Our Discovery Institute colleague Cornelius Hunter has offered a response to Dr. Joshua Swamidass’s arguments for human-ape common ancestry. Hunter points out that Darwinian evolution doesn’t really predict any particular degree of genetic similarity between rats and mice, or between humans and apes, undercutting Swamidass’s argument for common ancestry. While Dr. Hunter’s post was focused on the theoretical side of things, there are evidential arguments to make as well, many of which we’ve seen in the past here at Evolution News.
(1) Swamidass argues that humans and chimps are only about 1.5 percent different. However, by some metrics human and chimps are much more genetically dissimilar.
As we’ve discussed before (see here or here), depending on how you measure genetic similarities, human and ape genomes might be far greater than 1 percent (or 1.5 percent) different. For example, Casey Luskin wrote:
While recent studies have confirmed that certain stretches of human and chimp DNA are on average about 1.23% different, this is merely an estimate with huge caveats. A recent news article in Science observed that the 1% figure “reflects only base substitutions, not the many stretches of DNA that have been inserted or deleted in the genomes.” In other words, when the chimp genome has no similar stretch of human DNA, such DNA sequences are ignored by those touting the statistic that humans and chimps are only 1% genetically different. For this reason, the aforementioned Science news article was subtitled “The Myth of 1%,” and printed the following language to describe the 1% statistic:
- “Studies are showing that [humans and chimps] are not as similar as many tend to believe”;
- the 1% statistic is a “truism [that] should be retired”;
- the 1% statistic is “more a hindrance for understanding than a help”;
- “the 1% difference wasn’t the whole story”;
- “Researchers are finding that on top of the 1% distinction, chunks of missing DNA, extra genes, altered connections in gene networks, and the very structure of chromosomes confound any quantification of ‘humanness’ versus ‘chimpness.'”
Indeed, due to the huge caveats in the 1% statistic, some scientists are suggesting that a better method of measuring human/chimp genetic differences might be counting individual gene copies. When this metric is employed, human and chimp DNA is over 5% different. But new findings in genetics show that gene-coding DNA might not even be the right place to seek differences between humans and chimps.
Far from being “but a hand-breadth away from our evolutionary cousins at the DNA level,” the evidence shows that the genetic differences between humans and chimps amount to “35 million base-pair changes, 5 million indels in each species, and 689 extra genes in humans.” Such a level of differences is not known to exist for human/human intraspecies genetic variation. Moreover, as geneticist Richard Buggs has explained, the genetic similarity between humans and chimps may be even lower than 95%.
Swamidass further writes: “As predicted by common ancestry, human and chimpanzee genomes are extremely similar (greater than 98% similarity in coding regions), much more similar than we would expect without common descent.”
At first glance this seems to be a standard mistaken argument that a high degree of similarity indicates common ancestry. It would seem to ignore common design as a perfectly good explanation for human/chimp common ancestry.
But Swamidass isn’t only making the typical simplistic argument that 1.5 percent genetic similarity between humans and chimps (if real) implies we’re related. Rather, he’s arguing that humans are more similar to chimps than are species we already agree are related, like mice and rats.
Here, Swamidass echoes Dennis Venema who argued that human intraspecific genetic variation is as much as 2 percent, and yet everyone believes that all humans are related. So if apes are only 1.5 percent genetically different, why not think we’re related to them? Casey Luskin has responded:
Initially, I must point out that Venema’s post provides no citation for his claim that human intraspecies genetic variation is as much as 2%. I contacted Dr. Venema privately and asked for citations that back this claim, but neither of the two papers he provided proposed an official percentage of human intraspecies genetic variation. One paper suggested that there are about 10 million single nucleotide polymorphisms (SNPs) and about 1.5 million insertion and deletion polymorphisms (INDELs) in the human genome. But the other paper, published in 2011, provided the highest estimates for human intraspecies genetic variation, stating: “Like known SNPs, which affect ~15 Mb of DNA … our INDEL variants affected 11.9 Mb of the human genome.”
Looking at the latter article, if we add those amounts together and round up to about 27 million base pairs, that implies that only about 0.9% of the human genome is known to vary — not 2%. This statistic is corroborated by the National Institutes of Health website which states “Human DNA consists of about 3 billion bases, and more than 99 percent of those bases are the same in all people.”
Swamidass probably realizes that extrapolating from human intraspecific genetic variation to common ancestry of humans and chimps doesn’t work as an argument. Why? Because human intraspecific similarity is probably greater than human-chimp genetic similarity. Thus, he changes his argument and looks instead at more tractable genetic statistics: mice and rat genetic similarities. He asks: If we’re more genetically similar to apes than mice are to rats, and if we believe that mice and rats share a common ancestor, then why not reason that humans and chimps do as well? This brings us to our next error.
(2) There’s good evidence that genetic differences between humans and chimps could not arise by standard evolutionary mechanisms.
Let’s assume for the sake of argument that Dr. Swamidass is correct that mice and rats are about 15 percent genetically different whereas humans and apes are only 1.5 percent different. The much bigger problem is that even in humans and chimps, those few percent differences still equate to tens of millions of base pair differences, plus millions of larger differences that entail indels, etc. And when we’re looking at humans and chimps, if just two of those specific differences were required before giving any functional advantage, then that would have taken over 200 million years to arise. So even very small genetic differences can still take a long time to arise. Casey Luskin explains here, writing in response to Dennis Venema’s similar argument:
Venema has badly overstated the case for Darwinian evolution, as no one has any idea if these changes are “easily accessible to evolutionary mechanisms.” In fact, we still don’t understand what the vast majority of these differences mean. When speaking of human/chimp genetic differences, David Haussler, a biomolecular engineer at UC Santa Cruz, writes “To sort out the differences that matter from the ones that don’t is really difficult.” And there are good reasons to believe that some (or many) of these differences might encode features not amenable to stepwise Darwinian evolution under known timescales.
In 2008, Michael Behe’s critics Rick Durrett and Deena Schmidt tried to refute him in the journal Genetics with a paper titled “Waiting for Two Mutations: With Applications to Regulatory Sequence Evolution and the Limits of Darwinian Evolution.” But Durrett and Schmidt found that to obtain only two specific mutations via Darwinian evolution “for humans with a much smaller effective population size, this type of change would take > 100 million years.” The critics admitted this was “very unlikely to occur on a reasonable timescale.”
In other words, if any of the 35 million base pair changes between humans and chimps entail adaptive changes that require two or more specific mutations before providing any advantage, then they would be extremely unlikely to evolve by random mutation and natural selection in the mere 6 or 7 million years since we shared our alleged most recent common ancestor with chimps.
Durrett and Schmidt found that it would take too long to wait for two specific mutations to gain an advantage. How does Venema know that there aren’t many differences between humans and chimps that would require two (or perhaps many more — dozens, hundreds, millions?!) mutations before any benefit arises?
Venema’s claim that the genetic differences between humans and chimps are “easily accessible to evolutionary mechanisms” is difficult to accept, because frankly, no one knows that this is true. We should not assume that naturalistic evolutionary explanations are correct; we should test them, and hold on to them only if they are valid.
Humans and chimps supposedly shared a common ancestor 6-8 million years ago (Ma) yet there are millions of genetic differences. If only 2 of those were required to give some advantage, then it would take over 200 Ma to become fixed into a hominid population. This means there’s not enough time for humans and chimps to have evolved via selection and mutation.
Now it’s true that this rebuttal undercuts Darwinian processes, and Swamidass isn’t arguing specifically for an “unguided” origin of humanity. He’s only arguing for common descent. But how does he establish common ancestry? He tries to establish it by assuming that the genetic similarities between humans and chimps (or between mice and rats) arose by unintelligent evolutionary mechanisms — random mutations that were passed from a common ancestor on to diverging lineages through unguided evolutionary processes — with no interference by any intelligent agency. In other words, his argument is predicated upon the assumption that our genetic code diverged from apes through strictly undirected mechanisms. His case for common ancestry isn’t based upon design; it’s based upon random mutation and unguided evolution.
But if humanity’s origin isn’t unguided, and if there has been design in the genome, then there’s no basis to argue that any particular genetic similarity between humans and chimps necessarily reflects an undirected evolutionary history — this includes both Darwinian/neutral evolutionary processes and its implication of common ancestry. Why? Because of the possibility of common design — where a designer re-uses similar functional components (i.e., genes) in different designs (i.e., species).
(3) Swamidass doesn’t appreciate the power of arguing from common design.
Swamidass asks, “What design principle can explain why humans are 10 times more similar to chimpanzees than mice are to rats? No one knows.” Again, ID proponents have long-identified such a principle: common design. As Paul Nelson and Jonathan Wells explain:
An intelligent cause may reuse or redeploy the same module in different systems, without there necessarily being any material or physical connection between those systems. Even more simply, intelligent causes can generate identical patterns independently.
(Paul Nelson and Jonathan Wells, “Homology in Biology,” in Darwinism, Design, and Public Education, (Michigan State University Press, 2003).)
Intelligent agents re-using parts and components that perform common functions can readily explain how different systems come to have similar properties. Swamidass doesn’t ignore the possibility of common design — in fact he tries to demean it by calling it a “lawyerly objection” to common ancestry.
Clever though his rhetorical retort may be, he never attempts to explain why common design can’t account for functional genetic similarities between species, such as humans and chimps.
(4) Swamidass never establishes that mice and rats are related.
Swamidass’s main argument is rhetorical. He claims that “humans are about 10 times more similar to chimpanzees than mice are to rats,” and since (in his own words) “we readily accept mice and rats are related,” we must therefore also agree that humans and chimp are genetically related. He thus writes that we appear “more closely related to chimpanzees than mice are to rats, as if we share a common ancestor with them.”
Despite his protests against “lawyerly” arguments, it’s Swamidass who sounds a little like a lawyer here. Again, his argument goes like this: Humans and chimps are more similar than mice and rats. You accept the shared evolutionary ancestry of mice and rats. So why don’t you accept the common ancestry of not humans and chimps?
Perhaps mice and rats do share a common ancestor, but when exactly did we “readily accept” this? Indeed, molecular clock data is all over the map when it comes to explaining the degrees of genetic similarity between mice and rats, indicating their supposed divergence dates. Papers gives dates ranging from as high as 33 or 35 million years ago to as low as 12 million years ago. That uncertainty doesn’t necessarily mean that mice and rats aren’t genetically related. But it does reflect much uncertainty about how to properly interpret the meaning of mouse-rat genetic similarities and differences.
More importantly, a careful analysis of mice and rat genomes shows that huge patterns of genomic similarities cannot always be explained by common ancestry. A few years ago here on Evolution News, evolutionary biologist Richard Sternberg discussed unexpectedly similar “lineage specific” SINE DNA in the mice and rat genomes which cannot be explained by common descent:
Let’s do a thought experiment. Consider the surfaces of two moons that were once part of the same planetary body 22 million years ago. Since their separation, both have been subjected to independent collisions with asteroids, meteorites, and other pieces of space debris. Question: Would you expect the scar patterns on both to be different or identical? (It may seem like a silly question, but bear with me.)
Replace now the word “moons” with the “mouse and rat genomes” and “asteroids and meteorites and other pieces of space debris” with SINEs, and you will see what I am asking. So I’ll rephrase my question. What should we expect regarding the linear distribution of independent SINE impacts along mouse and rat chromosomes?
A. Completely independent patterns — like meteorite impact sites on moons;
B. A few overlapping patterns, due to chance; or
C. Nearly identical patterns.
And the Mystery Signal Is…
This is a second figure from the article co-authored by Francis Collins. (From Fig. 9c of Ref. 1.) The scale on the x-axis is the same as that of the previous graph — it is the same 110,000,000 genetic letters of rat chromosome 10. The scale on the y-axis is different, with the red line in this figure corresponding to the distribution of rat-specific SINEs in the rat genome (i.e., ID sequences). The green line in this figure, however, corresponds to the pattern of B1s, B2s, and B4s in the mouse genome.
Was it what you expected from a degenerative process? Why?
At this point the theistic evolutionist might say — Silly Rick: Common descent explains this pattern!
Wrong, wrong, wrong.
Let me repeat — each graph denotes only lineage-specific mutational insertions.
The mutational signal from mouse B1s, B2s, and B4s is equivalent to the mutational signal of rat IDs. It almost looks as if, say, the rat graph was copied, slightly redrawn, labeled “mouse,” and then pasted above the previous line. (Of course, it wasn’t.) How strange that two independently-acting degenerative processes — affecting mostly “junk DNA” — would lead to the same higher-order pattern.
It’s a bizarre pattern. And this correlation occurs throughout both genomes.
Not a Secret, Folks: Collins et al. Discussed This “Unusual” Finding
The Rat Genome Consortium — and thus Francis Collins — apparently thought it worthy to devote a whole section to the phenomenon. Titled “Co-localization of SINEs in rat and mouse,” the section states:
Despite the different fates of SINE families, the number of SINEs inserted after speciation in each lineage is remarkably similar: ~300,000 copies…Figure 9c displays the lineage-specific SINE densities on rat chromosome 10 and in the mouse orthologous blocks, showing a stronger correlation than any other feature. The cause of the unusual distribution patterns of SINEs, accumulating in gene-rich regions where other interspersed repeats are scarce, is apparently a conserved feature, independent of the primary sequence of the SINE and effective over regions smaller than isochores. [Italics mine.]
The potential signal in these two genomes, then, should be obvious. If not, I will belabor the point:
- The strongest correlation between mouse and rat genomes is SINE linear patterning.
- Though these SINE families have no sequence similarities, their placements are conserved.
- And they are concentrated in protein-coding genes.
Am I suggesting that extraterrestrials were fiddling with rodent DNA? No. Am I implying that we are seeing the “language of God” in rodent-script? I haven’t the foggiest notion. What I am saying is that we know a lot about the genome that is being glossed over in the popular works that the theistic evolutionists write.
Sternberg’s point is exactly applicable to Joshua Swamidass’s post, which vaguely asserts a single general statistic of genetic similarity between mice and rats and assumes that those similarities must be explained by common ancestry. But as Sternberg says, these similar patterns cannot be explained by common ancestry because they are not related to similar sequences.
Rather, the similarities in the signal show similar densities of SINE elements across a huge stretch of over 100 million base pairs. Similar genomic structure, very different sequences. Because of the differences in sequence, these highly similar patterns in the mouse and rat genomes cannot be explained by common ancestry. This should make us cautious before we “readily” interpret genetic similarities between mice and rats as necessarily indicating common inheritance.
If not common ancestry, then what does explain these similar broad-scale genomic patterns in mice and rats? Sternberg continues to explore this question:
That Strange SINE Signal…Again
The almost one-to-one correspondence of mouse-specific and rat-specific SINE insertion events along homologous regions of the two genomes is almost as remarkable as the matching geographical distributions of the monoliths in the analogy of the two moons. Remember the graph (from Figure 9c of Ref. 1):
We have two genomes that went their separate ways 22 million years ago. We have two lineages that have been subjected to different historical events. Yet, when we compare the chromosome locations of mouse B1s/B2s/B4s with those of rat IDs, they look almost the same. Where the ID SINEs rise in density, so do the B1s/B2s/B4s SINEs; where the ID SINE levels decrease, so also do the B1s/B2s/B4s SINE levels. Independent mutational events have generated equivalent genomic patterns. How can we causally account for this striking pattern?
In the paper written by Francis Collins and his colleagues, under the heading “Co-localization of SINEs in rat and mouse,” we read:
The cause of the unusual distribution patterns of SINEs…is apparently a conserved feature, independent of the primary sequence of the SINE… [Italics mine.]
Let’s unpack this part of the sentence. We have:
1) A cause of some sort.
2) A cause that is conserved between the mouse and rat.
3) A cause that is independent of SINE primary DNA sequences.
That’s all very well and good, but the specific cause is never mentioned. Where, then, can we find it?
Sternberg doesn’t identify a cause, other than to say that whatever it is, it’s “outside the BioLogos box.” But we have already seen such a cause that we can identify: common design. Such a high level of genetic convergence is much better explained by common design than by common descent. And this is no mere “lawyerly” rhetoric. It’s a principled argument based upon our knowledge that intelligent agents can independently generate the same pattern in different systems that are not genetically related — perhaps including the similar SINE patterns between mice and rats.
(5) Swamidass tries to rule out the option of responding to him through “common design.”
Despite his protests against “lawyerly” arguments, Swamidass sounds rather lawyerly himself in his response to Cornelius Hunter. He tries to rule out contrary arguments before they can be made. He thus calls for a response from skeptics of human/ape common ancestry to explain why humans and apes are so genetically similar — but he places this restriction on what can be said in response:
All papers that simplistically conclude that “design” explains that humans and chimps are similar will be rejected.
This is like an attorney who tries to exclude contrary evidence from a courtroom trial.
Swamidass then qualifies his requirement and says you may argue for common design so long as you explain why humans and apes are more similar to one another than mice are to rats. This is basically a “no designer would do it that way” argument. In fact, if a designer wants to make some systems more similar than others, there’s no problem with that. In human-designed technology, we see all sorts of variations on themes all the time — iPhones are more similar to one another than they are to an Android, and Androids are more similar to one another than they are to iPhones. Differing degrees of similarity don’t refute design.
(6) Swamidass points to pseudogenes as evidence for common ancestry, even though many pseudogenes show evidence of function, including the vitellogenin pseudogene that Swamidass cites.
Swamidass repeatedly cites Dennis Venema’s arguments for common ancestry based upon pseudogenes. However, as we’ve discussed here in the past, quite a few pseudogenes have turned out to be functional, and we’re discovering more all the time. It’s only recently that we’ve had the technology to study the functions of pseudogenes, so we are just at the beginning of doing so. While it’s true that there’s a lot about pseudogenes we still don’t know, an RNA Biology paper observes, “The study of functional pseudogenes is just at the beginning.” And it predicts that “more and more functional pseudogenes will be discovered as novel biological technologies are developed in the future.” The paper concludes that functional pseudogenes are “widespread.” Indeed, when we carefully study pseudogenes, we often do find function. One paper in Annual Review of Genetics tellingly observed: “Pseudogenes that have been suitably investigated often exhibit functional roles.”
One of Swamidass’s central examples mirrors Dennis Venema’s argument that the vitellogenin pseudogene in humans demonstrates we’re related to egg-laying vertebrates like fish or reptiles. But a Darwin-doubting scientist was willing to dig deeper. Good genetic evidence now indicates that what Dennis Venema calls the “human vitellogenin pseudogene” is really part of a functional gene, as one technical paper by an ID-friendly creationist biologist has shown.
(7) Swamidass lauds Dennis Venema’s arguments from synteny, ignoring the strong genomic evidence that the 3D structure of the genome is important.
Synteny refers to large-scale similarities between genomes. Venema and Swamidass argue that large-scale 3D features of the genome carry no functional significance. This is simply incorrect. Many papers show that the 3D large-scale organization of the genome is vital for genomic function. As the revolution in epigenetics has taken hold, molecular biologists now know that the structure of chromosomes, and their 3D arrangement(s) within a cell, are important parts of genomic regulation.
Casey Luskin has recounted asking an ID-friendly scientist for papers showing that chromosomal structure and gene ordering can be functionally important. The response was impressive:
- Jachowicz et al., “Heterochromatin establishment at pericentromeres depends on nuclear position,” Genes & Development, 27: 2427-2432 (2013);
- Verdaasdonk et al., “Centromere Tethering Confines Chromosome Domains,” Molecular Cell, 52: 1-13 (December 26, 2013);
- Filion et al., “Systematic Protein Location Mapping Reveals Five Principal Chromatin Types in Drosophila Cells,” Cell, 143: 212-224 (October 15, 2010);
- Giacomo Cavalli, “From Linear Genes to Epigenetic Inheritance of Three-dimensional Epigenomes,” Journal of Molecular Biology (2011);
- Justin M. O’Sullivan, “Chromosome Organizaton in Simple and Complex Unicellular Organisms,” Current Issues in Molecular Biology, 13: 37-42 (2011);
- Dirar Homouz and Andrzej S. Kudlicki, “The 3D Organization of the Yeast Genome Correlates with Co-Expression and Reflects Functional Relations between Genes,” PLoS One, 8: e54699 (January, 2013);
- Stephen A. Hoang and Stefan Bekiranov, “The Network Architecture of the Saccharomyces cerevisiae Genome,” PLoS One, 8: e81972 (December, 2013).
(8) Dr. Swamidass lauds Venema’s arguments regarding chromosome 2 fusion, wrongly thinking that fusion in modern humans or Denisovans shows we’re related to apes.
As Casey Luskin has shown, the “fusion” evidence only shows that there was a fusion event along the human line; it does not show our lineage leads back to apes. We now know that the nearly modern-human Denisovans also had a fused chromosome 2. But for the same reasons, the fusion evidence doesn’t show that they were related to apes either. We might be just as unrelated to apes as the Denisovans were.
Postscript: Dr. Swamidass has now appended a Q&A to his original post. After some debate, he seems to accept “common design” as a possible explanation, writing that: “At a high level, just looking at the >98% similarity between humans and chimp coding regions, design is a plausible explanation.” He then quickly adds: “However, design does not explain why humans and chimps are 10 less different than mice are from rats.”
His comment makes no sense. Why is it so hard to conceive that humans and chimps are more similar than mice and rats? Humans and chimps are also more similar than jellyfish and bananas. So what? Different designs can have varying levels of similarity and difference — we see this all the time in technology and none of it bears upon (a) the merits of whether those systems were designed, or (b) whether those systems share some genetic naturalistic common ancestry.
Dr. Swamidass also attempts to respond to a less sophisticated version of this sort of argument:
Out of the gate, in what world is 6 million years a short time? That is an absurdly long time. Our subjective understanding of what is short and long, anyways, is beside the point. It is really just a matter of math. Let me show you the formula.
The basic pattern is that we multiply the number of generations between us and chimpanzees (by way of the common ancestor) times the number of mutations we see per generation along each leg. To be clear, we can directly measure the number of mutations we see per generation in human and chimps as a starting point, so we are not just pulling numbers out of thin air. Of course generation time is more complicated, but for most of human history was probably less than 20 years. One can get substantiated estimates for all of these terms from the literature, and compute the expected difference between genomes.
You can work this the other way around. Using your numbers (which are a bit off), we can see that there are 12M (2 times 6M) years of mutations between us and chimpanzees. We observe 30M mutations between us, so that means there are 2.5 mutations per year (30M divided by 12M). For generation times of about 20 years, we compute an upper bound of about 50 mutations per generation (2.5 times 20),7 which compares favorably (within an order of magnitude) to the experimentally determined generation mutation rate for humans (about 10 per generation). Our computed expectation is absurdly close to the experimentally observed mutation rate. For a back-of-the-envelope calculation, this is remarkable agreement between theory and experiment. Any discrepancies are easily explained by errors in the numbers, or shifts in these numbers of over 6M years of time. Better formulas and numbers only improve on these results. Why did God make our genomes so consistent with this math?
This is not exactly how population genetics works. Population genetics must explain how particular mutations get “fixed” into populations, not just how they appear within individual lineages. It takes a very short time for mutations to accumulate in individual lineages. It takes a very long time for mutations to become fixed into populations. This is why we have pointed out that if any two of those 30 million base pair differences were required to provide some advantageous trait that got fixed into our populations, then it could never arise in 6 million years. If you don’t believe us, ask anti-ID scientists Richard Durrett and Deena Schmidt, writing in the journal Genetics:
[I]n humans, a new transcription factor binding site can be created by a single mutation in an average of 60,000 years, but, as our new results show, a coordinated pair of mutations that first inactivates a binding site and then creates a new one is very unlikely to occur on a reasonable timescale.
To be precise, the last argument shows that it takes a long time to wait for two prespecified mutations with the indicated probabilities.
(R. Durrett and D. Schmidt. 2008. “Waiting for Two Mutations: With Applications to Regulatory Sequence Evolution and the Limits of Darwinian Evolution.” Genetics, 180: 1501-1509)
So if you just want to say that 30 million unspecified mutations can arise in two lineages of primates in 6 million years, then perhaps that’s correct. But that’s not the issue here: If only two of those 30 million mutations are required for some advantageous trait, then suddenly the math becomes a fatal barrier to Darwinian evolution.
Update: Six million years is not an “an absurdly long time” to fix two specified mutations into a typical hominid population. It would be more accurate to say that it’s an absurdly short time.
Image credit: © creativenature.nl — stock.adobe.com.