Does Luria-Delbruck really mean that mutations are random? Or might it indicate something else?
Luria-Delbruck is often cited as the de facto experiment showing that mutations are random, and providing a mathematical/experimental basis for such a claim.
The original paper is a fabulous application of mathematics and experiment to answer a class of biological questions. The exact question that Luria-Delbruck was attempting to answer is “does a given change occur in response to selection or does it occur independently of selection?” and the test is known as the fluctuation test.
This has often been used to imply that mutations are random with respect to fitness. But I think in fact it shows something else. I think that genetic change in general may be random with respect to individual selection, but directly related to population fitness.
The thing to keep in mind is the Luria-Delbruck was written before Watson/Crick discovered the double-helix, and long before Nirenberg actually cracked the genetic code.
Now that we know the genetic code, in order to say that a change is random with respect to fitness, the best way to do that is to (a) examine the number of nucleotides changed, and (b) compare that to the size of the search space within the genome to accomplish this change. So, if a change requires two amino acid changes to occur, the search space within the genome has about 17-18 digits behind it, depending on how you do the calculation, in the absence of selection (which is what we are looking at — change in the absence of selection — some mutations may kill the organism, which will limit the search space in that direction — but most will not). So, with a given mutation rate and search space, it will be nearly never that you will find a sequence of mutations to produce a specific phenotype that requires multiple amino acid residues.
Therefore, if you are continually getting a given mutation periodically, you can’t assume that the mutations are random, but rather the genome is biased to produce that mutation.
But what Luria-Delbruck does say is that these mutations are produced independently of selection. Why might that be important for an organism? The answer: survival.
Imagine that you had a population of bacteria that grew from a single parent. If this bacteria was just clones of the parent, then any change in the environment could easily kill the entire colony. So what does the colony do? The same thing that money managers do — establish a hedge. If the colony provides a set of variants which have different metabolic parameters than the parent population, it can be sure to retain at least some members of the colony in the face of strong future selection. In such a selection scenario, the mutants would then become the primary metabolic method, and the revertants would now be the deviants.
So, what is the best way to determine which ones are deviant? The answer — stochastically, with a low level production of variants. This will make the population produce “hedge” deviants without having to have colony-wide communication about who is going to do it.
So, it’s “random” only in the idea that they are produced stochastically from a set. It is non-random in that the set is a highly reduced set from the total potential mutations in the genome, geared toward those that are stable and provide good hedges against environmental change. So, indeed, evolution can “look ahead”.
In addition, just to point out, some mutations are certainly in direct response to selection.