The synapsids were an incredibly successful and diverse group during the Permian period, but after the devastating “Great Dying” mass extinction event 252 million years ago only three lineages survived into the Triassic – the cynodonts (close relatives and ancestors of modern mammals), the dicynodonts (beaked tusked weirdos who briefly took over the world), and the therocephalians.

Therocephalians were close relatives of cynodonts, and convergently evolved several very mammal-like anatomical features in their skulls, teeth, and limbs. But unlike their cousins this lineage never fully recovered in the Triassic, and they ultimately disappeared completely around 242 million years ago.

Ericiolacerta parva was one of these short-lived Mesozoic therocephalians, known from the early Triassic (~252-247 million years ago) of South Africa and Antarctica, in regions that were connected at the time as part of the supercontinent of Pangaea. It was a fairly small animal, about 20cm long (~8″), with small sharp teeth that indicate it mainly fed on insects, and semi-opposable thumbs and inner toes that suggest it was also a capable climber.

Holes in the bones of its snout would have carried numerous nerves and blood vessels, which may be evidence of sensitive fleshy lips and possibly whiskers. And while there’s no direct evidence of fur in therocephalians, they do appear to have been active warm-blooded animals – and possible fossilized synapsid hair from the Permian period suggests fuzziness might have been ancestral to all of the “protomammal” lineages that survived into the Triassic.

Retro vs Modern #15: Dimetrodon limbatus

With its prominent sailback Dimetrodon is one of the most iconic prehistoric animals – and one that still frequently gets mistaken for a dinosaur, despite being closer related to modern mammals.


The first known Dimetrodon fossil was an upper jaw fragment found in Canada in the 1840s, but at the time this specimen was thought to represent a dinosaur. It wasn’t until the late 1870s that species like Dimetrodon limbatus (initially called Clepsydrops limbatus) from the Midwestern and Southern United States were recognized as belonging to a much older and different group of animals given the name “pelycosaurs“.

While some paleontologists did propose pelycosaurs as being ancestral to mammals quite early on, for several decades the prevailing view was actually that they were an ancient branch of rhynchocephalian reptiles closer related to modern tuataras. From the 1910s onwards pelycosaurs were finally linked back to mammals, with their similarities to the therapsids placing them as early members of the synapsid lineage – although all these early mammal-relatives were still considered to be derived from reptiles, and “mammal-like reptile” became a commonly-used term for them. 

As a result reconstructions of Dimetrodon during this time period usually depicted a highly reptilian and heavily scaled lizard-like animal, with a sprawling belly-dragging pose, protruding crocodilian-like teeth, and a highly shrink-wrapped sail on its back modeled on those of some modern lizards. Some earlier images also showed a short stumpy tail, since Dimetrodon‘s longer tail proportions weren’t confirmed until the 1920s.


During the late 20th century new classification techniques led to the messy concept of “reptiles” being properly redefined as sauropsids, and synapsids being recognized as an entirely separate non-reptilian lineage of amniotes. Along with new studies and discoveries this has resulted in our understanding of Dimetrodon changing a lot in the last few decades, moving away from a heavily reptilian interpretation and instead letting it be its own weird “protomammal” thing. 

We now know there were at least a dozen different species of Dimetrodon living during the early-to-mid Permian, about 295-272 million years ago. Most of them are known from North America, but an additional species discovered in Germany suggests this genus ranged further across Pangaea than previously thought.

Dimetrodon limbatus was one of the larger species, about 3m long (10′), and like other members of the genus it had a tall narrow skull with high-set eyes and two distinct types of teeth in its jaws. The structure of its nasal cavities suggest it had a good sense of smell, and like the related synapsid Ophiacodon it may have had a closer to “warm-blooded” metabolism than previously thought.

It would have had a very poor sense of hearing, however, and probably didn’t even have any visible ears on its head. It may have been functionally deaf to air-borne sounds entirely, only able to detect vibrations by pressing its lower jaw to the ground.

No skin impressions are known for Dimetrodon. Scaly reptile-like skin has been found on varanopids, a group traditionally classified as very early synapsids – but some recent studies have suggested they were actually part of the true reptile lineage, so their extensive scaliness probably doesn’t apply to synapsids like Dimetrodon after all. There is some possible evidence of rows of square or rectangular scale-like scutes on the underside of the belly and tail in pelycosaur-grade synapsids, but otherwise the next-closest known synapsid skin comes from the distantly-related therapsid Estemmenosuchus, which seems to have had smooth glandular skin similar to a hairless mammal.

The characteristic back sail, formed by highly elongated neural spines on the vertebrae, is now thought to have been covered in a different pattern of soft tissue than older reconstructions depicted. The texture of the bone along the spines’ length shows that at the base they were deeply embedded in the back musculature, then further up they were covered by skin webbing, but then at the tips they may actually have been unwebbed and free-standing, giving a much spinier profile.

While the sail was traditionally assumed to be used for temperature regulation, more recently this has started to seem less likely. The sail doesn’t seem to have been quite as well-supplied with blood vessels as previously thought, and there’s a lack of direct correlation between sail size and body size in different Dimetrodon species and age classes. Instead this structure may actually have been used for visual communication and display, and could therefore have been quite flashy and brightly-colored.

Fossilized trackways also suggest that Dimetrodon didn’t move with a low lizard-like sprawling gait but instead with something more like a crocodilian “high walk”, with its limbs much closer to upright. It was probably a fairly active terrestrial predator and would have eaten a wide variety of other smaller Permian animals, with its teeth having been found in association with the remains of the amphibians Eryops and Diplocaulus and the freshwater shark Xenacanthus.


Some of the earliest large terrestrial herbivores on Earth were the edaphosaurids – a very early-branching group of synapsids, the evolutionary lineage whose only modern surviving members are mammals. Like their more famous cousin Dimetrodon these animals sported huge elaborate sails on their backs formed from highly elongated vertebral spines, but despite the similarity in appearance they actually seem to have evolved these structures completely independently.

Known from a single partial skeleton discovered in southern New Mexico, USA, the edaphosaurid Gordodon kraineri dates to around the very end of the Carboniferous or the very earliest Permian, about 299 million years ago.

It was fairly small for an edaphosaurid at about 1.5m long (~5′), and seems to have had transitional anatomy between earlier and later members of the group. Its sail spines were thicker than those of earlier species but still less heavyset than those of later forms, and while each spine had numerous side projections these structures were small, thorn-like, and randomly distributed, unlike the more organized thick crossbars seen in Edaphosaurus.

Its head was proportionally small compared to its body, but still relatively large for an edaphosaurid, and it had an unusually long neck for an early synapsid. But its most distinctive features were its jaws and teeth – it had a narrow snout with a pair of large incisor-like teeth at the front of both its upper and lower jaws, followed by a large toothless gap (a diastema) and then a short row of small peg-like teeth. Like Edaphosaurus it also would have had batteries of grinding tooth plates inside its upper and lower jaws, but probably not as extensively.

Overall its tooth arrangment looked more like a modern herbivorous mammal than an early synapsid, much more highly specialized than anything else known to be alive at the time – the next synapsid known to convergently evolve similar teeth lived around 90 million years later!

It probably had a very different diet to its relatives, with its specialized teeth and fairly slender body suggesting it may have been a selective feeder, cropping the softer more nutritious parts of plants like the fleshy seeds and cones of gymnosperm plants.

Its discovery also hints that herbivorous edaphosaurids in general were much more diverse than we previously thought, and there may be even more surprising forms out there still to be discovered.


If there’s any equivalent to carcinization in mammals, it’s turning into an otter-beaver-like semi-aquatic form.

Because it just keeps happening.

Modern examples alone include otters, beavers, muskrats, giant otter shrews, desmans, aquatic genets, yapoks, lutrine opossums, and platypuses – and in the fossil record there were early pinnipeds, remingtonocetids, pantolestids, stagodontids, and Liaoconodon going as far back as the early Cretaceous. Even outside of the true mammals there were also Castorocauda, Haldanodon, and Kayentatherium during the Jurassic, and much further back in the late Permian there was the early cynodont Procynosuchus.

So a non-cynodont synapsid doing the exact same thing really isn’t all that surprising.

Perplexisaurus foveatus was a member of the therocephalians, a group of synapsids that were close evolutionary “cousins” of the cynodonts-and-true-mammals lineage. Similar in size to a modern rat, about 20cm long (8″), it lived in Western Russia during the Late Permian about 268-265 million years ago.

At the time this region was a river plain with a tropical climate, experiencing seasonal floods that turned the whole area into what’s known as “viesses” (a name based on the abbreviation “V.S.S.” standing for “very shallow sea”), vast shallow lake-seas that persisted for weeks or months at a time.

So this little animal has been interpreted as being semi-aquatic, swimming around and feeding on aquatic invertebrates and tiny fish and amphibians. Its skull had numerous pits around the front of its face, suggesting that it had a highly sensitive snout – probably whiskery, allowing it to hunt entirely by touch in dark murky water, but it’s also been proposed to have possibly had an electroreceptive sense similar to modern platypuses.


Despite having a genus name that sounds more like it should belong to a cartoon dinosaur mascot for dental hygiene, Smilesaurus ferox was actually a real gorgonopsian, a predatory synapsid distantly related to modern mammals.

Living in South Africa during the Late Permian, around 259-254 million years ago, Smilesaurus was comparable to a medium-sized dog at around 1m long (3’3″). It had some of the longest sabre-like canine teeth of any known gorgonopsian, proportionally comparable to those of sabertoothed cats – and it may have hunted in a similar manner, using powerful grasping limbs to pin down struggling prey and then dispatching it with slashing bites.

…And it also turns out that when you don’t horribly shrink-wrap a gorgonopsian, you end up with something that looks rather like a bear-hippo.

(For some similarly chonky gorgonopsians, check out Tas’ @i-draws-dinosaurs reconstructions here. Bullet Man was definitely a bit of an inspiration in this.)


The docodonts were a group of mammaliaformes (close relatives of the earliest true mammals) which lived across North America, Europe, and Asia from the Middle Jurassic to the Early Cretaceous. Originally only known from teeth and jaw fragments they were traditionally thought to be fairly generic shrew-like insectivores, but more recent discoveries of better fossils have revealed they were actually much more diverse, occupying ecological niches ranging from squirrel-like tree-climbers to mole-like diggers to beaver-otter-like swimmers.

Most of the more complete fossil material of these animals comes from the mid-Jurassic of China, but one species from elsewhere is also known from a partial skeleton.

Haldanodon exspectatus here lived in central Portugal during the Late Jurassic, about 155 million years ago. Around 15-20cm long (6-8″), it had small eyes and short chunky well-muscled limbs with the front paws adapted for digging. Since it inhabited a very swampy environment it probably wasn’t a pure mole-like burrower – extensive tunnels would have constantly flooded – but it may have instead been a similar sort of semi-aquatic animal to modern platypuses and desmans, foraging for invertebrates in the water and excavating burrows in the banks.

Roughened areas of bone on its snout may also have supported a patch of tough keratinous skin, which would have helped protect its face while digging.


Even before Lystrosaurus briefly took over the world, the dicynodont synapsids were a highly successful group. These herbivorous beaked-and-tusked mammal-cousins were abundant all over the supercontinent of Pangaea during the Late Permian and occupied a wide variety of ecological niches, ranging from rat-sized to almost bear-sized. (And later during the Triassic some got even bigger.)

Towards the smaller end of that size range were species like Diictodon. Living around 259-254 million years ago in Southern Africa (but with fossils also found in northern China, suggesting a much larger geographic range) this dicynodont grew up to about 45cm long (1’6″) and was a gopher-like creature adapted for digging, with a tubular body and short muscular limbs.

It was a very common animal, making up around half of all vertebrate fossils in some locations. Numerous preserved spiral-shaped burrows have been found concentrated in small areas, going down as much as 1.5m (5′) into the ground.

Several different species have been named within the Diictodon genus, but currently they’ve all been lumped together under the single name of Diictodon feliceps. There’s a lot of anatomical variation between specimens, though, with some notably being smaller and lacking the distinctive tusks seen in others – which may be evidence of sexual dimorphism, with the tuskless individuals possibly being females. (Although differences in inner ear anatomy may also indicate they were a separate species entirely, in which case female D. feliceps might instead be represented by fossils showing smaller tusks.)

I’ve illustrated one of the tuskless forms here, since they don’t generally get as much attention as the tusked ones. It’s also speculatively fluffy and iridescent similar to modern golden moles.


Echinerpeton intermedium here was one of the earliest known members of the synapsids, the lineage that includes all mammals along with other “reptile-like” stem-mammals such as the famous sailbacked Dimetrodon.

Living during the Late Carboniferous in Nova Scotia, Canada, this 60-70cm long (2′-2’4″) distant cousin to modern mammals was previously known only from the fossilized remains of juveniles – with all known specimens showing slightly elongated spines on their vertebrae that gave it a sort of high-backed “proto-sail” appearance.

But a newly described fossil has completely changed what we know about this animal.

A single vertebrae identified as belonging to Echinerpeton shows a much much longer spine than anything we’ve ever seen before, and confirms that this species actually had a large elaborate true sailback – making it the earliest known tetrapod to experiment with this type of anatomy.

This individual seems to have been older than the other known specimens, but still not fully grown, leaving the possibility that fully mature Echinerpeton may have had even larger sails than this.

Weird Heads Month #30: Lumpy-Faced Synapsids

Among the synapsids (“proto-mammals”), head ornamentation evolved multiple times in the therapsids, from basal members of the group like Tetraceratops, burnetiamorphs, and dinocephalians to later lineages like dicynodonts and gorgonopsids.

But these sorts of structures don’t seem to have really ever developed in one of the lineages most closely related to the ancestors of modern mammals, a group known as therocephalians.

…With the exception of Choerosaurus dejageri.

Living in South Africa during the late Permian, around 259-254 million years ago, this small synapsid was only about 35cm long (1’2″) but sported some large bulging bony bosses on the sides of both its snout and lower jaw.

The bosses would have been covered by tough skin in life, similar to modern giraffid ossicones.

A study of Choerosaurus‘ skull found that its head was rather delicately built, and the bosses were relatively fragile and lacked the sort of reinforcement needed to resist impacts, suggesting that these structures weren’t used as weapons for fighting each other but were probably more for display – so they may even have been brightly colored.

The upper jaw bosses were also well-supplied with nerves and blood vessels, and would have been quite sensitive to touch.

Weird Heads Month #10: Permian Crowns

The tiny-headed Cotylorhynchus we saw earlier in this series wasn’t the only synapsid with a weird head.

A little more closely related to modern mammals, the dinocephalians were a a diverse group that were found across Pangaea during the middle of the Permian period. Many of them had thickened skulls that may have been used for headbutting each other, and some also developed bony horn-like projections around their faces.

And Estemmenosuchus mirabilis here was particularly elaborately ornamented, earning it a name meaning “wondrous crowned crocodile”. It lived in the Perm region of Russia during the mid Permian, about 268-265 million years ago, and was one of the largest dinocephalians, reaching at least 3m long (9’10”).

It had two big antler-like structures on its head, two wide cheek flanges, and a small nose horn, almost looking like the synapsid version of a ceratopsid dinosaur – and with its big bulky body, fairly erect-legged posture, and herbivorous-or-omnivorous diet it may have been a fairly close ecological equivalent to them, too.

But it’s also possible it was semi-aquatic, and it certainly does have a very hippo-like appearance when reconstructed with a decent amount of soft tissue.

One specimen of Estemmenosuchus even preserved skin impressions around its face, which were described in Russian in the early 1980s. They show scaleless glandular skin with a slightly bumpy texture, similar to that of hairless mammals or some amphibians. Since it occupied a point in the synapsid family tree close to where hair may have originated (somewhere in the Permian therapsids), it’s not clear if it was entirely hairless or if it had just secondarily lost some of it.