Building Bodies, By Instalments

Perhaps the most interesting session at Euro Evo Devo was, for me, the one on the origin of segmentation - that is, the propensity for the body of an organism to be divided up into a series of more or less repeating parts. When William Bateson was studying the anatomy and development of acorn worms in the mid-1880s, he supposed that segmentation originated as a kind of oscillation, as of a train of waves. I wish I could pin down, now, precisely where he said this, but the point was made amply in his 1894 polemic book Materials for the Study of Variation, in which he classed segmentation as the basis for a kind of variation in repetitive parts he termed 'meristic'. Looking at the tree of life as it is conventionally viewed today, jut three among the thirty-something recorded phyla are regarded as segmented - annelid worms, arthropods and chordates. Now, in the Old Days, when Cromercrox was taught zoology, it was thought that arthropods evolved from an annelid-like ancestor, in which case the segmentation seen in earthworm and earwig were homologous - that is, evolutionarily related.

That all changed following a meeting I had in 1996 in the Wyndham Hotel near Los Angeles International Airport, with Professor James A. Lake of UCLA, who proposed a new molecular-based phylogeny of metazoa in which annelids and arthropods were fundamentally separated. Arthropods would be united with the determinedly unsentimental unsegmental nematodes and some other stuff in the Ecdysozoa, a clade of 'molting animals'. Lake and colleagues had already published (in Science) a molecular phylogeny linking annelids with molluscs and brachiopods into a clade called the Lophotrochozoa - the revelation of the Ecdysozoa paper was that the arthropods and the annelids were fundamentally separate. (The chordates, of course, were and still do belong to a quite separate group, the deuterostomes). At first, many people, including me, thought that the whole idea was loopy. But hey, what do I know? I am a mere hack. The molecules say different; the Ecdysozoa was published in my esteemed organ, and the Deuterostome/Ecdysozoa/Lophotrochozoa split of the animal kingdom has been borne out by much subsequent work.

What this means for segmentation is this: that each major animal group has one phylum within it that is habitually segmented. In the Lophotrochozoa, it is the annelids, standing amid a squishy morass of unsegmented molluscs and brachiopods. In the Ecdysozoa, it is the arthropods, rising above the unsegmented nematodes and a load of other stuff you've probably never heard of, and even if you did, you'd probably want to forget about it immediately afterwards. And among the Deuterostomes, it is the chordates - including we vertebrates - with our mesoderm divided into somites, each with its own vertebra, innervation and muscle block. If you don't believe me, just ask a salmon steak.

Now, this poses a nice conundrum - which my father says is different from two elephants sitting on a bagel, who have a bunundrum - and this conundrum is this: did the habit of segmentation originate in each phylum independently, in which case the common ancestor of all these animals was unsegmented, or was it a more general feature of animals that most creatures have lost, all except for the annelids, arthropods and chordates? This, I discovered, is a matter still ripe for debate, and what made the symposium at Euro Evo Devo so exciting.

To posit an independent origin for all three instances of segmentation would be by far the easiest option, given that the three phyla concerned are widely sundered by evolution, and the alternative is to suppose a wholesale loss of segmentation in virtually every other kind of animal. The ugly fact that spoils this easy idea, though, is that in all three cases, the process of segmentation utilizes pretty much the same network of genes - cognates of genes isolated long ago in the fruit fly and bearing names such as engrailed and wingless. These days it is fashionable to invoke the concept of deep homology in which structures in widely separated animal groups, although they look very different from each other and might have indeed appeared independently, are based for their engenderment on a similar cassette or module of interacting genes. This idea allows the invocation of segmentation in each of the three phyla to appear separate, even though it is based on fundamentally the same genetic substructure.

This, however, raises the question of why segmentation has not happened in all the other phyla, given that they all must share the same module or cassette of genes, and, to follow this line of thinking, why the ancestor of all three major animal groups - which would have also shared this cassette or modeule of genes - might not also have been segmented.

Truly, a question that would cross a rabbi's eyes, noch?

The solution, it seems, is to loosen what we think of as segmentation. For it is not the case that all animals other than annelids, arthropods and chordates are utterly without any sign of segmentation.  True, their bodies might not  be so rigidly divided into compartments in accordance with our usual requirements for segmentation - but they do, very often, show a repetition of parts along the body axis. Primitive molluscs, such as chitons and some other forms, show repetitive arrangements of shells, gills and so on, even though they are not usually thought of as segmented. Non-chordate deuterostomes such as acorn-worms show repeated gill slits. Even humble flatworms show repeated arrangements of gonads, gut diverticulae and so on. In the widest sense, therefore, there is a tendency for creatures to divide their bodies, to a greater or lesser extent, into a series of repeated structures, forming a continuum from the completely unsegmented to the fully segmented.

Perhaps Bateson had it right all along, if only figuratively. Perhaps there is a tendency within animal bodies to create distinct domains by means of wavelike morphogenetic gradients, whose results are not distinguished by a simple division into those animals that are segmented against those that are not - but simply by those animals in which the wave crests of such oscillators are higher or lower, more distinct or less. The common ancestor of Ecdysozoa, Lophotrochozoa and Deterostomes might not have been strictly segmented, or strictly unsegmented - it is probably impossible to know - but it would have had an oscillatory system of body partitioning that would have fallen out naturally from the interaction of cassettes or modules of genes. Like many revelations in science, it is not the data that change - but the way you look at them in the light of new evidence.