Is natural selection omnipotent or are there developmental constraints to what is possible and it is only from a limited range of possibilities that natural selection has to choose? The tension betwen two schools of thought (sometimes thought of in terms of pro-Gould and anti-Gould, as he has written much about developmental constraints and against vulgar adaptationism) is still alive and well. It is nice to see someone actually do an experimental test of the thesis:
Why are there no unicorns? Perhaps horses develop in a way that cannot be easily modified to produce a unicorn, so such creatures have never arisen. Or maybe unicorn-like animals have been born in the past but because there is no advantage for a horse to have a horn, such creatures did not thrive and were weeded out by natural selection.
The problem highlights a general issue in evolutionary biology of what determines the range of plants and animals we see compared to those that might have evolved theoretically. To what extent does observed biodiversity reflect the rules of development or the action of Darwinian selection?
To address this problem, Enrico Coen at the John Innes Centre and Dr. Przemyslaw Prusinkiewicz and colleagues at the University of Calgary analysed not Unicorns, but a more tractable system, the evolution of flower branching displays, or inflorescences. Flowering plants have three basic types of inflorescence - racemes, cymes and panicles.
Theoretically there are many other possible branching arrangements so why has nature chosen only these three? The researchers showed how the three types arise quite naturally from a simple mathematical model for how growing tips switch to make flowers. The model was supported by experimental studies on genes in the garden weed Arabidopsis.
That was the basic theoretical background. Now, what did they actually do?
The poet Dylan Thomas wrote, "The force that through the green fuse drives the flower drives my green age." Now, a team of international scientists has unlocked some of the secrets of that force: it has described the rules that govern how plants arrange flowers into branching structures, known in technical terms as 'inflorescences.' Nature has literally thousands of examples of inflorescences, which include the flower clusters of Mountain Ash, the tiny filigreed blossoms on Lilac and the stalkier inflorescences in Fireweed.
Dr. Lawrence Harder, a University of Calgary biologist and co-author of the paper, says one of their model's key features is that it is able to anticipate regional variations in inflorescence structures and recognizes that some developmental patterns are impossible.
Nice. I guess Gould was right after all. He would be pleased with this study, I bet. I am.
Who needs unicorns when there are narwhals?
Perhaps the morphospace of possible forms could be further refined by delimiting the smaller set of probable forms within it by applying the constructal principle, symmorphosis, and life history theory.
Rupert Riedl discussed impossible horns in Order In Living Organisms (1978).
I think this "morphospace constraint" idea is bullshit, not to put too fine a point on it. NS is an improbability generator, but that doesn't mean that all morphologies are equally probable from a given starting point. If a lineage has horns that are bilaterally triggered, then unless there is a convergence of those growth fields in successive developmental cycles, it's unlikely that NS will lead to a single combined horn growth field. And in fact rhino horns are, I think, based on the convergence of growth fields for hair along the median line.
The "horse unicorn" is a relatively modern invention. The medieval unicorn was a goat. And goats have been known to occasionally have the developmental "defect" of a single horn. So sorry to rain on everyone's parade, but unicorns do exist.
I heard there were unicorns once upon a time, but a roving band of squat, limp-dicked, Oxycontin-addicted conservative radio hacks slaughtered them on a rumor that ground unicorn horn would give them superboners.
"I think this "morphospace constraint" idea is bullshit..."
"NS is an improbability generator, but that doesn't mean that all morphologies are equally probable from a given starting point."
True. Similarly, nobody believes that selection is "omnipotent." (If anyone is tempted to cite Dawkins and Dennet, see the former on kaleidoscope embryos and the latter's remarks on developmental constraints.)