Vintana and the vain search for the clades Allotheria and Gondwanatheria

Figure 1. Vintana as originally illustrated. I added colors to certain bones. Note the high angle of the ventral maxilla and the deep premaxilla. Lateral view reduced to scale with other views.

Figure 1. Vintana as originally illustrated. I added colors to certain bones. Note the high angle of the ventral maxilla and the deep premaxilla. Lateral view reduced to scale with other views.

Earlier we looked at Vintana (Fig. 1, Krause et al. 2014a, b). To Krause et al. Vintana represented the first specimen in the clades Allotheria and Gondwanatheria to be known from more than teeth and minimal skull material.

To Krause et al. 
Allotheria included Multituberculata and nested between the clade Eutriconodonta (including Repenomamus and Jeholodens) and the clade Trechnotheria (including the spalacotheres Maotherium and Akidolestes) and Cronopio, Henkelotherium, Juramaia, Eomaia, Eutheria and Metatheria.

Taxon exclusion issues
The large reptile tree (LRT, 1005 taxa) did not recover the above clades or relationships. Alotheria does not appear in the LRT.

  1. Multituberculata, Henkelotherium and Maotherium nest within Glires (rats and rabbits and kin) in the LRT.
  2. Repenomamus and Jeholodens nest within the pre-mammalian trityllodontid cynodonts in the LRT.
  3. Akidolestes nests within basal Mammalia, close to Ornithorhynchus in the LRT.
  4. Cronopio and Juramaia nest within basal Mammalia between Megazostrodon and Didelphis in the LRT.
  5. Eomaia nests at the base of the Metatheria in the LRT.
  6. Vintana nests with Interatherium among the derived Metatheria (marsupials), with wombats, like Vombatus and Toxodon in the LRT.

Despite a paper in Nature
and a memoir of 222 pages in the Journal of Vertebrate Paleontology; despite CT scans and firsthand examination with electron microscopes; despite being examined and described by many of the biggest name and heavy hitters in paleontology… Krause et al. never understood that Vintana was just a derived wombat, evidently due to taxon exclusion problems.

Figure 3. Interatherium does not nest with notoungulates or other purported interotheres. Rather cat-sized Interatherium nests with wombats, between Vombatus and the giant Toxodon.

Figure 2. Interatherium does not nest with notoungulates or other purported interotheres. Rather cat-sized Interatherium nests with wombats,with Vintana,  between Vombatus and the giant Toxodon

The large reptile tree now includes
1005 taxa, all candidates for sisterhood with every added taxon. Despite the large gamut of 74 taxa employed by Krause et al. they did not include the best candidates for Vintana sisterhood. Perhaps the fault lies in the reliance of prior studies and paradigms. Perhaps the fault lies in the over reliance by Krause et al. and other mammal workers, on dental traits. Perhaps the fault lies in the absence of pertinent sisters to the above-named taxa, including Interatheriium for Vintana.

In any case
Vintana does not stand alone as the only taxon in its clade represented by skull material. Based on its sisterhood with Interatherium, we have  pretty good idea what its mandibles and post-crania looked like. Yes, Vintana is weird. But Interatherium is also weird in the same way, just not as weird.

The LRT has dismantled and invalidated
several other clades, too, Ornithodira and Parareptilia among them.

References
Krause DW, Hoffmann S, Wible JR, Kirk EC, and several other authors 2014a. First cranial remains of a gondwanatherian mammal reveal remarkable mosaicism. Nature. online. doi:10.1038/nature13922. ISSN 1476-4687.
Krause DW et al. 2014b. Vintana sertichi (Mammalia, Gondwanatheria) from the Late Cretaceous of Madagascar. Journal of Vertebrate Paleontology Memoir 14. 222pp.

wiki/Vintana
pterosaur heresies – Vintana

Danjiangia: not a chalicothere, not a brontothere…

It’s an early Eocene pig
according to the large reptile tree (LRT, 1003 taxa). A large gamut minimizes inclusion set bias and gains greater authority with every added taxon. It also reduces the average phylogenetic distance between taxa, all of which are species and individuals, not suprageneric taxa.

Figure 1. Danjiangia nests with the extant pig, Sus, in the LRT.

Figure 1. Danjiangia nests with the extant pig, Sus, in the LRT. Note the very low naris and nasal. The lost skull could have been elevated, as imagined here after phylogenetic analysis.

Danjiangia pingi (Wang 1995; early Eocene) was originally described as a basal chalicothere with brontothere traits. Then Hooker and Dashzeveg (2003) nested it as a basal brontothere (without including any other brontotheres). Mihlbacher 2008 and others used it as an outgroup to the brontotheres. The posterior skull is not known, but note the rise over the orbits suggesting a tall cranium, as in Sus (Fig. 2). Also note the very low naris below the low nasals. Usually you don’t see nasals so low, and perhaps that is due to taphonomic shifting.

Figure 2. Skull of the extant pig, Sus in several views.

Figure 2. Skull of the extant pig, Sus in several views. Note the elevated cranium and squamosal.

The long fused dentary 
of Danjiangia is a trait also shared with pigs and other taxa, like chalicotheres, by convergence.

Figure 1. Skeleton of Sus, the pig, a taxon commonly used as an outgroup for whales. In the LRT it is a sister to other even-toed ungulates, like Giraffa, not Odontoceti nor Mysticeti.

Figure 3. Skeleton of Sus, the pig. It provides good clues as to the missing postcranial skeleton of Danjiangia.

Sus the pig
(Fig. 3) provides good clues as to the missing postcranial skeleton of its sister taxon, Danjiangia. The other model for post-cranial details is the basal artiodactyl, Cainotherium (Fig. 4).

Fig. 1. Cainotherium nests at the base of the Artiodactyla or even-toed ungulates. I wonder if it had five fingers, even if vestiges, given that Ancodus, a derived artiodactyl, retains five fingers.

Fig. 4. Cainotherium nests at the base of the Artiodactyla or even-toed ungulates. I wonder if it had five fingers, even if vestiges, given that Ancodus, a derived artiodactyl, retains five fingers.

Why was the pig connection missed by others?
For the same reason that modern workers continue to include pterosaurs with archosaurs. It’s a tradition. Nobody wants to do the extra work of testing other candidate taxa. Nobody wants to acknowledge contrarian studies. Paleontology tends to run very slowly as we learned earlier here. Hail, hail the status quo!

Figure 1. Subset of the LRT focusing on ungulates, which split into three clades here.

Figure 5. Subset of the LRT focusing on ungulates, which split into three clades here. Note the nesting of Sus together with Danjiangia.

References
Beard KC 1998. East of Eden: Asia as an important center of taxonomic origination in mammalian evolution; pp. 5–39 in Beard and Dawson (eds.), Dawn of the Age of Mammals in Asia. Bulletin of Carnegie Museum of Natural History 34.
Mihlbachler MC 2004. Phylogenetic systematics of the Brontotheriidae (Mammalia, Perissodactyla). PhD dissertation. Columbia University. p. 757.
Mihlbachler MC 2008. Species taxonomy, phylogeny and biogeography of teh Brontotheriidae (Mammalia: Perissodactyla). Bulletin of the American Museum of Natural History 311:475pp.
Wang Y 1995. A new primitive chalicothere (Perissodactyla, Mammalia) from the early Eocene of Hubei, China. Vertebrata Palasiatica 33: 138–159.

Stylinodon may be a giant herbivorous mink

In 1873 
O. C. Marsh 1874) found an extinct Eocene (50.3 to 40.4 Ma) mammal “of great interest. The lower molar teeth, all essentially alike, and inserted in deep sockets” were the most striking feature. He named it Stylodon mirus (Figs. 1,2). All the teeth grew with “persistent pulps” and had a thin layer of enamel. The specimen was considered close to Toxodon with some edentate affinities (Marsh 1897). Stylinodon was placed under the family Stylinodontidae and the order Tillodontia. According to Schoch 1986 (first issue of JVP!) its ancestors were like Onychodectes.

Stylinodon mirus (Marsh 1874; middle Eocene, 45 mya; Figs. 1-2) was originally considered a taeniodont, perhaps derived from Onychodectes. Here it nests with Mustela, the living European mink, among the Carnivora. There were twice as many molars (4), each with a single root, as in the two double rooted molars of the mink. Large claws and certain forelimb traits indicate that Stylinodon was a digger, not a cursor.

The present nesting
of Stylinodon mirus (YPM VP 011095, Marsh 1874; Figs. 1, 2) in the Carnivora occurred when I realized it was a poor fit at the base of the Condylarthra/Paenungulata, despite its herbivorous dentition and tusk-like teeth (canines, not incisors).

Figure 1. Stylinodon skull. Note the transverse premaxilla, a trait of the Carnivora.

Figure 1. Stylinodon skull. Note the transverse premaxilla, a trait of the Carnivora.

Distinct from condylarths
Stylinodon has a transverse premaxilla, essentially invisible in lateral view. The lower canine is the anteriormost tooth on the dentary. These traits are shared with other members of the Carnivora. In the present taxon list Stylinodon shares more traits with Mustela, the European mink (Fig. 1) despite the loss of molar cusps and increase in size. They both were diggers. Together they nest with Phoca, the seal, and Palaeosinopa, the amphibious piscivore, all derived from a sister to Procyon, the omnivorous raccoon (Fig. 2).

Figure 1. Stylinodon compared to Mustela, the European mink to scale.

Figure 2 Stylinodon compared to Mustela, the European mink to scale.

As in the earlier issue
with indricotheres, related taxa can have distinctively different types of teeth, one more reason to not weight dental traits too heavily, unless that’s all you have.

Figure 2. Mustela the European mink is an extant relative to Stylinodon.

Figure 3. Mustela the European mink is an extant relative to Stylinodon.

Mustela lutreola (Linneaus 1761; extant European mink; up to 43cm in length) is a fast and agile animal related to weasels and polecats. Mustela lives in a burrow, but it also swims and dives skilfully. It is able to run along stream beds, and stay underwater for one to two minutes. Mustela is derived from a sister to Phoca and other seals, all derived from a sister to Procyon. With this close relationship, Stylinodon (Fig. 2 a giant weasel with simple teeth.

Schoch and Lucas 1981
and Schoch 1983 considered Stylinodon and kin derived from a sister to the long-legged basal condylarth, Onychodectes. The large reptile tree (LRT, Fig. 2) does not support that nesting. Onychodectes has a long premaxilla lacking in taeniodonts.

Figure 2. Subset of the LRT showing the Carnivora nesting at the base of the Eutheria (placental mammals).

Figure 4. Subset of the LRT showing the Carnivora nesting at the base of the Eutheria (placental mammals).

Schoch and Lucas 1981
determined that Stylinodon had two upper incisors (one lower), a giant canine, four premolars and three molars, as in Onychodectes. That may be so, but the premolars and molars look alike.

 

Figure 6. Wortmania as drawn freehand by Schoch compared to bones Photoshopped together.

Figure 5 Wortmania as drawn freehand by Schoch compared to bones Photoshopped together.

Wortmania (Hay 1899, Williamson and Brusatte 2013; above) and Psittacotherium (Cope 1862; below) are related to Stylinodon. All are among the largest taxa in the early post-Cretaceous, derived from smaller weael-like basal mammals in the Cretaceous.

Figure 7. Psittacotherium in various views.

Figure 6.  Psittacotherium in various views. Overall it is elongated to more closely match related taxa.

It is rare but not unheard of
for members of the Carnivora to become omnivores and herbivores. Think of the giant panda and certain viverrids. Now the stylinodontid taeniodonts join their ranks.

References
Linneaus C von 1761. xxx
Marsh OC 1874. Notice of new Tertiary mammals 3. American Journal of Science. (3) 7i: 531-534.|
Marsh OC 1897. The Stylinodontia, a suborder of Eocene Edentates. The American Journal of Science Series 4 Vol. 3:137-146.
Rook DL and Hunter JP 2013. Rooting Around the Eutherian Family Tree: the Origin and Relations of the Taeniodonta. Journal of Mammalian Evolution: 1–17.
Schoch RM and Lucas SG 1981. The systematics of Stylinodon, an Eocene Taeniodont (Mammalia) from western North America. Journal of Vertebrate Paleontology 1(2):175-183.
Schoch RM 1983. Systematics, functional morphology and macroevolution of the extinct mammalian order Taeniodonta. Peabody Museum of Natural History Bulletin 42: 307pp. 60 figs. 65 pls.

 

wiki/Stylinodon
wiki/Mustela

Necrolestes: 125 year-old assessment beats recent analysis.

As usual
I had second hand (academic papers and figures) rather than firsthand access to the specimens. It doesn’t matter how good your players are if you don’t show up on the right field at the proper hour. Here you’ll see, once again, how excluding the actual sister to an enigma taxon is the major problem, solvable by second-hand phylogenetic analysis in a large gamut study, the large reptile tree (LRT) that minimizes the problem of taxon exclusion.

Figure 1. Necrolestes skull. Note the scale bar problems. DGS colors the bones here.

Figure 1. Necrolestes skull. Note the scale bar problems. DGS colors the bones here. The lacrimal and infraorbital are enlarged here, providing a large opening for large facial nerves. Note the larger lower incisors as compared to the drawing above.

Necrolestes patagonensis  (Ameghino 1891; early Miocene, 16mya; Fig. 1; YPM PU 15065, 15384, and 15699) has been argued about for over a hundred years. Originally (Ameghino 1891) it was described as the only known extinct placental “insectivore” from South America and allied to Chrysochloris (Fig. 2), the extant golden mole.

Well done Ameghino!

Unfortunately, as time went on…
Saban 1954 considered Necrolestes a palaeanodont (Ernanodon was previously considered one). Patterson 1958 considered it a borhyaenoid metatherian. Asher et al. 2007 looked at several candidates and could not make a firm conclusion. Ladevèze et al. 2008 supported metatherian affinities. Goin et al. 2008 also could not be specific with regard to a closest known sister taxon.

The latest paper on the subject
Rougier et al. 2012 reported, “earlier studies leaned toward placental affinities and more recent ones endorsed either therian or specifically metatherian relationships.” Ultimately they nested Necrolestes with Cronopio (Fig. 4) which they considered a non-therian mammal. That is correct. They considered an earlier Van Valen 1988 statement  inspired, “…the enigmatic Miocene genus Necrolestes, usually thought to be a marsupial, is [conceivably] a late surviving Gondwantherian pantothere.” That is incorrect.

Figure 2. Chrysochloris skull lateral view. Note the many similarities to Necrolestes, including a ventral naris, expanded bulla, and similar shapes for the other bones.

Figure 2. Chrysochloris skull lateral view. Note the many similarities to Necrolestes, including a ventral naris, dorsally expanded bulla, and similar shapes for the other bones. Note the orbit is very tiny in this burrowing taxon. I don’t see an infraorbital foramen. here, distinct from Necrolestes.

Asher et al. 2007 report,
“Characters that support [Necrolestes] status as a therian mammal include a coiled cochlear housing of the inner ear. Necrolestes shows similarities to eutherian mammals, such as small incisive foramina and possibly three molars. Consistent with its status as a metatherian is the presence of five upper incisors, transverse canal foramina, and a broad proximal fibula. However, we cannot confirm other characters claimed by previous researchers as evidence for affinity with marsupial or nonplacental mammals, such as the presence of an inflected mandibular angle and epipubic bones.”

Asher et al. report,
“The external digital flexor in Chrysochloris ossifies along nearly the entire forearm, from the humeral medial epicondyle to the carpus. Necrolestes shows a similarly elongate element stretching proximally from the carpus.”

Asher et al. report,
“The idea that [Necrolestes] is related to golden moles was favored in the first two publications describing its anatomy (Ameghino, 1891; Scott, 1905). We do not believe Patterson’s contention that the status of Necrolestes as a marsupial is ‘‘virtually assured’’. We admit that the list of possible taxonomic affiliations for this animal still remains long.”

Figure 1. The Golden Mole (Chrysocloris asiaticus) nests with the tree shrew and elephant shrew in the large reptile tree, not the common mole. Image copyright Digimorph.org and used with permission.

Figure 3. The Golden Mole (Chrysochloris asiaticus) nests with the tree shrew and elephant shrew in the large reptile tree, not the common mole. Image copyright Digimorph.org and used with permission.

 

 

The large reptile tree
(920 taxa) tested Necrolestes against a wide gamut of mammal candidates and nested it securely with the golden mole, Chrysochloris. To shift Necrolestes next to Cronopio adds 22 steps.

Distinct from sister taxa
Necrolestes had five upper incisors and four lowers. That is closer to the primitive numbers for mammals and two more than in Chrysochloris. The molars are also primitive in having fewer cusps, but that also happens in whales and armadillos… and golden moles… with their simplified zalambdodont teeth… so let’s focus on other traits. Dental traits are plastic and can lead analysis astray.

Rougier et al. report,
“the first upper and lower premolars are double rooted and the following five molariform elements are single rooted, a condition shared only with the recently described meridiolestidan mammal Cronopio.” Convergent dental traits might be leading these workers so far afield the neglected to add Chrysochloris to their analysis, which seems odd and dangerous based on the long list of shared traits and overall similarity, not by convergence.

Figure 4. Cronopio nests between Juramaia and Didelphis + Ukhaatherium in the LRT. Rogier et al. nest this taxon with Necrolestes, contra the LRT. This taxon has an anterior naris, not a ventral one.

Figure 4. Cronopio nests between Juramaia and Didelphis + Ukhaatherium in the LRT. Rogier et al. nest this taxon with Necrolestes, contra the LRT. This taxon has an anterior naris, not a ventral one.

Rougier et al. gave us straw dogs
when they compared the basicrania of several sister candidates, but NOT that of Chrysochloris, to that of Necrolestes. Here I add a basicranium Rougier et al. chose to not show. Chrysochloris more closely matches the morphology of Necrolestes than any of the other three candidates. I don’t see Chrysochloris listed in the Supplemental Information for Rougier et al. which appears to test non-placental mammals only. So this is what I mean by another case of taxon exclusion. Ameghino (1891) got it right originally. Rougier Wible, Beck and Apesteguía 2012, for some reason, dropped the ball.

Figure 3. Necrolestes basicrania compared to three candidates by Rougier 2012. Here I add the basicranium for Chrysochloris for comparison and it's a better match.

Figure 3. Necrolestes basicrania compared to three candidates by Rougier 2012. Here I add the basicranium for Chrysochloris for comparison and it’s a better match. The blue element is the posterior mandible, which is not shown on the Rougier et al. drawings. Not how the lower (posterior) element curls over the basicranial element in only two candidates here. This is a placental trait. The LRT uses no petrosal traits, but image speaks for itself. Excluding the actual sister taxon was done for reasons unknown in this flawed study.

 

Deleting Chrysochloris from the LRT
nests Necrolestes with the remaining basal Glires, but resolution is lost. Not sure why, but Necrolestes has a history (see above) of being a confusing taxon when not nested with Chrysochloris.

Deleting all placentals from the LRT,
except Necrolestes, nests it between Didelphis and Asioryctes a node apart from Cronopio. So taxon exclusion doesn’t recover what Rougier et al. recovered.

Now that we have golden moles in Africa and South America
this is evidence that golden moles first appeared before those continents split apart 118 to 115 mya, long before the end of the Cretaceous. Video link here. Naish reports, “Golden moles and tenrecs appear to be close relatives, forming a clade usually termed Afrosoricida Stanhope et al., 1998 (though this is essentially synonymous with Tenrecoidea McDowell, 1958, see Asher (2001)“. That relationship is not supported by the LRT. Golden moles probably first appeared in the Early Jurassic, given that other Glires, multituberculates, split from rodents about the same time and are found as early as Middle Jurassic strata.

Rougier et al. tested earlier studies and found them flawed
Similarly, I tested Rougier et al. and found it flawed. Perhaps someday someone will likewise test this test and present additional insight into this former enigma taxon.

References
Ameghino F 1891. Nuevos restos de mamíferos fósiles descubiertos por Carlos Ameghino en el Eoceno inferior de la Patagonia austral. Especies nuevas, adiciones y correciones [New remains of fossil mammals discovered by Carlos Ameghino in the lower Eocene of southern Patagonia. New species, additions and corrections]. Rev Arg Hist Nat 1:289–328. Spanish.
Asher RJ, Horovitz I, Martin T and Sanchez-Villagra MR 2007. Neither a Rodent nor a Platypus: a Reexamination of Necrolestes patagonensis Ameghino. American Museum Novitates 3546:1-40.
Ladevèze S, Asher RJ, Sánchez-Villagra MR 2008. Petrosal anatomy in the fossil mammal Necrolestes: evidence for metatherian affinities and comparisons with the extant marsupial mole. J Anat 213(6):686–697.
Patterson B 1958. Affinities of the Patagonian fossil mammal Necrolestes. Breviora Mus Comp Zool 94:1–14.
Rougier GW, Wible JR,  Beck RMD and Apesteguía S 2012. The Miocene mammal Necrolestes demonstrates the survival of a Mesozoic nontherian lineage into the late Cenozoic of South America.
Saban R 1954. Phylogénie des insectivores [Phylogeny of the insectivores]. Bull Mus Natl d’Hist Nat. Ser 2 26:419–432. in French
Van Valen L 1988. Faunas of a southern world. Nature 333(6152):113.

Tetrapod Zoology on golden moles

When claws became hoofs and hoofs became claws

The cladogram
represented by the large reptile tree (LRT) shows that at more than one time within the clade Mammalia, sharp, clawed unguals became broad, often hoofed, unguals. And then several times broad unguals once again became sharp claws. Here’s the pattern:

  1. Small basal mammals all had sharp claws. So did basal monotremes, marsupials and placentals.
  2. Ernanodon, among the marsupials, had broad, round unguals
  3. Toxodon, among the marsupials, had short, anteriorly concave unguals, perhaps supporting hoofs.
  4. Basal placentals, including Carnivora and Glires, had sharp claws. Primates, like Proconsul, developed fingertip pads surrounding their unguals with only the keratinous nails exposed.
  5. Alcidedorbignya, among pantodont placentals (basal herbivores), had proto–hoofs (tips nearly as wide as bases). In the same clade, so did Pantolambda.
  6. Bayrlambda, among basal xenarthrans, had short unguals, barely longer than wide.
  7. Paramylodon, a ground sloth, had long sharp claws, but Glyptodon, a sister had truncated unguals, more like hoofs.
  8. Bradypus, a tree sloth, had long, curved, sharp claws. So did Orycteropus, Dasypus and kin.
  9. Titanoides, a basal phenacodont had sharp claws.
  10. Phenacodus, another basal phenacodont, had round hoof-like unguals. Many subsequent large mammalian herbivores had similar hoof-like unguals, except…
  11. Mysticete whales, like Balaenoptera, had long sharp unguals hidden within their flippers.
  12. Homalodotherium had sharp claws, as did Protypotherium, but
  13. Miocochilius had hoofs and it was a sister to those two.
  14. Dusisiren the sea cow, had sharp unguals beneath the mitten of its flipper, but…
  15. Procavia, the rock hyrax and Elephas had tiny unguals.
  16. The remaining mammals, like Sus, the pig, are all ungulates and they all have hoofs, except…
  17. Chalicotherium, which apparently had claw-like hoofs.

So you can see,
the sharp claw – broad hoof morphology comes and goes within the clade Mammalia, according to the LRT, with only one clade developing padded unguals with keratinous nails.

Fleshing out Andrewsarchus, the giant tenrec

All we know of Andrewsarchus
is its skull — without a mandible. A few days ago the dentary of a sister taxon, Sinonyx, was added to the skull of Andrewsarchus ((Osborn 1924; middle Eocene, 45 mya; AMNH 20135; 83cm skull length; also see Fig. 1) just to see if it would fit.

Before that…
everyone forever has always fleshed out Andrewsarchus like a giant bear/dog, moving the eyeballs to the top and giving it a bear/dog nose. Image googling Andrewsarchus will give you an idea what a widespread and accepted tradition that has been. I even followed that tradition back in 1989 in the book Giants, which you can see here as subset 1 of a larger pdf of the entire book.

Unfortunately,
Andrewsarchus does not nest with bears, dogs or mesonychids. It nests with tenrecs and Rhynchocyon (Fig. 2.), one type of elephant shrew. (The other type of elephant shrew is unrelated, as we learned here, Fig. 2). Tenrecs have a long flexible nose.

So, here, without further adieu
is a first shot at adding tenrec soft tissue to the skull of Andrewsarchus (Fig. 1). Is it close to being correct? I hope so, given the present evidence.

Figure 1. Andrewsarchus restored as giant tenrec alongside, Canis, the wolf to scale. Note the small and low-set eyes on Andrewsarchus. The mandible comes from Sinonyx. Note the natural tilt of the canid skull permitting binocular vision. Andrewsarchus had low-set eyes, rather un-canid-like. We have to give up the bear-dog restoration of Andrewsarchus.

Figure 1. Andrewsarchus restored as giant tenrec alongside, Canis, the wolf to scale. Note the small and low-set eyes on Andrewsarchus. The mandible comes from Sinonyx. Note the natural tilt of the canid skull permitting binocular vision. Andrewsarchus had low-set eyes, rather un-canid-like. We have to give up the bear-dog restoration of Andrewsarchus.

Now, just imagine the post-crania…
and the best clue we have is the living tenrec, Rhynchocyon (Fig. 2) with long legs, robust torso and short tail, only ten times bigger.

Figure 6. Rhynchocyon (above) and Macroscelides (below) compared. Though both are considered elephant shrews, they nest in separate major mammal clades in the LRT.

Figure 3. Rhynchocyon (above) and Macroscelides (below) the other type of elephant shrew compared. Though both are considered elephant shrews, they nest in separate major mammal clades in the LRT.

Maybe it’s time to 
give up the bear-dog restoration for Andrewsarchus and give it the giant  tenrec restoration it deserves based on phylogenetic bracketing and phylogenetic analysis.

Figure 3. The skull of Andrewsarchus mated to the body of Leptictis to make a chimaera.

Figure 3. The skull of Andrewsarchus mated to the body of Leptictis to make a chimaera.

References
Osborn HF 1924. Andrewsarchus, giant mesonychid of Mongolia. American Museum Noviattes 146: 1-5.

An imaginary mandible for Andrewsarchus

All I did
was take the mandible from sister Sinonyx and scale it to Andrewsarchus (Fig. 1; Osborn 1924). I also added a patch to extend the apparently broken and missing posterior nasals over the fontanelle between the frontals because that’s how far the nasal extends in Sinonyx.

See how sometimes
you don’t ‘see’ something until after you see it in a sister?

Figure 1. Andrewsarchus with Sinonyx mandible. The lower canine helps constrain the shape of the missing upper canine. 

Figure 1. Andrewsarchus with Sinonyx mandible. The lower canine helps constrain the shape of the missing upper canine. Note the transparent extension of the posterior nasals to cover up the fontanelle between the frontals, as in Sinonyx.

BTW
it bothered me that Sinonyx and Andrewsarchus were so much larger than their sisters, especially their closest sister, a type of elephant shrew, Rhynchocyon. Moreover, several traits appear to be homologous. So I retested the relationship of Sinonyx and Andrewsarchus with mesonychids and I retested them with prejudice. Any traits that could relate Sinonyx and Andrewsarchus with mesonychids I scored that way.

In the end,
I was not able to nest Sinonyx and Andrewsarchus with mesonychids.

Furthermore
when I removed all tenrec and odontocete sisters from the tenrec clade (see Fig. 2), leaving only Sinonyx and Andrewsarchus alone they still did not nest with mesonychids, but kept their node unchanged between the Ptilocercus clade and Onychodectes.

Figure 3. Tenrec-Odontocete clade with Leptictis now nesting with the elephant shrew Rhynchcyon and the long-tailed tenrecs nesting with the short tailed tenrecs, basal to Pakicetus.

Figure 2. Tenrec-Odontocete clade with Leptictis now nesting with the elephant shrew Rhynchcyon and the long-tailed tenrecs nesting with the short tailed tenrecs, basal to Pakicetus. This tree moves Sinonyx closer to Pakicetus. Indohyus has already been associated with pakicetids.

Testing like this
brought certain problems to the surface. The current tree has been improved over earlier versions.

Here’s how the tenrec clade now stands:
(Fig. 2) Leptictis and the elephant shrew Rhynchocyon now nest together. They are both similar in size and build.

Giant Andrewsarchus and smaller Sinonyx still nest together. Would still like to see some post-crania for  these two.

The two living short-tailed terrestrial tenrecs, Hemicentetes and Tenrec now nest with two extinct long-tailed aquatic tenrecs, Lepticitidium and Indohyus. The latter has already been associated with pakicetids in the literature  (Rao 1971, Thewissen et al. 2007.)

Likewise Sinonyx and Andrewsarchus have already been associated with the origin of whales in the literature. The new tree topology brings them closer to Pakicetus.

Early members of the tenrec clade
were insectivore speedsters with long slender legs, based on the habits of Rhynchocyon. More derived tenrecs like Tenrec, are not speedy and Hemicentetes is protected with spinesLeptictidium had much longer hind limbs than fore limbs and was likely bipedal. Indohyus had subequal limbs so likely remained a quadruped. Tradtionally Indohyus has been considered an artiodactyl, but given the opportunity to nest with artiodactyls in the LRT, it does not do so.

Perhaps the most convergent clade
By all the present evidence, some tenrecs converged with rabbits and elephant shrews, some with mesonychids, others with artiodactyls and still others with mysticete whales. It’s a pretty amazing and woefully under appreciated clade.

It is interesting to consider the possibility
that since both elephant shrews and tenrecs have a proboscis that extends beyond the jaw line, it is possible that early land whales, Andrewsarchus and Sinonyx, might have had a similar long nose. Some of these taxa might have used such a snorkel to breathe while underwater, just below the surface — or — the long nose was the first soft tissue to disappear during the transition, because whales have no such nose.

References
Osborn HF 1924. Andrewsarchus, giant mesonychid of Mongolia. American Museum Noviattes 146: 1-5.
Rao AR 1971. 
New mammals from Murree (Kalakot Zone) of the Himalayan foot hills near Kalakot, Jammu and Kashmir state, India. Journal of the Geological Society of India. 12 (2): 124–34.
Thewissen JGM, Cooper LN, Clementz MT, Bajpai S and Tiwari BN 2007. Whales originated from aquatic artiodactyls in the Eocene epoch of India. Nature 450:1190-1195.

wiki/Leptictidium
wiki/Indohyus