Constructive Neutral Evolution

This is a good conversation starter on how neutral evolution alone can build complexity.


See this from Doolittle:

As we pointed out previously, machines of marvelous complexity such as lightharvesting antennae in photosynthesis, RNA and DNA polymerases and their attending initiation, elongation, and termination complexes, apparatuses for import, folding, and degradation of proteins, or the cytoskeleton and its motors, all might have grown to their current form through a process of CNE accretion. The same argument could apply to large and complex regulatory networks, which are often described as being ‘‘finely tuned’’ but might be better interpreted as ‘‘runaway bureaucracy’’ or biological Rube Goldberg machines where what could be a relatively simple task is performed though many steps by an unnecessarily complex machine.

See this @colewd (but don’t recycle any arguments please), from Moran:

For example, there is good reason to think that the evolution of the complex spliceosome that removes introns has evolved by mainly non-adaptive evolution.

1 Like

Some comments on Larry’s post do raise interesting questions. How likely is the scenario laid out in his post and in the graphic you shared above to occur? The argument is IC is unlikely to evolve. Not that it can’t. Moran seems to think it isn’t unlikely at all. But I’m curious how he reached that conclusion

1 Like

I would agree.

Neutral evolution is happening at a very high rate, and there is nothing to prevent unnecessary complexity from arising. So it does. They cite evidence from protein interaction networks. I’d agree.

Looks like Behe actually responded to one of the papers presented here:


From Kimura:

Is 4N a loooooong time for a neutral mutation to become fixed like Behe said it is?

Except it doesn’t need to be sequential. Mutation rate equal fixation rate.

That depends on whether you are calculating the odds of any such structure to result from the accumulation of mutations, or the probability of a particular example of it.

What was the likelihood that you were born with the particular set of 100-150 mutations that you were? It’s something like 1 in 10^1000

Looking after the fact at some particular combination of mutations that “would have had to occur”, in order for some particular example of molecular complexity to have happened, is meaningless. You’re always going to get some extremely unlikely combination when you specify the set of mutations you want to see evolve. But those ~100 mutations you were born with really happened. And we would have calculated that the odds of those ~100 mutations were approximately 1 in 10^1000, as any such a set is.

Now compound the probabilities going back 10.000 generations. Ten thousand consecutive generations of sets of mutations each with an independent probability of ~1 in 10^1000 is approximately 5.7x10^-10080000

That’s about six times ten to the negative ten millionth power.

What does this number tell you?

We can’t know beforehand what mutations will happen, and if we pick out some imaginary combination and try to calculate the odds, we get an impossibly low number. What are the odds that gene 7 on chromosome 19 will be contained in some duplication by retroposition and end up under control of a promoter downstream of gene 582 on chromosome 4? What are the odds that it will later have a transition mutation from C to T in position 613?

Any particular instance of such changes will look unfathomably unlikely after they occurred. How did this one happen out of all the possible ones? It’s meaningless.

What we can say is that similar such events will occur with some regularity, but any particular instance of it will look incredibly unlikely after the fact.


Is CNE then evolution that is within the reach of pure chance?

This model provides an explanatory counterpoint to the selectionist or adaptationist views that pervade molecular biology

Pervasive as in the reigning paradigm?

Conventional wisdom holds that complex structures evolve from simpler ones, step-by-step, through a gradual evolutionary process, with Darwinian selection favoring intermediate forms along the way.

Conventional wisdom as in … paradigm?

Almost everyone who writes about constructive neutral evolution understands that it poses a problem for those who cling to adapatationist or selectionist views of evolution.

Is he talking about IDists?

Evolution is not pure chance. I’m not sure pure chance is even a real thing.

When you say “similar such events” what are you referring to?

What do you think @mung? It is clear from context.

Do you ever press ID advocates and creationists this way?

Probably things like protein interactions

1 Like

Consider the difference between these two questions:

What is the probability that this specific gene G will be duplicated, and then later mutate through this specific loss of function mutation L, and that this particular other protein P will buffer against the loss?

What is the probability that some gene will be duplicated, and then later will mutate by some loss of function mutation, and that some other protein P will buffer against the loss?


You start out with something that looks is extremely unlikely. Say a change from one base to a different base at a specific location. So it only looks unlikely because we are restricting ourselves to one location. But changes just like that are taking place throughout the genome, so it’s always happening somewhere.

But this is rather beside the point, because we are supposed to be talking about constructive changes. It is your position that neutral constructive changes are taking place all the time so there’s nothing improbable about them?


I don’t know what “all the time” means here. My point is that for the probability argument against such scenarios to make sense, we’d have to know at least the approximate frequency with which the general case happens, as opposed to the odds of the specific.

How often are genes duplicated? How often do duplicates suffer loss of functions mutations? How often do duplicate genes interact in buffering ways with other proteins? If we can answer these general questions we can calculate the frequency with which such events in general would be expected to occur given population size and so on.


Good questions @mung. I do think that constructive and neutral changes are very common. Any individual one is rare, the chance of having one from the whole class, however, is perhaps even quite high, depending on the precise definition.

Ironically this goes hand and hand with the ENCODE data. We see an immense amount of biochemical function that is essentially neutral. As we study this in populations we will see variation in this function, some which is getting fixed over time. Most of it is neutral, but it is also constructive, increasing complexity as it goes.

So how many different types of routes are there to IC structures. You have your more direct route with a mullerian two-step event where there’s positive selection at every step (if I’m understanding that event correct correctly, a more indirect route with an instance of exaption, and completely by chance as laid out by CNE. Good rough summary?

1 Like

Mullerian two step does not require selection (see my notes on it). CNE is exactly the same thing, but on a system level.

Start with a system. Add a new interactions/parts somewhere (very easy to do), at some point along the line make some of those interactions/parts necessary (very easy to do). CNE is a mullerian two step. Very easy to do.

On easy way to add a new part is duplication with divergence.

Interactions are very easy to evolve because it is the chance times the square of the number of parts. It is very likely to happen some where some how in a complex system.


Not sure I agree with that. How do you make a part “necessary” without involving selection?

1 Like

Neutral drift.