There’s something fishy about Archegosaurus decheni.

Living in the Czech Republic and Germany during the Early Permian, about 299-295 million years ago, this temnospondyl amphibian was a tropical freshwater predator occupying a similar ecological niche to modern crocodilians.

Hundreds of fossils have been found of this species, from 15cm long larvae (6″) all the way up to 1.5m long adults (5′), so we’ve got a very good idea of its life history and anatomy. Larvae had external gills and shorter blunter skulls, and as they matured they developed internal gills and lungs, and their snouts elongated into more crocodile-like shapes. Every life stage was fully aquatic, with very limited ability to venture onto land, and gut contents show their favored prey was Acanthodes fish.

But despite how much Archegosaurus looked like a salamander-croc, a detailed study of its physiology has estimated that its metabolism and body functions were actually much more similar to those of air-breathing fish like bichirs and lungfish than any modern amphibian.

This suggests that its whole evolutionary lineage had retained a lot of physiological traits from their earlier fish-like tetrapod ancestors, and many other early aquatic temnospondyls may also have been much less amphibian-like than we usually think of them.

(And since one hypothesis places modern caecilians as the descendants of this fishy lineage of amphibians, they may even still have living representatives around today!)


We have a fairly good picture of the evolutionary origins of most groups of aquatic mammals – except for the pinnipeds. The fossil record of early seals is still rather sparse, and for a long time the earliest known species was Enaliarctos, an animal that was already very seal-like and didn’t help much in figuring out whether seals’ closest living relatives are bears or musteloids.

But then Puijila darwini was found in the late 2000s, a transitional form with a near-complete skeleton, filling in a gap in our understanding so conveniently it almost seems too good to be true.

This is the equivalent of Archaeopteryx for seals.

Discovered in Nunavut, Canada, Puijila dates to the early Miocene, about 23-20 million years ago. It was a small freshwater otter-like animal, about 1m long (3’3″), with a long tail and webbed feet adapted for paddling with all four of its limbs.

It lived at around the same time as the more specialized Enaliarctos, so it wasn’t a direct ancestor of modern seals, instead being part of an early offshoot lineage that retained more basal characteristics – but it does gives us a clue as to what the earliest pinnipeds looked like. Along with genetic studies it also helped to clarify that seals’ closest relatives are indeed the musteloids, although they’re estimated to have last shared a common ancestor around 45 million years ago so there’s still a lot of time unaccounted for in the proto-seal fossil record.

Several other fossil species that were previously thought to be musteloids have now also been recognized as close relatives of Puijila, and it seems that they were a fairly widespread group basically filling the ecological niche of otters at a time before true otters existed.

Most surprising and frustrating of all, however, is that some of these other otter-seals actually survived all the way into the Pleistocene, only going completely extinct sometime in the last 2 million years.

We barely missed having them still alive today!


Between about 9 and 7 million years ago, the modern regions of Tuscany, Corsica, and Sardinia were once part of a single island in the ancient Mediterranean Sea.

And since evolution often goes in weird directions on isolated islands, it’s no surprise that some unusual species developed there.

One of which was a very odd duck.

A map of the Mediterranean region during the Late Miocene, showing the location of the Tusco-Sardinian island.
From fig 4 in Williams, M. F. (2008). Cranio-dental evidence of a hominin-like hyper-masticatory apparatus in Oreopithecus bambolii. Was the swamp ape a human ancestor?. Bioscience Hypotheses, 1(3), 127-137.

Bambolinetta lignitifila lived during the Late Miocene, about 7.5 million years ago. Known from a single partial skeleton discovered in the mid-1800s, it was initially thought to be a fairly normal dabbling duck and wasn’t properly re-examined until 2014, when its strange features were finally recognized.

It was a medium-sized duck, probably around 50cm long (1’8″), but it had much chunkier wing bones than its relatives, with noticeably shortened forearms – looking much more like the wings of an auk or penguin, and suggesting that it was a similar sort of wing propelled diver. This is incredibly weird for a duck, since every other known diving species uses feet for propulsion instead, and so Bambolinetta may be the only known waterfowl to ever develop this type of underwater locomotion.

It’s not clear whether it was still capable of flying or not. There were few predators in its habitat, so it may well have become completely flightless – and that could also be the reason it later went extinct. Sea levels in the region began to drop around 7 million years ago, reconnecting the Tusco-Sardinian island to the European mainland, and Bambolinetta‘s high level of ecological specialization and its potential island tameness would have given it little defence against an influx of new unfamiliar predators.


Coelacanths are represented today by just two surviving species, one in East Africa and one in Indonesia, both very similar in appearance and ecology to each other.

For a long time their lineage was thought to be all “living fossils“, retaining the same basic body plan for the last 400 million years – but more recent discoveries have revealed that these fish were actually much more diverse over the course of their evolutionary history.

Holopterygius nudus was a fairly early member of the group, living during the mid-Devonian about 385 million years ago. The only known fossil specimen was discovered in Germany in the 1970s, but it was originally thought to be a different type of fish entirely and wasn’t identified as being a coelacanth until over 30 years later.

And compared to its living relatives it was tiny, just 7cm long (2.75″), with a distinctive tapering eel-like tail. Its convergent close resemblance to modern cusk-eels suggest it may have occupied a similar ecological niche, living near the sea floor and hiding in tight spaces like crevices and burrows.