Alternative explanation for nested hierarchies of genetic mutations?

I have a question about one of @DennisVenema’s blog articles on the Biologos website. Perhaps Dennis or someone else who is also knowledgeable of genetics could answer it?

The article is this one about the evidence for common ancestry for humans and other apes due to nested hierarchies of genetic mutations:

I’m no expert on any of this, but this is the most compelling evidence for common ancestry I’ve ever seen.

But I do have a question about the main argument. Could the nested hierarchy pattern be explained by the degree to which the species that share the mutations also share overall DNA similarity? So, for instance, suppose my Honda and my wife’s Toyota are 95% mechanically similar to each other but only 90% similar to my neighbor’s Kia. Then suppose that we found out that the Honda and Toyota shared six mechanical problems, but of those six only four are shared with the Kia. In this sort of case, a plausible explanation would be that the commonalities in mechanical problems are due to broader commonalities in overall mechanical structure. That is, the four problems shared across all the cars is due to the 90% shared structure, and the two that the Honda and Toyota share are due to the extra 5% similarity in structure that they share that isn’t shared by the Kia.

The point is that for the nested hierarchy pattern to be evidence for common ancestry, we’d need to know that the pattern isn’t explained by the very features that already gave us the hierarchical structure to begin with: the much broader similarities in DNA and phenotypic traits. We’d need to know that the mutations specific to two species on a “branch” of the hierarchy “tree” are not caused by the commonalities in virtue of which they share the same branch.

But maybe there are good reasons to think that genes can’t bring about mutations in other genes, or that common phenotypic traits can’t contribute to common genetic mutations. I’m just too ignorant of genetics to know.

Any thoughts would be much appreciated. Or if this has been addressed this before, even a link would be great! Thanks!

[Mod tag: @Swamidass @Joel_Duff @glipsnort]


You raise a really good question, Paul. It just so happens that we had a very deep and highly informative discussion about this very question about 2 weeks ago. You can find the thread here. It is still open, so you can participate, ask questions, raise issues, etc. to your heart’s content on that thread.


Hi @Chris_Falter, thanks for the tip, but I don’t see my question addressed anywhere on the other thread you link to. If you could tell me which part of the thread you’re thinking of, that might help me.

Note that my question is just whether the nesting pattern of genetic mutations Venema describes could also be explained by the other features the species have which already account for the phylogenetic relationship they stand in. The issue under discussion in the thread you link to seems very different to me, focusing instead on the adequacy of current phylogenetic classifications as opposed to alternative classifications in terms of species similarity. The sort of alternative explanation I’m wondering about is consistent with the sort of phylogenetic classification Dr. Swamidass is defending there. If the alternative explanation I"m wondering about can’t be ruled out, all it would undermine is Dr. Venema’s claim that the nested hierarchies of mutations are themselves evidence for common ancestry. Which of course is consistent with there being a great deal of other evidence for common ancestry.

At least, this is how it seems to me, but I know next to nothing about evolutionary genetics, bioinformatics, etc., so feel free to set me straight.


Hello Paul,

One important point: we’re not looking at nested hierarchies of mutations, we’re looking at nested hierarchies of sequences. [quote=“PaulNed, post:1, topic:34487”]
The point is that for the nested hierarchy pattern to be evidence for common ancestry, we’d need to know that the pattern isn’t explained by the very features that already gave us the hierarchical structure to begin with: the much broader similarities in DNA and phenotypic traits.
No, we wouldn’t. i think the distinction I made above is important in correcting this misunderstanding.


I think you should try to start over from scratch. A few things:

  1. We’re not looking at mutations, but rather fixation of alleles.
  2. The most compelling part is that the hierarchy for components (protein sequences) repeatedly and predictably replicates the hierarchy for organisms, allowing for minor systematic errors. This is why your car analogy isn’t relevant.

Would engaging with the software and data online, constructing trees, help?

1 Like

I think I understand your question. The thing is that not all genetic information contributes to phenotype. There are three-letter bits of DNA that each code for an amino acid. These amino acids get strung together and depending on the sequence, fold up differently into proteins. Most proteins only have a few little surface areas that actually interact in the cell to do stuff.

But there are more three-letter possibilities than there are amino acids used in cells, so they double or triple up and different DNA letters can code for the same amino acid. Also, it’s entirely possible to swap out amino acids and still have your protein fold the same way and perform the same function. And finally, there’s a lot of DNA that doesn’t even code for proteins at all, not to mention proteins that float around and don’t do anything.

So phenotypes that match up are one thing. Finding out that the genotype underneath matches up is a whole 'nother level. It’s like taking one of your car parts out and taking it apart and discovering that the scratches from manufacturing it are in the exact same places.

Does that mean that the car parts were actually made in the same factory, or just made according to the exact same process by the same design? Well, you can argue about standards of certainty for a while if you want, but what it comes down to is exactly what you said: it’s really good evidence for common descent.

Hi Paul,

In this post to the previous thread, our friend @Swamidass explains that nested hierarchies provide much stronger predictive power than genetic similarity measures. Why do they provide much better explanatory power? The scientific answer is that nested hierarchies are a good model of the real world facts.

Hope that helps,


@benkirk I’m not sure you’re thinking about the article I’m asking about. It’s the one by Venema I linked to in my OP. He is talking about nested hierarchies of mutations, even if you’re not. Perhaps you are talking about the same thing he is, but if so you’re using different terminology (“sequences” as opposed to “mutations”). That’s of course fine as far as it goes, but since I don’t know much about evolutionary genetics it would help me if you stuck to the terminology Venema used. He explained things well enough for me to feel like I understand the point he’s making.

I did read the other thread you and Chris mention, and if it helps with my question, the way in which it does needs some spelling out. See my first response to Chris to see why I don’t see why the other thread helps.

I think you’re getting hung up in the irrelevant parts of my car analogy. Let me try again. Here is the key part of the argument Venema makes (quoting):

“…in the data set from this study, twelve identical mutations were found to be shared between primates in a particular pattern…Three of these mutations are shared between humans and chimpanzees only; a further three are shared between humans, chimpanzees and gorillas; and a further six identical mutations are shared between humans, chimpanzees, gorillas and orangutans. This pattern is readily explained by the following: the common ancestral population of all four species has six mutations that occurred prior to the lineage leading to orangutans branched off; three more that occurred before the lineage leading to gorillas branched off; and another three that occurred before the lineages leading to humans and chimpanzees separated…Put another way, in this data set if we see a mutation shared between humans and gorillas, we see those exact mutations in chimpanzees without fail. Likewise, if we see a mutation shared between humans and orangutans, we see those exact mutations in gorillas and chimpanzees, again without fail. The shared mutations make a precise pattern that is exactly the pattern we expect if indeed these species share common ancestral populations that progressively divided into four lineages—a pattern known as a nested hierarchy.”

This is supposed to be an argument for common ancestry, I take it, because the best explanation for why these mutations would nest down the lineage tree in this way is that the lineage pattern they follow really does capture the historical lineage of humans, chimps, gorillas, etc.

My point is this: this nesting of mutations is evidence for common descent only if there are not other properties of the genomes of humans, chimps, etc. that follow the same nesting pattern and explain the nesting pattern visible in these mutations. And there must be other properties of the genomes that follow this same nesting pattern, otherwise there wouldn’t be grounds for positing the lineage pattern that Venema thinks gains additional support from this mutation nesting. In other words, there are reasons Venema and others think humans, chimps, etc. stand in this lineage, independent of nesting mutations. I assume the reasons Venema and others place these species in this lineal relationship is partly due to similarities in genome that also follow this nesting pattern. I mean, that’s how they got that pattern to begin with, right?

So the question just is: could any of these other shared genomic properties which also follow this nesting pattern explain the nesting pattern of mutations? And if so, how likely or unlikely would this be?

Without dealing with this issue, the sort of argument Venema is making here doesn’t work. I’m not saying his argument doesn’t work–maybe there are great reasons to think the sort of explanation I’m asking about can’t happen. I’m just asking for someone to explain why this sort of alternative explanation isn’t available.

(Also note: what I’m asking about is not “common design,” or whether these shared mutations could have randomly occurred in just the right pattern across these genomes of the species, or any other alternative explanation Venema addresses in his post).

Thanks again to all who are trying to help me out here! Any thoughts appreciated!

I think I see where you are getting at. I think this is your key question…

Not that I know of. What specific genomic properties do you think might do this?

Similarity does not necessarily produce a nested clade pattern of mutations. It is possible to construct sequences that fall into nested clades differently than one would place them if just forming the tree by overall similarity. Yes, there is a correlation between similarity and clade membership, but similarity is not enough to cause the nesting pattern we see. The other thread gives some good evidence for this point. Empirically speaking, on real DNA sequences hierarchies constructed by similarity are often different than hierarchies constructed from nested clades.

If you have another genomic property in mind, I would be happy to consider it.

I think your conjecture is interesting and it points to real and testable hypotheses. If your goal is to make precise hypothesis and test them against the data, I support that. But, if your goal is to just raise the possibility without actually seeing if it can be demonstrated, that is a problem.

Until we are able to identify these “genomic properties” and demonstrate that they would produce the same nested clade pattern (with the right number of exceptions to the rule), then common descent really does remain the best scientific explanation of the nested clade pattern.

So what exact genomic properties do you think could cause the nested clade pattern?

[quote=“PaulNed, post:8, topic:34487”]
I’m not sure you’re thinking about the article I’m asking about. It’s the one by Venema I linked to in my OP. He is talking about nested hierarchies of mutations, even if you’re not.[/quote]
No, he really isn’t, even though it looks that way from your perspective. The nested hierarchies are constructed from sequences. Venema is discussing individual differences between the sequences and where they appear in the hierarchy.

We are talking about the same thing. Venema is pointing to specific features (mutations) in the hierarchy derived from sequences. I’m hoping that by pointing this out I can help you understand it better.

I’m a geneticist, and I think your feeling is incorrect.

The data set is sequences!

And that hierarchy is the same as the one from the whole sequences. Venema is trying to get you to look at the trees. I’m pointing out that the forest (the data set) says the same thing, which seems to be what you’re asking about!

It’s a granular one, but the entire sequences give the same picture in a more rich, but also more difficult to illustrate, way.

I don’t see your point here. The “if there are not” throws it off.

That’s why I’m pointing out that the entire sequences have the same relationship. If you look at the cladogram, the length of the horizontal lines should be proportional to the total number of differences (mutations).

There are many more of these.


[quote]So the question just is: could any of these other shared genomic properties which also follow this nesting pattern explain the nesting pattern of mutations? And if so, how likely or unlikely would this be?
[/quote]No, but really I have no idea what you mean by “shared genomic properties.”

That is a better way of explaining what I am trying to explain.

The car analogy is defective because the biological hierarchy is much, much more precise than similarity. Another difference is that if we look at every protein in these organisms, each hierarchy is superimposable the whole and on each other, even for nonfunctional differences. That is in no way true for the cars; for example, they may use identical 10-mm bolts from the same supplier.

Dr. Swamidass, thank you for your charitable reply. Up until now it has been difficult to find someone who both understood the genetics and could see my point from the perspective of a non-scientist.

A few follow-ups, if you don’t mind:

First, I take your point about similarity not being the ultimate basis for clade membership. But I think my point remains just the same: there are some properties sequences have in virtue of which Venema and others place them in the nesting pattern. Call these the ‘Fs’. To know that the mutations in question are evidence for this nesting pattern being correct we’d need to know that the mutations aren’t being caused to occur in that pattern due to the Fs. Same point.

Second, I of course don’t have any specific genomic properties in mind that might do this, because I am not a geneticist and thus wouldn’t know enough about the genome to have a hypothesis about this “common cause.”

Third, I certainly don’t understand why pointing out a potentially confounding variable for Venema’s explanation is problematic if I don’t myself intend to see whether it can be demonstrated. It’s not like an explanation is any better or worse if those it is presented to plan to test for confounding variables or not. No–the explanation either is able to explain what it aims to or it isn’t.

Put it this way: if you proposed an explanation, and others asked whether you had checked for a certain sort of possibly occurring confound, it wouldn’t do to reply “No I didn’t, but my explanation is still good until somebody else does, and your asking about this potential confound is problematic unless you yourself intend to check on the confound.” Right? That just doesn’t seem at all to be in the spirit of good science.

Or at least this is what it sounds to me like you’re saying here. But again, I’m no scientist, and may have the circumstances wrong.

Finally, you point out that unless and until someone demonstrates my hypothesis to be true common descent remains “the best scientific explanation of the nested clade pattern.” I totally agree. And in fact, even if someone did demonstrate my hypothesis (which I don’t expect), I wouldn’t take that to be sufficient to dethrone common descent as the best scientific explanation here.

I see your point. However, knowing what I know of genetic sequences, I cannot imagine what these “Fs” are exactly. I understand conceptually your point, but unless we can actually enumerate these possibilities, there is no way to test them. It is not possible to unspecified and known things.

I’m not being argumentative here, but I just can’t really imagine what could do this other than common descent.

The only example I can come up with is if God is creating new species from the genomes of other species with modifications to their genomes that look like what we expect from mutations. This is consistent with some versions of the Reasons to Believe special creation model. But even in this case, the nested clade pattern is inferring a real history of descent. Other than this contrived example, I can’t see another solution.

I understand what are you saying, but the only proposals that anti-evolutionists have put forward are:

  1. SImilarity is caused by common design. As the other thread explained, this misses the point on several levels. First off, similarity does not exactly correlate with nested clade, and function correlates more closely with clades the similarity.

  2. Alternatively, Walter Remine proposed in the Biotic Message that God did this to make it clear that He designed us all. The exact reasons for this are somewhat esoteric, but key to his argument is two things. (1) This pattern is supposed to be absolute with no exceptions; and the lack of exceptions is what is supposed to make it a clear signature of God’s action. (2) Evolution on the other hand is predicted to produce nested clade patterns, but not always; sometimes there are exceptions. It turns out that ReMine is correct about #2 (because of things like incomplete sorting, drift, etc.), but the data itself shows predominantly nested clades with rare exceptions (just as we predict from evolution).

So none of these end up explaining the data. A lot of people have thought about this for a long time, but we have yet to see an argument much different than this.

The problem is that you haven’t actually pointed out a confounder. You are just supposing that maybe one exists. Who knows? Maybe one does, but it is not at all obvious what it is. I can’t really imagine what it could be.

But we have tested all the known confounders. As other potential confounders are identified, we will test them too. But there isn’t enough specificity to your notion for it to be testable.

In science, this is all we are left with: the best explanation we can come up with, acknowledging that all our explanations are provisional and maybe there are better theories out there. Now, in the case of common descent, nested clades are not the only piece of evidence pointing this way too. So it will take an explanation that will explain nested clades along with all the other patterns we see.

Just propose a specific hypothesis and it might be testable. Positing the possibility that there is an explanation is just an instinct. We need more than that.

Well we agree there =).


This topic was automatically closed 6 days after the last reply. New replies are no longer allowed.

“Let your conversation be always full of grace, seasoned with salt, so that you may know how to answer everyone.” -Colossians 4:6

This is a place for gracious dialogue about science and faith. Please read our FAQ/Guidelines before posting.