On the ‘Evolving Complex Adaptations’ thread, a side-discussion arose with CharlieM over Behe’s ‘CCC’ argument. In summary, Behe places an event of probability 
as the upper bound or ‘Edge of Evolution’. If a specific single mutation has a probability of 
of arising in any one replication event, a specific double mutation has a probability of 
, a triple 
, and a quadruple – that is, if four independent changes must happen simultaneously before a particular step is achievable, then that step cannot realistically have occurred in the history of life on earth. Behe thinks he has found a case with a probability in the resistance to chloroquine in the malaria parasite Plasmodium falciparum.
He cites as support for this this review by Professor Nicholas White, in which he makes the following statement:
Resistance to chloroquine in P. Falciparum has arisen spontaneously less than ten times in the past fifty years. This suggests that the per-parasite probability of developing resistance de novo is on the order of 1 in 1020 parasite multiplications.
According to Behe, given that this requires two specific mutations in the transport gene PfCRT in the first instance, we require two 
events to get one ‘CCC’ (a ‘Chloroquine Complexity Complex’). Two CCCs, ie 4 simultaneous changes, and you hit the edge of evolution. If it takes an organism with a population numbering in the trillions to gain a single CCC, how much less likely is it for populations numbering in the thousands or millions, such as multicellular eukaryotes?
Numerous blogosphere heavyweights have weighed in on this argument – Dawkins, Coyne, Myers, Ken Miller, Moran – and much of the discussion was done and dusted in 2007. But as late as 2014, Behe seemed still to be claiming to remain unrefuted (“show us your numbers”), and evolution critics continue to turn up and say the same. You explain why the argument doesn’t fly, and the response is always “… but you haven’t refuted him”.
There are several arguments that can be made against the analysis – the possibility of multiple pathways and the dubiousness of deriving universal principles from a single selective scenario in one organism, to name but two – but I wish to focus on the probabilistic argument, and its relationship to population factors.
White himself says that the per-parasite probability of developing resistance is 
. He doesn’t (AFAIK) explain how he derives this figure, but if he treats the two mutation probabilities as independent, he makes the same mistake as Behe. How those replications actually occur – serially, simultaneously or the combination of both – is crucial, and the raw number is insufficient. Pace White, I don’t even think a per-cell rate is really meaningful except for the unrealistic scenario of a single round of replication for a population of 
individuals (giving rise to 
in that step). To take other extreme possibilities, a lineage of populations each one cell in size replicating times, and another allowing unrestrained expansion until cells had existed – all would experience the same number of double mutants. But the likelihood of the double mutant existing in the final population would be substantially different, and that is what ultimately matters. Chloroquine changes the game somewhat, but let’s ignore selection for now.
There certainly is a per-cell probability of both mutations arising during the same genome-replication, and for all I know it is . But this is not the sum of all sources of that particular chloroquine resistance genotype except in two very restricted scenarios – either the single-step replication of a population of individuals referenced above, or a more realistic population in which the single mutants were lethal. Only if they are lethal can they not build to any frequency in the population, and so must both happen in the same round of replication. It is then an independent probability, that of sampling a DNA strand from a pool to experience mutation A, replacing it, then of sampling the same strand again to get mutation B. If you sample and mutate until you have one double mutant, you will have also generated 
dead singles. All other scenarios involve the complexities of dependent probability, and depend upon population considerations.
Real populations include drift, recombination and repeat mutation. Even if the single mutants were mildly deleterious, they can be promoted both by drift and mutation to an appreciable frequency within the current population – subpopulations A and B. The total size of this single-mutant population A+B will be the deciding factor in the probability of the second mutation existing in the same genome copy as the first. The relative sizes of the individual A and B subpopulations will also be a determining factor in the possibility of the double combination arising by recombination in meiosis, because these organisms are sexual.
The probabilities are therefore dependent. It is way beyond my mathematically-challenged brain to work out what the dependent probability is, but it will be the contributed to by the double-mutant-in-one-replication scenario, the probabilities that the single mutants will get to particular frequencies, the probability that the second mutant will occur given that frequency, the probability of recombination between alleles a specified chromosomal distance apart in these subpopulations, and the local selective environment wrt chloroquine and virulence. That is, the only probability Behe will allow is by far the most trivial of the lot. Do we find the single mutants in the population at any frequency? If we do, then even if the first mutant has probability , the double is clearly not , in that population.
What we can actually say is that the probability of developing chloroquine resistance in 50 years, on the evidence, is likely to approach 1. It happened c10 times in 50 years, so it’s a pretty safe bet the 5-year probability is more than 0.5. That or we’ve got at least 10 heads in a row. If we have, we wouldn’t want to start deriving universal evolutionary principles from it!
Effective Population Size.
Behe uses the high population sizes of this single-celled eukaryote to try and disbar smaller populations of larger organisms from evolving. He suggests you need big numbers to get Close to the Edge. But how big are they? Yes, they produce vast numbers of cells. But so do we – they just don’t all survive. And neither, it turns out, do most of the cells of Plasmodium falciparum. Essentially, an inoculum of about 10 sporozoites is injected to start the infection; these replicate asexually to produce numbers of the order of 
, and then a small sample – a few tens – is taken up by the next one or a few biting mosquitoes. Various things happen in the mosquito, but not much increase.
In any mature single infection, mutants A and B will both have been produced about 100 times each. In every billion mature infections, the double mutant will also have been produced once. But in neither case are they guaranteed to persist. First, they have to get out. Getting out depends on your host being bitten again before dying or recovering, and on your frequency in the local population – in the human in which you first arose. Without selection, if the mutation happens on the first division, it will be in 50% of late-stage cells, but will only happen on the first division in 
people. Second division, it will be in 25% of cells and this will happen in 
people. And so on. The earlier it happens, the more likely it is to get out – but the less likely it is to happen. Of course, this means that the single mutants are less likely than in an uncontained, non-bottlenecked population. But it also means that drift is stronger. The single mutants are more likely, at least periodically, to reach significant numbers, even if they crash again.
Much the same happens with the multicellular germ line. A series of mitotic replications rapidly amplifies the diploid genome, which then undergoes a reductive division (meiosis) to produce a gamete. In a human female, there are about 7 million; over the lifetime of a male, about 5 billion. During the course of this, a significant number of mutations is produced, but that larger set is effectively randomly culled. There is a ‘gamete population bottleneck’ in every individual. As luck would have it, one of the malaria population bottlenecks is also based upon human individuals. It is not legitimate to equate the census population size of Plasmodium cells and that of human individuals, due to the effect of bottlenecking. The correct parameter for comparison is effective population size. This is not a personal eccentricity; it was introduced by Sewall Wright in 1931, to try to discount those census individuals that do not contribute to evolution, and arrive at a more ‘correct’ value for population size than simply counting numbers. Plasmodium discovered a CCC 5-10 times in 50 years with an effective population size (something which can be approximately measured) of the order of 15,000-50,000, not LateStageCells * InfectedIndividuals = VeryBigNumber. Choosing an organism that is bottlenecked by humans to try to assert evolutionary limits on humans is a subtle way to shoot oneself in the foot.
The probability of chloroquine resistance is approximately 1, on the evidence.
I use Darwinian because I think the term is apt. Resistance appears because random changes to the organism give it an advantage which facilitates its spread throughout the population. Random changes are selected due to a changed environment. If you think this is not a classic example of Darwinian evolution in operation, can you explain why?
We both agree that like Newtonian physics, Darwinian evolution is limited in scope. Newtonian physics cannot account for the orbit of Mercury and Darwinian evolution cannot account for the evolution of life on the larger scale.
Resistance is a good example of unguided evolution. I would say the evolution of the bones of the inner ear are a good example of guided evolution. It is a feature of animals at a higher stage of evolution that they can reveal their inner feelings through sound. In order to achieve this they need to be able to both receive and transmit sound. But humans do not just express their inner feeling through sound. We can also convey rational thought and abstract concepts through sound. In order to do this we possess the vocal equipment to produce an infinite variety of sounds, the hearing equipment to distinguish all the subtle variation, and brains capable of interpreting these inputs. Three separate systems working in such a harmonious way which they must do if any physical organism is to objectively contemplate itself and the world around it. Every human zygote begins with the potential to develop into a free thinking adult and earthly life began with the potential to produce free thinking creatures.
This is not a haphazard process of separate parts evolvinging blindly in their own direction, it is a co-ordinated development of systems within the whole.
There are so many to choose from. Behe’s example of the bacterial flagellum makes a very good example, or if you prefer something more macroscopic, a bird’s wing.
Although we need to keep in mind that we cannot consider the wing on its own detached fron the whole organism. It belongs to the nature of an eagle that it wishes to free itself from earthly heaviness, it’s whole form is to that end. It uses the basic form of the pentadactyl limb to fashion a structure that is in keeping with the nature of the whole. The eagle expresses the archetypal form of the pentadactyl limb in a way that accords with its own nature.
As such, can you make a prediction using your idea about something that is currently unknown? Postdictions are easy.
To believe that mutations shape life comes from a reductionist mindset. As you will have gathered I come at it from a more holistic angle.
Which is great, but is your angle productive?
I am doing the opposite. The evolution of resistance is a case of breaking things down. In order for life to develop form has to be built up. I’m not extrapolating one from the other I’m contrasting one with the other.
Behe used the example of a few protein to protein binding sites. Do you have any examples of unguided evolution producing anything as complex as four or five protein to protein binding sites where any intermediary does not have a selective advantage?
You are right, its not the probability of CR occuring. It is an estimate of the actual number of times it has occured throughout the population.
What evidence do you have that the single mutatnt K76T in PfCRT is anything but lethal?
From The article – A critical role for PfCRT K76T in Plasmodium falciparum verapamil‐reversible chloroquine resistance.
They could not get parasites with only the single mutant K76T to reproduce so how is is going to multiply by drift?
Do you have a reference for the first statement?
No, it’s not that either. We know how many times it occurred to detectable levels: 5-10. White’s number is (or appears to be) an estimate of the number of cellular replications that occur between detected instances.
If you were to take a fresh nutrient-filled flask every morning and seed it with a dozen E coli cells from yesterday’s flask, then incubate during the day, looking out for some rare event that needs 2 serial mutations, there are numerous things to consider
– the number of times that each individual event occurs
– its retention in the population in descendants
– the number of cellular replications that occurred between detections.
– The size of the daily samples relative to the flask population.
White is talking of the third as if the others have no effect on the probability of detection. They clearly do.
Note that K76T does not have to happen first. Both mutations have to be lethal before you can reduce the probability to that of both mutations happening in the same replication round. For any particular ‘double mutation’, the probability of it occurring is conditioned by the current size of the single populations, in which the second can occur and between which recombination can occur. If K76T is truly lethal, you can discount its subpopulations from the serial scenario, and both subpopulations from the recombinant one. But you can’t discount them all.
Your own quotation. They found no reduced viability with mutations other than K76T. And Behe himself acknowledges the presence of K76T alone in populations, though he immediately tries to back-pedal its significance: “The result — widespread, chloroquine-sensitive K76T PfCRT in a population — is medically and epidemiologically interesting, but would have nothing to do with how resistance originally arose”. He misses the point that if K76T can survive alone, then it can drift in the first place, your quote notwithstanding. It may require other cryptic mutations to help support it – who knows? How can we draw lessons in general evolution from such a specific case?
I trust you aren’t proposing that even if chloroquine resistance had lethal intermediates blocking all serial paths, this applies to all double-mutant evolution in every other species? That’s pretty much what Behe is trying to do.
There must be a reason why chloroquine resistance occurs with reasonable frequency despite Pf’s low effective population size due to bottlenecking. I would suggest that the reasons are the unaccounted forces of drift and recombination.
The obvious inference is that God likes to kill children with malaria, and spares no expense to make sure malaria survives.
Aside from that, what does this have to do with the evolution of complexity?
Charlie: do you have a response to Darwin’s challenge? Can you name something that could not have evolved incrementally? Can you provide the actual history of the object or feature?
This analogy bears no comparison to the situation under discussion. Your flask is full of nutrients but you have left out the most important ingredients. Where are the components that are deadly to E coli and will ensure that the bacteria do not propagate? Your E coli are placed into a land of milk and honey where all their needs are met, on the other hand malaria parasites have to compete for nutrients in a hostile environment. White takes account of anti-malarial blood concentrations. He calculates that 100,000–300,000 parasites would encounter residual anti-malarial agents in the blood of an individual. He gives the average number of parasites in the blood of symtomatic patients as lying between 5,000 and 50,000 per microliter of blood depending on whether they occur in areas of low or high transmission intensity. He reasons that, “those patients with high parasitemias who survive their infection to transmit viable gametocytes carry a significant proportion of all the world’s ‘potentially transmissible’ malaria parasites”. He considers host immunity, inoculation rates and drug concentrations among other factors.
A more fitting analogy would be to take several million nutrient filled flasks also containing chemicals which are fatal to the E coli that do not have the mutations. That would give a better comparison to this:
Its not that K76T does not have to happen first, it is that it cannot happen first.
Please give the exact quote where Behe acknowledges the presence of K76T alone.
CharlieM,
Even if this were true, all that is required is that the other one happens first. Both mutations must happen in a single individual IF AND ONLY IF both mutations are individually lethal. Not detrimental, lethal.
But further, it is simply not true that K76T cannot survive alone.
CharlieM,
??? I gave the exact quote immediately after the terminating colon in the passage you abstracted! Here it is again: “The result — widespread, chloroquine-sensitive K76T PfCRT in a population — is medically and epidemiologically interesting, but would have nothing to do with how resistance originally arose”. Chloroquine sensitive K76T is K76T without the second mutation. Both mutations confer resistance, ie the double mutant is not chloroquine sensitive; only the single one is.
CharlieM,
Irrelevant. One does not need to simulate all the challenges faced. The simple fact is that Plasmodium populations grow from 10 or so sporozoites to 10^12, and are then sampled in another bite back to a few dozen or so. The conditions of the organisms are irrelevant if they both produce, under those conditions, equivalent numbers of cells, and the sampling is of a similar order in both cases. Plasmodium thrives in people. That’s what makes ’em ill, don’cha know.
Large numbers are only produced in successful infections, which would tend to be those without the challenge of chloroquine – a land of milk and honey. The mutational environment is not the chloroquine-dosed population, but the entire population. Chloroquine selects, but it does not mutate.
Charlie – are you seriously suggesting that, even if K76T were lethal on its own, this would be applicable to every equivalent double mutation in every species ever – one of the pair is always lethal?
Putting aside the bickering over detail, this nicely exposes the ridiculousness of Behe’s argument.
The probabilistic regime under which the CCC arises is specific to that particular change in that particular set of circumstances, within a very unusual lifecycle. You cannot simply square such a one-off figure (even if you calculated it properly) and draw widely-applicable evolutionary rules from it.
The argument is more hole than material.
As I have mentioned, Behe has had to struggle to find edgy material. In his entire career he has only found a few hard cases.
That’s just plainly false. You are in fact trying to extrapolate the case of the difficulty of evolving chloroquine resistance to the rest of the biosphere. That is your entire point with this discussion, to show that because this single thing is difficult to evolve, that is must be an even greater problem to evolve more complex things that require even more mutations, interactions and so on
No*, I also don’t have a videotape recording of the transition from fish to amphibians.
We are discussing whether the evolution of the biosphere without divine intervention is “too improbable” and whether one can draw any conclusions about what is or is not possible regarding the origin of extant biodiversity from the single case of evolving chloroquine resistance.
We are not discussing whether I or anyone else have seen them happen personally.
* Though I’m not even sure, because it seems likely to me that one can dig up multiple references from the literature where immunoglobulins have found new antigen binding sites, and if one were to try and trace out their mutational history it is quite likely some of those mutations were neutral or even slightly deleterious. Another alternative would be bacterio phage evolution, which is also incredibly fast, so that’d be the areas to look.