I think it's an excellent example of the difficulties that many creationists will face when they try to come to grips with modern evolutionary biology.
Here's what he wants to say ...
Since Professor Moran has kindly addressed the questions I have raised, I feel obliged to respond here. But I must add that my response will be restricted to one topic only, and therefore this reply should not be construed to have the same purpose as the earlier discussion.
I find only a few minor contentious issues in Professor Moran´s post, so I prefer not to bother the readers with details about them. In my opinion, it is important that we have started to look at the core issue – the average fixation of about 100 mutations per generation, or 22,000,000 in 5,000,000 years, according to the genetic drift model - in terms of close to real-life conditions. Now we see that the simple genetic drift model needs extension, to include population splitting and recombining that leads to the postponement of fixation (I thank Professor Felsenstein for improving the clarity of this point). Furthermore, we see that in expanding populations the fixation rate is lower than the average. Thus, in today´s human population, the fixation rate per generation is close to zero. In order to compensate for the lower than the average fixation rates in some generations, it is necessary to postulate higher than the average fixation rates in other generations, if one wishes to account for the 22,000,000 fixed mutations in 5,000,000 years.
Let us consider the most dramatic case, mentioned in the comments, leading to the maximal fixation rate: the shrinking of a whole population to just a single couple, 2Ne = 2. Today we know that two unrelated human individuals differ in 1 nucleotide per about 1,000 nucleotides, so that each one of us carries about 3,000,000 SNPs. This is the maximal number (actually the maximal number is smaller because a fraction of SNPs is always in the heterozygous state) that can be fixed in this dramatic case. An interesting thing comes out to light now: the average of 100 fixed mutations per generation may result from the values that span a range of over six orders of magnitude, from less than 1, to over 1,000,000.
This raises the question of the meaning of the term “genetic drift model”. Can we maintain to be talking about the genetic drift model if the essential postulate of that model – mutation rate equals fixation rate – does not hold, because while the mutation rate changes little, the fixation rate can vary over six orders of magnitude? I think not. Drastic scenarios, known as “bottlenecks”, do not belong to the genetic drift model, in my opinion.
Now the important question is this: During the 5,000,000 years, what is the number of mutations that would have been “delayed to fix” because of the expansion of the human population, and/or due to the splitting-recombining, so that we must postulate dramatic events (“bottlenecks”) to account for their fixation? In other words, how many mutations were fixed in “bottlenecks” and how many by the ordinary course of genetic drift, in percentage? I doubt anyone can provide a verifiable answer (I kindly ask Professor Felsenstein to correct me if I am wrong). If for a “bottleneck” we take 2Ne significantly larger than 2, then many more “bottlenecks” need to be postulated in order to account for the same number of “delayed to fix” mutations. Is it possible to account for all the fixed synonymous mutations found in the human and chimp genomes by invoking fewer than two “bottlenecks” with 2Ne = 2? So that only 6 Million mutations are fixed in the “bottlenecks”, while 16 Million are fixed by genetic drift? I do not know.
And here is an additional complication. According to Wikipedia (since Professor Moran has relied on that source in his post, I follow suit):
Early humans (before Homo sapiens)If correct, this information means that the time available for fixation of 22,000,000 mutations is reduced by about 2 Million years - to just about 3 Million years - because after migrating out of Africa different human sub-populations fixed different neutral mutations, due to the stochastic nature of the process. All specific human-chimp genetic differences (= all humans have them, no chimp has them, or vice versa) must have been fixed before the out of Africa migration. Is it possible to construct a genetic drift model (without “bottlenecks, with reasonable numbers) able to account for all the fixations, now within 3 Million years? I doubt it, but am willing to review a model that could dispel my doubts.
Early members of the Homo genus, i.e. Homo ergaster, Homo erectus and Homo heidelbergensis, migrated from Africa during the Early Pleistocene, possibly as a result of the operation of the Saharan pump, around 1.9 million years ago, and dispersed throughout most of the Old World, reaching as far as Southeast Asia. The date of original dispersal beyond Africa virtually coincides with the appearance of Homo ergaster in the fossil record, and the associated first emergence of full bipedalism, and about half a million years after the appearance of the Homo genus itself and the first stone tools of the Oldowan industry. Key sites for this early migration out of Africa are Riwat in Pakistan (1.9 Mya), Ubeidiya in the Levant (1.5 Mya) and Dmanisi in the Caucasus (1.7 Mya).
Let´s suppose that there are indeed 22 Million fixed synonymous mutations between the two genomes. I have no principal problem with that, or any other, experimentally established number. Whatever the exact number may turn out to be, scientists will continue looking for a model that fits the data best. In my opinion, no model should be rejected a priori. In order to contribute more constructively to this discussion, I ask: Why not test a model that uses 2Ne = 2 for the starting human population? With this model, for example, in the first generation 15 Million synonymous mutations might be fixed. Therefore, this model does not require multiple “bottlenecks” (perhaps just one) to account for a large fraction of the fixed mutations; while a smaller fraction - the 7 Million remaining mutations – could then be fixed in many subsequent generations in an expanding and splitting-recombining human population according to the population genetics theory.
One could argue that the starting Ne = 2 model is preferable in view of the principle known as Occham´s razor. But I would be the first one to disagree with such argumentation. Only in view of other experimental data found in the sequenced genomes one should decide which model is the preferred one; if the genome sequence data contradict one of any two models, the bad model should be rejected; and if the data contradict both, both models should be rejected.
I hope the above makes clear my thinking on this topic.