Reviewing “Replacing Darwin” – Part 6: Jeanson’s Fulcrum Fails

Reviewing “Replacing Darwin” – Part 6: Jeanson’s Fulcrum Fails

To those who have been waiting 12 months for this, I can only apologise that it’s being published so much later than I planned. I was swamped by my real-life work, and the sheer amount of material there is to cover in this part was a daunting task. It took me a while to get started on writing this, never mind to get through it all. As you might guess from the length of this article, I wanted to really do the whole chapter justice, as this is where most of the real meat of Jeanson’s arguments are found, and it forms the foundation of his arguments in the rest of the book.

In the last chapter Jeanson introduced the idea that the timescale in which life has diversified might not actually be as long as “evolutionists” claim, and in the 3rd and final Part of the book, entitled “Dawn of a New Era”, he attempts to substantiate this claim using cutting-edge “creation science” that he’s been working on for the best part of the last decade.

I’ve done my best to write this coherently and make each important point clear in the main body of the text, but I will also include a summary of the key points at the end of this post.

Chapter 7 – Turning the (time)Tables

This chapter will centre around mitochondrial DNA (mtDNA) and mutation rates, so Jeanson begins by explaining the basic principle of mutations that occur over generations. These can be studied by sequencing the DNA of parents and offspring to see what differences have appeared in the offspring that weren’t present in the parents. He points out that this methodology obviously doesn’t just document the fact that mutations happen, it is also capable of documenting the rate at which mutations occur (per generation), which has implications for the timescale that DNA differences arise in.

Simply put, there are 3 main relevant pieces of data: the mutation rate, how long ago a single species diverged into 2 new species, and the number of DNA differences between those 2 species. If we know any 2 of these numbers, we can make a simple calculation to estimate the third. For example, if a species diverged 1000 years ago and the mutation rate is 1 mutation per year (or 5 per generation with a generation time of 5 years), we should expect to find 2000 DNA differences between the species. It’s 2000 rather than 1000 because mutations accumulate in both descendent lineages simultaneously. Unfortunately, it’s not quite as simple as this in the real world, but we’ll get to that.

This will be the theme of the rest of the book though: as we have a lot of data on mutation rates and DNA differences, inferences about timescales can be made. Jeanson says:

“Perhaps not surprisingly, the results of these tests turned several fields of science upside-down.”

This is quite a bold claim, and one that won’t be substantiated.

The nested hierarchy of mtDNA differences

Before getting into the subject of timescales, Jeanson broaches the topic of phylogeny for the second time, noting again that as evolution would predict, the pattern of sequence differences in mtDNA forms a nested hierarchy:

“As we observed in chapter 5, evolutionists see this pattern as evidence of common ancestry. Darwin expected life to form a branching, tree-like structure. This groups-within-groups (nested hierarchical) mtDNA pattern fits this expectation.
Furthermore, in light of the discovery of the fact of DNA mutation, it would seem that evolution had a ready-made mechanism by which to explain this pattern – the accumulation of DNA differences with time. As species diverged from common ancestors, they would continue to mutate their DNA. The longer two species had been separated, the more DNA differences would accumulate between them. The mtDNA patterns that we just observed fit the evolutionary model of the mammal ancestor evolving first, then the ancestors to Artiodactyl and Perissodactyl species diverging from one another, and then the various ancestors to modern Perissodactyl families diverging from one another, with the tapir-rhino split occurring first, followed by the rhino-equid split. Lastly, the species within each of these families would have diverged from one another.
If evolution were the only explanation for nested hierarchies, then we might conclude that all mammals from echidnas to equids (Figure 7.2) and every mammal in between have a common ancestor.”

Of course, there’s a “but” coming:

“However, at least one other competing explanation exists. Again, as in chapter 5, the mtDNA hierarchy shows strong parallels with the hierarchy present within the Linnaean classification system.19 Since this system is based on biological function, the parallel between the mtDNA hierarchy and the Linnaean categories suggests that the mtDNA hierarchy has something to do with function. In other words, the creation/design model predicts the fact of mtDNA nested hierarchies as much as evolution does.”

As I described in my review of chapter 5, a design model doesn’t really fit the Linnaean hierarchy. A design model is based on function while the Linnaean hierarchy is based purely on morphology. There’s a subtle but crucial difference. Many morphological differences between species are neutral, not functional. A design model also wouldn’t prohibit extensive “re-use” of modular parts by the designer in distant branches (e.g. why not give feathered bird-like wings to bats?).

“More specifically, I have taken these design expectations one step further and derived a very detailed, testable model on mtDNA genome function.20 For mtDNA differences among species within families, my model treats these as functionally neutral changes — the result of mutation over time. However, with respect to the differences between families — those mtDNA positions that are identical among species within a family but different to species outside of the family — my model views these as having been created. Thus, my model predicts that these mtDNA differences play a functional role specific to each family.”

So, the argument is that mtDNA sequences fit into a hierarchy which matches the hierarchy based on morphology. There’s no reason to believe that there are any functional reasons this would be the case. Why should mitochondrial function match up with morphology? Why should we expect a cow’s mtDNA to be more similar to that of a whale than to that of a horse, or the mtDNA of shrews to be more similar to bats than to rats, under a design model based on function? Regardless, Jeanson is making a genuine prediction: if animals were created as separate kinds, then most mtDNA differences within each kind should be functionally neutral, while most mtDNA differences between each kind should result in functional differences. I know he says “family” in the text but of course he actually means “kind”. Humans are in the same family as chimps and gorillas but AIG wouldn’t classify humans as being in the same “kind” as those apes, that would mean they were related! I will give him some credit for making a semi-quantitative prediction here. Jeanson says that evolution would make the following prediction:

“In contrast, since the evolutionary model treats all mtDNA differences as the result of mutation, it generally predicts that most mtDNA differences are functionally neutral.”

This is tricky. Mitochondria aren’t entirely self-sufficient, they’re not completely cut off from their intracellular environments – they coevolve with the nuclear genome since they interact with each other (Hill, 2016). A mouse mitochondrial genome might function in mouse cells in the same way as a human mitochondrial genome functions in human cells, but because of mutations in both the mtDNA and nuclear genome in the respective species, they would only work in their respective cells. The mutations in the mtDNA (and nuclear DNA) might have been initially neutral, but they were later cemented into what would appear to be more functional roles as a result of coevolution between the mtDNA and the nuclear genome.

Because of this interplay with the rest of the cellular and nuclear DNA context, it’s very difficult to describe “functional differences” in mtDNA in the way that Jeanson wants. For example, in a simple hypothetical world where we discover that a single nuclear gene co-evolves with the mitochondrial genome, such that the mouse version of this nuclear gene matches the mtDNA of mice in such a way to be optimally functional, the human version of this nuclear gene matches the mtDNA of humans, etc, it may well turn out that most sequence differences between the mtDNA of families tend to result in functional differences in experiments in which the mtDNA sequence of mice is altered. It wouldn’t necessarily follow from this observation that the families were individually created though, as it might be the case that if you “transplanted” the mtDNA of any species along with the nuclear gene of that species into any other species, it would work in exactly the same way as it did in the first species. For example, you could examine the mtDNA of 2 species from different families, and identify a particular difference – in species 1 there is an adenine (A) at a particular position, while in species 2 there is a guanine (G) in the corresponding position. If you wanted to test whether or not this difference was functionally relevant, you could edit the mtDNA sequence of species 2 to change that G into an A. If you observe a decrease in some metric of fitness in these edited organisms, you might conclude “aha, this G is functionally important to this organism”, and Jeanson would say “OK, this was probably a difference in the initial creation of these families.” However, if you change a particular base in the nuclear genome at the same time as you change the G into an A in the mtDNA, there might be no fitness difference – in this case the G wasn’t really a “functional difference” in the way that Jeanson would want to use the term, as it wouldn’t make sense for a creator to create species 2 (or the progenitor of the family it’s in) with those 2 base pairs (1 in the nuclear genome, 1 in the mtDNA) being different, as there’s no functional reason to do so. In other words, some nucleotide differences in this dynamic can be neutral and yet result in fitness differences when changed in isolation.

All this is in addition to the local (intra-mtDNA) epistatic effects that would have to be taken into account. To make a long story short, this is a much more complicated subject than Jeanson suggests, so his prediction really only serves the purpose of throwing out the observation of a nested hierarchy of mtDNA patterns among species as evidence against creationism or for evolution. He suggests a very obscure way in which this pattern could maybe fit with young-earth creationism (YECism) rather than evolution, then dismisses the pattern as evidence either way until the data exists to truly test his prediction:

“Thus, the nested hierarchy of mtDNA differences among mammals doesn’t reveal anything new about species’ ancestry. Rather, it represents an experiment waiting to be performed.”

This is actually quite an interesting technique for taking certain evidence off the table. Jeanson isn’t going to be the one to make the colossal effort to properly test his prediction, so he gets to throw out a random alternative interpretation and suddenly the evidence is voided until his high standard of evidence is met. He’s not even willing to say “yeah, that’s quite good evidence of evolution, but it’s not absolute because there’s a chance it fits the YEC model”, he jumps straight to “an alternative ad hoc interpretation might be possible if certain future data comes in, therefore this current data is completely worthless and shouldn’t be considered evidence for either model”. I don’t think I need to explain in detail why this sort of reasoning is flawed. The fact that evolution predicted the nested hierarchy of mtDNA differences before they were found, and creationism (via Jeanson) is only just now coming up with a potential model to explain the hierarchy already indicates that the hierarchy itself lends support to evolution rather than being neutral.

A much simpler test of whether or not DNA sequences forming a nested hierarchy supports evolution and universal common ancestry or creationism and separate origins has already been done, and I’ve blogged about it before. Some of the DNA sequences analysed were mtDNA, and the researchers found that the reconstructed sequences that represented the sequence present in the common ancestor of clades (above the family level of classification) like Neoaves and Galloanserae were significantly more similar to each other than the consensus of the extant sequences of each clade were, indicating that these large clades shared a common ancestor as well. I explain it in more detail in the previous post, so I highly recommend reading it if you want to understand what I’m getting at here. The take-home point is that this is another successful prediction by evolution about patterns of differences in DNA sequences, and another dismal failure of creationism.

Establishing a human mtDNA mutation rate

Now Jeanson gets to the main subject of this chapter: mtDNA diversity and how it supposedly demonstrates YECism when combined with mutation rates. First, the rates:

“Furthermore, pedigree-based measurements of the rate at which human mtDNA mutates have been the subject of over 15 studies — which span two decades.24 Though the results of these studies have been hotly contested,25 part of the reason for this controversy is the fact that mutations are statistically rare events. Consequently, mutations require a certain minimal sample size to detect. Not surprisingly, the 15+ studies differed dramatically in their statistical resolving power. Nevertheless, once each of these results were weighed appropriate to their statistical resolving power, a clear result emerged: One mtDNA base pair mutates every 5 to 8 generations.26 Could this rate inform the question of human ancestry?”

This mutation rate is introduced so suddenly, and we readers are expected to just accept it. It reminds me of one of the important tricks of the trade used by magicians: they might start by asking you to pick a card, look at it and place it back in the deck, then spend 5 minutes doing all kinds of shuffling before the big reveal where they identify your card in the deck. Everyone is impressed by the 5 minutes of shuffling, thinking how skilled the magician must have been to have found your card after all that, but in reality that innocuous moment at the start where they asked you to pick a card was when the trick really happened – a card was forced onto you without your knowledge, and the rest of the routine was just hand-waving and showmanship.

Far from being an accepted mutation rate in the field of human genetics, Jeanson actually obtained this “clear result” himself in 2 “papers” published in Answers Research Journal (AIG’s “scientific journal”). In the first, he surveys the literature to come up with an average mutation rate reported for the so-called D-loop, a short sequence representing approximately 7% of the entire mitochondrial genome. This region is known to be more variable than the rest of the mtDNA as a whole. Nevertheless, in the second “paper”, he generalises this D-loop mutation rate across the entire mitochondrial genome on the basis of a single paper (Ding et al., 2015). Ding et al. sequenced entire mitochondrial genomes from the lymphocytes (a type of white blood cell) of a series of Sardinian mother-offspring pairs, but this research had nothing to do with trying to determine the mtDNA mutation rate. Jeanson noticed that the data in the paper included several mutations that were found in the offspring that weren’t present in the mother, and used these to calculate a whole-mitochondrial genome mutation rate. This approximately matched the D-loop average rate he obtained in his first paper, so he concluded that this mutation rate actually represents the mutation rate of the whole mitochondrial genome, and that’s where the figure of 1 mutation every 5-8 generations (~0.16 mutations per mitochondrial genome per generation) comes from.

One glaring problem with this is that he’s basing his per-generation mutation rate on the results of sequencing somatic cells in mother-offspring pairs, rather than trios. If you sequence the mtDNA of some, say, skin cells of a mother and her child, you have no way of knowing whether the differences that you found are the result of mutations that took place in the germline, or mutations took place in somatic cells over the course of the many cell divisions that occurred since the mother or child was an embryo. The latter mutations won’t be passed on to future offspring, so they aren’t relevant to a per-generation mutation rate at all. Figure 1 describes this concept. Using trios (grandmother, mother, offspring) can overcome this problem, because mutations that are shared between the mother and offspring but not by the grandmother very likely occurred in the germline of the grandmother, being passed onto the mother and by extension the offspring, but wouldn’t be present in the somatic cells of the grandmother. This is why using trios in mutation-accumulation pedigrees is the gold standard.

For a more in-depth description of the problem with Jeanson’s counting here, I recommend reading this article. It outlines the concepts of heteroplasmy and homoplasmy, which I will discuss shortly, so it’s really required reading to understand what comes next.

Figure 1 | Gametic versus somatic mutations. Somatic mutations occur in somatic tissues and will only be present in a subset of all cells in an individual (mosaicism), while gametic mutations occur in the germline (sperm or egg) and will be present in all cells as it is inherited by all the cells of the body as they “descended” from the fertilised egg. Image from

Jeanson is aware of this problem, and it earns a whole 2 sentences of mention in his second “paper”:

“The only remaining caveat to the present results is whether the mutation rate reported in Ding et al. (2015) represents a germline rate rather than a somatic mutation rate. To confirm germline transmission in the future, the DNA sequences from at least three successive generations must be sequenced to demonstrate that variants were not artifacts [sic] of mutation accumulation in non-gonadal cells.”

Interestingly, this awareness doesn’t stop him from presenting his mutation rate in a book aimed at the general public as though it were a brute fact. I called him out on this in the Q&A portion of a debate he had with Dr. Herman Mays of Marshall University, and Jeanson said he felt justified in concluding that his rate was really the correct germline mutation rate for 4 reasons:

  1. It agrees with other estimates of human mtDNA mutation rate from other pedigree studies.
  2. A 2016 paper from Mark Stoneking’s lab looked at the human mitochondrial germline transmission bottleneck, which can be used as an indirect measure of mutation rates, and found a similar mutation rate.
  3. It makes predictions about demographic events like the transatlantic slave trade, which Jeanson has verified as accurate.
  4. It produces a timescale in agreement with mtDNA mutation rates from other species.

First, I can’t evaluate point number 3 since Jeanson hasn’t released his “paper” yet – he says it will be out (in Answers Research Journal, no doubt) in the next few months. We’ll get onto point number 4 later in this review when I look at his data for the other species he lists.

So, in reference to his first claim that it matches other studies: his whole mitochondria mutation rate approximately matches mutation rates other studies have obtained for the D-loop/control region. Let me just briefly recap what Jeanson did: he surveyed pedigree studies looking at mutations in the D-loop, weighted them by statistical power, and got a mutation rate of 1.08×10^-5 mutations per base pair per generation for the D-loop. Then he looked at whole-mitochondria pedigree studies and found a rate of 9.55×10^-6 (entirely driven by his interpretation of Ding et al. 2015). It’s true that these approximately match, but given the methodological issues with Jeanson’s treatment of Ding et al., 2015, we really can’t conclude much from it at all. What we have here is a single whole-mitochondria study with a flawed analysis (by Jeanson) – that’s not good enough evidence for extrapolating the D-Loop mutation rate over the entire mitochondrial genome. Whether Jeanson likes it or not, there is abundant evidence that the D-loop region mutates about an order of magnitude faster than the rest of the mitochondria. This evidence comes both from studies “assuming” an ancient human-chimp divergence (e.g. Soares et al., 2009) and from pedigree studies (e.g. Howell et al., 2003Santos et al., 2008). I know Jeanson is aware of these papers, he’s cited them before. He’s just not willing to accept their results.

Regarding point number 2: I read the paper from Mark Stoneking’s lab: Li et al. (2016), published in Genome Research. Here’s a quote from it that Jeanson seems to be referring to:

“The heteroplasmy incidence is significantly correlated with the estimated mutation rate for each site (Pearson’s rho = 0.421, P = 7 × 10−14), as observed previously (Li et al. 2010), which is in accordance with expectations under neutrality.”

Jeanson says that the heteroplasmy incidence (related to the transmission bottleneck) is an indirect measurement of the mutation rate – if we know one, we can know the other. However, it’s not clear to me how he obtained a mutation rate from the data in the paper. This quote basically says (by referring to Li et al., 2010) that mtDNA sites with higher mutation rates are more likely to be found to be heteroplasmic. Li et al. (2010) established this fact by obtaining data about the prevalence of heteroplasmy and then correlating that dataset with a dataset of relative mutation rates of each site in the mtDNA. So where did that particular mutation rate dataset come from? Soares et al. (2009), the very paper that Jeanson has dismissed (in the comments section of this article, among other places) as “assuming an ancient timescale and then fitting DNA differences to this timescale”!

In other words, Jeanson’s claim that Stoneking’s 2016 article supports his mutation rate is directly based on the data from Soares et al. (2009), which is mutation rate data that explicitly disagrees with Jeanson’s mutation rate. Soares et al. estimated the mtDNA mutation rate to be about 35-70x slower than the figure Jeanson is using!

By applying the D-loop mutation rate across the entire mitochondrial genome on the basis of faulty evidence, Jeanson immediately inflates the overall mutation rate by an order of magnitude (about 10x). Even without this inflation, it’s important to note that there is a well-known difference of about 10x between pedigree mutation rates and mutation rates inferred based on phylogeny (i.e. based on calibration using a human-chimp divergence ~7 million years ago). In other words, Jeanson increased an accepted difference of 10x all the way up to a difference on the order of 100x by extrapolating from the D-loop mutation rate. Why did he feel the need to do this? Because while the known ~10x difference in rates would be enough for him to argue that that the evolutionary model is in trouble, it wouldn’t be sufficient for him to argue that YECism fares any better. Later in this post, I will explain the reason for the genuine 10x difference between pedigree- and phylogenetic-based mutation rates. Spoiler alert: it’s not actually a problem for evolution. For now, the important point to bear in mind moving forwards is that Jeanson’s human pedigree-based mutation rate is about 10x too high.

Mutation rates must be constant… Right?

With his highly suspect mtDNA mutation rate in hand, Jeanson continues:

“Could the mtDNA clock elucidate the timescale over which our own species appeared?

To answer this question, we have to answer a more specific, technical question: Has the clock always ticked at the same rate?”

One of the assumptions of the crude molecular clock calculations Jeanson will present is that the mutations rates measured today are constant, being the same throughout history. The problem is that creationists are usually loath to assume constant rates for things like the speed of light and radioactive decay, or even fairly constant rates for things like the movements of tectonic plates.

“This assumption is critical for a number of reasons. First, without it, the geologic and astronomical arguments for millions and billions of years collapse. Second, as already mentioned, these discussions largely predate the discovery of DNA clocks.

Third, creation scientists have long questioned this assumption. Specifically, young-earth creationist (YEC) geologists* propound the idea that the universe and earth are just 6,000 to 10,000 years old, and that species have formed within this timeframe. They also hold to a world-wide flood (the Flood of Noah) about 4,500 years ago. In theory, this world-wide flood would dramatically alter global rates of geologic change. At a minimum, these creationists would argue that the rates at the Flood were very different from rates we can measure today.

To be consistent with evolutionary practice, we should assume constant rates of change in genetics. In other words, when using the mtDNA clock to trace a species’ history, we should assume that the clock has ticked at a largely constant rate.”

So Jeanson argues that things like the speed of light aren’t constant, referring to the work of his fellow creationists that supposedly demonstrates this, but then turns around and uses a constant rate for mutations “to be consistent with evolutionary practice”. This implies that “evolutionary practice” is to just blindly assume that everything always occurs at the same rate, which simply isn’t true. To basically say “evolutionists assume the speed of light is constant so they should also assume that mutation rates are constant” is just ridiculous. There is a huge difference between assuming that the speed of light hasn’t changed throughout history and that mutation rates haven’t changed throughout history. The former is tied to fundamental physical constants, and we have no evidence to suggest that the speed of light was, or even could have been, different in the past. The assumption of a constant speed of light is perfectly justified. On the other hand, mutations rates are not so constrained. We know mutation rates can change because we can observe such changes in experimental evolution and mutation accumulation experiments (Barrick and Lenski, 2013). Even without these kinds of experimental results, it’s long been known that mutation rates differ between different taxa (Britten, 1986), so it should be immediately obvious that by “evolutionists’ logic”, mutation rates must have changed since all those species shared a common ancestor with a single mutation rate. There is a rich literature out there concerning the evolution of mutation rates (e.g. Lynch, 2010 and Thomas and Hahn, 2014). Converting “mutations per generation” rates into “mutations per year” rates also relies on knowing the generation time (the average age of parents when their offspring is born), which again is something that can obviously change during the course of evolution. Jeanson seems to accept this fact and uses a wide range of possible generation times in his calculations.

It has to be emphasised that when scientists “assume” constant or near-constant rates of processes in the past, these aren’t baseless, despite what Jeanson might have you believe. These assumptions are usually rigorously tested and cross-confirmed by other independent measurements. A perfect example is how radiometric dates and the GPS-measured motion of tectonic plates came together to cross-confirm each other, described in an excellent blog post by Joel Duff.

It’s true that constant molecular clocks have been used in many published molecular analyses, mostly older ones because this is just the simplest thing to do. Now, however, relaxed molecular clocks are much more popular, which often use Bayesian calculations to evaluate which evolutionary rate (for our purposes, mutation rate) is most probable for each branch on the phylogeny. As soon as we obtained the statistical techniques and computational power to go beyond simple models, mainstream evolutionary biologists did so. Meanwhile, Jeanson is living in the same year as the rest of us but insists on using the most primitive models. Perhaps this isn’t surprising, given that Jeanson is probably mostly self-taught when it comes to molecular evolution, phylogenetics, and evolutionary biology in general. His PhD in Cell and Developmental biology doesn’t give him any expertise here.

Jeanson’s test of evolutionary timescales with human mtDNA

So now, finally, we can start to get into the results of Jeanson’s analyses. He’s got a (inflated) mutation rate in hand (about an average of 1×10^-5 mutations per bp per generation, or 0.16 mutations per generation, with 95% confidence values of 0.119 and 0.197), and decided that he’s justified in assuming that this mutation rate must have been constant throughout all of evolutionary history (at least as far back as our divergence with chimpanzees).

“Evolutionists” have proposed that humans and chimps diverged somewhere between 4.5 million years ago and 17 million years ago. This is the range, but most estimates tend to cluster around 7 million years ago. Jeanson also considers a range of possible generation times, between 15 and 50 years per new generation. Using all of these different values, he calculates how many mtDNA differences there should be between humans and chimps, given his mutation rate. He calculates an expected range between 21,480 and 447,368 differences. For example, 1 generation per 50 years * 4.5 million years * 0.16 mutations * 2 lineages = 28,800 differences. In reality, however, there are only 1483 mtDNA differences between humans and chimps. Jeanson shows these numbers in a graph (Figure 2).

Reproduction of Figure 7.3.png
Figure 2 | Jeanson claims evolution predicts too many Human-Chimp mtDNA differences. A reproduction of Figure 7.3 from Jeanson’s book.

Remember, Jeanson’s human germline mutation rate is on the order of 10x faster than the one accepted by scientists in the field, so whenever you see one of these graphs, you have to mentally reduce the magnitude of the “prediction” bar by about 10x. In this case, even if we were to push it and reduce the “prediction” bar by a factor of 35x, only the very tip of the lower 95% confidence limit bar would capture the actual number of differences. This would be based on a relatively recent divergence time and unrealistically short generation time, so I think it’s fair to say that there really is a disagreement between Jeanson’s “evolutionary prediction” and the actual number of mtDNA differences between humans and chimps.

Some of you might be wondering how we could ever find tens or hundreds of thousands of differences between the human and chimp mitochondrial genomes, given that the mtDNA is only about 16,500bp in total. Jeanson says:

“In practical terms, the 447,000 result is the number of predicted mutations. Since the total mtDNA genome size is far less than 447,000 base pairs, each mtDNA position would have been mutated multiple times over. In other words, the mtDNA genome would have been mutationally saturated. Today, a comparison of human and chimpanzee mtDNA reveals two genomes that are far from mutational saturation – the 1,483 differences represent just 9% of the total human mtDNA genome length.”

I think it’s very odd that Jeanson doesn’t take a couple of sentences here to really explain to his readers the practical implications of this mutational saturation effect. If 447,000 mutations occurred in a 16,000-bp sequence of DNA, when you did a sequence comparison to the original sequence using Jeanson’s criteria, it would look about 25% identical just due to chance (assuming, as Jeanson does, that the mutations are spread evenly over the mtDNA). In other words, it would appear as though on the order of 12,000 mutations had occurred – there would be no way to distinguish between 12,000 mutations occurring and 447,000 mutations occurring, because of saturation. After 12,000 differences accumulate, any additional ones would fail to be detected.

Imagine a new mutation occurring in a mtDNA sequence that has already accumulated 12,000 differences. There’s a 25% chance (4,000/16,000) that this new mutation will occur in a base that hasn’t already been changed. There’s a 75% chance that it will occur in a position that has already been mutated. If the original base was an A, and it’s now a T, when it gets mutated again there are 3 options: it can mutate into a C, a G, or back into an A. In other words, there’s a 25% chance (75/3) that the new mutation will be a back-mutation, restoring some of the sequence’s similarity. The remaining 50% of mutations won’t change the similarity of the sequences – if the original base was an A, and then it mutated from a T to a C, it’s still different to the original. If 25% of mutations make the sequence less similar, 25% make it more similar, and 50% don’t make a difference, there is an equilibrium, so the sequence won’t become more or less similar than 25% over extended periods of time. Of course, this assumes completely unbiased mutation frequencies, which isn’t realistic, but so does Jeanson’s calculations, so it’s a point of internal critique.

12,000 differences are still far too many compared to the observed number, I just think it’s interesting that Jeanson uses these silly visuals of hundreds of thousands of mutations in big bar charts when it’s pointless to show the numbers in this way. It’s just another excuse to make the “evolutionary prediction” look much bigger than the observed value. If he wanted to present the “prediction” in a form that would actually be meaningful to compare to the observed data, in the sense that it’s a prediction of “what we should actually expect to see” rather than “how many mutations might have happened”, his bar chart would look more like one I mocked-up in Figure 3. This is a trick he repeats later in the chapter with the yeast, water flea, roundworm and fruit fly datasets, but to an even more egregious degree.

7.3 realistic.png
Figure 3 | A more meaningful presentation of Jeanson’s results. It doesn’t have quite the same impact as his actual presentation in figure 7.3 of his book (Figure 2).

Next up is a human-Neanderthal comparison. Performing the same calculations as last time, only with a divergence time of 400,000-700,000 years ago provided by “evolutionists”, Jeanson calculates that we should expect to see somewhere between 955 and 9,211 mtDNA differences. The actual number? 213. To Jeanson, this indicates there’s a problem with the evolutionary timescale, but if we reduce his “evolutionary predictions” by a factor of 10 for reasons we saw earlier, the range of expected differences (95 – 921) does in fact overlap with the actual number. That being said, it’s still on the lower end, so I’d say that this “evolutionary prediction” is just about met, but only just. By Jeanson’s standards at least, it would be a successful prediction.

Finally, Jeanson looks at just modern humans. Same story as before, except this time the “evolutionary” date is 200,000 years ago – the approximate origin of modern humans in Africa. In other words, this is based on “mitochondrial Eve” living 200,000 years ago. He estimates that between 477 and 2,632 mtDNA differences should arise in this time. In the book, Jeanson presents the “actual” number as 2 separate values: the actual average in African populations, and the actual average in non-African populations, being 79 and 39, respectively. The maximum value seems to be about 123 differences between the 2 most divergent human mitochondrial genomes (Kim and Schuster, 2013). Using a 10x lower mutation rate to decrease Jeanson’s predictions by 10x, we find that the actual number of mutations fits in the range (47-263) quite nicely.

Jeanson comes away from this section with what appears to his readers like a great case against evolution: far too few mtDNA mutations are found between species and human populations that “evolutionists” claim have been accruing mutations for hundreds of thousands or millions of years. However, just by correcting his mutation rate, we actually find that human populations and Neanderthals actually can fit inside the reasonable range of predicted values. Humans and chimps, on the other hand, don’t seem to fit. Why not?

What’s wrong with Jeanson’s “evolutionary” predictions?

Jeanson’s point is clear: something must be wrong with the “evolutionary model”, based on his predictions with his inflated mutation rate. Jeanson runs through a few options to see if there are any obvious solutions (he doesn’t consider that his mutation rate might be wrong). First, it could be that the divergence times between humans and chimps, humans and Neanderthals, etc, really are much more recent than currently thought. However, Jeanson correctly points out that this wouldn’t fit the geological and palaeontological data (as interpreted by “evolutionists”). Changing the timescale to have humans and chimps diverging only a couple of hundred thousand years ago just isn’t going to fly. Next, it could be that the mutation rate actually isn’t constant, as mentioned earlier. If mutation rates were slower in the past, then Jeanson’s analysis would be overestimating how many mutations should have accumulated over time. He says (my emphasis):

“In a similar vein, perhaps the assumption of constant rates of change could be altered. However, as we observed above, evolutionists have insisted for years that changing rates must not be invoked to explain the majority of phenomena observed in geology and astronomy. Instead, they have claimed that present rates are the key to the past, and that the world we see today has arisen primarily by slow, constant rates over time. Invoking changing rates in genetics would be logically inconsistent with the practice of evolutionary geology and astronomy.

I really can’t overstate how ridiculous this is. He’s repeating his earlier assertion that it’s logically inconsistent to suggest that something like the speed of light is a constant while mutation rates can change. We know mutations rates can change, we’ve observed it happening! Modern molecular clock analyses typically don’t work based on the assumption of a constant rate – they do allow for rate heterogeneity.

Finally, he considers natural selection:

“Perhaps the explanation involves natural selection. At first pass, this might seem plausible. After all, mtDNA encodes proteins with critical functions in the cell. If you interrupt basic metabolism, cellular death is sure to result. Surely most of the thousands of mtDNA mutations that have occurred over the last several million years of evolutionary time were lethal to the possessors of these mutations. Consequently, natural selection would surely have eliminated these mutations (and individuals) from the mtDNA pool.”

This is an explanation I imagine many of you already thought of as you read the preceding passages. Of course, it’s intuitive that at least some of the mutations we observe in pedigree studies are deleterious, so while they might not be lethal (meaning the individual can be included in the studies), they wouldn’t persist over longer timescales – natural selection would get around to weeding them out (Figure 4). This means that the mutation rate obtained through even the most accurate pedigree studies will always be higher than the substitution rate that we would get by looking at longer timescales. Like, say, the millions of years since the human-chimp common ancestor. The only question is to what extent this rate will be higher.

Figure 4 | Visualising the difference between mutation rate and substitution rate. “Germline mutations” represent all the mutations that occur. “Pedigree mutations” are estimated from mutations observed in pedigrees (or mutation accumulation experiments), while “Evolutionary substitutions” represent the mutations that actually accumulate over evolutionary time. As you can see, fewer substitutions are observed than mutations from pedigrees, because most of the deleterious mutations are lost. Adapted from Barrick and Lenski (2013).

Despite the fact that Jeanson seems to accept that natural selection would impact his results, and could go some way to explaining the discrepancy between the “evolutionary predictions” and reality, he dismisses it. Why? Because, Jeanson says, this explanation doesn’t predict mutation rates, ergo it’s not scientific:

“The elimination of thousands of mtDNA mutations by natural selection might seem plausible. But to be scientific, this explanation would have to make testable predictions. For example, the mtDNA mutation rate in the most divergent African people groups (San peoples, Biaka peoples, etc.) has not yet been measured. Can the evolutionary explanation of natural selection predict what this rate will be? In other words, before the rate is actually measured, will evolutionists publish a guess as to what it will be? If not, is the evolutionary explanation scientific?”

However, it doesn’t logically follow that invoking natural selection means that this model should be able to predict mutation rates in particular people groups. The ability to make predictions is important, but Jeanson isn’t the arbiter of what specific data must be predicted for a model to be scientific. Natural selection does make some relevant predictions that have in fact been tested.

The fact that natural selection eliminates some mutations over evolutionary time, influencing molecular clock analyses, is part of a model known as “time-dependent mutation rate slowdown”. According to this model, mutation rates are time-dependent, meaning that the estimated rate will vary depending on the timescale we analyse (Figure 5). In the very short term, such as in pedigree studies, rates will appear fast, because many deleterious mutations will be included. In the long term, such as in phylogenetic studies, they will appear to be much slower, as most of those deleterious mutations will have disappeared thanks to natural selection. In medium timescales, just some of the deleterious mutations will have been removed by natural selection so the mutation rate will appear to be intermediate. So, even with a constant mutation rate, selection and other factors can make it seem, at first glance, that mutation rates were slower in the past, just because fewer differences exist between lineages that diverged long periods of time ago than might be expected. The data supporting this has been around for a long time, so it’s not as though Jeanson is shocking the world and all of this is a new ad hoc response. No one in the field today is surprised that spontaneous mutation rates don’t match the number of mtDNA differences observed between populations/species that diverged long ago. Jeanson really hasn’t contributed anything new here, his results have been known about and explained long before he came along (e.g. Ho and Larson, 2006). That’s not the impression you would get from Jeanson’s book, however.

Screen Shot 2018-09-11 at 00.50.45.png
Figure 5 | Time-dependent estimates of mutation rate. With increasing time, estimated mutation rates (not actual mutation rates) decline as the spontaneous mutation rate becomes the evolutionary substitution rate. From Ho et al. (2011).

For some reason, Jeanson doesn’t mention this time-dependent rate slowdown at all in this chapter but he does in the next chapter (my emphasis):

“In the realm of mtDNA, evolutionists have already discussed a phenomenon termed time dependency.20 In other words, evolutionists have suggested that the mtDNA clock ticks at different rates at different points in history. Specifically, evolutionists have argued that mtDNA mutation rates have been slower in the distant past — an explanation which could, in theory, reconcile the erroneous predictions of the previous chapter with actual mtDNA differences.”

Given what he said in these few sentences, Jeanson doesn’t understand what time-dependency actually is. Despite citing Ho et al. 2005 in footnote 20, a paper which explains quite clearly what it is, Jeanson seems to think time-dependency means that mutation rates themselves were actually lower in the past. He repeats this several times in later chapters too. This really is a big oversight on his part. Either he is genuinely misinformed about this, or he’s giving his readers that false impression deliberately or by using sloppy language. None of these options would reflect well on Jeanson. To reiterate: time-dependency doesn’t mean mutation rates were lower in the past, it just means that they will appear this way when using ancient calibration points to estimate mutation rates, most notably because of natural selection removing mutations over evolutionary time.

Anyway, he’s at least heard of time-dependent mutation rate slowdown in the context of mtDNA, but he doesn’t mention it at all as a potential evolutionary explanation for his results in this chapter. Even when he briefly brings it up in later chapters, as you can see, it only gets a passing mention. It’s dismissed, of course, this time because it’s apparently inconsistent with evolutionary explanations for nuclear mutation rates, but I’ll cover that when I review chapter 8.

It gets better: the very data that Jeanson used bears the pattern of time-dependent mutation rate slowdown (Figure 6). Jeanson, knowingly or not, even indirectly comments on it in the text when he notes that there is less of a discrepancy between the “evolutionary prediction” and the actual number of mtDNA differences in his within-human calculation (diverged 200,000 years ago) than in his human-chimp calculation (diverged 4.5-17 million years ago). This pattern is precisely what is predicted by the slowdown hypothesis – note that Figure 6 matches the trend in Figure 5 from Ho et al. (2011) above.

Jeanson slowdown.png
Figure 6 | Jeanson’s data shows time-dependent rate slowdown. I used the “actual” number of mtDNA differences and the average divergence times Jeanson cites for his “evolutionary predictions”, or 7 million years ago for a more reasonable estimate of human-chimp divergence, to calculate the constant mutation rate that would be predicted by these dates, assuming a generation time of 20 years. For example, humans and Neanderthals diverged approximately 550,000 years ago, and given that there are 213 mtDNA differences between them, a mutation rate of 1.17×10^-8 mutations/bp/year would be estimated. The mutation rate for the “Jeanson Human Pedigree” point was approximately that used by Jeanson (4.8×10^-7 mutations/bp/year). The “Realistic Human Pedigree” rate point was 10x lower than Jeanson’s. A simple power curve was fitted to the data points. The Y-axis is logarithmic.

At this point, it’s worth mentioning one of the other important implications of Jeanson’s earlier acceptance of Li et al. 2010’s finding that sites with higher mutation rates are more likely to be found to be heteroplasmic in pedigree studies. It means Jeanson accepts that different sites in the DNA sequence have different mutation rates. Since sites with higher mutation rates are going to be overrepresented in pedigrees, the mutation rates estimated from them are inevitably going to be biased upwards, in favour of the higher mutation rates at those specific sites. When Jeanson extrapolates mutation rates from pedigrees over things like all the mtDNA differences between humans and chimps, he’s not accounting for these “hotspots” of mutation. Differences in these hotspots would saturate over long time scales, masking their presence. I’ll try and illustrate this with a hypothetical (and extreme) example:

Imagine a sequence that is 10,000bp long. The mutation rate for 9,999 of those sites is 1×10^-5 (0.00001) mutations per bp per generation. The mutation rate for the 1 remaining site (“site X”) is 1×10^-1 (0.1) mutations per generation. If you searched for mutations over 10 generations in a pedigree study, you’ll probably find a total of 1 mutation, and it will probably be in site X. From this, you conclude that the mutation rate over the entire sequence is 0.1 mutations per generation. Now Jeanson comes along and decides to apply this mutation rate to evolutionary timescales. He says the population originated 200,000 years ago (according to “evolutionists”), which he calculates to be 10,000 generations ago. So he calculates that there should be 2,000 differences (10,000*0.1*2) in this sequence between the 2 most divergent individuals if they really did diverge 200,000 years ago. But he only finds 101 differences between the sequences. At this point, Jeanson concludes there must be a problem with evolutionary timescales. The problem is that Jeanson’s analysis doesn’t consider which sites the differences are actually in. In this example, of the 101 observed differences, there is only 1 difference in site X, and the other 100 differences are spread throughout the rest of the sequence. This perfectly matches the evolutionary prediction of how many differences should be present when taking into account the relative mutation rates. Over the 10,000 generations, 2,000 mutations indeed happened, but 1900 of them occurred in a single site because of its higher mutation rate: site X. Since it’s only 1 site, it will only show up as a single difference in a comparison with another sequence, no matter how many mutations it actually experienced throughout history.

This is one of the other key reasons for time-dependent mutation rate slowdown aside from natural selection: sites with higher mutation rates will become saturated faster, meaning they show up more in short-term measurements (e.g. pedigrees) than they will over longer timescales (e.g. in species-species comparisons). Sites or sequence regions with higher mutation rates can also be influenced by purifying selection, meaning that even though many mutations are occurring in a particular region, they could be being removed almost as consistently.

As I said, this model can continue to be tested, for example with studies estimating mutation rates by calibrating with medium-term events known from archaeology. If estimated mutation rates keep falling on that curved line (Figures 5 and 6), this is further support for the time-dependent rate slowdown hypothesis. For example, consider the results of Rieux et al. (2014). They estimated mutation rates using calibration points of ancient DNA sequences that were independently dated to between 2,500 and 64,500 years old. Not only did their overall estimated mtDNA mutation rate come out to a value intermediate between the fast rate from pedigree studies and the slow rate based on more ancient divergences, but they also observed a time-dependent trend within their own dataset (my emphasis):

“We observed a subtle but significant negative linear correlation between the age of the ancient sequence used for calibration and the substitution rate estimated, […] which is consistent with this prediction.

Another prediction this hypothesis can make is that synonymous sites should display far less signature of time-dependence than non-synonymous sites, as mutations in these sites would be less likely to cause any fitness effect that could be acted upon by natural selection. For example, mutation rates estimated from 3rd codon positions should be fairly similar regardless of timescale, while the estimates from 1st and 2nd codon positions should differ, following the trend in Figure 5. This is a prediction that has been fulfilled (Endicott and Ho, 2008; Subramanian et al., 2009). Put more simply, deeper (older) branches in the phylogeny should display a lower ratio of non-synonymous to synonymous mutations than more recent branches, which is what we find (Kivisild et al. 2006).

To conclude then, there is extensive evidence for this time-dependent trend in mtDNA substitution rates, not just in humans, but in multiple animal taxa (Molak and Ho, 2015, Ho et al., 2015). Correcting for the causes of this trend, it’s possible to resolve the difference between pedigree-based and phylogenetic-based rate estimates (Soares et al., 2009). Now we’ll head back into Jeanson’s world where this data doesn’t exist.

Creationism: the superior model?

Content that “evolutionists” have a real problem on their hands explaining the observed data, Jeanson triumphantly unveils his alternative explanation:

“Predicting mtDNA differences for Homo individuals over 6,000 years exactly captures both the average mtDNA differences among non-Africans and among Africans (Figure 7.6).”

He shows that his inflated mutation rate, combined with a 6,000 year (YEC) timescale, fits the observed number of mtDNA differences among modern humans. Just. African populations are very genetically diverse so he has to use a generation time for African populations of 15 years to squeeze their average diversity into this timeframe, which he justifies with UN census data showing that African women got married younger than non-African women did in the 1970s. Almost 33% of African women between the ages of 15 and 19 were married according to 1976 census data (compared to 11% of non-Africans), and from this Jeanson concludes that the average age of a mother giving birth in Africa is/was/could be 15 years old. Yes, you read that right and no, it doesn’t make sense.

He also hypothesises that African populations could have up to a 2x higher mtDNA mutation rate than non-Africans and that that could help explain their great diversity arising in just 6,000 years. Again, you read that right. After all the song and dance earlier in the chapter about having to assume a constant mutation rate, Jeanson breaks this convention at the first hint of a problem for his timescale. Remember that this is on top of his use of a human mtDNA mutation rate that is on the order of 10x faster than the evidence suggests. Without that faster rate, Jeanson wouldn’t even be able to get anywhere near the actual African diversity using his timescale.

Next, Jeanson delves a bit deeper into the human data to talk about mitochondrial haplogroups:

“Another fact is readily observable in the human mtDNA tree (Figure 7.8): Three major groupings are visually identifiable. In fact, these nodes are so iconic that the evolutionary community has assigned specific labels to them. In technical terms, they are referred to as haplogroups. The L, M, and N haplogroups59 have been recognized for years.60”

A high-resolution version of the human mtDNA trees that he shows in this chapter can be found here, on the AIG website. Take a quick look (or see Figure 7A below). After a few pages about the supposed geological evidence for the flood (which I will cover briefly in the next section), Jeanson says that these 3 haplogroups must represent the descendants of the 3 wives of Noah’s 3 sons, as they repopulated the Earth post-flood.

It’s true that the 3 major haplogroups have long been recognised by mainstream science, but there’s a problem. I reproduced Jeanson’s human mtDNA analysis and the resulting phylogenetic tree (matching Jeanson’s) is shown below in Figure 7A. To a lay reader, it certainly does appear to result in 3 nodes (red arrows). However, anyone familiar with how to read phylogenetic trees should immediately realise that you can’t choose 3 nodes on an unrooted radial tree like this and decide that these represent 3 contemporary individuals that gave rise to the 3 clades (haplogroups, in this case). Lines (branches) on a phylogenetic tree represent ancestor/descendant relationships. The 3 nodes in Figure 7A are directly connected to each other by lines, so there must necessarily be some kind of ancestor-descendent relationships between these individuals.

Final tree fig arrows
Figure 7 | Human mtDNA phylogenetic trees. A and B are identical phylogenies, only differing in how they’re presented. A) Radial tree with no clear root (reproduction of Jeanson’s tree), B) rectangular phylogenetic tree rooted on the midpoint. The red arrows point to each of the 3 nodes Jeanson proposes as the wives of Noah’s sons. Black branches represent the L haplotype, purple represents the M haplogroup, blue represents the N haplogroup, and green represents the R haplogroup. I reproduced Jeanson’s phylogeny collecting the same set of 369 human mtDNA sequences (supplemental table 2 of Jeanson’s 2016 ARJ article), aligning them, removing the positions containing gaps, and using the “draw tree” function in ClustalX (neighbor-joining). Trees were visualised in FigTree v1.4.3. High-resolution version can be found here.

We can use Jeanson’s own methodology to estimate the number of generations that separated these individual women. In his 2016 article in ARJ, Jeanson says there are 2-8 mtDNA differences separating each of these nodes (or 2-10, depending on whether you believe the text or the figure of his article). We can use his own mutation rate and generation time from earlier to calculate the time between each of those nodes. It works out to be about 400-1500 years, which, using Jeanson’s generation times, represents about 8-100 generations. Put simply, Jeanson’s results and reasoning actually indicate that two of Noah’s sons were married to the great, great, great… 2-94x more greats… great-granddaughters of the wife of Noah’s remaining son. I don’t think I’ve ever heard a creationist defend that idea before.

To get an idea about the directionality of the ancestor-descendant relationships involved, a “rooted” tree is required. One simple way to do this is to root the tree on the “midpoint” of all the branches. Jeanson uses this method in Chapter 10 to root mtDNA phylogenies for other “kinds”, so he shouldn’t mind it being applied here too. The resulting human mtDNA tree is shown in Figure 7B above. It looks quite different, doesn’t it? Suddenly it becomes clear that the M and N haplogroups really are “descendants” of the L haplogroup, so the L, M, and N haplogroups couldn’t possibly have arisen from 3 contemporary maternal ancestors. Notice how the red arrows marking Jeanson’s claimed 3 maternal ancestors are positioned in Figure 7B, and that none of these women are ancestors of most members of the L haplogroup alive today. Also note that Figure 7B matches Figure 8 below, an illustrative cladogram of human mtDNA haplogroups from the literature.

Figure 8 |  Simplified rooted phylogeny of the human mitochondrial haplogroups. The 3 “main” haplogroups are L (orange and yellow), M (violet), and N (blue and green). Haplogroup R arises from within N, just as M and N arise from within L. From van Oven and Kayser (2005).

It’s also not clear what criteria Jeanson used to classify L, M, and N as the “main” haplogroups while excluding the R haplogroup. In his Figure 7.7 of an unrooted mtDNA phylogeny, the R haplogroup (green in my Figure 7A above) is labeled as a “subgroup”, without further explanation. I suspect it’s simply because he needs 3 main haplogroups to fit the narrative of them representing 3 wives of Noah’s sons, and that L, M, and N can fit the bill if he uses an unrooted phylogeny to present the data. The R haplogroup can’t fit into this model, no matter how he plays around with the tree topology.

Jeanson also claims that the relative branch lengths of this phylogeny support the YEC narrative. As he apparently doesn’t understand how to read his own phylogenetic tree, Jeanson thinks the branches connecting the 3 haplogroup nodes represent the “pre-flood” ancestry of Noah’s son’s wives, while the radiation from each node represents “post-flood” lineages. So, since the time ratio between pre-flood and post-flood is approximately 1:3 (1600:4400 years), Jeanson says he expects the branch lengths to follow a similar pattern. But they don’t. The ratio of branch lengths in his own figure is closer to between roughly 1:5 and 1:30. However, Jeanson doesn’t go into any specific calculations after calculating the 1:3 ratio, and simply says that there are “many more” mutations within each haplogroup than between them and that that’s good enough. He does say that the longer pre-flood generation times could mean that fewer mutations would have accumulated pre-flood, potentially accounting for the discrepancy, but doesn’t offer anything in the way of details.

This concludes his human mtDNA arguments:

“Thus, on three counts the mtDNA tree fits the young-earth creation model. First, the absolute DNA differences in this genetic compartment fits the 6,000-year timescale. Second, the existence of three nodes fits the fact of three surviving maternal ancestors at the time of the Flood. Third, the relative lengths of the branches connecting to and radiating out from these nodes, match the temporal expectations of Genesis 1–11, and the genealogies of the New Testament.”

As I’ve explained, the number of differences only fits the 6,000-year timescale if you use a grossly inflated mutation rate and ignore natural selection and the overwhelming evidence for time-dependent rate slowdown. The evidence for the 3 maternal ancestors and the branch length argument are entirely based on Jeanson’s misunderstanding of how to read a phylogenetic tree. Not exactly good evidence for YECism.

A Geological Tangent

As I mentioned earlier, in the middle of his claims about mtDNA haplogroups corresponding to the 3 wives of Noah’s sons that repopulated the earth post-Flood, Jeanon takes a few pages to try and justify the idea that such a catastrophic event actually occurred. He basically just throws out a lot of classic creationist canards, not spending much time on the details of the claims in the main text, instead referring to lots of different articles on the AIG website. Since he didn’t bother to put much effort into making these points, I won’t expend much debunking them. Time for a lightning round:

“The fossil record is filled with the remains of once aquatic creatures.62”

This obviously isn’t controversial, and Jeanson doesn’t directly try to make an argument for the flood with it. I just thought it was funny that Jeanson’s reference for this claim in the footnotes reads:

“62. Derived from my own analysis of the fauna in the Paleobiology Database (”

Thanks, Dr. Jeanson, I’m sure we’re all glad that you’ve personally analysed the fauna in the fossil database and come to the conclusion that there are lots of fossils of aquatic animals. Good to know. Let nobody ever say creationists can’t do ground-breaking research.

“The cause must also have been catastrophic in nature, rather than slow and gradual. Numerous peculiar fossils have been found that defy a tranquil fossilization event. For example, an Ichthyosaur mother in the process of giving birth was preserved in the rock layers (Figure 7.9). As another example, a fish was swallowing another fish before death ended its culinary endeavors (Figure 7.10). It’s difficult to imagine that these creatures froze in position, slowly sank to the ocean bottom, and then were slowly covered with sediment in a manner that avoided scavenging or disturbance. Instead, a rapid catastrophic burial seems much more likely.”

Yes, some animals can die by being quickly buried. Even sea creatures, for example, when storms or earthquakes rapidly displace large amounts of sediment. It’s also possible for animals to die in the middle of performing an action like giving birth or eating an oversized meal. Yes, believe it or not, sometimes fish try to eat a fish that’s too big and end up “choking” on it. Once these animals die, they can sink to the sea floor and be rapidly buried. The rapid burial of animal remains (not the rapid burial of live animals) is the most common way that remains become fossils. To quote from Robert L. Caroll’s 1988 textbook “Vertebrate Paleontology and Evolution”:

“The bodies of most animals are consumed or scattered by predators and scavengers soon after death, and their bones are broken up and decompose. Perhaps no more than one in a million are so quickly buried that they may become fossilized.”

Coincidentally, and definitely not in a contrived manner at all, the cover photo of Caroll’s textbook is of the same Ichthyosaur fossil in the process of giving birth that Jeanson shows. Next, and along the same lines, Jeanson says that we have data suggesting that entire fossil beds, not just individual fossils, were the result of catastrophic burial. This is true, some fossil beds formed in catastrophic events. There aren’t many ways of fossilising lots of remains in one go without something like a large ashfall or avalanche of sediment.

“Together, these conclusions appear to conflict with traditional geologic interpretations of the earth’s crust.”

Jeanson provides no justification for this claim at all. Nothing about the fact that “catastrophic events happen” conflicts with “traditional interpretations of the earth’s crust”.

“Nevertheless, recent geologic research has augmented the above paleontological results with geologic evidence for catastrophic burial. For example, some of the layers in the Grand Canyon that were thought to be iconic examples of desert deposition now appear to be the result of catastrophic aquatic processes.66”

Here Jeanson is referring to the Coconino sandstone, but he neglects to mention that this conclusion is a fringe one reached by creationists publishing primarily in the “Answers Research Journal”, not accepted by “mainstream” geology. See rebuttals to this creationist interpretation here, here, and here.

“The Mount St. Helens landscape is a 1/40 scale model of a much more familiar geologic landscape — the Grand Canyon (Color Plate 78). Though conventional geologic thinking puts the ages of the layers in the Grand Canyon at millions of years, the events at Mount St. Helens demonstrate the speed with which catastrophes can accomplish geologic work. The Grand Canyon doesn’t require millions of years to form. It just requires a large enough catastrophe.”

The mudflow at Mt. St. Helens cut through layers of loose sediment rather than the solid rock of the Grand Canyon, and the resulting small canyon it formed is clearly in the shape of a washout rather than a meandering river like that of the Grand Canyon. More details here and here.

“The catastrophic forces that formed the fossils layers around the world appear to have operated in the recent past. For example, soft tissue is recoverable from fossils.68 Some of these fossils are dated by conventional means to hundreds of millions of years ago. Yet, we know from experience that soft tissue decays rapidly. It’s difficult to imagine that dinosaur tissue could remain pliable after 65 million years of repose in the earth’s crust.”

Difficult to imagine, and yet clearly indicated by experiments. I’m not going to open the huge bag of worms that are the YEC soft-tissue arguments here. Entire books have been written debunking this interpretation, and a good article can be found here.

Neanderthals and ancient DNA

Now it’s time for him to mention how Neanderthals fit into his model:

“Up to this point, you may have noticed that I have said nothing about the YEC predictions for Neanderthal DNA. I did so deliberately. The explanation for these differences follows from what I just discussed. When Neanderthal and modern human sequences are visualized together in tree format, the Neanderthal sequences branch off of the sub-Saharan African lineages (Figure 7.12).75 From the YEC perspective that I’ve just outlined, it would appear that this lineage derived from ancient Africans.”

It only appears this way because Jeanson doesn’t use Neanderthals or even a midpoint to root his phylogeny, and instead insists on using an unrooted phylogeny. I suppose it comes down to what YECists think Neanderthals actually are. I’ve heard them referred to as early “pre-Flood” humans in some creationist literature (e.g. here), but now Jeanson is suggesting they’re just a very derived group of African humans that diverged post-flood.

“Since some African people groups might mutate their mtDNA faster than non-African people groups do, Neanderthal DNA might simply represent a hyper-mutating lineage — which eventually went extinct.”

For the second time, after warning “evolutionists” against invoking changing mutation rates, he’s now proposing that Neanderthals (in addition to modern Africans) were a hyper-mutating lineage in order to squeeze the data into his narrow timescale. Alternatively, Jeanson says that Neanderthal DNA sequences, being ancient, might simply be too degraded to analyse:

“In short, when I perform DNA sequence analyses in the lab, I tend to throw away DNA sequences that are older than a year. Despite storing them at -20° C, being 12 months removed from their normal cellular environment appears to do irreversible damage to DNA. How much more so when DNA sequences sit in fluctuating temperatures and environmental conditions for thousands of years. (My evolutionary colleagues disagree with my assessment regarding DNA degradation — which is why I still made predictions for Neanderthal DNA under the evolutionary model.)”

Here Jeanson makes it sound as though “evolutionists” think Neanderthal DNA is pristine and can be trusted at face value, and that he is the lone voice of reason piping up “in my experience in my lab, DNA seems to get damaged over short periods of time”. Nothing could be further from the truth, of course. Everyone working with ancient DNA knows their DNA samples are heavily damaged and degraded. As such, researchers are incredibly cautious and apply several methods to account for this. This is especially doable with mtDNA as the sequence is so short and abundant – it’s fairly easy to assemble based on even the very short, fragmented sequencing reads you get from ancient samples. For example, if you can get past the paywall, read how Briggs et al. (2009) obtained 5 independent Neanderthal mtDNA sequences from different samples.

What about other species?

Jeanson recognises that his human mtDNA mutation rate might just be an anomaly – it’s not very meaningful on its own. YECism says that all species arose recently, not just humans, so more species need to be considered to come to more confident conclusions. One immediate obstacle arises: of the 3 non-human vertebrate species to have their mutation rates estimated in pedigrees (mice: Hardouin and Tautz, 2013, chickens: Alexander et al., 2015, and penguins: Millar et al., 2008), none of them match Jeanson’s predictions (my emphasis):

“To date, only three vertebrate species possess a published mtDNA mutation rate — mice (Mus musculus),77 chickens (Gallus gallus),78 and Adélie penguins (Pygoscelis adeliae).79 At current rates of mtDNA mutation,80 neither the 6,000-year timescale nor the evolutionary timescale81 captures mtDNA diversity among species within these families (Table 7.3).82”

They’re not off by a little bit, either. The mutation rates are about 10-50x too slow to fit Jeanson’s 6,000-year timescale. Jeanson says that these results might simply be anomalous and that once more studies are done on these organisms an average mutation rate might emerge that will be in line with his predictions. However, across the human studies he worked with there was really only a 3-5x range of results, so it seems unlikely that the variance in the mutation rates in these other species across studies could be as high as >50x. The statistical power of the animal studies cited were actually quite good, at least on par with the best of the human studies Jeanson used to arrive at his average human mutation rate. In the case of the chicken study, the statistical power (by Jeanson’s metric) was 20x higher than the best of the human homoplasmy pedigree studies Jeanson cites, so it should be pretty trustworthy. Millar et al. (2008) only looked at Highly Variable Region I (HVR I) of penguins. As the name suggests, this region is known to mutate more rapidly than the rest of the mtDNA (it’s part of the D-loop), so is already going to overestimate of the whole-mtDNA mutation rate if extrapolated uniformly. Jeanson is hoping that the average overall mtDNA mutation rate will be 10x higher than the estimated rate in the fastest 2% of the mtDNA!

All 3 studies mention that their rate estimates are indeed higher than those estimated phylogenetically, and 2 mention time-dependent rate slowdown as an explanation. The exception was Millar et al., which only looked at the 384bp HVR I. They didn’t find evidence for time-dependence in this region, but in a paper published the following year they did find evidence for time-dependence in the rest of the penguin mtDNA (Subramanian et al., 2009).

While the 3 studies note that their rate estimates are higher than phylogenetic (“evolutionary”) estimates, they’re not “off” by as much as Jeanson reports. Jeanson says the mouse rate estimate is ~30x higher than the “evolutionary prediction” based on mtDNA differences between species, but the very study he cites says this difference is only 6x, referring to the phylogenetic rate calculated by Goios et al. (2007). He says the chicken rate is “off” by a factor of ~20-600x, but according to Figure 2 of Alexander et al. (2015), it’s only “off” by a factor of ~4-100x. Jeanson says the penguin rate is “off” by ~100-300x – this may well be correct, but again, this is so high because the rate Jeanson is using here is the mutation rate at HVR I.

We “evolutionists” have an explanation for this discrepancy in rate estimates, while Jeanson doesn’t offer any explanation for why they don’t match his YEC prediction. Not in this chapter anyway. He hints at a potential explanation in chapter 10, but we’ll cross that bridge when we get to it (spoiler alert: he invokes a higher mutation rate in the past).

Moving on from the vexing vertebrates, Jeanson brings up 3 invertebrate genera (Daphnia, Drosophila, and Caenorhabditis – common names: water fleas, fruit flies, and roundworms), and a yeast (Saccharomyces), for which mutation rate data and mtDNA sequences are available. For his calculations, he needs to know the generation time of these species, but he points out that these are mostly known from artificial laboratory conditions, and rates in the wild could be significantly slower due to things like diapause (slowing of development in response to stress). Jeanson slows all the lab-based generation times by 10x to estimate the “natural” rate.

Again, he runs the numbers using the “evolutionary” dates of divergence (millions of years ago) within the genera and finds that the number of observed mtDNA differences are ~5-100x lower than evolution would supposedly predict, in the best case scenarios. Here, as I mentioned earlier in this review, Jeanson employs frankly ludicrous bar charts to visually communicate these numbers to his audience. The invertebrates all have a mtDNA genome that’s approximately 15,000bp long, similar to that of the vertebrates we saw earlier, and yet Jeanson’s bar charts show the “evolutionary predictions” that go up to as many as >10 million differences. For the reasons I mentioned earlier (saturation), there’s just no way that it’s possible to “see” 10 million differences in these mtDNA comparisons, even if 10 million mutations actually happened. Jeanson comments on the fact that these numbers would necessitate that many mtDNA bases would have mutated multiple times, but again, nowhere does he make it clear to his lay audience the practical implications of this. Even worse than the invertebrates is his bar chart for yeast (mitochondrial genome of 74,000bp) which I’ve reproduced in Figure 9. His “evolutionary prediction” goes as high as 200 million mtDNA differences!

Reproduction of Jeanson's figure 7.19
Figure 9 | Jeanson claims evolution predicts too many Baker’s yeast mtDNA differences. A reproduction of Figure 7.3 from Jeanson’s book.

Despite the fact that practically, only a maximum of about 55,000 differences (75% of 73,963) could ever be detected in this case due to the effects of saturation, Jeanson still chooses to portray his calculations in the most visually striking manner possible, rather than the most practically applicable. He describes the numbers in the text, the only reason I can think that he’d want to reiterate them in these bar charts is to really drive home to his primary readership of creationists: “look at how big the difference is between what evolution predicted and reality!” By telling his readers that “evolution predicts 100x too many mtDNA differences”, he’s giving them the impression that the number of differences “evolutionists” expected to see is 100x larger than the number we actually find, when in reality, in the case of this yeast, “evolutionists” would expect to see a maximum of about 55,000 differences, so the “prediction” is only “off” the observation by ~3-11x (55,000 compared to 5,042-19,670).

Jeanson cites Haag-Liautard et al. (2008) for his fruit fly mtDNA mutation rate. Once again, he completely disregards natural selection despite the fact that the authors make it clear in the abstract that selection will reduce the long-term substitution rate (my emphasis):

“Our high mitochondrial mutation rate estimate largely comes from mutations at nonsynonymous major-strand G sites; these are subject to strong purifying selection in nature, and thus contribute little to between-species divergence.”

They also comment in the abstract that they were only able to quantify mutation rates in part of the mtDNA, excluding the slowest evolving part of the sequence, so their overall mutation rate is again expected to be an overestimate:

“Finally, parts of the Drosophila mitochondrial genome are so A+T-rich that we were unable to amplify and scan these regions for new mutations. The single-nucleotide mutation rate to single nucleotide events is expected to be lower in these regions than in the relatively G+C-rich regions that we were able to scan.”

It’s a similar story with Denver et al. (2000), the source of the C. elegans (roundworm) mtDNA mutation rate (my emphasis):

“A nonrandom distribution of substitutions with respect to coding function is expected in animal mtDNA evolution. Substitution patterns in the mitochondrial protein-coding genes among natural isolates of C. elegans display the typical bias toward synonymous sites (21). For example, between N2 and RC301, 26 synonymous substitutions and 3 replacements are observed. By contrast, in the MA lines 9 of the 15 mutations in nonhomopolymeric regions of mitochondrial protein-coding genes alter the amino acid encoded. The significant decrease in the proportion of synonymous mutations (P < 0.001) in the MA lines suggests a dominant role for purifying natural selection in the evolution of the mtDNA protein-coding genes in natural populations (22).”

Jeanson also cites Molnar et al. (2011) as using a mutation accumulation experiment to estimate the mtDNA mutation rate of another species (from a different genus) of roundworms: Pristionchus pacificus. He doesn’t include this genus in his analyses as P. pacificus is the only species with a mtDNA sequence available, so he can’t do within-genus (within-kind) analyses with it. However, Molnar et al. actually performed a within-species analysis with their available data. They worked out an observed mtDNA mutation rate for the species in the lab, then collected mtDNA sequences from 9 wild strains of the species from around the world and used their observed mutation rate to try and estimate the divergence times between the strains. They found the same result as I mentioned earlier – divergence time estimates using 3rd codon positions were consistently older than those based on the 1st and 2nd codons, likely as a result of selection. These estimates put the common ancestor of all of the wild P. pacificus strains living between 470,000-1,200,000 generations ago. Using a realistic wild generation time of 1-10 generations per year, Molnar et al. estimated that this common ancestor lived approximately 47,000-1,200,000 years ago, most likely somewhere in the ballpark of the low hundreds of thousands of years ago. Another counterexample to Jeanson’s predictions. I wonder why he didn’t mention it…

For the Daphnia mutation rate, he cited Xu et al. (2012). The authors made it clear that about 75% of the mutations they detected in their mutation accumulation lines were insertions and deletions (INDELs) rather than single nucleotide substitutions. Jeanson explicitly says in footnote 106 that he’s aware of this and that he decided to use the overall mutation rate (INDELs+substitutions). However, footnote 106 indicates that he used this combination of rates to predict the number of single base substitutions only, because when he compared the mtDNA sequences of the 3 Daphnia species to work out how many mtDNA differences there were, he says he excluded all the gaps, which would have represented INDELs! In other words, he’s saying “evolution predicts X number of mutations but we only find Y number of mutations”, while artificially reducing Y by ignoring all the INDELs – potentially up to 75% of all the mutations. On the other hand, in the caption of his Figure 7.20, he says that the “actual” number of mtDNA differences reported includes both single base differences and INDELs. Who are we to believe? Jeanson, writing in a figure description, or Jeanson, writing in a footnote? If he does count INDELs, he doesn’t explain anywhere how he did it.

To arrive at his upper limit of mutations predicted over 6,000 years, using a generation time of 73 per year (5 days per generation – the generation time in the lab in optimum conditions), Jeanson uses a mutation rate of 4.07×10^-7, which comes from the upper 95% confidence limit (CL) value from one of the asexual MA experiments reported in Xu et al., and is based on 5 mutations accumulating over 116 generations. Four of these mutations were INDELs in the D-loop – notice again how Jeanson’s analysis relies almost exclusively on D-loop mutations?

The other experiments, on both asexual and sexual lines of Daphnia reported significantly lower overall mutation rates – the next highest upper 95% CL was 2.53×10^-7, from the other asexual line. It’s fair to say then, that in choosing the upper rate of 4.07×10-7, Jeanson really was going for the absolute best case scenario. To account for the observed number of differences between Daphnia species using a generation time of 73 per year over ~6,000 years, Jeanson requires a mutation rate of at least ~3.5×10-7/bp/gen. The average reported in the paper, even for the faster-mutating asexual lines of Daphnia was only half that: 1.73×10-7/bp/gen. Remember that this is combining the substitution mutations and INDELs into a single larger mutation rate, the rate only for substitutions (which comprise most of the differences between Daphnia species) is about 5x lower still: about 3.75×10-8/bp/gen.

Looking at the alignments by eye, I’d estimate that approximately 120 INDELs of varying sizes are present between the mtDNA sequences of the 3 Daphnia species. Removing these gaps, and counting the substitutions, there are about 4,300 (out of 15,000bp). This means the ratio of INDELs to substitutions in the sequence comparison is about 1:36, while Xu et al. report a ratio of more than 2:1. This is an immediate red flag, indicating that you can’t simply extrapolate the overall mtDNA mutation rates from the mutation accumulation lines in Xu et al. to evolutionary timescales. Either the mutation dynamics have changed over time, or, more likely, natural selection and saturation are removing most of the INDELs over time, resulting in the severely reduced number of INDELs seen in the sequence comparisons within the genus.

Now back to Jeanson’s narrative. He claims his results are a big problem for evolution:

“Thus, in at least five independent measurements of the mutation rate — in humans, roundworms, fruit flies, water fleas, yeast — the evolutionary model failed to predict modern DNA differences. Furthermore, these species represented a very diverse swath of life. From a classification perspective, not only were separate phyla represented — Chordata (humans), Nematoda (roundworms), Arthropoda (fruit flies, water fleas) — but separate kingdoms (Animalia = humans, roundworms, fruit flies, water fleas; Fungi = yeast) were as well.

Consequently, the evolutionary failure becomes all the more difficult to dismiss as a statistical or biological anomaly. Instead, the results suggest a systematic problem with the evolutionary model.”

And, once again, despite Jeanson’s claim to the contrary, the evolutionary model adequately explains this discrepancy between mutation rates observed in pedigrees and over long evolutionary timescales through time-dependent rate slowdown. In fact, since these invertebrate species (and yeast especially) have very high effective population sizes and fast generation times, the effects of purifying selection and therefore the importance of considering time-dependence would be even greater in these species.

Moving on, Jeanson naturally offers the YEC timescale as an alternative explanation:

“In contrast, the YEC model successfully explained mtDNA differences among these animal and fungal species.108 Just like the results for modern humans, I found similar congruence between prediction and fact. After 6,000 years of mtDNA mutation, a few thousand mtDNA differences should separate roundworm species from one another. This is exactly what we see today (Figure 7.22). The same agreement between prediction and fact held true in fruit flies, water fleas, and yeast (Figures 7.23, 7.24, 7.25).”

There is one major caveat though. Jeanson was using the widest possible spectrum of mutation rates and generation times to come up with his 6,000-year expectations of how many mtDNA differences we should see in these species. This included the laboratory generation times that Jeanson estimated earlier were 10x faster than in the wild. So while the range of his predictions did encompass the actual number of mtDNA differences, even Jeanson recognises that in reality (the wild) this range will be more narrow. So narrow in fact, that in at least one case (the fruit flies), the “actual” number of mtDNA differences exceeds the number predicted by Jeanson’s 6,000-year timeframe.

Jeanson was comparing mtDNA differences between members of the entire genera as a proxy for “kinds”. In the case of the fruit flies, he was comparing different members of the genus Drosophila. So, Jeanson decides upon an easy out (my emphasis):

“a more conservative conclusion would be that mtDNA differences are explicable over the 6,000-year timescale at, perhaps, the subgenus level.”

In other words, if there are too many differences in a “kind” to fit into your 6,000-year timescale, just reduce the size (and therefore mtDNA disparity) of the group you classify as a “kind”! Jeanson has made himself an elastic ruler. Rather than having the groups called “kinds” be found to match the mtDNA disparity predicted by a YEC timescale, Jeanson has decided that any level of classification whose members have the mtDNA disparity predicted by the YEC timescale should be called a “kind”. “If the prediction works, great, if not, change the dataset until it does.” This might be a useful technique for the study of baraminology (deciding what “kinds” are), but it means Jeanson is abandoning the idea of making predictions that could falsify his model.

Jeanson hand-waves this objection away, basically saying “classifications are fuzzy”, and then moves swiftly on to say that his results are consistent with the pattern seen earlier in humans: the evolutionary timescales fails to make accurate predictions (retrodictions, really), while the YEC timeline did.

Jeanson’s final remarks

Jeanson wraps up by claiming that he’s shown solid evidence against evolutionary timescales and supporting YEC timescales (my emphasis):

The mtDNA findings contained in this chapter called into question the entire foundation of the evolutionary timescale. For example, in the fields of geology and astronomy, the entire millions-of-years paradigm rests on the assumption that rates of change have been largely constant.112 Yet, in the field of genetics, the assumption of constant rates of change (i.e., of mutational change) yields a 6,000-year timescale, not an ancient one. If constant-rate assumptions reject a millions-of-years timescale in the field of genetics, why should these assumptions be the only ones allowed in the fields of geology and astronomy?113”

After reading this paragraph carefully, you might be tempted to raise an objection: Jeanson seems to be mixing up constant and variable rate processes in his model. He’s happy to invoke variable rates in geology and astronomy but uses constant mutation rates to “call into question the entire foundation of the evolutionary timescale”. Surely that’s a contradiction?

Well, Jeanson anticipated this and discusses it at some length in footnote 113. In short, he says that Genesis 1-9 requires variable rates of change in geology and astronomy, but is less clear about variable rates of change in genetics. There’s wiggle room. Since the bible allows for constant rates in genetics, it’s not inconsistent for creationists to invoke it. On the other hand, he says, “evolutionists” arbitrarily insist that the present is the key to the past, so if we do that just to help our paradigm, it seems unfair to allow us to arbitrarily insist that genetic rates must not be constant just to help our paradigm again.

Of course, “evolutionists” certainly do not just “arbitrarily insist” on largely constant rates of change in fields like geology and astronomy. This is one of Jeanson’s most repeated points, but as I said earlier, these are in fact conclusions that have been reached on the basis of evidence, as these hypotheses make testable predictions which have been met. I already mentioned one in geology, outlined by Joel Duff here, and here’s another example from the field of astronomy. Decades if not centuries of scientific inquiry have shown that geological and astronomical processes are largely constant, while decades of genetic research show that mutation rates can be somewhat variable. There’s nothing contradictory there. Even if we assume mutation rates are largely constant between recently-diverged groups, the evidence and reasoning behind time-dependent rate slowdown show that Jeanson’s “evolutionary predictions” are based on faulty premises.

Jeanson argues that he’s shown that a YEC timeline makes testable predictions, and that “evolutionists” have always denied this. However, I don’t think anybody would ever deny that the hypothesis “the universe, the earth, and life are 6,000 years old” makes testable predictions. The problem is that all the significant ones unique to this hypothesis that have been tested thus far have been found to be incorrect. That’s why we reject YECism in favour of a superior model, not because it’s entirely unfalsifiable. Make no mistake though, aspects of YECism are unfalsifiable. Miracles can always be invoked to explain away inconvenient evidence. While Jeanson and some of his colleagues might tell the world “we’re testing Genesis”, they’ve already committed in writing that their belief is unfalsifiable. Read the AIG statement of faith (my emphasis):

“By definition, no apparent, perceived or claimed evidence in any field, including history and chronology, can be valid if it contradicts the scriptural record.”

Finally, Jeanson summarises his mtDNA results for the 8 genera:

“Similarly, though the results in this chapter involved mtDNA mutation rates from only six species, they were consistent across a very diverse set of biology. Again, in classification terms, these six species belong to two separate kingdoms (Animalia, Fungi), several animal phyla (Chordata — humans, Nematoda — roundworms, Arthropoda — fruit flies, water fleas), and two major arthropod divisions (Insecta — fruit flies, Crustacea — water fleas). Aside from the puzzle of the three vertebrate species (mice, chickens, and penguins), these results suggested that the 6,000-year timescale would apply in general across life.

Jeanson considers 8 “kinds” (genera) in total in this chapter: humans, mice, chickens, penguins, roundworms, fruit flies, water fleas, and Baker’s yeast. Just for fun, let’s quickly add one more result to Jeanson’s dataset before addressing his conclusion.

Chimps (Pan troglodytes) and bonobos (Pan paniscus) are the same “kind”, according to Jeanson. In fact, he makes it clear in the next chapter that he actually considers all great apes to be the same “kind”: chimps, bonobos, gorillas, and orangutans.
Let’s run the same calculation Jeanson did for the human-chimp expected mtDNA differences, this time on 2 species he believes did diverge from one another a maximum of 6,000 years ago. Chimps and bonobos have a similar generation time to humans, around 25 years, but let’s use the same range of generation times that Jeanson did for humans: 15-50 years. Jeanson was content to use his human mtDNA mutation rate as an estimation of the chimp mtDNA mutation rate, so I think it’s reasonable to extend it to bonobos as well.

We now have all our parameters to make a prediction (or retrodiction): a generation time of 15-50 years and an (inflated) mutation rate of 0.112-0.197 mutations per mitochondrial genome per generation gives us a rate of change of 1 base pair per 76 to 420 years. (6,000/76)*2=158 and (6,000/420)*2=29. Those are the upper and lower bounds of the YEC retrodiction for how many mtDNA differences should be found between chimps and bonobos. The reality? About 700. The YEC retrodiction is off by at least a factor of 5 (Figure 10). If we use a more specific and realistic generation time of 25 years to get a more precise retrodiction, that comes out to a mutation every 127-210 years, so 57-94 differences over 6,000 years of divergence. Less in the 4,400 years since the flood. Jeanson’s timescale is off by about 10x. If we assumed a similar mutation rate for gorillas and orangutans and did the same analysis among all the great apes, the problem would become far worse as Jeanson would have to explain several times more differences arising in the same time period. For example, chimps and orangutans differ at approximately 2,400 positions so Jeanson would have to invoke a great-ape mutation rate on the order of 40x higher than his calculated human mtDNA mutation rate.

Figure 10 | A 6,000-year timescale predicts far too few mtDNA differences between chimpanzees and bonobos. I used Jeanson’s human mutation rates of 0.119-0.197 mitochondrial mutations per generation along with generation times between 15-50 years to generate a retrodiction for the YEC timescale using Jeanson’s calculations. The “actual” number of mtDNA differences was obtained with a pairwise BLAST of the chimp and bonobo mtDNA sequences, and counting the number of mismatches.

So, of these 9 genera, 4 outright didn’t agree with Jeanson’s YEC predictions (chimps, mice, chickens, and penguins), and at least 1 of the invertebrates (fruit flies) didn’t agree with his predictions when using a realistic generation time, so he had to change that particular dataset from a genus down to a subgenus to get it to fit. Two of the remaining “kinds” (humans and Daphnia) took a good deal of contorting to get to fit the predictions. That only leaves roundworms (the Caenorhabditis genus, not the Pristionchus genus) and yeast as even half-decent demonstrations of Jeanson’s predictions. When more than half of the data doesn’t match a model’s predictions, it suggests there’s something very wrong, not that the model is correct and can be applied to every living organism.


So, let’s quickly recap the key points:

  • Jeanson argues that the nested hierarchy pattern of differences in mtDNA might not indicate common descent, and instead fit a design model. He offers no evidence to support this idea but proposes a way it could be tested. The only problem is that that test would be prohibitively difficult to actually perform, and even though Jeanson won’t be the one striving to do it, he says that we should suspend all judgment until the test has been done. That’s not how science works. We go with the best evidence available at the time, and in this case, it doesn’t support him.
  • Instead of paying attention to patterns of nested hierarchy, Jeanson says that a simpler test is to use observed mutation rates and available genetic sequences to calculate the ages of groups of organisms. Starting with humans, he introduces a human mitochondrial mutation rate that’s approximately 10x too fast because he generalises the mutation rate in a particularly rapidly-mutating region of the mitochondrial genome to the entire sequence. Next, he argues that we “evolutionists” must assume constant mutation rates to be internally consistent with our own approach, but gives himself a fair bit of wiggle room in this department.
  • Using his faulty mutation rate, a wide range of generation times, and data on the number of mtDNA sequence differences between human populations, humans and Neanderthals, and humans and chimps, Jeanson concludes that evolutionary timescales are far too long to accurately match the data. Instead, he shows that, with some significant massaging (invoking unrealistic generation times, increased mutation rates, and the supposed unreliability of ancient DNA sequences), he can get the human mtDNA diversity to fit into a 6,000-year timeframe.
  • Jeanson claims that mitochondrial haplogroups provide evidence of all modern humans tracing their maternal ancestry back to 3 women and that these 3 women were the wives of Noah’s sons that repopulated the world after the flood. This conclusion was based on a catastrophic misunderstanding of phylogenetic tree topology (it wouldn’t be the first time).
  • He performs similar calculations for 7 other animal genera (“kinds”) with data available and says that 3 of them matched his predictions. Even if I generously count the human and Daphnia data as matching his predictions, ignore the P. pacificus data, and add in my own analysis of the chimps and bonobos, Jeanson’s model still has a less than 50% success rate. I don’t think that’s very impressive, but Jeanson is confident he’s on the right track.
  • Throughout the discussion of the discrepancies between observed mutation rates and evolutionary timescales Jeanson consistently avoids the explanation that the mainstream literature provides: time-dependent mutation rate slowdown. He does mention natural selection as a possible explanation but dismisses it because it’s not predictive and therefore unscientific. Except, as I show, it does make predictions which been found to be accurate. He will bring up his critique of natural selection as an explanation again in the next chapter, and I will expand more on this subject then.

What this chapter boils down to is reporting the work Jeanson has been publishing in “Answers Research Journal” for the last few years: applying a 1st-year undergraduate-level equation to simple mutation rates and mtDNA sequences to calculate the age of mitochondrial Eve for several species/genera. He “rediscovers” observations of discrepancies between mtDNA mutation rates estimated using pedigrees/mutation accumulation lines and phylogenetic divergence times that have been published in the literature for over 15 years (and makes them a few times worse by exaggerating the human mtDNA mutation rate). In doing so, Jeanson tries to convince his audience that there’s serious evidence against evolutionary timescales by ignoring everything that “evolutionists” have found to explain this observation for over a decade.

By refusing to take into account purifying selection, rate heterogeneity, sequence saturation, base composition, and mutation biases, all of which are observable phenomena, Jeanson manages to contort the data into fitting into a 6,000-year timescale instead. Well, a small fraction of the data.

Jeanson has published all of this work in “Answers Research Journal”, AIG’s flagship “peer-reviewed journal”, and he has a legitimate doctorate from Harvard University. Despite this, when I take a step back and think about it, this level of work is less than what I would expect from an undergraduate thesis project. The data collection was simple, just counting differences in matrices of mtDNA sequences – no attempt was made to make any more detailed analysis of the types of substitutions, for example. Then he applied an equation found in introductory textbooks to mutation rates and generation times he found by reading a handful of papers. The analysis was even easier, just comparing a couple of bars on a bar chart. Do the 95% confident intervals overlap, yes or no?

And that’s essentially it. No further analysis, and most importantly, no detailed discussion of alternative explanations or interpretations of the data. Now, I didn’t go to Harvard, but I would hope that my university wouldn’t have awarded me even a Bachelor’s degree if I’d handed in such a simplistic and shoddy piece of work as my thesis. A shortened version wouldn’t even fly as a piece of homework. I’m really not exaggerating.

Jeanson himself describes this chapter as the “fulcrum” of his book – it’s the essential foundation for the YEC model that he lays out in Replacing Darwin. Now that it’s clearly been refuted, it’s almost not worth me reviewing the final 3 chapters. Almost. I’ve come this far, I’ll see it through. Just for you, dear reader.

In the next chapter Jeanson takes a swing at nuclear DNA instead of mitochondrial DNA, so look forward to that. I promise the wait for this next part of my review won’t be as long as the year-long wait for this one.

Comments and queries are welcome.


18 thoughts on “Reviewing “Replacing Darwin” – Part 6: Jeanson’s Fulcrum Fails

  1. As always, I admire your work. As always, I think you are working far too hard, and wonder how many readers will do more than glance at what you have written.

    If we are comparing Linnean, DNA-derived (mtDNA or otherwise), and design-predicted relationships, design would place manatees, seals, and whales close together, whereas Linnean anatomical classification, the fossil record, and molecular phylogenies agree in placing manatees with elephants, seals with lions and tigers and bears, and whales with hoofed mammals. (Placing whales anatomically was a bit more difficult than the other cases, but if I recall correctly the resemblance between the internal anatomy of whales and of hoofed animals had been observed as far back as the 19th-century.) That, surely, should suffice.

    So please keep up the good work, but I do feel that you’re giving more detailed attentionto one particular tedious piece of creationist nonsense nonsense than it deserves

    Liked by 1 person

  2. Jeanson does what critics of yEC always say. he does a exastive investigatiuon of claims about these trees etc based on dna concepts. That alone recommends this book for those interested.
    these are obscure subjects and not likely to gain audiences but its a excellent schoraly work.
    i’m glad the host here reveiwed it. it is indeed worthy.


  3. With regards to supposed constancy of mtDNA mutation rates, there are a couple of plant lineages (Plantago pro parte and Pelargonium pro parte) that have highly elevated mtDNA mutation rates. (Plant mitochondria in general have lower mutation rates than animals.)


  4. Really extraordinarily excellent. I had always suspected that Jeanson was cherry-picking data, applying inconsistent rate estimates, and ignoring the impact of selection pressure, but it’s good to see it all laid out from someone with the chops to do it properly.


    1. Jeanson is not chery picking. He shows that the genetic data can be interpreted easily in a creation framework, even in a YEC framework. He also shows that the evolution framework only works with several unproven, rather unlikely assumptions.


  5. “Now that it’s clearly been refuted, it’s almost not worth me reviewing the final 3 chapters.”

    This review looks like a scientific refutation of the creation science model, but as usual the focus in only on that part of the genomes that is shared between organism, also in Jeanson’s book. The shared part of nested hierarchy is completely irrelevant, however. It is not even interesting. What requires to be addressed is the non-shared, unique part that falls out of the nested hierarchy. That is, the 10-30% orphan genes that refute the common descent hierarchy. I would like to have that addressed by the evolutionary community. As long as they do not address it, the best explanation for nested hierarchy is not common descent but the combination of common design plus common mechanisms of genetic change, as I described in my book. IMHO, a frontloaded evolutionary theory is the better explanation. I have come to understand that evolutionary theory cannot be scientifically refuted: traits of organism can always be explained as the results of common descent plus modifications. Still, I would like to have this issue addressed. Is there any input from your side?

    Peter B.


    1. This series of blog posts is specifically a rebuttal of Jeanson’s book, not of all possible creationist models or arguments. I’m sorry I didn’t address your pet argument in this series, perhaps a future article might address the subject in detail.

      I’m not sure why you’re acting as though someone like me has to “address” this “problem” though – reading the scientific literature you can find plenty of discussion about the origin and distribution of taxonomically restricted genes. There’s no indication that the subject is in fact a problem for evolutionary mechanisms. If you want to discuss this subject feel free to make a post on, so I and others can chime in. Before doing so, I recommend you read through some of the existing threads on this subject, so as not to retread old ground.


  6. Your explanations cannot be scientfically tested. They look very sophisticated and scientfic but they cannot be falsified. For instance you say “…it would appear as though on the order of 12,000 mutations had occurred – there would be no way to distinguish between 12,000 mutations occurring and 447,000 mutations occurring, because of saturation.” This is just nonsense. As if this would biologically even be possible. What you have to explain is the data as they are, set up a hypothesis and test it. In science, observations have to be explained, not non-observations or possibilities. Since Darwin, the evolutionist is obsesses with possibile, imaginative stories. Evolutionary theory is refuted every the time we treat is as science. So, the creationists are right in claiming we are dealing with two competing worldviews, not with science. And that is why the debate is never ending.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s