Evolution has a knack for confronting us with strange and unexpected questions. One of them echoed through the halls of the Collections Centre of the National Museum of Scotland, not too long ago:

"Why does a fish need a sacrum!?"

Lauren Sallan was peering through her microscope, studying a fossil specimen of Tarrasius, when she noticed something odd. Its spine was divided into five sections, as it is in every tetrapod, the group of four-limbed, land-dwelling animals that includes humans. The only problem: Tarrasius was a fish.

Modern fish don't have a sacrum, nor do they have a neck or thorax. Their vertebral column is a monotonous series of vertebrae strung together, with only minor modifications from head to tail. For us tetrapods it's a different story. Travel down our spine and you will encounter five distinct types of vertebrae, each with their own place, function and shape. First up are the cervical vertebrae of the neck, followed by the thoraic vertebrae that carry our ribs, the lumbar vertebrae, the sacral vertebrae wedged between our hip bones and finally the caudal vertebrae of the tail, for those of us that have one.

Since segmented backbones evolved in the early tetrapods that exchanged water for land, paleontologists assumed it was a specific adaptation to a terrestrial life. The enigmatic Ichthyostega, an ancient amphibian that lived between 375 and 360 million years ago, was amongst the first to waddle around with such a fully 'regionalized axial column'. Ahlberg, who reconstructed the Ichthyostegaskeleton in 2005, pointed out that Ichthyostega could probably flex the lumbar region of its spine up and down. This was a departure from the side-to-side wrenching that is so typical of fish. Ahlberg speculated Ichthyostega could walk, or at least crawl.

Update: computer models indicate Ichthyostega could not walk. Like the Tarrasius paper, this research was published today.

Case closed, but not quite. Tarrasius tells a rather different story: not all animals with necks and sacrums were landlubbers. Far from it, Tarrasius was a small and slender eel-like fish. It lived between 359 and 318 million years ago, in the shallow waters of what is now Scotland. Here it stalked the reefs, crushing the shells and carapaces of its prey with its molars, just like modern wolf eels do.

Tarrasius crushed its prey with its molars, just like modern wolf eels do. Photo Dan Hershman.

While Tarrasius might have looked and lived like a modern eel, it certainly didn't swim like one. Its lumbar vertebrae were compact, its sacral vertebrae huge and locked into each other, whereas the vertebrae in its tail were tiny, thin and scattered. Sallan think this unusual anatomy allowed Tarrasius to thrust through the water like a giant tadpole, sweeping its flexible tail from side to side. The rigid trunk would have controlled the forces generated by the beating tail. This is why this fish needs a sacrum.

Put this way, it makes perfect sense for a fish to have a segmented vertebral column. Yet this particular arrangement of vertebrae only evolved in Tarrasius and in tetrapods. How did two distantly related lineages stumble upon the same solution? Was there a common template, an ancient developmental program that unfurled in Tarrasius and tetrapod, but was ignored in others? Or did similar selection pressures shape their backbones in similar ways? In other words, is homology or convergence to blame?

There are signs pointing in both directions. For example, even though dogfish have no segmented spine, regulatory Hox genes in the vertebral column of developing dogfish embryos are activated in the same order as they are in tetrapod embryos, where they demarcate the boundaries of the segments in the tetrapod spine. Perhaps the recipe for a segmented spine was already present in the common ancestor of cartilaginous and bony fish.*Tarrasius and tetrapods could both have drawn upon this developmental potential, an axial patterning system more ancient than themselves. Now might be too early to verify or reject this possibility: developmental biologists have only just begun to chart the diversity of Hox gene activity in the animal kingdom. At the moment, Hox gene expression data is only available for a handful of creatures, few of them fish.

Acanthostega, one of the first tetrapods with limbs.

Regardless how deep the roots of these genes run, the segmented spine evolved twice, on independent occasions. What made Tarrasius and tetrapods converge on this particular body plan? Sallan suggests it was their similar way of lifen. Both the first tetrapods and Tarrasius united a short torso and heavy skeleton with a broad an strong tail. Tarrasius propelled itself forward with its paddle-shaped pectoral fins, tetrapods used their webbed limbs for the same purpose. And finally, both were bottom dwellers, navigating reefs and floodplains.

If this scenario is true, our distant ancestors evolved a segmented spine to be adept swimmers in shallow waters. Only later did their swimming rod modified to bear our limbs on land and make walking possible. This would not be the first trait our ancestors coopted and repurposed. The gait we think is so typical of land-dwelling animals evolved underwater. Lungfish can walk too.

Whether these speculations will hold up in the future is up to science, but I can tell you I find this more complex and detailed explanation immensely more satisfying than any simple story in which our ancestors ‘crawled on land’. There is no direction in evolution, no creature with the foresight to realize that opportunities await outside of the water. The first tetrapods were animals in their own right, with adaptations for their own sake and an uncertain future ahead of them. Their path was never clear-cut. The segmented spine that was vital to the terrestrial success of our ancestors, turned out to be a dead end for Tarrasius. Wrong place, wrong spine. Rest in science, little buddy.

*: Confusingly, tetrapods are bony fish. Blame the taxonomy.


Tarrasius reconstruction from first reference.

Wolf eel by Dan Hershman

Acanthostega from second reference.


Sallan, L., (2012). Tetrapod-like axial regionalization in an early ray-finned fish Proceedings of the Royal Society B DOI: 10.1098/rspb.2012.0784

Ahlberg, P., Clack, J., & Blom, H. (2005). The axial skeleton of the Devonian tetrapod Ichthyostega Nature, 437 (7055), 137-140 DOI: 10.1038/nature03893