The pycnodonts were a diverse group of ray-finned fish that were found in shallow coastal waters from the late Triassic to the late Eocene (~215-37 million years ago). They usually had deep but very narrow body shapes with a disc-like appearance, convergently similar to modern reef fish like marine angelfish or butterflyfish – but some looked much weirder, with elaborate horns and spines, long snouts, or vertically-stretched bodies.

Most of them also had jaws full of round flat teeth used to crush hard-shelled prey, but some may instead have been herbivorous grazers similar to parrotfish.

And a couple of lineages even became carnivores.

Serrasalmimus secans lived in what is now Morocco during the late Paleocene, about 59 million years ago. Although only known from its jaws, the size of the fossil material suggests it was fairly large for a pycnodont, possibly around 80cm long (~2’8″).

It had sharp flesh-cutting teeth similar to those of modern piranha, but with a surprising evolutionary twist. Unlike any other known ray-finned fish, Serrasalmimus‘ teeth were true shearing carnassials anchored into bony sockets, with new replacement teeth forming directly below each current tooth – a very specific arrangement of features previously only known in mammals.

This is an especially remarkable example of convergent evolution because on land placental carnivorans were developing their own carnassials at the same time, just a few million years after the K-Pg mass extinction. Both mammals and pycnodonts were simultaneously taking advantage of the vacant predatory roles in their respective ecosystems, and ended up with incredibly similar tooth adaptations as a result.


Platycepsion wilksoni was a temnospondyl amphibian that lived during the early-to-mid Triassic (~251-242 million years ago) in what is now New South Wales, Australia.

A single partial skeleton discovered in the 1880s is the only known record of this species, and represents a juvenile that would have been around 15-20cm long (6-8″). We don’t know exactly what it would have looked like as an adult, but it was probably quite similar to other closely-related members of the brachyopid family – mostly-aquatic salamander-like animals with short but wide toothy jaws, eyes set towards the front of the head, small limbs, and paddle-like tails.

A recent re-analysis of the Platycepsion specimen found evidence of soft-tissue preservation of external gills, showing that it wasn’t just a juvenile but a true larva, a sort of temnospondyl “tadpole”.

Distinct larval stages have been found in a few other types of temnospondyls, but this is the first definite example from the stereospondyls, a major Mesozoic lineage that survived all the way into the Early Cretaceous.


Gliding has convergently evolved multiple times within mammals, from the Jurassic-aged haramiyids and volaticotheres to numerous species of modern marsupials, rodents, and colugos.

And yet despite the huge diversity of gliding mammals, and their particular prevalence in tropical forests, there’s an entire continent famous for its rainforests that’s somehow completely lacking any modern examples: South America.

It’s not clear why the gliding lifestyle never took off in South America, but the continent is surprisingly devoid of any other gliding vertebrates, too. The only exceptions are a few species of flying frogs in the northwestern tropical forests around Colombia.

But back in the early Eocene, about 53-50 million years ago, there was at least one South American gliding mammal. Some fossil limb bones found in the Itaboraí Formation in southeastern Brazil look very much like those of a gliding mammal – long and thin, with a locking elbow joint, knees adapted for jumping, and flexible ankles typical of tree-climbers.

These remains haven’t been given a new scientific name, however, because there’s a good chance they belong to an already-described species. Fossils from Itaboraí are found disarticulated, broken, and with bones of multiple different species jumbled together, so most fossil mammals named from the site have been based on their more easily distinguishable teeth and jaw fragments.

The problem is matching those teeth with these bones.

Currently the best identity guess based on size is Gaylordia macrocynodonta. This mammal would have been around 30cm long (1′), about the size of a modern rat, and had distinctive large canine teeth. It used to be classified as a marsupial related to opossums, but more recent studies have found it to have actually been a marsupialiform metatherian instead, much more closely related to Pucadelphys and sparassodonts than to any modern true marsupials.

Gaylordia‘s crushing molars suggest it was carnivorous, able to crunch through bones or hard-shelled invertebrate prey. This would be a very unusual diet for a gliding mammal, since most other mammalian gliders are herbivores or omnivores – the only other known predatory examples were the volaticotheres over 110 million years earlier.


The first definite ichthyosaur fossil found in Colombia was a single well-preserved skull, found in Early Cretaceous deposits dating to between 130 and 112 million years ago.

Although first discovered in the 1970s, this marine reptile wasn’t described until the late 1990s, at the time being named as a species of Platypterygius. But since then more pieces of the skeleton have been recovered, and the Platypterygius genus has been found to be a wastebasket taxon in need of revision, so in 2021 the Colombian ichthyosaur got a more detailed redescription and its own distinct name: Kyhytysuka sachicarum.

Kyhytysuka was a mid-sized ichthyosaur, about 5.5m long (18′) – about the size of a small modern orca – with a large head and a long robust snout. Its teeth varied in size, shape, and spacing along its jaws, with several different regions that were specialized to catch, slice, and crush its prey.

It could also open its jaws very widely, possibly up to an angle of 75°, suggesting it was able to tackle particularly large prey such as other marine reptiles. Possible soft tissue preservation around its lower jaw might also be evidence of elastic connective tissue that would have allowed its throat to expand out while swallowing big prey items.

This makes Kyhytysuka the first known example of a Cretaceous-aged ichthyosaur with an apex predator lifestyle, convergently evolving a similar ecological role to some earlier Triassic and Jurassic species.


Typhloesus wellsi has been a mystery for a long time.

First discovered in the early 1970s, in the mid-Carboniferous Bear Gulch Limestone deposits (~324 million years ago) of Montana, USA, it was initially mistaken for the long-sought-after “conodont animal” due to the presence of numerous conodont teeth inside its body. But just a few years later well-preserved eel-like conodont animals were found elsewhere, and it became apparent that the conodont teeth inside Typhloesus had actually just been part of its last meal.

But if it wasn’t a conodont… then what was it?

Up to about 10cm long (4″), Typhloesus had a streamlined body with a vertical tail fin and paired “keels” along its sides. It had a mouth and a gut cavity, but no apparent anus, and it also didn’t seem to have any eyes or other sensory structures. And in the middle of its body there was something very weird – a pair of “ferrodiscus” organs, disc-shaped structures which contained high concentrations of iron but whose function was completely unknown.

This anatomy just didn’t match any other known animals, so much so that it gained the nickname of “alien goldfish”.

For the next few decades it remained a bizarre enigma, at best tentatively considered to represent an unknown lineage of some sort of metazoan that left almost no other fossil record due to being entirely soft-bodied.

But now, 50 years after its initial discovery, we might just finally have a clue about Typhloesus’ true identity.

Recently something new was discovered in some Typhloesus specimens – a radula-like feeding structure that was probably part of an eversible proboscis. This would mean that Typhloesus was a mollusc, possibly a gastropod that convergently evolved a swimming predatory lifestyle similar to modern pterotracheoids.

It’s not a definite identification yet, and even if it was a mollusc it was an incredibly strange one, with features like the ferrodiscus still lacking any explanation. But this discovery at least shows that there are still new details waiting to be found in the “alien goldfish” fossils, and gives us a start towards bringing its classification back down to earth.

It Came From The Trash Heap (We Don’t Talk About Kholumolumo)

A wastebasket taxon is what happens when species can’t be easily classified and instead get hurled into a “catch-all” category.

…But that’s not the only kind of taxonomic tangle that can befall a new discovery.

When a scientific name is assigned to a new species, but it isn’t given a corresponding formal description and type specimen, it becomes a nomen nudum – a “naked name”. Without a proper description and assigned holotype the name isn’t valid, and the new species isn’t technically accepted by the wider scientific community.

This has even happened to some surprisingly famous names. In the 1920s Velociraptor mongoliensis was briefly given the nomen nudum “Ovoraptor djadochtari” before getting its much more familiar name when it was officially described. Meanwhile the giant pterosaur Quetzalcoatlus northropi was stuck as a nomen nudum for decades, only finally getting a proper published description in 2021.

And there’s another particular long-standing nomen nudum that became mildly infamous – “Thotobolosaurus”, the “trash heap lizard”.

An illustration of Kholumolumo (previously known as "Thotobolosaurus"), an extinct prosauropod dinosaur. It's a bipedal dinosaur with a small head, long neck, chunky arms, thick bird-like legs, and a long counterbalacing tail. It's colored dark blue, with lighter striping on its underside and a speckled red-and-orange pattern along its back. Its face is brighter teal-blue with a mask-like streak of orange than continues down onto a fleshy throat dewlap. There are also reddish quill-like filaments on the back of its neck and in a tuft at the tip of its tail.
Kholumolumo ellenbergerorum

Discovered next to a literal trash pile in the village of Maphutseng in Lesotho, a few scattered and broken bones of this “prosauropod” sauropodomorph dinosaur were first found in 1930. But it wasn’t until the mid-1950s that a more extensive bonebed began to be unearthed at the site, and over the next decade over 1000 fossil fragments were collected.

In the mid-1960s the remains were initially classified as belonging to Euskelosaurus browni (which is now considered to be a wastebasket taxon), but just a few years later in 1970 the “Maphutseng Beast” was re-evaluated as a species new to science. It was referred to as “Thotobolosaurus mabeatae” – based on the local name of the discovery site, “Thotobolo ea ‘Ma-Beata” (trash heap of Beata’s mother) – but this name was never actually formally published.

Despite “Thotobolosaurus” being an undescribed nomen nudum it nonetheless went on to be repeatedly referenced in scientific literature over the next few decades, and appeared in several popular dinosaur books (even as recently as 2020!).

In the mid-1990s it was alternatively named “Kholumolumosaurus ellenbergerorum” in a Ph.D. dissertation, with this name derived from the kholumolumo, a reptilian creature in Sotho mythology, and the Ellenberger brothers who worked on the site. But this also didn’t count as a formal publication and instead became a second nomen nudum for the species.

Eventually, 90 years after the first bones were found and 50 years after the debut of the name “Thotobolosaurus”, this long-neglected sauropodomorph was finally given a proper published full anatomical description in 2020.

And it also got a third name, this time officially valid, based on the second one from the 1990s: Kholumolumo ellenbergerorum.

For something associated with trash for so long, Kholumolumo is actually now one of the most completely-known prosauropods. At least five different individuals were present in the collected fossil material, possibly as many as ten, and between them most of the full skeleton is represented – with the exception of the skulls, which are only known from a couple of small fragments.

We now know Kholumolumo was rather heavily-built, with chunky limb bones and unusually short shinbones. It would have been one of the biggest animals around in the Late Triassic (~210 million years ago), measuring at least 9m long (~30′) and weighing around 1.7 tonnes (1.9 US tons), but despite its size it seems to have still been bipedal.

Due to the highly disarticulated nature of the bones the fossil site may have been a “bone accumulation area”, a place where dismembered bits and pieces of different carcasses were regularly carried to be eaten by a predator or scavenger  – essentially a trash heap, fittingly enough. A couple of “rauisuchian” teeth have actually been found among the remains, which might indicate what was chomping on these particular Kholumolumo.

It Came From The Wastebasket #20: Colossally Convoluted Condylarths

“Insectivora” was a wastebasket taxon so bad it had to be revised multiple times, but there’s another particularly infamous case in mammal taxonomy that’s still in the process of being resolved – the “condylarths.

This group was first created in the early 1880s, during the Bone Wars, and initially was just a subgroup of odd-toed ungulates containing only the phenacodontids. But just a few years later Condylarthra was promoted up to its own order, and groups like the periptychids and hyopsodontids were added in too.

Then over the next few daceds century various groups were added and removed from the condylarths, most notably with the mesonychids and arctocyonids being brought in from their previous position with the creodonts.

By the mid-20th century the condylarths had become a big convenient dumping ground for any and all “primitive” ungulate-like mammals that didn’t easily fit into any modern groups, ranging in age from the early Paleocene through to the early Oligocene. But it soon became apparent that they had the same problem as the “insectivores” – there weren’t really any unique anatomical features that united all these animals together.

They generally had rounded-cusped molar teeth and hoof-like toes, but they also had rather generalized “primitive mammal” features and a diverse range of ecologies. Some were small herbivores, but others were coati-like or dog-like omnivores, and some were even bear-sized carnivores.

An illustration showing five different "condylarths". On the top row is Hyopsodus, a small guinea-pig-like animal with a long horse-like head; then Meniscotherium, a slightly larger animal that somewhat resembles a capybara; then the much larger Arctocyon, which has a slightly bear-like body, hoofed towes, and a dog-like head. On the bottom row is Ectoconus, a small animal with a tapir-like body and a blunt rectangular snout; then Mesonyx, a much larger dog-like animal with hoofed toes and a long tail.
From left to right, top row: Hyopsodus lepidus (hyopsodontid), Meniscotherium chamense (phenacodontid), Arctocyon primaevus (“arctocyonid”).
Bottom row: Ectoconus ditrigonus (periptychid), Mesonyx obtusidens (mesonychid)

It wasn’t even clear how the various different condylarth groups were actually related to each other. The best guess was that arctocyonids had arisen from within the “insectivores”, with a Protungulatum-like form as the common ancestor of all the other condylarths. Where exactly modern ungulates had then evolved from within the condylarths was also still uncertain.

Cladistic analysis in the 1980s began to tackle the confusing pile of assorted condylarths, and showed that they weren’t the single ancestral source of all modern ungulates, but instead a loose collection of several unrelated groups from all over the ungulate evolutionary tree. Arctocyonids, periptychids, and hyopsodontids were placed as early “primitive” lineages, phenacodontids were loosely linked with the ancestors of odd-toed ungulates once again, and mesonychids were considered to be the ancestors of whales.

An image of a diagram from a 1994 academic paper, showing the proposed evolutionary relationships of various "condylarths" with main ungulate groups. It differs majorly from modern understanding by its inclusion of elephants, sirenians, hyraxes, and their extinct relatives, along with showing whales as descending from mesonychids instead of artiodactyls.
…And if you know modern mammal phylogeny you’ll probably see some big problems here. 🐘
(Image source:

And, once again paralleling the mess of the “insectivores”, it wasn’t until genetic methods became available in the late 1990s that larger-scale ungulate relationships began to be properly resolved. The paenungulates (elephants, hyraxes, and sirenians), which had been traditionally considered to be a major branch of ungulates, were removed entirely and reclassified as afrotheres. And, along with some new fossil discoveries, whales were recognized as having actually evolved from within the even-toed ungulates instead of from mesonychids.

This shake-up threw the still-problematic “condylarth” classifications back into question – with some “condylarths” turning out to also be afrotheres instead of true ungulates.

Today the actual relationships of the main “condylarth” ungulate families are still in the process of being figured out, and there’s a lot of remaining uncertainty and disagreement about them.

Phenacodontids seem to have mostly maintained their traditional position as early odd-toed ungulates, and hyopsodontids may potentially be part of this group too – possibly as members of the hippomorph lineage, closely related to horses and brontotheres. Arctocyonids might be a wastebasket themselves, with some studies finding them to be a mix of several different archaic ungulate lineages. Periptychids may have links to the even-toed ungulates. The mesonychids, meanwhile, are now generally considered to be a separate order from the traditional “condylarths”, and may be either an early branch of the even-toed ungulates or much more basal ungulates closely related to the “arctocyonids”.

Since the term “condylarth” no longer has any real taxonomic meaning some paleontologists have proposed replacing it with “archaic ungulate” to distance from the historical messiness of the old name. But this hasn’t really caught on, and many papers still use “condylarth” in a very loose sense to refer to an “evolutionary grade” of early ungulates of unclear evolutionary affinities.

A cladogram showing the modern classification of several different "condylarth" families. They're shown as potentially occupying positions throughout the branches of the even-toed and odd-toed ungulates.

And while that’s the last main entry for this month, we’re not quite done yet. There’s still one weekday left in October, and after digging through so many taxonomic garbage cans there’s only one place we can go now.

…See you in the trash heap.

It Came From The Wastebasket #19: The Pterrible Fate Of Ptychopariida

The Ptychopariida were some of the earliest known trilobites, first appearing in the early Cambrian about 521 million years ago and surviving until the end of the Ordovician about 444 million years ago. They included some of the most numerous and common trilobite species, and were probably ancestral to multiple other major lineages – including the very last trilobites at the end of the Permian – making them incredibly important in understanding the overall evolution of trilobites as a whole.

…But this group is also one of the biggest wastebaskets in paleontology.

First established in the early 20th century, the ptychopariids seemed to have some fairly good defining characteristics based on their facial sutures, large thoraxes, and relatively small pygidia. But the group quickly became a dumping ground for a massive amount of Cambrian trilobites, eventually containing numerous different families, hundreds of genera, and many more individual species.

Actually figuring out their internal evolutionary relationships also turned out to be extremely difficult – so much so that some paleontologists working on them just gave up trying and arranged the genera names alphabetically instead!

Even cladistic studies from the 1970s onward struggled to make sense of these highly “problematic” trilobites, and any larger-scale analysis was a daunting task due to how huge and diverse the ptychopariid wastebasket had become over the years. Worse, some of the anatomical features the group had been based around were starting to look more like the result of a lot of convergent evolution across multiple lineages than any actual shared ancestry.

Efforts were still made at breaking up the mess, however, with better-understood sub-groups like the Proetida, Harpida, Asaphida, Trinucleida, and Olenida being gradually split off into their own separate orders over the course of the last few decades.

An illustration of Ptychoparia, an extinct trilobite from the Cambrian period. It has a large semicircular head with a pair of antennae, small eyes, and a bulbous "forehead" region. Its body has 13 segements, and its "tail" is small and shaped like a blunted triangle.
Ptychoparia striata

But even by the early 2010s what remained of the Ptychopariida was still paraphyletic at best, more of an “evolutionary grade” of early trilobites than a single lineage, with most of its constituent families also rather poorly defined. There was even a proposal to abandon the group entirely, stating that “it serves no scientific purpose” and that its orphaned contents should be considered “order uncertain” until their actual relationships can be untangled.

Today the “ptychopariids” are in dire need of a full revision – since they were the ancestors of many other major groups they’re still crucial for understanding early trilobite evolution. There may be a salvageable single lineage somewhere in the remains of this wastebasket, even if it’s restricted to just close relatives of the genus Ptychoparia, but until somebody tackles them properly they’re stuck in taxonomic limbo with their name only being used in a loose sense.

It Came From The Wastebasket #18: Lots And Lots of Lepidotes

Lepidotes was a ray-finned fish that lived during the Mesozoic, found in both freshwater and shallow marine environments. It was a member of the ginglymodian fish lineage, related to modern gars, and along with distinctive thick enamelled scales on its body it was also one of the earliest types of fish able to protrude its jaws for suction feeding.

First discovered in Jurassic-aged fossil deposits in Europe in the 1830s, this genus was quickly turned into a notorious wastebasket taxon for any similar-looking fossil fish. Over time dozens of different Lepidotes species were named, many of them rather dubious, from locations all around the world and spanning a time period of over 100 million years.

An illustration of Lepidotes gigas, an extinct fish related to modern gars. It has a somewhat carp-like shape, with a body profile resembling an elongated oval, a relatively large head region, small fins, and a forked tail. Its scales are rhombus-shaped and are arranged in tightly-packed rows along its body.
Lepidotes gigas

But despite Lepidotes being a wastebasket for almost two centuries, it wasn’t until surprisingly recently that any real progress began to be made on cleaning it all up.

In the early 2010s a large-scale review of ginglymodian relationships found that many “Lepidotes” species were either invalid or polyphyletic, belonging in completely different genera or families. True Lepidotes were restricted down to just the original type species Lepidotes gigas and a few of its closest relatives, all from the early Jurassic of Europe, while some other forms were moved into the newer genera Scheenstia and Callipurbeckia. Since then some other “Lepidotes” have also been reclassified, creating new names like Macrosemimimus, Occitanichthys, and Quasimodichthys.

There’s still work needing to be done on untangling all these Lepidotes-like fish – Scheenstia might actually now represent several different lineages, for example – but at least Lepidotes itself is now in a much better situation than it was just a couple of decades ago.

It Came From The Wastebasket #17: Getting Ornithomimus In Order

The ostrich-like “bird-mimic” dinosaur Ornithomimus was named in 1890, based on some hand and foot bones from Late Cretaceous-aged fossil beds in Colorado, USA.

The first ornithomimid known to science, it was initially thought to be a ornithopod, but then a few years later more fossil material revealed it was actually a theropod – and then it spent some time classified as a “megalosaur” before ornithomimids were finally recognized as being coelurosaurs in the early 20th century.

And for nearly a century after its discovery it was treated as a wastebasket taxon for any similar-looking fossil material from North America and Asia, with around 17 different species named within the genus. One of these was split off into Struthiomimus in 1917, but it wasn’t until much later that the rest began to get sorted out.

A review of known Ornithomimus fossils in the early 1970s renamed a couple more species into the new genera Archaeornithomimus and Dromiceiomimus, and dismissed most of the remaining species as dubious or invalid. Just two valid species now remained: the original Ornithomimus velox from Colorado, and Ornithomimus edmontonicus from Alberta, Canada.

An illustration of Ornithomimus, an extinct feathered dinosaur. It has a a small beaked head atop a long slender neck, two wing-like arms with three clawed fingers, long ostrich-like legs, and a counterbalancing tail with longer feathers towards the tip.
Ornithomimus edmontonicus

Since then opinions have gone back and forth about some of the other Ornithomimus species. For a while Dromiceiomimus was merged back into Ornithomimus, but more recently it’s been found to have distinct limb proportions and was probably actually a separate genus after all. Another species that’s usually considered to be part of Struthiomimus is also sometimes instead classified as an Ornithomimus instead.

Really all of the North American ornithomimids are in need of a modern taxonomic revision – especially since Ornithomimus edmontonicus shows enough anatomical variation that it might actually represent a species complex of multiple very similar forms, which might get split apart in the future if anyone can figure out how to reliably distinguish them.