Meet the outgroup sister to all eutherian mammals: Monodelphis, and its ALIVE!

Updated August 10, 2022
with the addition of taxa from 764 to 2124.

Okay, I know this comes as no surprise…
a mouse-like marsupial at the origin of placental (eutherian) mammals. That’s old news.

Figure 1. The marsupial, Monodelphis domestica, nests as a sister to Eomaia, the oldest known placental.

But this EXTANT mouse-like marsupial
has traits found in basal eutherian mammals like Eomaia… not found in other tested marsupials… which is exactly the way it should be. We’re always looking for a gradual accumulation of traits in the large reptile tree (LRT, 764 taxa (then, 2124 now), subset Fig. 3).

So far among marsupials
Didelphis, the Virginia opossum nests basal to all other tested marsupials. Three marsupials nest together in a clade, Thylacinus, Dromiciops and Macropus. One marsupial, Monodelphis, nests between these taxa and  basal placentals. So marsupials (Metatheria) are forming a grade, not a clade.

A few more mammals added to the LRT

The last few days
have been spent out of town on family business and updating the LRT (large reptile tree, now 761 taxa) with a few more mammals (Fig. 1). Sorry to be gone. Now I’m back.

Figure 1. Subset of the large reptile tree focusing on mammals and their immediate ancestors. Here a few more taxa have been added, many from the Early and Late Cretaceous. Vincelestes, Repenomamus, Ernanodon, Asioryctes, Liaoconodon, Proconsul and Jeholodens are among them. There are several errors here revealed by the addition of more taxa. See figure 2.

Some new basal taxa have been added
and none of them have upset the basic tree topology that just keeps growing. Changes in red due to additional taxa.

1. Vincelestes – a basal carnivore with a short face and giant canines, Early Cretaceous, 125 mya. Now nests with Thylacosmilus in the creodont marsupials.
2. Repenomamus – large enough to eat baby dinosaurs, this carnivore sister with smaller canines nested basal to bats, flying lemurs, pangolins and primates, but definitely on its own branch yet to be filled, Early Cretaceous, 125 mya. Now nests with Gobiconodon and Liaoconodon in the Trithelodontidae.
3. Ernanodon – close to lemurs (primates), but basal to pangolins, but again on its own branch yet to be filled, Paleocene, 61 mya. Now nests with Hyaenodon in the creodon marsupials. 
4. Asioryctes – a tiny basal insectivore with a full arcade teeth, Middle Late Cretaceous, 85 mya. Now nests with Perameles, a basal omnivorous marsupial.
5. Liaoconodon – a sister to Asioryctes, Early Cretaceous, 120 mya. Now nests with Repenomamus in the Trithelodontidae.
6. Jeholodens – a basal multituberculate, close to rodents, Early Cretaceous 125 mya. Now nests with Spinolestes in the Trithelodontidae.
7. Proconsul – a primate nesting between lemurs and humans,

Figure 2. Subset of the LRT focusing on the Kynodontia and Mammalia. Non-eutherian taxa in red were tested in the LRT but not included because they reduce resolution. Eutherian taxa in red include a basal pangolin and derived xenarthran, clades that extend beyond the bottom of this graphic. The pink clade proximal to mammals was considered mammalian by Lautenschlager et al. due to a convergent mammalian-type jaw joint.

Remember 
each fossil is likely not found at the origin of each genus, but probably later, at its fullest and most populous time.

These nestings indicate
that the radiation of basal carnivorous and insectivorous mammals occurred much earlier than 65 million years ago, prior to 125 mya. The large herbivores listed above (Fig. 1) likely appeared later.

The multituberculate/triconodont issue.
Traditional paleontologists consider (eu)triconodonts and multituberculates basal mammals (= Mammaliaformes) that radiated prior to the origin of the Monotremes, Metatherians and Eutherians. By contrast, and using far fewer dental characters, the large reptile tree found these two clades were sisters to rodents (multituberculates) or nested at various places elsewhere within the Mammalia. Basal mammals from the Early Cretaceous listed above (Fig. 1) typically (but see Sinocondon) have a full arcade of teeth and other traits, not found in multituberculates and triconodonts. On the other hand, multituberculates share a long list of traits with rodents and kin, from their skulls to their toes. That should not happen if they were indeed non-mammals mammaliaformes.

Why don’t we see rodents, rabbits and other taxa earlier than we do?
Perhaps those taxa did not inhabit substrates that encourage fossilization. This tree does indeed upset current thinking regarding the emergence of certain clades and interrelationships. I don’t want it to. It simply does this on its own, as it has done identifying pterosaurs within a new clade of lizards, the Tritosauria, etc. etc.

Epipubic bones
In the large reptile tree epipubic bones continue in several Cretaceous clades and disappeared several times by convergence.

All sister taxa here
look similar to one another and share long lists of traits that lump them together. Short lists split sisters apart. I’ll start listing traits and showing reconstructions over the next few weeks and you’ll see where the data have produced the above cladogram (Fig. 1).

 

Two Cretaceous mammals join the LRT: Asioryctes and Repenomamus

This post is updated. See August 11, 2016 for new data on Repenomamus. 

As in traditional reptiles
I’m seeking basal relationships of basal taxa in mammals. For those you have to go to the Middle and Early Cretaceous, not the Paleocene.

Asioryctes (mid-Cretaceous, 85 mya, Kielen-Jaworowska 1975, 1984, Fig. 1) was a basal eutherian mammal, not far from Eomaia (Ji et al. 2002), the basalmost eutherian. In the large reptile tree (LRT, subset Fig. 5) Asioryctes nested at the base of Tupaia + Glires (including Multituberculates). It retained most of the original arcade of teeth, but with smaller canines. The overbite is a trait of this clade.

Fig. 1. Asiorryctes skull on the tip of someone’s finger. Overlays include a traditional tracing that does not have the same proportions, a stretched out version of the same and a DGS tracing from which data was scored. Note the reduction in the canines (yellow), which are lost in this herbivorous clade.

Figure 2. Asioryctes in life. Post crania is imagined. Image out of “From the Beginning – the Story of Human Evolution”.

The second Cretaceous mammal
is the predator, Repenomamus (Li et al. 2000, Hu et al. 2005; Early Cretaceous, Figs. 3, 4). One specimen of R. robustus was the size of the opossum, Didelphis, and included digested baby dinosaur bones, R. giganticus was half again as large, one of the largest known mammals of the dinosaur era. Despite its predatory nature, note the reduction in the canine teeth.

Figure 3. Repenomamus giganticus in situ. Here the overlooked coracoids are identified.

Repenomamus
is traditionally considered a non-therian gobiconodontid mammal. Here it nests as a stem mammal along with Pachygenelus.

You have to go with
phylogeny first, and then identify your clade traits, rather than the other way around. And be prepared for change, which has always been the nature of Science.

Chronology and Phylogeny
The nesting of Repenomamus indicates that certain mammal clades had already diverged prior to its appearance.

Yixian Formation, Early Cretaceous, 125 mya yields Eomaia and Repenomamus from this list. Other Early Cretaceous mammals include:

1. Astroconodon – 2mm triconodont molars, likely a predator
2. Jugulator – 5mm triconodont molars, likely a predator
3. Pinheirodontidae – clade of multituberculates
4. Spinolestes – 24 cm long triconodont teeth, skeleton and fur along with xenarthrous vertebrae.
5. Vincelestes – a 4 cm  skull with large canines described as a stem-therian, I’ll test it next to see how it nests.

The presence of
Early Cretaceous multituberculates implies the presence of more primitive coeval stem rodents and rabbits and perhaps stem tenrecs.

References
Hu Y, Meng J, Wang Y-Q and Li C-K 2005. Large Mesozoic mammals fed on young dinosaurs. Nature 433:149-152.
Li J-L., Wang Y.Wang Y-Q and Li C.-K. 2000. A new family of primitive mammal from the Mesozoic of western Liaoning, China [in Chinese]. Chin. Sci. Bull. 45, 2545–2549).
Ji et al 2002. The earliest known eutherian mammal, Nature 416:816-822.
Kielan-Jaworowska Z 1975. Preliminary description of two new eutherian genera from the Late Cretaceous of Mongolia. Palaeontologia Polonica 33:5-15.
Kielan-Jaworowska, Z 1984. Evolution of the therian mammals in the Late Cretaceous of Asia. Part VII. Synopsis. Palaeontologia Polonica 4: 173-183. online pdf

wiki/Asioryctes
wiki/Eomaia

Tenrecs and Echolocation

Just a short note before the weekend…
Earller here, here and here we looked at the traditionally overlooked morphological traits that link little tenrecs to larger whales.

Well, here’s one more
that I just became reacquainted with. Tenrecs, like whales, hunt by echolocation. I think we all heard that somewhere in our past. I did, but forgot it in the last fifty years. Here’s the reference:

References
Gould E 1965. Evidence for Echolocation in the Tenrecidae of Madagascar
Proceedings of the American Philosophical Society 109 (6): 352-360. online here.

Let’s talk about mammal interrelationships

Now that a wide gamut of mammals
has been added to the large reptile tree (LRT, subset Fig. 1), the tree topology has become distinct from prior studies, many of which depend on DNA, which does well in many cases, but makes untenable sisters of several genera.

Figure 1. Mammal subset of the large reptile tree with clades named. This needs to be updated with Onychodectes nesting closer to Ectoconus + Pantolambda and Maelestes taking its place. See the large reptile tree for other changes due to added taxa.

Case in point: Macroscelidea
(elephant shrews, Fig. 2). Stanhope et al. (1998) proposed the clade Afrotheria based on molecular evidence. Their clade members included elephants and elephant shrews. That’s difficult to accept on the face of it, and the large reptile tree does not recover that relationship.

Figure 2. Macroscelides proboscideus, the elephant shrew or sengis is NOT more closely related to elephants with the purported ‘Afrotheria.’ but instead is related to Tupaia, the tree shrew.

Rather
the large reptile tree recovers:

1. Monotremata (Ornthorhynchus) as the most basal mammal clade.
2. Didelphis (opossum) is basal to both Metatheria (so far only three genera) and Eutheria.
3. The first eutherian split occurs between small carnivores and smaller insectivores
4. Carnivora also splits into small insectivores: Chiroptera + (Dermoptera + Primates, including Manis)
5. The two tree shrews, Tupaia and Ptilocercus, are not sister taxa.
6. The former clade Insectivora is resurrected. It includes Tupaia + elephant shrews, Trogosus + Apatemys and Glires.
7. Glires includes the traditional rabbits and rodents, but also shrews, moles and multituberculates
8. Condylartha is resurrected and includes ungulates, xenarthrans and paenugulates.
9. Maelestes gives rise to tenrecs, which give rise to giant tenrecs and whales.
10. Onychodectes + (Pantolambda + Ectoconus) give rise to Xenarthra (sloths, anteaters), Paenungulata (elephants and kin) and Ungulata (hoover mammals).

Not only do these relationships make more sense
on their face, they provide a gradual accumulation of derived characters down to the toes. You can’t do that with ‘Afrotheria’ and other odd-bedfellow sisters that have become widely accepted within paleontology, despite the fact that they make no sense at several nodes. At other nodes, some DNA clades do match morph studies.

Once again,
we need to look at the results, put our thinking caps on, and toss out DNA results that do not make sense and cannot be supported with morphological studies.

Don’t take my word for it.
I’m reporting results, like Galieo looking through a telescope for the first time or dropping balls off the leaning tower of Pisa. Because this is Science, you can repeat the experiment and discover the mammal family tree for yourself. If you do, let us all know what you recover.

References
Stanhope MJ, Waddell VG, Madsen O, de Jong W, Hedges SB. Cleven GC, Kao D and Springer MS 1998. Molecular evidence for multiple origins of Insectivora and for a new order of endemic African insectivore mammals. Proceedings of the National Academy of Sciences 95 (17): 9967–9972. Bibcode:1998PNAS…95.9967S. doi:10.1073/pnas.95.17.9967. PMC 21445. PMID 9707584.

The human occiput and palate

We looked at the facial portion
of the human skull earlier. Today we’ll look at the occiput and palate (Fig. 1).

Figure 1. Human occiput and palate. On most tetrapods these two are usually set at right angles to each other, but an upright stance has rotated the occiput to a ventral orientation.

There’s nothing new here. 
This is just an opportunity to educate myself on the human palate and occiput. Only the endotympanic (En) is a novel ossification. The occiput is a single bone here, the product of the fusion of several occipital bones. Can you find the suborbital fenestra? It’s pretty small here.

The asymmetry is interesting here.
Sure, this is an old adult, missing some teeth, but you’ll see other examples elsewhere.

Let me know
if you see any errors and they will be corrected. As you already know, everything I present here was learned only 48 hours earlier — or less.

The vague persistence of ‘absent’ bones in the human skull

Humans
(genus: Homo) are, of course mammals, and basal mammals lose the postorbital bar found in cynodonts like Chiniquodon. But even that postorbital bar does not include two bones found in more basal therapsids, the postfrontal and postorbital. In Chiniquodon the postorbital bar is created by a process of the frontal meeting a process of the jugal.

So I was surprised to find
what appear to be vague apparitions that look like those lost bones in the frontal of the human skull. There are 17 frames in this animation (Fig. 1). The last one holds for five seconds and reveals where I think I see vague outlines of the the prefrontal, postfrontal and postorbital bones all fused to the frontals, which are themselves fused medially.

Figure 1. Labeled skull bones in Homo sapiens. The last frame appears to identify the lost prefrontal, postfrontal and postorbital bones last seen in our ancestors from the Permian. Note the dislocated jaw articulation.

The last part of the postorbitals

to ‘disappear’ are the posterior processes, which seem to laminate to the cranial bones in several therapsids, including dinocephalians. But there is also a vague portion that appears on the duckbill platypus, Ornithorhynchus.

The prefrontals
were the last to fuse and so appear to be the easiest to see now. Like gills and a tail, humans retain the genes for these, but they get resorbed or fused during embryonic ontogeny.

There is a difference between losing a bone
and fusing a bone. Phylogenetically losing a bone is usually marked by a reduction to disappearance. Fusion is, well, fusion. And that can be more readily reversed.

The pterygoid and lateral sphenoid
are both bones typically seen in palatal view, but in mammals and humans they rise, tent-like, to appear in the nasal and orbital cavities.

In mammals the bones sometimes get new names
but they’re still the same bones. The jugal becomes the zygomatic arch, for instance. The bones that make up the occiput fuse to form the occipital.

We’ll look at the palate and occiput soon. 
And a fetus.

 

Origin of turtles linked to digging? That’s not the problem here.

From the Lyson et al. 2016 highlights:
“Recently discovered stem turtles [in this case Eunotosaurus] indicate the shell did not evolve for protection. Adaptation related to digging was the initial impetus in the origin of the shell. Digging adaptations facilitated the movement of turtles into aquatic environments. Fossoriality likely helped stem turtles survive the Permian/Triassic extinction.”

From the Lyson et al. summary:
“Developmental and fossil data indicate that one of the first steps toward the shelled body plan was broadening of the ribs (approximately 50 my before the completed shell.” 

Figure 1. The origin and evolution of turtles. Here Meiolania and Niolamia nest as the most basal hard shell turtles. Odontochelys is the most basal soft shell turtle.

Unfortunately
Lyson et al. have been beating this dead horse [or, dead turtle] for several years. Eunotosaurus is not a stem turtle, but they keep trotting it out. Currently it is the last of its kind and is most closely related to Acleistorhinus.

The real stem turtles
(Fig. 1) are small pareiasaurs, Elginia and Sclerosaurus according to the LRT, which tested all currently proposed candidates, including Eunotosaurus. It is important to have the right phylogeny or you’re sunk in a morass of fiction and faith that will never make sense.

Lyson et al. are unaware
that living turtles had dual origins among small pareiasaurs. Among domed hard-shell turtles, tabular and supratemporal horns were the first distinct traits, along with a reduced size. We don’t have post-crania for Elginia, but that’s where and when the high-domed shell first appeared if not slightly earlier or slightly later. Sister taxa, including the late surviving Meiolania, are already full shelled and with a long armored tail.

Among low-domed soft-shell turtles, once again tabular and supratemporal horns were the first steps. A pattern of ossified scutes preceded broader ribs in Sclelrosaurus.

Protection from Permian predators
was likely the raison d’être for the origin of turtle shells among phylogenetically miniaturized pareiasaurs. Large pareiasaurs were slow and otherwise defenseless so they depended on their bulk. Smaller forms survived with greater armor, both over the torso as a carapace that also covered the limbs and with greater armor over the head and tail with horns, clubs and spikes. Fossoriality and aquatic submergence are additional strategies for defense. To that end, it is easier for small tetrapods to hide whether underground or underwater. Moreover, a low metabolism helped basal turtles survive without food for long periods.

But what about Eunotosaurus?
Unfortunately the sisters of Eunotosaurus are known chiefly from skulls. The closest taxa with post-crania are the caseasaurs, including Oedaleops, and Milleretta, which has slightly expanded ribs.

Figure 2. Diadectes phaseolinus showing off those very broad anterior dorsal rib costal plates, by convergence similar in shape to what is found in the pre-pareiasaur/pre-turtle Stephanospondylus and Odontochelys. Diadectids and pareiasaurs grew large by convergence and are not directly related except through tiny ancestral taxa.

At the same time
Some Diadectes specimens and Stephanospondylus were developing greatly expanded ribs beneath their pectoral girdles. I don’t know if Lyson et al. 2016 discuss these taxa, but in the past they have not done so.

References
Lyson et al. (8 co-authors) 2016. Fossorial Origin of the Turtle Shell.
Current Biology (advance online publication)
DOI: http://dx.doi.org/10.1016/j.cub.2016.05.020
http://www.cell.com/current-biology/fulltext/S0960-9822(16)30478-X

Haramiyidans and Multituberculates are rodents, not pre-mammals.

Updated Janurary 5, 2018 with the addition of more taxa.

Wikipedia reports
“Haramiyidans have been known since the 1840s, but only from fossilized teeth and a single partial lower jaw. However, several features of the teeth have shown for many years that haramiyidans are among the most basal of mammaliaforms. Megaconus (Middle Jurassic, Zhou et al. 2013, Fig. 1) is a member.”

Wikipedia also reports
“Multituberculata is is an extinct taxon of rodent-like mammals. At least 200 species are known, ranging from mouse-sized to beaver-sized. Multituberculates are usually placed outside either of the two main groups of living mammals—Theria, including placentals and marsupials, and Monotremata—but closer to Theria than to monotremes. The oldest known species in the group is Rugosodon from the Jurassic.” 

Figure 1. Megaconus in situ. Original tracing and DGS color tracing which appears to show that both hind limbs and the vernal pelvis have been displaced posteriorly — unless their is a counter plate that preserves skeletal parts that don’t appear to be present here.

Recently added taxa
to the LRT (749 taxa then, 1370 taxa when updated) include the purported haramiyid allothere mammaliaform. Megaconus mammaliaformis (Zhou et al., 2013) the mutituberculates Rugosodon (Yuan et al. 2013) and Kryptobaatar (Kielan-Jaworowska  1970). Unfortunately most of the other known haramiyid allothere mammaliaformes are known from too few traits to test in the LRT. So far as I know, only Kryptobataar (Fig. 8), Rugosodon (Fig. 9) and Megaconus (Fig. 1) are known from complete skeletons. There may be others, but these three are enough to test the nesting. In the LRT they nest together with Rattus (the rat. Fig. 3). Additional taxa nest Haramiyavia with pre-mammal trithelodonts. 

Figure 2a. Mammals include rodents. Haramiyidans and multituberculates nest with rodents. This is a cladogram from 2016. Compare to a more focused look at primates + glires in Figure 2b 

Figure 2b. Subset of the LRT focusing on Primates + Glires.

By contrast…
Zhou et al. 2013 report: “Here we describe a new fossil from the Middle Jurassic that has a mandibular middle ear, a gradational transition of thoracolumbar vertebrae and primitive ankle features, but highly derived molars with a high crown and multiple roots that are partially fused. The upper molars have longitudinal cusp rows that occlude alternately with those of the lower molars.” The first three traits put Megaconus among the pre-mammal cynodonts. The last three traits are specialiizations. The broader traits employed by the LRT put Megaconus in the rodent clade. Rattus (the rat) and Oryctlagus (the rabbit) were included taxa in both studies. The primitive traits appear to be atavisms (reversals).

So now we have a phylogenetic problem.
Do Megaconus and Rugosodon nest more primitively than monotremes? According to Zhou et al. they do. The Zhou team employed more taxa, more traits and more dental traits — by far.

Figure 3. Megaconus mammaliaformis in situ and reconstructed. Compare to the similar, but smaller Vilevolodon.

Unfortunately,
Megaconus otherwise looks so much like a rodent that it has been given the nickname, ‘the Jurassic squirrel.’ The LRT tests only the relatively easy traits to see, not dental details. In the LRT, shifting Megaconus and Rugosodon to Juramaia adds 37 steps. That’s a big hump to get over. I do not know how many additional steps would be added by shifting Megaconus to Rattus in the Zhou et al. study.

Figure 4. Megaconus mandible showing cynodont-like posterior mandible bones, not tiny mammal-like ear bones. Unfortunately this key trait cannot be confirmed with present photo resolution. Mammals and reptiles call the same bones different names in some cases and some of these are labeled here.

If the Zhou et al. team is correct
then we have a problem. If the Zhou et al. team is not correct, they have a problem. They have identified an angular/ectotympanic where there is none. Rugosodon and Kryptobataar likewise do not have pre-mammal-type posterior jaw bones prior to their evolution into tiny ear bones.

Figure 5. Brown Rat (Rattus norvegicus) skull showing how lower incisors are used to scrape away and sharpen upper incisors The ear bones are located inside the circular ectotympanic posterior to the mandible and below the skull.

How can we reconcile this problem? 

1. If Megaconus indeed nests with Rattus, then the ankle, posterior jaw and other traits may represent reversals to a more primitive state. 
2. If Megaconus is indeed primitive, then it anticipates and converges on a long list of traits with Rattus under the LRT, Given that living monotremes have a long list of special traits, it is not unreasonable to accept that Megaconus diid likewise. The only caveat to that hypothesis is that monotreme special traits are not shared with or converge with those of other mammals.
3. If Megaconus parts have been misidentified, then (no exceptions) all other traits indicate it is a rodent sister.

Figure 6. Haramiyava dentary showing what a more typical stem mammal dentary and teeth look like. Earlier studies linked this clade to multituberculates, but this dentary was cause to reject that association. Just the appearance of that poster medial groove is enough to indicate the presence of tiny jaw bones that had not transformed into ear bones. From Luo et al. 2015.

Stem mammals have lots of teeth
(Fig. 6) and Megaconus does not have lots of teeth. It has rodent-like teeth and everything else is rodent-like, too. And check out that overbite!

Figure 7. Eomaia skull traced and reconstructed. Eomaia nests between marsupials and placentals. Note the unspecialized skull and dentition. Megaconus has a very specialized dentition.

The skull of
Kryptobataar, (Fig. 6) another purported multituberculate, likewise shows no trace of tiny post-dentary bones, either here or in a Digimorph scan.

Figure 8. The skull of the multituberculate Kryptobataar, which now nests as a rodent in the LRT. B&W image copyright Digimorph.org and used with permission.

The skull of
of Rugosodon (Fig. 9) likewise shows no trace of long, gracile post dentary bones, either here or originally.

Figure 9. The skull of Rugosodon. There are no tiny post dentary bones present here according to the original authors or my own tracings.

References
Kielan-Jaworowska Z 1970. New Upper Cretaceous multituberculate genera from Bayn Dzak, Gobi Desert. In: Kielan-Jaworowska (ed.), Results of the Polish-Mongolian Palaeontological Expeditions, pt. II. Palaeontologica Polonica 21, p.35-49.
Luo Z-X, Gatesy SM, Jenkins FA Jr., Amarai WW and Shubin NH 2015. Mandibular and dental characteristics of Late Triassic mammaliaform Haramiyavia and their ramifications for basal mammal evolution. PNAS 112(41) E71010-E7109. doi: 10.1073/pnas.1519387112
Wible Jr, Rougier GW 2000. Cranial anatomy of Kryptobaatar dashzevegi (Mammalia, Multituberculata), and its bearing on the evolution of mammalian characters. Bulletin of the American Museum of Natural History 247: 1–120. doi:10.1206/0003-0090(2000)2472.0.
Yuan CX, Ji Q, Meng QJ, Tabrum AR and Luo ZX 2013. Earliest evolution of multituberculate mammals revealed by a new Jurassic fossil.. Science 341 (6147): 779–783. doi:10.1126/science.1237970.
Zhou CF, Wu S, Martin T, Luo ZX 2013. A Jurassic mammaliaform and the earliest mammalorian evolutionary adaptations. Nature 500 (7461): 163. doi:10.1038/nature12429.

wiki/Rugosodon
wiki/Megaconus

Trogosus, the tillodont, is a giant tree shrew!

Updated August 1, 2022
with a new tracing for Trogosus (Fig 1).

Figure 1. Trogosus, the tillodont, shown to scale with tiny Ptilocercus.

Pretty obvious
when you put the two of them together. Tupaia and Ptilocercus are regular tree shrews. Trogosus (Leidy 1871) is a really big one as recovered by the large reptile tree (LRT, (745 taxa then, 2123 taxa in August 2122.)

Figure 2. Ptilocercus, pen-tailed tree shrew

Over the past five years
the LRT created new clades among basal reptiles. Now it’s deleting and combining clades within the Mammalia. Today we’ll talk about Trogosus castoridens (“beaver-toothed gnawing-hog”, Fig. 1), a purported member of the Tillodontia (Early Paleocene to Late Eocene). This is just one more overlooked taxonomic sister relationship recovered by the LRT.

Wikipedia considers tillodonts
close to Pantodonta (Pantolambda and kin), which are similar in size.

By contrast, the large reptile tree
nests the largest tillodont, Trogosus, close to Tupaia. Trogosus had a 37cm skull, so it was the size of a bear.

Re: tillodonts, OC Marsh 2013 wrote:
“These animals are the among the most remarkable yet discovered in American strata, and seem to combine characters of several distinct groups, viz: Carnivores, Ungulates, and Rodents. In Tillotherium (=Trogosus), the type [specimen] of the order, the skull has the same general form as in the Bears, but its structure resembles that of Ungulates. The molar teeth are of the ungulate type, the canines are small, and in each jaw there is a pair of large scalpiform incisors faced with enamel, and growing from persistent pulps, as in Rodents.”

As in rodents, indeed!
Rodents arise from these basal taxa.

References
Leidy J 1871. Remains of extinct mammals from Wyoming. Proceedings of the Academy of Natural Sciences 23: 113–116.
Marsh OC 1875. New Order of Eocene Mammals. American Journal of Science 9: 221.