Reader Comments

Post a new comment on this article

Explicit Landscape Details?

Posted by cwnelson on 07 Jan 2009 at 21:10 GMT

Very insightful article!

I have some questions concerning experiment 2:

(1) How is fitness determined under this scheme? It seems at times in the article that fitness is the result of the number of A or B instructions, whereas at other times it seems that the number of A / B instructions contributes to the mutation rate, which in turn has a fitness value.

(2) If the former is the case, how does the number of each letter contribute to fitness; if the latter, how do the letters determine the mutation rate?

(3) How is a smooth terrain enforced? It is unclear how a succession of ascending mutation rates could contain fitness valleys in one experiment and not in another, if the fitness is the result of the mutation rate. It seems that the fitness is not the result of the rate, but rather is being manipulated artificially?

Any insight would be grand.

RE: Explicit Landscape Details?

jclune replied to cwnelson on 09 Jan 2009 at 17:10 GMT

Thank you for your interest in our work.

In the experiments with explicit landscapes, fitness depended solely on the match between the environment and the number of a key instruction that organisms had in their genomes. In season A (left column) the key instruction was deleterious while it was beneficial in season B (center column). Rugged fitness landscapes with maladaptive valleys (rows 2–4) were
introduced by setting the fitness of organisms with intermediate numbers of the key instruction to the minimum fitness level of one.

The mutation rate never has a direct effect on fitness. It only indirectly affects fitness because it changes the rate at which genomes are mutated.

Does that answer your question?

Yours,
Jeff Clune

RE: RE: Explicit Landscape Details?

cwnelson replied to jclune on 12 Jan 2009 at 15:55 GMT

Thank you, this makes sense. I mistakenly thought that the x-axis indirectly represented mutation rate through the key instruction's effect on the rate, which actually does not exist.

I do have a few follow-ups:

(1) Why, under the explicit landscape scheme, were only organisms with the optimal rate placed on the higher peak (with proximal beneficial mutations possible)? Do you think it would be interesting to observe the evolution of those with the depressed rate from that starting point (or why was this excluded)?

(2) When starting from the higher peak, effectively 50% of mutations will be very beneficial; moreover, the selection coefficients of these will always exceed 11/10 (1.1). Do you feel that the unrealistic nature of the emergent mutation distribution and selection coefficients here effect your results' applicability to biology? If not, why?

Yours,
Chase

RE: RE: RE: Explicit Landscape Details?

jclune replied to cwnelson on 22 Jan 2009 at 17:42 GMT

Hello Chase-

In science, it is a customary and powerful approach to isolate and manipulate specific variables. The point of performing the experiments on explicit landscapes was to investigate whether the ability of evolving populations to attain the mutation rate that would maximize their long-term average fitness depended on whether the landscape was rough or smooth, all else being equal. Our results showed that was the case, with populations becoming "stuck" at suboptimal mutation rates on rough, but not on smooth, landscapes.

Best,
Jeff

RE: RE: RE: RE: Explicit Landscape Details?

cwnelson replied to jclune on 28 Jan 2009 at 21:11 GMT

Jeff,

Thank you; indeed, I learned a lot from the methods of the study.

Regarding the questions I asked, I'll try to be more succinct:

(1) Did you consider whether the optimal rate would evolve from the suboptimal rate, if the organism with the suboptimal rate was placed at the delta sign in Figure 3? (If not, that's okay; I am just wondering. The most intuitive reason I can think of for not doing this is that too few beneficial steps were available; the environment would have needed to fluctuate wildly.)

(2) 50% of all mutations will be relatively highly beneficial (always >10%) on the smooth landscape you implement. Do you think that a less favorable distribution with smaller selection coefficients would yield different results? (This seems possible, because even when beneficial mutations are always available, the proportion of mutations they constitute can affect how quickly they arise. Of course I could be misunderstanding something.)

Thanks for your time :-)
Yours,
Chase W. Nelson