You can simulate just about anything in a computer. Doing it with *real* DNA means running fairly close to real time.
Data point: Each new pesticide germicide antibiotic what-have-you is easy to defeat in the lab. Not the ideal case, but it helps establish a baseline. Here’s how:
1. Lay out a growth medium for the pest species to eat and thrive in, and seed it with the germ-to-be-killed.
2. Make many fine rows of the pesticide, say very very little in the outside rows, a bit more, etc. up to a good dose down the middle row.
3. Come back in a day and see what’s going on. Usually the middle row will be dead, and the outside rows will be so-so, then graduating toward dead in the middle.
4. Replenish the growth medium’s nutrients but leave the pest species in place; replenish the pesticide, also in place.
5. Rinse, cycle, repeat. At the end of a few dozen cycles, the center row may show some effect, but by then the constant addition of pesticide will make it toxic for just about anything. The outer rows will show a new version of the germ species that thrives on the stuff.
In short, you have sped up evolution, not to create a new species, but to armor-plate a private version of the pest species.
Researchers go yum-yum-yum when they get their hands on one of these, because there will be so much more to learn about its metabolism and how it has responded to the pesticide. It helps to learn how to tell, out in the field and five years from now, what kind of resistance has cropped up in the real world.
But new-species evolution tends to include things like taking chromosomes apart, making new ones etc. *That* level of evolution takes tens of thousands to millions of generations, not dozens.
Speed like that can be simulated, in a deeply crude fashion, via computer algorithms. You need to presume some approach to how many mutations occur naturally per generation per billion DNA codons, and you would need to know a great deal about how the non-gene parts work to shape, expose, express, etc. govern the operation of the gene parts.
We also have epigenetics, which operate like a cheat-sheet or cook’s adjustments to a generic recipe: “IF AT HIGH ALTITUDE MAKE THE FOLLOWING ADJUSTMENTS TO LIQUIDS AND TEMPERATURE.” Epigenetics appear to ‘take notes’ on the environment at hand, then fine-tune which genes are turned down or off, and which are turned on or up.
Human mothers who are pregnant during times of famine bear children which are likelier to become obese as adults. It’s easy to see the strategy behind that – if food is scarce, program the next few generations to eat and save all they can. Epigenetics is the mechanism that does that.
(Warming up to a neat problem to solve) The simulation models we can make today have no way tool that can handle a real genome for something and begin to stress it, shoot mutations at it, and play out whatever effect that might have on the next generation. But once we can, – –
Most mutations will be failures. A few can confer an advantage, a new way of reacting to the world. In earth’s historuy, evolution that creates novel species has tended to happen when some significant geologic or climatic event isolates a species into multiple disconnected populations. Contiguous populations may experience drift, but will tend to arrive at a state which fits the current environment so well that ‘good’ mutations just don’t happen that much. Their skeleton, skin, musculature, yada and yada stop changing, and their fossil record stabilizes.
A change that separates one species into multiple non-interbreeding populations tends, one can see intuitively, to destabilize each of their environments at the same time. Each population should reach its own new optimal state – the duration of this in generations is huge, but measured by geologic time is pretty small. This phenomenon has a name: Punctuated Equilibrium. That’s why so few “missing link” fossils occur. The fossil record has very little opportunity to save mid-change forms simply because they happen too fast for the geologic fossil record to capture them.
Or another way, the Galapagos Islands acquired some starter population of finches at a time when no birds existed there. Over time the parent set separated into ‘occupations’ like seed eater, insect eater, nectar eater, fruit eater, and so forth. Each one developed its own bill shape, body size, etc. to handle that niche.
When Darwin arrived he found that they had adapted into populations which could interbreed but which did not, because each self-selected population had its own adaptation for seeds, fruit, bugs, etc. as its primary diet. An interbred clutch of eggs would produce ill-adapted chicks. They could not survive, so the tendency to interbreed died with them. Given a good geologic time span they would have diverged to the point where interbreeding would simply fail.
Can we model evolution in the lab? Maybe once we get good enough at simulating DNA on a computer. It’s easy to observe in retrospect via the fossil record and in tiny one-or-two gene adaptations. Doing that in the lab, to generate brand new species, would take hundreds of human lifetimes; to do it in a computer would make at least a little more sense, but would also require us to model working DNA, protein-folding, complex enzyme interactions, and a thousand yada’s.
Computers keep getting smarter – in twenty years we might ask one to think about it.
Oh, and Another Thing 