Disclaimer: I write this post as a layman in the area of biology. If possible, I want to make some of the debate over the alleged impossibility of an historical Adam available at the level of laymen. The scientific issues involved are, of course, highly controversial. If I make cut-and-dried errors of factual statement or terminology, I welcome correction.
I have recently found the book Science and Human Origins to be extremely helpful as a starting place for examining some of these issues. Even more interesting is the Internet debate about the contents of the book. Paul McBride, who at least was at the time an evolutionary biology PhD student in New Zealand, wrote a 2012 series of posts criticizing the book. That series appears in six parts here. (That is part 5. It includes links to parts 1-4.) The authors of the book and other ID advocates have written many responses. Casey Luskin posts a link round-up here. See also here. I will be discussing several of Ann Gauger's responses in this post in more detail. Anyone who is interested in the issue of the alleged overwhelming evidence for undesigned human evolution from ape-like ancestors has no lack of material to study; nor have the advocates of intelligent design theory made a one-sided statement of their case without responding to critics. So if you want to read up, this is the Internet age; go for it.
This post will be about Ann Gauger's work concerning population genetics. Population genetics has become a major Darwinian argument of choice, allegedly putting a nail in the coffin of the historical Adam, and it was chiefly for Gauger's essay that I originally got the book. Now, to business:
Gauger's chapter and Ayala's 1990's argument
For a foil in her book chapter, Ann Gauger chooses an argument made by a scientist named Francisco Ayala from gene called HLA-DRB1. Ayala obtained samples of chimp, human, and macaque DNA from a specific region of HLA-DRB1 and constructed conjectural phylogenetic "trees" using alleles (i.e. variants) of this gene. In essence, one takes alleles of the genes, notes similarities and differences among present-day sequences, and then one constructs a "tree" which is supposed to show how these alleles are related. By using estimated mutation rates and various algorithms of population genetics, one calculates far back into time as to how many alleles are supposed to have existed at some period of interest--say, at an estimated time when, according to evolutionary theory, humans and chimps last shared a common ancestor. This procedure is supposed to give an idea as to how big the initial population of humans must have been.
Ayala went back to a particular estimated time when humans and chips allegedly last shared a common ancestor (Ann Gauger points out that estimates on this vary a good deal, but his was that it was four to six million years ago) and stated that, by his calculations, there must have been a minimum of 4,000 individuals alive at that time in the human ancestry. He also argued that the actual population must have been much larger still.
The idea is that this estimated population size shows the impossibility of an historical Adam and Eve, because our ancestors would have had to carry more genetic diversity than would be found in the genome of merely two people.
I cannot summarize Gauger's whole article. (This post is going to be long enough as it is.) But I will summarize a few of her points which seem to me utterly devastating to Ayala's argument regarding that particular portion of the genome: First, Ayala focused on the exon 2 region of the HLA-DRB1 gene, which was a poor choice. When one uses a different region (introns 1-4) of the gene, one gets a tree that sorts out, sensibly enough, by species. Macaque alleles look like other macaque alleles, human alleles like human alleles, and chimp alleles like chimp alleles. When one uses the exon 2 region one instead gets an incompatible tree in which the alleles are all mixed up. Sometimes human alleles are more similar to macaque alleles or chimp alleles than to other human alleles, macaque alleles are sometimes more like human alleles than like other macaque alleles, and so forth. In other words, using the exon 2 region to make a phylogenetic tree creates, to use technical scientific terminology, an unholy mess.
Let me emphasize that the variants of the exon 2 region do not sort by species right now, meaning that something very weird is going on in exon 2. Presumably there is some cause of hyper-variability in that region. In fact, it appears that exon 2 has a much higher mutation rate than do other parts of the gene. A tree based on exon 2 using a mutation rate that is too low will, understandably enough, give a false picture of the number of variants present in a population millions of years ago.
Ayala did attempt to correct for the fact that natural selection will tend to inflate genetic diversity for exon 2, since people with different versions of exon 2 in their genome will tend to have more surviving offspring. Whether his fudge factor sufficiently corrected for this effect of natural selection or not, he disregarded the apparently much higher mutation rate in exon 2 leading to the failure of sorting by species.
Gauger also notes that a different late 90s population genetics argument based on neighboring DNA sequences (not exon 2) came to a very different conclusion from Ayala's, concluding seven versions of the HLA-DRB1 gene existed four to six million years ago, whereas Ayala had concluded that thirty-two existed. (This was the basis for his population estimate.) Let me be clear: I am not saying, and I'm sure Gauger is not saying, that this study was simply right while Ayala's was wrong. As I shall discuss later, Gauger is questioning the reliability of population genetics altogether for estimates about population sizes long ago. The point I would make here is simply that, when we get conflicting trees from these methods, we are justified in asking why we should accept the reliability of the underlying methodology.
Gauger also points out that Ayala's calculations included other assumptions that would have also led to an overestimate of the number of ancestors needed to carry the relevant information. The larger HLA-DRB1 region does not do a lot of recombining when the DNA is reproduced. This is kind of unusual. Different versions of sections of DNA that tend to be inherited as a unit are sometimes called haplotypes. You can think of a haplotype as a chunk of genes that gets inherited together. Now, if the inheritance of DNA in this region is "chunky" like this, that (in addition to the other considerations) throws all of Ayala's calculations into a cocked hat. Pause here to contemplate how easy it is to throw population genetics estimates into a cocked hat.
To do population genetics, one has to make double handfuls of assumptions about rate constancy, equiprobability, and biological independence in order to make the math workable. Sometimes these assumptions turn out to be wrong, and wrong in ways that cause overestimates.
Gauger points out in the chapter that there are just five basic groups of the HLA haplotype in the human population. She then points out that as many as four haplotypes could be carried even by a perfectly ordinary pair of individuals.
This point seriously calls into question the necessity for a minimal bottleneck of 4,000 to 100,000 in the human lineage.
Paul McBride's response to Gauger's chapter
Interestingly, McBride doesn't bother to defend Ayala's argument (as far as I can see). He seems pretty eager to drop Ayala's entire study down the memory hole and move on as quickly as possible, which I'm guessing means that the problems with Ayala's study are as decisive as they appear to be.
McBride does try to make much play out of the fact that, if we grant that the relevant haplotypes (even if only four) first showed up 4 to 6 million years ago, this was before the appearance of any of the genus Homo, including Homo erectus, and hence before any plausible candidate for the species of human beings. McBride thinks that this means that Gauger's point about four haplotypes and her conjecture that these may have been carried by Adam and Eve makes no sense as a defense of a possible first couple, because the first individuals carrying those lineages couldn't have been human at all. He points out that Casey Luskin's preferred candidate species (in a different chapter of Science and Human Origins) for the first true humans is Homo erectus, which are now thought to have showed up circa two million years ago, not four to six. So, says McBride, even the four haplotypes idea still "rules out" the existence of Adam and Eve as the first humans.
Gauger has clarified here that she was not saying that a first human pair was probably created four to six million years ago. An ordinary first couple could have carried as many as four versions of the gene. A fifth variant could have arisen by natural genetic change after humans came on the scene. (There is a reason why I keep saying "ordinary individuals," but you will have to read to the end to get to the explanation.) Since the discussion is over whether there is too much genetic diversity in mankind now to have arisen from a first couple long ago, this point is relevant.
Gauger also points out in this recent post that even some scientists who accept common descent from an ape-like ancestor estimate that there were four to five lineages of this gene long ago and that much of the genetic diversity in this gene, within lineages, has arisen much more recently--between 350,000 to 500,000 years ago. This is directly contrary to Ayala's conclusions.
Li and Durban's claims and Gauger's response
McBride seems eager to move on to discussing something other than Ayala. His favored population genetics study is one by scientists Li and Durban, using a different method from Ayala's. In particular, Li and Durban used much larger regions of DNA to try to rule out confounding effects specific to a narrow area (part of what tripped up Ayala's study). They concluded that a minimum bottleneck around the time of Homo erectus was about 15,000 individuals.
Gauger's initial post refers to problematic assumptions in Li and Durban's type of population genetics study, which is known as coalescent theory. McBride at first complains that she doesn't say what the allegedly problematic assumptions are; in her second post on the subject she goes into this in more detail. Coalescent theory, says Gauger, is based on the assumption that, in populations of the relevant size, "the vast majority of mutations are neutral and have no effect on an organism’s survival."
Let that sink in just for a minute. Can you say "natural selection"? Yes, that's right: Natural selection, a crucial, major driver of neo-Darwinian theory, is treated by coalescent theory in population genetics as having no significant effect upon population sizes over hundreds of thousands and even millions of years.
McBride insists that the assumptions made in population genetics tend to cause underestimates rather than overestimates concerning population size, but it should be fairly clear even to a layman that this is by no means necessarily the case when it comes to literally discounting natural selection as a force influencing population size! For example, as noted above, Ayala himself knew that natural selection would sometimes inflate genetic diversity. And Ayala's biological independence and mutation rate assumptions did not cause an underestimate of past population size.
Perversely, I'm tempted not to make too much of this matter of natural selection and then to sit back at a climactic moment and watch the sociobiologists and evolutionary philosophers squirm. Their goal is to "explain" complex sociological phenomena such as why gentlemen prefer blondes or why human beings are religious in terms of the effects of natural selection throughout human evolutionary history. It would be almost too much fun to wait for one of those folks to cite Li and Durban's study allegedly showing that Adam and Eve couldn't have existed and then to spring on him the unwelcome news that Li and Durban are treating natural selection as irrelevant to human evolution for the past three million years.
But actually, I won't. Because I have to admit that, again, speaking as a layman, it seems pretty implausible that most hominid mutations for the past three million years have been neutral in effect and that we can use that assumption to generate population census estimates regarding our long-ago ancestors. Did I say something above about double handfuls of equiprobability assumptions in population genetics? Yes, I believe I did. Gauger (who isn't a layman) apparently agrees, which is why she highlights the neutral mutation assumption of coalescent theory as problematic.
McBride sometimes gives the impression that the phylogenetic trees generated by population genetics are consistent and mutually confirmatory, at least as regards the claimed impossibility of Adam.
I shall reiterate: the possibility of a two-person, Adam and Eve bottleneck is simply not an open question in science. There cannot have been such a bottleneck.
Taken together, there is a bias in our estimates of actual past population sizes, but it is a bias towards underestimation. So, when we make long-term estimates of the human effective population size that vary by two orders of magnitude but bottom out at around 1,000, this lower estimate (and its corresponding census population size of 10,000) are almost-certain underestimates of the real picture. Even with the variation in different estimates, we are left with no room for a census population size of two--a further four orders of magnitude lower again.
Li and Durban, however, trace effective population sizes (evolutionarily relevant estimates of the adult breeding population size) over our evolutionary history as a species, using comparisons of entire genomes from across a range of ethnic groups. At no point is there anything approaching an Adam and Eve population bottleneck at any point that correlates to the genus Homo. Again, when an earlier chapter has drawn the human/non-human distinction between Homo habilis and Homo erectus (i.e. approximately 2 million years ago) we can be quite well assured that a literal Adam and Eve are unsupported by population genetics.The figure below, from Li and Durban, suggests an effective population size of about 15,000 at the origin of Homo erectus.
From all of these statements one might surmise that perhaps there is significant, detailed mutual support going on among various population genetics studies. McBride gives the impression that we can be quite certain scientifically that no Adam and Eve existed because, wherever else they might disagree, these population genetics estimates agree in not coming up with a two-person bottleneck.
There are several things wrong with such an argument. McBride doesn't seem to see that there are real reasons for questioning these estimates, all of them, altogether--in other words, for questioning whether population genetics (in its various incarnations) is a reliable way of finding out how many individual ancestors lived millions of years ago. That is definitely a major point Gauger is making. She says:
[I]n my opinion it is an open question whether present genetic diversity provides sufficient information from which to draw conclusions about ancient populations. Determining events in deep human history may be beyond the reach of population genetics methods.
From her book chapter (p. 119):
Where ancient genetic history is involved, dogmatic statements are out of place. We understand very little of our own genetic makeup--way too little to make accurate calculations about our distant genetic past.
Why should this not be correct? There is certainly no law that says that we have to have reliable methods of estimating how many of our ancestors were alive two or three million years ago! Indeed, to claim that we can make such reliable estimates would seem to require meeting rather a strong burden of proof.
That these various conjectural methods all "agree" in placing the smallest bottlenecks at a number greater than two has virtually no significance in and of itself as support for the enterprise of population genetics, because the agreement is much too broad. Consider: If you have three entrail readers who all agree that next year will be a prosperous year but all for different reasons, does this give you any good evidence for the reliability of entrail reading as a way of divining the future?
McBride does not even try to claim that Li and Durban's phylogenetic tree agrees in detail with other trees throughout millions of years of history. His entire focus is on their agreement that (thank goodness) there were never fewer than 1,000 human ancestors, and allegedly many more for an effective population size. But Gauger is right to point out that variations for the minimum within several orders of magnitude imply that very different trees that disagree with each other are being constructed.
We already saw this from Gauger's chapter: A tree constructed on the basis of one genetic segment was incompatible with a tree constructed on the basis of the rest of the same gene. They couldn't both be right.
According to Gauger, such problems are pervasive:
[W]e see evidence of tangled trees at all levels of phylogeny.
To my mind, this is the mark of a method characterized by guesswork and instability. That evaluation is also borne out by the points above about unjustified equiprobability assumptions. If we do indeed see conflicting and tangled phylogenetic trees at all levels, this raises a very real question as to why we should even care that various conjectures based on population genetics all "agree" with the broad claim that there was no time in deep history when we had only two ancestors.
Part of my point here, and I assume Gauger's as well, is that there is no known-to-be-reliable, independent way to check any of these phylogenetic trees for the distant past.
This points to the relevance of another complaint that Gauger makes:
In addition, all these calculations depend on assumptions of common descent as the only explanation for our origin.
[M]ore worrying to me are the hidden assumptions in evolutionary models. Population genetics is a theory-laden subject, based entirely on neo-Darwinian assumptions.
Population genetics as a scientific discipline makes use of mathematics and the principles of neo-Darwinism to try to understand how genetic variation spreads through populations and influences their evolution. It does so by assuming that all processes are purely natural and unguided, and that phylogenetic history is mainly the product of common descent, at least in multicellular organisms.
Gauger's point, simply put, is that all of these estimates assume that nothing was going on in the relevant millions of years but inheritance and mutation. No intelligent intervention, no creation, for example. Since these estimates are based on this assumption, and since there is no clear independent way to check that these guesses are right or even to confirm the reliability of the methods by independent means, why should we take the mere existence of these conjectures to be evidence against the intelligent design of man? In this situation, the accusation that the argument against an historical Adam from population genetics is circular has force.
Creation, genetic diversity, and ad hocness
Lurking behind the capitulation of some theistic evolutionists to the "no Adam" conclusion lies, I believe, something like the following thought process. The words are mine, but based upon conversations I'm convinced that something like this is worrying people:
You can't bring up the possibility of miraculous activity as an explanation of genetic diversity, because that would be ad hoc. After all, you can "explain" anything by saying that God set it up that way. We don't want to fall prey, just to save the appearances and keep our creation story, to the picture of a deceptive God who, for reasons unknown, made it look just like there weren't two original parents. If that's the best we can do, we should just conclude, however sadly, that the scientific evidence strongly disconfirms the existence of Adam.
I want to confront this concern and not let it be the elephant in the room.
I've recently been doing some philosophical work on the issue of ad hocness. (You can read the results here if your library has a subscription, or write to me for a scholar-to-scholar share of the e-offprint.)
My conclusion, in very brief terms, is that ad hocness doesn't arise unless the auxiliary hypothesis in question has low probability given the overarching theory in question. There are other necessary conditions as well, but that is one of them. Take a classic example of ad hocness: Junior discovers that "Santa's" handwriting on the Christmas present tags looks just like Mom's and confronts Mom about it. Mom tells Junior that Santa has deliberately made his handwriting look like hers to make things more difficult or to throw people off his (Santa's) trail. Why is this ad hoc? One obvious part of the answer is that, based on the assumption that Santa exists, one would have expected the opposite to be true. One would not have expected Santa to disguise his handwriting but rather to use his own handwriting. All the more so since children are supposed to know about Santa anyway, so evidently he's not trying to hide his existence.
An ad hoc hypothesis could come up in connection with God's activity. Suppose that I say that God created my table five minutes ago. Upon having it pointed out that everyone remembers the existence of my table for many years past, I could argue ad hoc that God created all of those memories in our minds. Again, why would we think that? It has very low probability given the initial hypothesis I was putting forward. On the contrary, we would expect God to show Himself by creating something, not to make it seem as if it existed previously for many years.
Now consider the hypothesis that Adam and Eve were specially created, perhaps even ex nihilo. And consider the auxiliary hypothesis that God gave Adam and Eve extra genetic diversity, more than is presently possessed and passed on via sexual reproduction by ordinary human beings. (This is my discussion of ordinary human beings, promised above.) Is this just something made up to try to rebut disconfirmatory studies of population genetics? Is it like the "Santa imitating Mom's handwriting" hypothesis? Not at all. For one thing, I've already argued that the population genetics studies don't give us anything that needs to be rebutted, because their arguments are poor.
But more: Consider what we already know about genetics, independently: We have reason to believe that close interbreeding causes the accumulation of harmful mutations. Therefore, if God were going to create Adam and Eve as the first couple and have all human beings descend from them (the overall hypothesis we are considering), we have independent reason to believe that God would do something to avoid the possible negative effects of in-breeding. One open possibility to fulfill that functional purpose is that Adam and Eve would be created with extra genetic diversity--that they would be different from modern humans in this respect. Their children might not even be genetic siblings. (V.J. Torley discusses this possibility here, among other places.)
If this were true, and if one analyzed Adam and Eve's genetic potential, there is a sense in which it would "look like" the genetic potential of more than two people. But this would not be God's pointlessly and deceptively making it "look like" we had more than two ancestors, because that design plan is not pointless. It has an important functional purpose for the first couple's offspring. Therefore, unlike an ad hoc hypothesis, it does not have a particularly low probability given an intelligently designed first couple.
Hence, even if we had more reliable methods than we do have for estimating distant past populations, and even if they did not happen to come up with a two-person first couple, this would not be serious evidence against the existence of an original couple.
Notice that since the issue is functional, the amount of extra genetic diversity that would be (to some degree) expected in Adam and Eve beyond what is normally found in two ordinary people will depend upon functional considerations--what range or level of genetic diversity would be valuable for a founding population? This should make it clear that the conjecture concerning extra genetic diversity is not being made to counter some study or studies that "seem to show" a high number as the lowest possible number of human ancestors. (As I have argued, the population genetics science is unconvincing.)
The epistemological point concerning function, creation, and genetic diversity should be borne in mind as a confounding factor if one is attempting to use genetic diversity estimates to argue against the existence of Adam and Eve.
What we're being told is that there is something analogous to a video of the past showing, via highly reliable methods, what appears to be an uninterrupted history of the natural evolution of mankind from non-human ancestors--a history of mankind that could not have begun with two individuals.
This is not true. Given the Scriptural evidence for the existence of an historical Adam and Eve, the first and only parents of mankind, Christians are fully justified in asking for strong counterevidence. It doesn't seem to have been forthcoming, despite all the fanfare. Even from the perspective of common sense, it should take a lot to convince us scientifically that there could not have been a first human couple as long ago as 600,000 to two million years. I have real doubts that any such negative can be proven, even within the weaker sense of "prove" that is appropriate to empirical endeavor.
One thing I am confident of: Population genetics has not given us strong evidence for that negative proposition. No, Virginia, science hasn't debunked Adam and Eve.