This is really a report from the frontal lobes. It' s not coherent yet, if ever, but I wanted to share.
I have been aware for some time of the work of Sergey Gavrilets
, at the university of Tennessee in Knoxville. He has been doing what seems like abtruse mathematical modeling of speciation and adaptation in various kinds of fitness landscapes, which he terms amusingly "holey", "rugged" and "smooth" and so on. But being a math-challenged individual (I blame my fourth form teacher, who managed to destroy my confidence totally. I am sorry to say that both my children have encountered his descendents), I was unable to make much of it, apart from seeing it as significant.
Last year, Gavrilets released a book entitled Fitness Landscapes and the Origin of Species
, through Princeton. I didn't go look for it, because speciation, the process whereby new species are formed, was tangential to my interests in species concepts for my own book. But last weekend, I was driving a bunch of philosophers of biology about, and one of them, Jonathan Kaplan
of Oregon State (the picture with the Koala was taken when he was here) happened to mention Gavrilets' work. I got interested again, and now I'm reading it. The book has some maths, but I can "bleep" over that, as Linus told Charlie Brown he did with the Russian names in War and Peace
. The result is some evocative ideas.
One of them is this. A fitness landscape
, a metaphor introduced by Sewall Wright
back in the 30s, is best conceptualised, says Gavrilets, as a hypercube of n
dimensions, one for each locus on a genome. Each point in the "space" (technically known as a "state space" or "phase space") represents a combination of some alleles in a population, and it has a fitness value assigned to it by the environment.
A biologically realistic fitness landscape will typically have millions, if not billions, of dimensions, for each possible gene and alleles. It follows, according to Gavrilets, that in a suitably complex space, there are going to be "ridges", or as I prefer to think of it, corridors in that space which are pretty much the same fitness values, and which are the most fit in that region.
The result is this: selection will tend to maintain a population at the fittest local "peak", but there is a way in which ordinary genetic drift - random collations of effects like mating chances, stochastic sampling of gene pools, and environmental noise - will enable a population (and hence the species made up of these populations) to wander about in the fitness landscape. In short, both
selection and drift cause biodiversity in a way I hadn't previously understood.
What this means is that selection keeps organisms more or less adapted (there are lag effects, "you can't get there from here" situations, and competing fitnesses for different genes that almost guarantee that no organism will be entirely fit or well adapted), but the form
of the adaptation suite will vary in a random manner.
These nearly-equivalent corridors also enable species to escape local adaptation peaks. If they are connected as networks, as Gavrilets shows they are, then once you get to another place in the "corridor", you may find a different branch that will enable you to ramp upwardly in fitness.
As I said, I'm still digesting a lot of this. Gavrilets also has a very nice conceptual mapping of the different processes of speciation. I'll address this later, I hope.