American Scientist

The Devonian fossil Palaeospondylus gunni is one of paleontology's enduring enigmas. At 20 to 60 millimeters in length, it is just large enough to show that it is a fish, but not much more. Within 10 years of its discovery by two amateur collectors, the cousins Marcus and John Gunn, hundreds of specimens had been collected from the Achannaras slate quarry in Caithness, Scotland. Equally quickly, rival scholars eagerly and acrimoniously assigned it to every possible class and order. By 1903, the debate became so heated that Bashford Dean (curator both of Ichthyology and Herpetology at the American Museum of Natural History and of Arms and Armor at the Metropolitan Museum of Art) concluded in frustration: "Professor [Thomas Henry] Huxley, the astute morphologist, is perhaps to be felicitated in having had no traffic with it." He urged that it be put aside "where it will do the least harm."

Of course, Dean was ignored. Retreat is unusual among paleontologists, who always love a puzzle and rarely back away from a good fight. And now it seems that the problem with Palaeospondylus was not just in the fossils but in the pal(a)eontologists. Most early authors were determined to make Palaeospondylus into something very important from the evolutionary sense: an ancient kind of jawless fish, a new line of jawed fish, or a tantalizing "missing link" between ancient placoderms and sharks. For the purpose of their arguments, therefore, the greater the mystery—the more different it was from other fishes—the better.

Something New Out of Achanarras

Palaeospondylus gunni, so named by Ramsay Heatland Traquair of Edinburgh in 1890 (after its discoverers and prominent backbone), is abundant at Achanarras Quarry but is found nowhere else except for two nearby localities where a very few additional specimens have turned up. Not only is it minuscule in size, but its structure, insofar as one can make it out, seems unlike that of any known creature, living or fossil. It has a strange basket-like apparatus on its snout but no teeth, a well-developed vertebral column but no fins. In a 1992 Marginalia essay in which I reviewed the sadly inconclusive story of this little fossil, my own meager contribution was an allometric study of body proportions showing that it was a larval form. But its anatomy and taxonomic position were unresolved. 

Oddly enough, the possibility that Palaeospondylus is a larva had been mooted early in the case. Huxley, in famously having nothing to do with it, dismissed it as a "baby Coccosteus" (a placoderm arthrodire, one of those weird extinct forms with a large head articulated onto massive neck armor). Dean himself noted the change of head-to-body proportions with size so typical of immature forms of vertebrates. J. W. Dawson (principal of McGill University) thought it might be a "primitive tadpole." Professor Graham Kerr of Glasgow suggested that it was a larval lungfish. However, interpretation as a larva was not taken seriously again until 1980, when Peter Forey and Brian Gardiner at the Natural History Museum in London repeated the suggestion. Now a new study shows that Palaeospondylus is definitely a larval lungfish, and we can even be pretty sure that the corresponding adult was the famous Dipterus valenciennesi, equally abundant at Achanarras. It is instructive to examine why it has taken so long to get this far.

The Technical Approach

One of the early major contributors to making Palaeospondylus a bigger puzzle than it needed to be was Oxford professor William Sollas. He perfected a serial grinding technique and apparatus for using thin sections of fossils to create enlarged wax-plate reconstructions. Fossil after fossil was dissected in this way by Sollas and his daughter Igerna at Cambridge, including a graptolite, a Triassic dicynodont reptile and a Cretaceous sea urchin. And, in 1903, Palaeospondylus. They revealed a wealth of anatomical detail not visible with the microscope alone. But Sollas also led everyone down a wrong track. Unable to homologize the head structures he found with those of other fishes, he gave each its own new name—in Greek. Ampyx, tauidion, hemidome and gammation are some of the most memorable. The result was that everyone then accepted that the anatomy really was uniquely different from that of other vertebrates.

A couple of years ago I discovered that all of Sollas's material was still in the Oxford University Museum. He had sectioned several specimens at about 0.25-millimeter intervals and photographed the cut and polished surfaces using superb lenses, so that each was preserved as a glass-plate negative. Naturally I had the idea that modern computer manipulations of his sections would give us a better reconstruction of the anatomy. My assistant Bethia Thomas set to work.

But I was wrong. Sollas had captured just about every detail that was available. However, postdoctoral fellow Mark Sutton had been working with my colleague Derek Siveter on magnificent computer reconstructions of far more minutely sectioned Silurian invertebrates. So I asked Mark if he would like to have a go at Palaeospondylus, making new sections at only 30-micron intervals. The Natural History Museum in London gamely offered up three specimens to be sacrificed—eventually to reappear in virtual form. This time the result was amazing. One of the specimens was preserved relatively uncrushed and undistorted, and at last we had a new look at its structure. There were the tauidion, gammation et cetera in all their glory. But still the question remained: how to interpret such a set of structures apparently unlike those in any known fish?

Head Case

The possibility that Palaeospondylus was a larval form had always been rejected because of the well-formed vertebral column and what seemed to be a good bony skeleton. A larval skeleton should have been cartilaginous and therefore would not have survived. Unfortunately, our new sections confirmed that, as Sollas had already discovered, no histological details are preserved in the bituminized material (although, significantly, cellular details are preserved in the bones of other, definitely adult, fishes from Achanarras). But several Palaeospondylus specimens show evident cartilage and even connective tissue in the curious apparatus at the front of the head. Some retain an outline of the body. So the fact that the "bones" were well preserved did not necessarily prove that they were made of "bone." 

A prominent feature of the head of every specimen is a pair of strange rods, the occipital lamellae, projecting from the back of the cranium alongside the vertebrae. In our new reconstruc-tion, the shape of the articular facets showed that (as Dean had guessed) in life they were oriented backward and downward. They are identical with the cranial ribs found elsewhere only in lungfish. Kerr had pointed that out in 1930, but no one had listened until Forey and Gardiner revived his suggestions.

The next breakthrough came with the curious apparatus on the snout that had confused so many authors, including the most recent ones. A reasonable explanation was that it had something to do with the nasal apparatus: either the complex nasohypophysial structure found in lampreys and hagfish or the nasal capsules of a gnathostome. However, I recalled that in 1879 Alexander Agassiz had described the development of the gar Lepisosteus, a primitive teleost that has an almost identical feature to the anterior structure in Palaeospondylus. It is an attachment organ, a larval adaptation lost in later life.

If the rostral apparatus is identified as a larval attachment organ, interpretation of the rest of the head falls neatly into place. The hemidomes are the nasal capsules and the computer images show that Sollas's hemidome septa were their perforated dorsal walls, just as in modern larval lungfishes. The ampyx is the paired premaxillae, and the tauidion is the vomer. In that case, the big, chunky gammation has to be the palato-quadrate. And the first post-branchial plate is the dorsal part of the hyoid arch. The new reconstructions also show that the enigmatic anterior trapezoidal bar articulates with the front of the gammation. If the gammation is the palato-quadrate, then the trapezial bar is possibly a rudiment of Meckel's cartilage—the embryological basis of the lower jaw. Suddenly, nothing present on the fossil was unaccounted for and it was all consistent with Palaeospondylus being a larval lungfish.

The Oldest Vertebrate Larva

One of the difficulties about Palaeospondylus has always been that obvious vertebrate features such as the dermal skeleton are missing. However, Palaeospondylus also lacks the teeth, proper jaws and all the gill apparatus behind the first arch. It has rudiments of the limb girdles, but no fins. We also have to explain why Palaeospondylus specimens do not become progressively more similar to adult lungfish as they become bigger (an early argument against larval status).

All this makes sense if Palaeospondylus is a true larva with a sharp metamorphosis in the life history, rather than the sort of juvenile that slowly morphs into an adult. As a true larval stage, it would require special adaptations—to compensate, for example, for the absence of teeth and jaws. Among those is, obviously enough, the anterior attachment organ. We may then suggest, although we cannot prove, that Palaeospondylus was a larval form spending its life in shallow waters, attached to either a substrate or a plant. Like larval Lepisosteus (and modern larval lungfish where the attachment organ is on the throat), it lived partially from the reserves in the yolk sac, and partially as a detritus feeder on the attached surfaces.

There is only one species of adult lungfish in the Achanarras beds—Dipterus valenciennesi, a handsome fish reaching to 30 centimeters or more—but it is never found smaller than about 60 millimeters. Somewhere there must exist very young Dipterus fossils or very old Palaeospondylus that would settle the question. Or are there yet other explanations still waiting to be explored? The anterior attachment organ might be consistent with Palaeospondylus being a parasite. In which case it would not represent the oldest known fossil vertebrate larva, but the oldest known vertebrate parasite. Or even both. One of the things that keeps paleontologists (and palaeontologists) going is the possibility that just around the corner is a deposit with some superb new material that will answer all our prayers. Or start another controversy.