3. The case for ID vindicated
In chapter 5 Dawkins describes various examples of natural selection of which, as I said in the preceding post, all but one are comparable with artificial selection: all that's happening is selection from an existing gene pool; there's no production of new or altered genes. The exception is a long-term experiment using bacteria (by Lenski et al. at Michigan State University) specifically designed to investigate the evolution of new characteristics.
The long-term evolution experiment (LTEE)
For more information about the overall long-term evolution experiment see myxo.css.msu.edu/ecoli/, and for the emergence of the ability to utilise citrate it’s useful to read their paper which is available from www.pnas.org/content/105/23/7899.abstract
The LTEE involves propagating 12 lines of E. coli bacteria, all initially identical (except that 6 had a genetic marker), but then each strain allowed to develop independently. They were grown in a glucose-limiting medium, and initially the experiment was designed to see how their growth adapted to this. The experiment has run since 1988, now exceeding 50,000 bacterial generations.
All strains improved their ability to utilise glucose, evidenced by increased initial rate of consumption, this effect reaching a plateau after about 20,000 generations (typical of selection from a gene pool). On the other hand, all showed reduced ability to utilise other sources of carbon such as maltose or lactose. This is similar to most instances of the acquisition of resistance to antibiotics, discussed in Evolution under the microscope pp235-244, where such resistance is generally at the expense of reduced overall fitness.
Utilisation of citrate
E. coli cannot normally utilise citrate under aerobic conditions, and a surprising discovery was that one strain evolved the ability to do this (citrate was present in the growth medium, and the ability to utilise it resulted in a marked increase in the bacteria’s growth), emerging after about 30,000 generations. Further investigation showed that this development was contingent on an earlier mutation, arising in this strain after about 20,000 generations.
The mutations have yet to be characterised. The authors suggest several possibilities, including enabling the expression of a carrier protein to enable citrate to pass through the cell membrane; as it is transport into the cell that normally limits its use - once inside the cell it is readily metabolised.
Because this evolution required at least two distinct mutations, Dawkins vaunts it as disproving the intelligent design concept of irreducible complexity. But that is merely his spin on the facts; a closer look shows that the results from the experiment actually support the ID case.
A typical mutation rate is approx 10-9 (1 in 109 per cell per generation) for any particular base in DNA (the authors cite the slightly lower figure of approx 5 x 10-10 for E. coli). As they say, despite this low rate, given the high numbers of bacteria and generations it can confidently be assumed that during the course of the experiment all possible single point mutations in the bacterial genome will have occurred, probably many times. Of course, as they comment, only a small number of these will become ‘fixed’ in the genome - even advantageous ones will not be fixed automatically but have only a small chance of spreading throughout the population (contrary to many of Dawkins’ comments, for an explanation see a text-book on population genetics).
It is therefore not surprising that not only have the same genes been affected in several of the strains, but that in some cases the same point mutations have been fixed, especially those that improved the ability to metabolise glucose. This too is analogous to the emergence of some antibiotic (and insecticide) resistance where the same point mutation has arisen independently.
Even where two mutations are required (i.e. neither alone confers resistance, so both must arise together, with a probability of doing so of just 1 in 1018), e.g. some resistance to penicillin, bacterial populations and their reproduction rate are so high that these can occur, albeit at a low frequency.
In a similar way, even after the potentiating mutation, the probability of the citrate-enabling mutation was estimated by the authors at about only 10-13, which they say indicates the change involves multiple point mutations or a rarer type of mutation.
And the preceding potentiating mutation also had a very low incidence - occurring in only one strain, even after efforts by the researchers to reproduce it. They suggest it may be a neutral mutation, which would have been be fixed only by genetic drift.
Geological time isn't enough
In Evolution under the microscope - based on the rate of occurrence of point mutations, the size of bacterial populations and their rate of reproduction, and supported by the observed instances of resistance - I suggest that the upper limit for multiple dependent mutations that could arise in the course of a year is likely to be 3 or maybe 4. And I went on to point out that, given the mutation rate is about 1 in a billion, this implies that the upper limit for mutually dependent mutations arising throughout the whole of geological time (e.g. a billion years) is only 4 or 5.
So, although it may be possible to switch a gene on or off with just a few point mutations, or modify its performance, this cannot be extrapolated to producing genes in the first place, as typically they require hundreds of specific base pairs e.g. to code for the many essential amino acids in the protein product.
In discussing the possible nature of the citrate-enabling mutation(s) the authors consider reactivation of a cryptic transporter, but think this unlikely because they would expect such a cryptic gene to have been degraded beyond recovery after millions of years of disuse. It begs the question - if they consider it so likely that the effect of random mutation is to degrade genes, how do they think useful genes arose in the first place?
Dawkins frequently emphasises the immense length of geological time, suggesting that this is more than enough to overcome the improbability of advantageous mutations. It is time he did a few simple calculations and started to look at geological time objectively. If he did so, he would realise that it is not the answer to evolution's problems that he makes it out to be.
And this is why, as I indicated above, far from defeating the case for intelligent design based on irreducible complexity, the LTEE results actually support it - because they demonstrate how limited is the ability of random mutations to generate useful sequences. And - unfortunately it needs to be repeated often - natural selection is dependent on being fed the right raw material on which it can work.
Further, it should be noted that the above-mentioned rate of finding useful mutation combinations applies to such as bacteria which have very large population sizes (at least billions) and high rates of reproduction (more than one generation per day). In organisms with smaller populations and slower reproduction rates, what can be achieved will be so much less.
Which is why, yet again, Dawkins misleads his readers by saying:
So whatever evolutionary change Lenski may have clocked up in the equivalent of a million years of bacterial generations, think how much more evolution might happen in say 100 million years of mammalian evolution (p119)
because the size of the human population will be so much smaller than the bacterial populations of the LTEE.
The ID argument is not that an advantageous mutation cannot occur, or even that a few mutually dependent ones cannot - but that the complexity and specificity of molecular biology would require so many mutually dependent ones that it is not credible they could have occurred.