which is much, but much, too sparsely sampled. This likely leads to long-branch attraction and similar phenomena all over the place.

Secondly, if there is a taxonomic scheme where the sister group of the species is a genus, then it needs to have generic status as well. To do otherwise would just destroy the beauty of the scheme.

In the grand scheme of things, the beauty of that system is the problem...

I realize that the diagnosis of a monospecific genus and that of its one included species are identical. But what if one discovers another species which belongs in that genus?

Whether that species belongs into that genus is not a fact (something that can be discovered), it's a decision.

If someone decides that another species should be put into the same genus, then they have to diagnose both species and the genus all anew. Happens all the time.

Afrotheria and Cetferungulata/Laurasiatheria (depending on preferred phylogeny) are two placental supergroups that seem likely to have "condylarth"-type origins.

Both of them have "insectivore" clades near their bases, so the classic hypothesis of an "insectivore"-type origin for Placentalia is at least as parsimonious as a "condylarth"-type one; but once we figure out where all the fossils fit, all that could of course change. Weirder things have happened.

And why isn't Hyracotherium a perissodactyl?

Because of the sparse taxon sampling in the outgroup.

Note how Carnivora, Pholidota, Chiroptera and Eulipotyphla are completely absent from the taxon sample. Remember their positions in the molecular trees?

The necessity of precise occlusion (resp., articulation of verts) probably has a lot to do with the amount of homoplasy.

Yeah. It seems that a lot of tooth characters are correlated to each other. Witness the Standard Herbivorous Archosauromorph Tooth™, which evolved something like 5 times alone, and its almost-occurrence in pareiasaurs and Iguana...

Andrewsarchus's jaws give the impression of being able to withstand only verticle bite forces rather than the sideways struggles of large and powerful prey.

Why? As far as I can tell, jaws built for vertical forces alone are tall, like those of most big theropods. Those of Andrewsarchus are almost tube-shaped, so they look resistant to torsion like a gharial's or spinosaur's.

Besides, the whale- and entelodont-like premolars suggest that Andrewsarchus did a lot of flesh cutting.

my theory

It's a speculation. A theory is something much, much, much bigger, like the theory of evolution (which explains all of biodiversity in space and time in one fell swoop) or the theories of special and general relativity (which explain... basically... everything from the size of a planet's orbit upward) or the theory of quantum electrodynamics (which basically explains all of matter at once). For the most part, the definitions given here are right (though that of "law" is completely wrong).

Speaking of speculation, can anyone speculate on why a single outgroup was chosen, and even odder why that single outgroup is such a highly apomorphic one as an aardvark (and one which makes pretty much all dental characters unrooted -- sorry, pun unintentional -- at that)? I don't have the paper, or I'd look it up myself.

Other Mesonychids like Sarkastodon

Sarkastodon is of course a creodont and not a mesonychian.

Have there been any discussions or analyses of this species aside from the original description in 1938? Since Granger compared Sarkastodon to the cat-like Patriofelis, I'm completely baffled why at least a couple illustrations portray S. as a brown bear with a long tail. I seem to remember this "reconstruction" in the oddity riddled Macmillan Illustrated Encyclopedia of Dinosaurs and Prehistoric Animals - famous for the 20 foot long fairy armadillo standing in for Peltephilus - so perhaps it is to blame.

Sarkastodon, BTW, would mean something like "flesh-tearing tooth". A sarcastic comment is etymologically a flesh-tearing one (cf. sarc- ( = sark-) meaning "flesh" in words like 'sarcophagous' and 'sarcoma').

Well, I've given the O'Leary & Gatesy paper a first skim. (No, I'm not a particularly fast reader: prose, as opposed to tabular data, occurs on only about half the pages.) I'm trying to think of intelligent comments to make, but meanwhile:

NOMENCLATURAL ODDITY: Basilosaurus isn't the only mammalian -saurus: there's an Anthracothere called Lybicosaurus.

GENERAL RANT: How in the HELL can you combine molecular data (O'L and G have 40,000 molecular "characters") with morphological (they have a bit over 600 morrph. traits) SENSIBLY? How can you weight the two kinds of character in a principled way? Ideally, I'd think, you'd want to have a morphological character count equivalently to the number of DNA bases that have to change in the genome in order to produce it, but I thinbk that (with maybe a few exceptions?) nobody has a CLUE as to how to BEGIN to ESTIMATE that!

O.k., I'll try to be serious now.

FURTHER WEIRDNESS. O'L and G's best trees make Protungulatum (a VERY primitive Paleocene "condylarth" sometimes classified as an Arctocyonid) clade with (two species of) Diacodexis (traditionally considered a basal Artiodactyl). I suspect this is an artifact. They do claim to have postcranial data on Protungulatum, but I'll bet its ankle bones aren't known! And, if you don't look at its Artiodactyl-like hind legs, Diacodexis is VERY primitive. (In their section on "support," p. 415, O'L and G note that the Prot-Diac association disappears in only slightly less parsimonious trees.) ... I am moved to note David Marjanovic's remark (comment 19) on scanty sampling of primitive types at or just outside the Cetartiodactyl complex: any study that lets Protungulatum be an Artiodactyl doesn't have enough "Arctocyonids" in it (i.e.: I think if there were more primitive types in the study, Prot would have clustered with them and so been kept out of Catart).

RANT: Arctocyonidae are the new "Insectivora." The "Family" seems to be a dumping ground for (dentally primitive?) "Condylarths" that don't fit into any more derived group. Case in point: the supposed Triisodontine "subfamily" of the Arctocyonidae. O'L and G include two of them, and one (Eoconodon) winds up at the base of the Cetacean/Mesonychian clade and the other (our hero, Andresarchus) winds up next to the Entelodonts. (Next to the Entelodonts?) ... Still on the topic of Arctocyonids, I note that everybody's favorite Paleocene pseudo-raccoon(*), Chriacus, isn't included: pity, since it has been proposed as being near the origins of the Artiodactyla. (*) It MUST have been raccoon-like: why else would artists reconstruct it with rings on its tail?

OH, HUM...: Molecules and morphology are still discordant. When O'L and G used only the morphological data, they got a tree in which Cetacea and Mesonychia clade together, but the Cetomesonychia(*) show up as a sister group to Artiodactyls rather than nested among them. (*) I just made that up: I don't think it's a name used in the literature.

DISCLAIMER: This is a rich paper, and the amount of work that went into it staggers the imagination. My flippant tone and critical comments should not be taken as implying disrespect!

FASCINATING COMMENT NEAR THE END (p.420): Suppose they're right and Cetomesonychians are nested deep within a larger Cetartiodactyla. Some of the Mesonychians (or, at least, more or less questionable primitive possible Mesonychians) are Paleocene forms. Suggesting that Cetartiodactyla must have originated ... oh, in the earliest Paleocene (or even, awful thought) before. Suggesting that the people who have looked for possible ancestors of Artiodactyla in the late Paleocene may have been looking in the wrong strata. I think THAT is a fascinating and important idea.

OH, ABOUT THE ANKLE BONES (p. 417): O'L and G have two "characters" representing the classic Artiodactyl double-pulley astragalus. They find the Artiodactyl ankle to be a synapomorphy of Cetartiodactyla... but with reversals in Mesonychia and some (not Artiocetus) basal Cetacea.

And, I'm REALLY grateful to Darren for drawing my attention to the paper!

I just made that up: I don't think it's a name used in the literature.

The name generally used for a Cetacea + Mesonychia clade when recognised has been Cete.

Speaking of speculation, can anyone speculate on why a single outgroup was chosen,

Oops.

Thanks for making me look at the tree again. Not having the paper, I had written my previous comment under the assumption that everything from Orycteropus to Rhinocerotidae is the outgroup, so that only Cetartiodactyla and Mesonychia are the ingroup. But there's no way Protungulatum could be part of the ingroup in such a scheme.

So, they really used just one outgroup, and...

and even odder why that single outgroup is such a highly apomorphic one as an aardvark (and one which makes pretty much all dental characters unrooted -- sorry, pun unintentional -- at that)?

It's by no means necessary that the outgroup lack missing data. PAUP* makes unrooted trees (that's possible because an apomorphy in one direction is the opposite apomorphy in the other direction), and then it displays the tree in such a way that the outgroup is shown on top.

The aardvark was probably chosen for being an afrotherian and a "pseudungulate". But in any case, it's so far away from the ingroup (in terms of branch length if nothing else) that it cries for long-branch attraction.

NOMENCLATURAL ODDITY: Basilosaurus isn't the only mammalian -saurus: there's an Anthracothere called Lybicosaurus.

Libycosaurus. O'Leary & Gatesy get it right. Because it's Libya. Not "Lybia". Libya.

Sorry, pet peeve of mine.

GENERAL RANT: How in the HELL can you combine molecular data (O'L and G have 40,000 molecular "characters") with morphological (they have a bit over 600 morrph. traits) SENSIBLY? How can you weight the two kinds of character in a principled way?

Not at all. All you can do is try a few weight ratios and see what happens.

How many of the 40,000 molecular characters are parsimony-informative...?

I must say the enormous morphological matrix is awe-inspiring. I hope the number of typos in it isn't too big.

Spammer alert: the preceding 3 comments posted on August 15, between 12.39 and 12.53 pm.

Could Scienceblogs provide a more efficient way for us to report spam? I could of course e-mail Darren with a list of links to spam comments, but if there is a way to flag comments for spamming, with eventual banning once a critical mass of reports has been exceeded, it might help in wiping out this scourge.

@ James O'Shea- I like your theory. And yes it is acceptable to use the word theory in a casual scientific context without it meaning the strict definition of whatever Wikipedia has down for what a scientific theory is. Perhaps "hypothesis" would be the term you would use if this were a scientific publication, although if it was, yes, it probably be something framed as "speculation." But this isn't a peer reviewed blog in that sense :)

Back to your idea, I want to respond to John Scanlon#42, who actually posted on Mesonychians II- after googling a picture of a thylacine skull, I can see the resemblance. Would that be why the zygomatic arch is so large, to allow sufficient freedom of movement of the lower jaw within the joint?

More importantly, why does the thylacine have such a large gape? Anything known what role this plays (played) in its biology/feeding ecology?

Perhaps Andrewsarchus could have snapped up the small gallopers of the time- early horses, etc (?)- pursuing them using a similar mode of locomotion (a nod to Mesonychians IV where Darren points out the mesonychids wouldn't have sprinted with a springy back like Carnivora), much like you see in the pictures of Diatryma terrorizing small ungulates?

Still, we have some interesting comparisons here- everything from Spinosaurus and Baryonyx to thylacines. A common thread would be interesting, although perhaps one or all of these comparisons is false homology (analogy, whatever)

why does the thylacine have such a large gape?

All the better to eat you with, my dear.

In other words, no idea... except that wide gapes are common in dasyuroids, and yawning is apparently used frequently as a threat display in a number of dasyurids species as well as Thylacinus. The Tassie Devil Sarcophilus is known for a lot of head-biting during fights and mating (with unfortunate recent consequences in the form of a transmissible cancer), which is presumably one fairly common form of selection for increased gape in this lineage. Possibly the main one; I'm not sure it would be much of an advantage in capturing, killing or ingesting prey.

Christopher Taylor (#25):
I knew "Cete" but wanted to avoid it because most of the people who have used it seem to have believed that it was a clade outside Artiodactyla... Which isn't, I suppose, a good reason not to use the same name for a clade with the same contents thought of as maybe outside and maybe inside Artiodactyla: I just didn't want to give even a hint of a suggestion that I was committed to the first hypothesis. (That and I just enjoy making up words.)

David Marjanovic (#26):
(1) apologies about misspelling L. (I **could** claim I was thinking of the ancient countries of western Anatolia, Lydia and Lycia, but actually I was just sloppy.)
(2) How many of the 40,000 molecular characters are informative? My copy of the paper is in my office, but I think it says about a quarter. ... The first few pages review previous studies: a couple have included about as many taxa, but, yes, the number of morphological characters is much higher than in previous studies! (Which makes it all the more disappointing that the morphological information STILL doesn't put Cetacea-- or Cete-- inside Artiodactyla where themolecular stuff says it belongs. My ***guess*** is that this is because autapomorphies of Cetacea mask its affinities, and that what we need is LOTS of information on LOTS of basal, probably Paleocene, forms so as to start attracting the more derived Cete into the right clusters. Which will take a while.)

David Marjanovic--
Re your question about how many of the characters were parsimony-infomative. O'L and G, p. 407: "The analysis showed that of the 41,563 characters in the matrix, 10,713 (24.48%) were parsimony informative[]. Five hundred and sixty-six (5.56%) of the informative characters were from the phenotypic partitions (e.g. osteology, soft tissue, behavior) and 9,607 were from molecular partitions."
And: "Over 90.68% of the characters that fossilized were parsimony informative, but only 23.47% of the molecular characters were informative." Which, since morphological characters are defined by investigators looking for phylosgenetic evidence and molecular ones come willy-nilly from sequences, may not be too surprising.

AD:

And yes it is acceptable to use the word theory in a casual scientific context without it meaning the strict definition

If we lived in a perfect world, maybe. Sadly however, creationists of all stripes constantly - and often deliberately - mis-use the word theory (as in: 'Evolution is just a theory'), which all too easily creates the impression among laypeople that evolution is only some vague, off-the-cuff speculation. Therefore, I'd strongly recommend that 'Thou shalt not use the word theory in vain'.

Perhaps "hypothesis" would be the term you would use if this were a scientific publication, although if it was, yes, it probably be something framed as "speculation."

Just to be clear; hypothesis and speculation are not synonymous terms either (as explained in David's link).

Well, reasonable people can disagree. Also, I know hypothesis and speculation aren't synonymous, that's why I made the distinction.

Let's just use "notion". "Hypothesis" should be restricted to formally (and not only trivially) testable notions, and "theory" to something that generates not just notions, but actual hypotheses. AAT, for example, doesn't qualify because the notions it generates are trivially falsified -- no new data needed. "String theory" doesn't because its notions aren't even formally expressible, never mind testable.

What about Andrewsarchus being semi-aquatic fish eater, a sort of a cross between kodiak bear and giant otter?

Would nicely fit aquatic hippo-whale clade, weak jaws and extreme size. Somehow, fish-eaters tend to be bigger than terrestrial predators.

There's a very different reconstruction of Andrewsarchus over here:

http://dibgd.deviantart.com/art/Andrewsarchus-revisited-135642493

Very interesting indeed.

I agree with Jerzy, long live the otterpotamus!

Ps - that sarkastodon is giving me the stinkeye!