About 370 million years ago, at a time when creatures like Tiktaalik were wallowing in the muddy shallows of freshwater lakes and rivers, the sea covered much of what is now North America and Europe. These waters were home to a diversity of fish, but among the most fearsome was Dunkleosteus, a 20-foot long terror with sharp plates around its moth that created a shear that could slice through the body of any animal unlucky enough to be its prey.
The creature we now call Dunkleosteus used to go by a few different names, however. Back in the late 19th century, when its remains were still incompletely known, some Dunkleosteus fossils were named “Dinichthys”. Along with other armored giants like Titanichthys, “Dinichthys” was at one time considered to be a very specialized lungfish closely related to modern forms like Lepidosiren and Protopterus.
This view was initially proposed by the paleontologist who named and carried out early research on Dunkleosteus (which he named Dinichthys at the time), J.S. Newberry, and it was hard to deny the similarities he pointed out.
At first glance the striking visage of Dunkleosteus might seem to have little in common with the small-eyed, placid expression of Lepidosiren, but naturalists were struck by similarities of the skull. The group to which Dunkleosteus belonged, the arthrodires, were a group of armored fish and much of what we know of them is because their thick body covering has been preserved. As stated above, the cutting “teeth” of Dunkleosteus are really modified armor plates (although it does have very tiny actual teeth within its maw). Lungfish are not covered in the same sort of armor plating, but if we were to remove the scales, skin, and muscle from a lungfish head we would see a skull superficially very similar to that of Dunkleosteus.
Indeed, the skull of Lepidosiren and its close relatives appeared to be formidable enough that rumors of gigantic, predatory lungfish inhabiting the Amazon were at least considered for a time. Naturalists searched in vain for the creature known as the minhocao, and while it is unlikely that any such creature existed, perhaps the connection of lungfish with extinct monsters like Dunkleosteus made rumors of its existence all the more tantalizing.
But what could account for such similarity? Arthur Smith Woodward thought he found the answer when he placed Dunkleosteus and its relatives within the group containing lungfish in his1898 Outlines of Vertebrate Palaeontology. Detailed accounts, like one made by C.R. Eastman affirmed a close evolutionary relationship between the arthrodires and the lungfish, and it was even possible that Dunkleosteus might even qualify as a transitional form important to the evolution of lungfish (and thus early amphibians). As a short report of a lecture Newberry gave on this subject said;
The determination of a relationship between this singular fish and the Dinichthys is of the highest interest, connecting one of the most powerful denizens of Devonian seas with the present time, and affording a new illustration of the decay which some orders once most perfectly developed have suffered.
Not everyone agreed that Dunkleosteus was a highly derived form of lungfish, however. In his comprehensive 1895 work Fishes, Living and Fossil Bashford Dean explained the “killer lungfish” hypothesis at length but emphasized that it was provisional. In a footnote at the end of the section describing the arthrodires he wrote;
The writer believes that the Arthrodirans may as well be referred to the sharks as to the lung-fishes; as far as existing evidence goes, they certainly differed more widely from the lung-lishes than did the lung-fishes from the ancient sharks. They may, perhaps, be ultimately regarded as worthy of rank as a class.
David Starr Jordan was likewise perplexed by the placement of Dunkleosteus and its kin. In his 1905 book A Guide to the Study of Fishes he wrote that the jaws of the great fish resembled those of a lungfish, that much was true, but this similarity was only skin deep. The bones of the skull of Lepidosiren and the armor plates of Dunkleosteus were not homologous, meaning that the jaw shape was not the result of a shared ancestry. They had instead been independently adapted into similar shapes in a case of evolutionary convergence.
Indeed, Dunkleosteus and Lepidosiren were not so closely related as Woodward had thought, and the striking similarity between their jaws was largely forgotten. Eventually, with continued study and the accumulation of more fossils, it became understood that Dunkleosteus and its close relatives were placoderms, armored fish that flourished from about 420 to 370 million years ago. Lepidosiren, by contrast, is a living representative of the fleshy-finned sarcopterygii, the larger group in which our tetrapod ancestry is rooted. This means that Dunkleosteus and Lepidosiren did share a common ancestor among the early jawed fish, but this was nowhere near the close relationship proposed on the basis of jaw structure.
Despite our modern understanding, however, it may be beneficial to use the debate over the affinities of Dunkleosteus as an instructive historical case. Were the resemblances between its jaws and that of the lungfish a result of convergence or shared inheritance? How could the relationship of these creatures be confirmed with so little available evidence? These questions vexed paleontologists at the turn of the 20th century. Rather than an embarrassing episode, however, the discussion surrounding these fish illustrate how science proceeds and self-corrects as new evidence becomes available.
*[The "degeneration" of certain lineages was part of a concept called senescence. Some naturalists thought that just as an individual ages and dies, so too did groups of organisms, in this case the vigorous stage being represented by Dunkleosteus and the "degenerate" stage by lungfish. This had more to do with subjective values and the search for an explanation of extinction, however, and work by 20th century paleontologists like G.G. Simpson trashed senescence during the time that the modern evolutionary synthesis was coalescing.]