Evolution News and Views (ENV) provides original reporting and analysis about the debate over intelligent design and evolution, including breaking news about scientific research.

Evolution News and Views
Responses NEWS
 

Does Natural Selection Leave "Detectable Statistical Evidence in the Genome"?

In Chapter 11 of Darwin's Doubt, Stephen Meyer responds extensively to critics who have claimed that the evolution of new genes is well understood. He even cites comments made by Nick Matzke who claimed that "[c]ompetent scientists know how new genetic information arises." Oddly, however, in Matzke's own response to Stephen Meyer and Chapter 11 in Darwin's Doubt, Matzke says very little about this topic. He largely rehashes old debates, writing:

[I]n 2010, we finally got a response from Casey Luskin, which basically amounted to the juvenile assertion that invoking mutations (which are known, regularly observed natural processes, which leave easily detectable and obvious evidence in the genome) and natural selection (which is also a known, observed natural process, and which also often leaves detectable statistical evidence in the genome, although it is not always quite as easy to detect as mutations) amount to "waving a magic wand" and telling a vague "just-so story."
DebatingDD.jpegNick must be referring to one of the eight parts of my 12,000+ word article, "The NCSE, Judge Jones, and Citation Bluffs About the Origin of New Functional Genetic Information," a long, detailed response to claims (like one Matzke made long ago) that "[c]ompetent scientists know how new genetic information arises."

I'll deal with Nick Matzke's misrepresentations of my arguments below. But for now, let's assess his substantive claim that "natural selection ... often leaves detectable statistical evidence in the genome." Can we definitively infer the action of natural selection upon genes?

Well, how do evolutionary biologists look for genetic evidence of "positive selection"? Typically, they infer selection through statistical techniques that compare the ratios of nonsynonymous to synonymous mutations in genes.

Because of redundancy in the genetic code, not all DNA mutations change the amino acid sequence. (In fact, the redundancy in the genetic code is finely-tuned to minimize mutations that might change amino acid sequence.) Synonymous mutations change the nucleotide sequence of a gene's DNA, but they don't change amino acid sequence in the expressed protein. Nonsynonymous mutations change both the DNA sequence and the amino acid sequence. According to proponents of these statistical techniques, if nonsynonymous changes have been fixed in the gene at a higher rate than synonymous ones, this implies that the gene underwent natural selection to preserve mutations that change amino acid sequence. According to this thinking:

  • An excess of nonsynonymous mutations implies "positive selection" is preserving mutations that change amino acid sequence.1
  • An excess of synonymous mutations implies selection is at work to "weed out" mutations that change amino acid sequence. This is called "purifying selection."
  • If synonymous and nonsynonymous mutations are fixed at a proportional rate, this indicates no selection pressure, and the gene is undergoing "neutral" evolution.
These tests were thought to derive directly from the work of the influential evolutionary biologist Motoo Kimura, who developed the neutral theory of evolution in the 1960s and 1970s. As an article in Annual Review of Genomics and Human Genetics explains:
Kimura's immensely influential formulation of the neutral theory of molecular evolution, which came in 1968, was based primarily on an argument about the magnitude of the genetic load that would be imposed by positive selection if it were the only driving force in protein evolution. The availability of α- and β-hemoglobin sequences from a variety of primate and mammalian species allowed him to estimate a rate of amino acid substitution and to model this substitution process under two opposing hypotheses, selection and neutrality. Kimura argued that the load was too great under selection, whereas it was practically nonexistent under near-neutrality. In this remarkable note to Nature, Kimura ... not only formulated his neutral theory of molecular evolution, he also pointed to a future direction in empirical testing of selection and drift hypotheses ... He also deduced a major role for selection by showing that the rate of amino acid substitution in hemoglobin was far lower than that predicted from reasonable estimates of the nucleotide mutation rate. He argued from this result that the prevalent form of natural selection acting on proteins was selective constraint, the elimination of deleterious mutations, and not positive selection. So auspiciously began the modern field of testing for neutrality and selection from protein and DNA sequences.2
In using these methods, evolutionary biologists make many assumptions. Two key assumptions are as follows:
  • First, they assume that synonymous mutations are selectively neutral because they don't modify protein sequence.3
  • Second, they assume nonsynonymous mutations which change the amino acid sequence are preserved because of selection.
Today, some 45 years after Kimura, there are good reasons to doubt many of the assumptions underlying these methods. Indeed, in his 2013 Oxford University Press book Mutation-Driven Evolution, Masatoshi Nei of Pennsylvania State University explains:
In recent years, however, a substantial number of papers claiming detection of positive selection at the protein level have been published. These papers are based upon statistical analyses of genomic data under various assumptions, which are not necessarily satisfied in the real world. It is therefore necessary to examine the validity of the assumptions and the statistical methods used.4

The recent epigenetic revolution in biology has shown that mRNA transcripts of genes commonly undergo important biochemical interactions before being translated into proteins. In particular, as the notion of junk DNA has collapsed,5 biologists have learned that non-protein-coding DNA produces many RNA molecules which interact with mRNA transcripts. Synonymous mutations which change the sequence of an mRNA transcript can influence these pretranslational reactions, strongly affecting gene expression. Synonymous mutations can also change the speed of translation, which can have significant phenotypic effects. As a 2010 paper in Science stated, "the discovery [reported in a research paper] that synonymous codon changes can so profoundly change the role of a protein adds a new level of complexity to how we interpret the genetic code."6 Contrary to the evolutionary literature, molecular biologists today know synonymous mutations aren't necessarily "more or less neutral"7 -- they can have selectable effects, invalidating the first assumption.

There are also good reasons to doubt the second assumption -- that nonsynonymous mutations are always subject to selection. While many amino acids are crucial to a protein's function, others are not. Some changes in amino acid sequence have no noticeable effect upon protein function. But papers that use these statistical techniques assume that all amino acid changes produce selectable changes in function, and rarely use experiments to independently establish that the amino acids changes claimed to result from "selection" actually are important for protein function.

Nonetheless, using methods that compare nonsynonymous to synonymous mutations, many papers purport to find evidence of natural selection. They rarely, if ever, provide a demonstration of the functional significance of any nonsynonymous changes. In some cases, evolutionary biologists invoke "natural selection" when they don't even know what specific function was being selected.8 Austin Hughes, an evolutionary biologist at the University of South Carolina, points out how uninformative such findings are:

[I]t is worth asking what can be learned from the conclusion that positive selection has acted -- even if that conclusion is true -- in cases where we have no knowledge of the biological basis of that selection. 9
In 2007, Hughes published a paper in the journal Heredity titled, "Looking for Darwin in all the wrong places: the misguided quest for positive selection at the nucleotide sequence level,"10 observing that the literature faces a current "saturation" of "numerous recent studies where it is claimed that positive selection acts on a given gene or certain codons within a given gene, but no biological hypothesis for the nature of this selection is provided."11 He recommends that researchers experimentally confirm whether specific amino acid changes are adaptive because "empirical testing provides far more convincing evidence of a trait's adaptive value than can any statistical argument."12 Unfortunately, few researchers have attempted to do this. And among the few that have, they often find these statistical methods fail.

One paper used statistical techniques to infer that the genes that confer trichromate vision on certain primates had undergone natural selection. But experimental work found that trait wasn't functionally adaptive as expected.13 Another study experimentally identified amino acid sites in genes for rhodopsin pigment-proteins critical for absorbing light in vertebrate retinas. The investigators expected that statistical techniques would confirm these amino acid sites were under selection -- but they found that "none of these predicted ['positive selection'] sites coincide with those detected by mutagenesis experiments."14 Moreover, the paper found that the sites that statistical methods predicted were under positive selection "do not seem to cause" any important changes in light absorption. The paper concluded that "statistical tests of positive selection can be misleading without experimental support."15

Such experimental work shows that these statistical methods for inferring natural selection lead not only to false positives, but also to false negatives. Sometimes they wrongly predict that portions of a gene underwent "positive selection," whereas at others they wrongly suggest parts of a gene aren't under selection.

"Hundreds of Natural-Selection Studies Could Be Wrong"

Because of such failures, these widely used statistical techniques have come under much criticism in the literature. In 2009, biologists at Pennsylvania State University and the National Institute of Genetics in Japan issued a statement titled "Hundreds of Natural-Selection Studies Could Be Wrong," contending that "several statistical methods commonly used by biologists to detect natural selection at the molecular level tend to produce incorrect results."16 Masatoshi Nei, one of the lead scientists who issued the statement, co-wrote in a technical paper that these methods are "not useful for identifying adaptive [amino acid] sites," and warned researchers "not to be overenthusiastic about statistical signatures of positive selection without biological confirmation."17 Nei expands on this in Mutation-Driven Evolution:

During the last 10 years, a large number of biologists have used these methods and reported detection of positive selection in many different genes from various organisms including humans, chimpanzees, and macaques. ... Recent theoretical and empirical studies have shown that these Bayesian methods are quite unreliable and generate a high proportion of false positives.18
Nei elaborates on why these methods not only generate false positives, but also false negatives:
[T]he likelihood ratio test (LRT) used in these methods is unreliable because unrealistic mathematical models are used. The number of nucleotide substitutions at a codon site is often too low for LRT to be used. In these cases, codon sites may be falsely identified as positively selected sites because of a high ω value generated by chance. ...

More importantly, these methods are dependent on the assumption that positive selection occurs at a codon site where the number of nonsynonymous substitutions exceeds the number of synonymous substitutions. In practice, this is not the correct assumption needed. Adaptive evolution often occurs by a single nonsynonymous (or amino acid) substitution without repetition. For example, red and green color vision in vertebrates are caused by single amino acid substitutions at positions 277 and 285 of a vision pigment protein. These amino acids remain the same for all vertebrates, though synonymous substitution may occur repeatedly. In this case, the ω value is excepted to be small.19

(Note: ω is the ratio of nonsynonymous mutations to synonymous mutations.)

In a 2010 paper in Annual Reviews of Genomics and Human Genetics, Nei and others argue that under the neutral theory of evolution, nonsynonymous mutations often become fixed due to genetic drift, and thus these statistical methods may be providing a mirage of selection that doesn't exist:
Because rEHH [another name for the statistical value used to infer selection] (or any other statistic) is affected by random events such as mutation, recombination, gene duplication, and genetic drift, as well as the amount and quality of SNP data, the statistic used is subject to substantial errors. Therefore, a high rEHH value may not necessarily imply selection. ... It is important to have some empirical evidence of selection for each putatively selected genomic region. Until this evidence is presented, the results of these studies remain mere speculations.20
Of course, such experimental, empirical work to show how specific amino acid changes provided some adaptive advantage is rarely, if ever, done. This means these "speculations" have not established natural selection at work in the genome.

In that same paper, Nei and his co-authors make a strong conclusion that even though these methods are widely used, they are flawed:

[M]any recently published papers claim the detection of positive Darwinian selection via the use of new statistical methods. Examination of these methods has shown that their theoretical bases are not well established and often result in high rates of false-positive and false-negative results. ... The finding of positive selection in MHC genes stimulated a number of theoreticians to develop statistical methods for identifying positive selection during the past two decades. These statistical methods have been used by many biologists, and there are now a large number of papers reporting positive selection. In our view many of these methods do not have solid statistical and biological bases. ... A critical review of these statistical methods has shown that their theoretical foundation is not well established and they often give false-positive and false-negative results.21
They conclude "detectability of selection with these methods is poor."22 If they are correct, then many of the papers Matzke cites have not in fact provided solid evidence of natural selection in the genome.

Austin Hughes and Masatoshi Nei are not obscure critics. Their joint papers in Nature in 1988 and 1989 was one of the first to actively apply statistical methods for detecting natural selection in genes.23 In essence, two of the fathers of this field have now become two of its chief critics.

Taking a Larger Perspective

At base, what these statistical studies are determining is the proportion, or ratio, of differences between two genes that yield changes in amino acids vs. those that don't. What does this mean? The ratio doesn't demonstrate which if any of the nonsynonymous changes might produce a functional change. In fact, unless researchers provide independent evidence that the changes in question actually provide a functional advantage, the sequence changes may mean nothing at all. Yet papers using these methods virtually never identify the adaptive advantage of specific amino acid changes, nor do they calculate the likelihood of the claimed adaptive changes occurring within any reasonable evolutionary timescale. Though commonly used, these methods are highly flawed, and do not provide direct, unambiguous evidence for Darwinian evolution of genes. According to a growing number of prominent critics, these statistical methods may not be telling us much of anything at all.

Matzke Misrepresents Our Arguments

Nick Matzke has a long history of badly misrepresenting our arguments, and refusing to engage what we actually say. Indeed, Matzke's entire discussion of Chapter 11 of Darwin's Doubt, "Assume a Gene," barely deals with Meyer's arguments, or mine. Instead, he refers to an article I wrote in 2010 about the origin of new genes, and dismisses it as "the juvenile assertion that invoking mutations ... and natural selection ... amount to 'waving a magic wand' and telling a vague 'just-so story.'"

Let's look at my article that Matzke seems to be referring to and ask if he has accurately characterized its arguments. In my article I go through a detailed analysis of many of the papers cited by Matzke, and and show that the evolutionary explanations he cites never address the likelihood that the evolutionary stories they tell actually could occur via blind evolutionary processes on a reasonable timescale. Here's what I argued:

In not a single case did the above papers cited by Long et al. [the review paper Matzke had cited] actually explain how new functional information arose. In no case was there an analysis of how natural selection could have favored mutational changes that were shown to be likely along each step of an alleged evolutionary pathway; never was any detailed step-by-step mutational pathway even given. At best, these studies offered vague and ad hoc appeals to duplication, rearrangement, and natural selection -- often in a sudden, extreme, and abrupt manner -- to form the gene in question. In many cases, natural selection was invoked to allegedly account for changes in the gene, even though the investigators didn't even know the function of the gene and thereby could not identify the advantage provided by the gene's function. In no case were calculations performed to assess whether sufficient probabilistic resources existed to produce the asserted mutational events on a reasonable timescale. In some cases, the original genetic material for the genes was unknown, or the studies asserted spontaneous "de novo" origin of genes from previously non-coding DNA. While they readily admitted that "de novo" gene emergence is rare, no attempt was made to assess whether such an unguided mechanism is even remotely plausible on mathematical probabilistic grounds. These papers play the Gene Evolution Game, but never ask the right questions to explain how neo-Darwinian mechanisms create new genetic information. ... To reiterate, in no cases were the odds of these unlikely events taking place actually calculated. Incredibly, natural selection was repeatedly invoked in instances where the investigators did not know the function of the gene being studied and thus could not possibly have identified any known functional advantages gained through the mutations being invoked. In the case where multiple mutational steps were involved, no tests were done of the functional viability of the alleged intermediate stages. These papers offer vague stories but not viable, plausibly demonstrated explanations for the origin of new genetic information.
Of course this is nothing like Matzke's characterization of my argument.

Likewise, Stephen Meyer argues in Darwin's Doubt:

[T]hese scenarios not only assume unexplained preexisting sources of biological information, they do so without explaining or even attempting to explain how any of the mechanisms they envision could have solved the combinatorial search problem described in Chapters 9 and 10. (pp. 217-218)
Matzke doesn't respond to these arguments. Instead, he points readers to his explanation of the evolution of a gene called Sdic, as if this settles the debate. Actually, Matzke and I had a back-and-forth about the origin of Sdic in 2011, the year after he posted his discussion of the gene. In a response to Matzke, "Leading Darwin Defender Admits Darwinism's Most 'Detailed Explanation' of a Gene Doesn't Even Tell What Function's Being Selected," I explained why Matzke's explanation failed:
You [Nick Matzke] just admitted that the most "detailed explanation" for the evolution of a gene represents a case where:
  • they don't even know the precise function of the gene,

  • and thus don't know exactly what function was being selected,

  • and thus don't know if there are steps that require multiple mutations to produce an advantage,

  • and thus haven't even begun to show that the gene can evolve in a step-by-step fashion,

  • and thus don't know that there are sufficient probabilistic resources to produce the gene by gene duplication + mutation + selection.
In effect, you have just admitted that Darwinian explanations for the origin of genes are incredibly detail-poor.
There isn't much more to say on this matter because in his response to Darwin's Doubt, Matzke devotes so little attention to rebutting Meyer's actual arguments -- and when Matzke does respond, he badly misrepresents Meyer's arguments. A final prime example of Matzke's misrepresentations comes when he states that Meyer's argument "is no better that the Omphalous argument that God made the Earth appear to be billions of years old, but actually it's only 6,000 years Old."

If we try to take Matzke's objection seriously, he seems to be insinuating that (a) if these genes look like they have a history, then their history must be unguided, and (b) Meyer is denying that history. But Meyer doesn't deny the evidence of history when it's actually there. And most of the evidence of "history" Matzke does cite ultimately amounts to little more than homology between parts of one gene with parts of other genes, and as Michael Behe has reminds us, this is insufficient evidence to establish a Darwinian pathway:

Although useful for determining lines of descent ... comparing sequences cannot show how a complex biochemical system achieved its function -- the question that most concerns us in this book. By way of analogy, the instruction manuals for two different models of computer put out by the same company might have many identical words, sentences, and even paragraphs, suggesting a common ancestry (perhaps the same author wrote both manuals), but comparing the sequences of letters in the instruction manuals will never tell us if a computer can be produced step-by-step starting from a typewriter. ... Like the sequence analysts, I believe the evidence strongly supports common descent. But the root question remains unanswered: What has caused complex systems to form?24

[M]odern Darwinists point to evidence of common descent and erroneously assume it to be evidence of the power of random mutation.25

It's an elementary point. Granting that these genes may have a history, evidence of history is not evidence of unguided mechanisms at work. Matzke has not established that unguided mechanisms like selection and mutation are capable of accomplishing the evolutionary feats he attributes to them, and in fact leading scientists in relevant fields now doubt that the statistical methods he cites can establish the action of selection upon a gene.

These statistical methods that Matzke cites not only don't demonstrate that selection has occurred, but they also do not demonstrate that stepwise evolutionary pathways are available. For that, experiments are needed. Matzke can call us "juvenile" all he wants, but the fact is, beyond a lot of hand-waving and name-calling, he has not provided an adequate account of how new genes might arise.

By the way, if readers want to see a revealing back and forth between Matzke and me on the topic of the evolution of genes, they should go back and read the comments at Richard Lenski's Long-Term Evolution Experiments with E. coli and the Origin of New Biological Information. This exchange is a good example which shows how Matzke often misrepresents our arguments about the origin of new biological information, and fails to respond to our actual positions -- a repeating theme in his criticisms of intelligent design.

References:

[1.] Austin L. Hughes, Masatoshi Nei, "Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection," Nature, Vol. 335: 167-170 (September 8, 1988). See also Austin L. Hughes and Masatoshi Nei, "Nucleotide substitution at major histocompatibility complex class II loci: Evidence for overdominant selection," Proceedings of the National Academy of Sciences USA, Vol. 86: 958-962 (February, 1989); Tim Massingham and Nick Goldman, "Detecting Amino Acid Sites Under Positive Selection and Purifying Selection," Genetics, Vol. 169: 1753-1762 (March 2005) ("An excess of nonsynonymous over synonymous substitution at individual amino acid sites is an important indicator that positive selection has affected the evolution of a protein between the extant sequences under study and their most recent common ancestor."); Yoshiyuki Suzuki and Takashi Gojobori, "A Method for Detecting Positive Selection at Single Amino Acid Sites," Mol. Biol. Evol., Vol. 16(10): 1315-1328 (1999) ("Selective forces operating at the amino acid sequence level have been detected mainly by comparing the number of nonsynonymous substitutions per site with that of synonymous substitutions per site. Generally speaking, the excess number of synonymous substitutions was considered to be the result of negative selection, whereas that of nonsynonymous substitutions was attributed to positive selection") (internal citations removed); Yoshiyuki Suzuki, Takashi Gojobori, Masatoshi Nei, "ADAPTSITE: detecting natural selection at single amino acid sites," Bioinformatics Applications, Vol. 17: 660-661 (2001) ("Natural selection may be detected by comparing the rate of nonsynonymous substitutions per nonsynonymous site per unit time (rN) with that of synonymous substitutions per synonymous site per unit time (rS). The observation that rN > rS suggests positive selection, whereas rN < rS suggests negative selection.") (internal citations removed).

[2.] Martin Kreitman, "Methods to Detect Selection in Populations with Applications to the Human," Annual Review of Genomics and Human Genetics, Vol. 1:539-559 (2000).

[3.] Austin L. Hughes, "Looking for Darwin in all the wrong places: the misguided quest for positive selection at the nucleotide sequence level," Heredity, Vol. 99: 364-373 (2007); Austin L. Hughes, Masatoshi Nei, "Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection," Nature, Vol. 335: 167-170 (September 8, 1988); Jianzhi Zhang, "Positive Darwinian Selection in Gene Evolution," in Darwin's Heritage Today, pp. 305-325.

[4.] Masatoshi Nei, Mutation-Driven Evolution (Oxford University Press, 2013), p. 81.

[5.] See The Myth of Junk DNA by Jonathan Wells (Discovery Institute Press, 2011).

[6.] Ivana Weygand-Durasevic and Michael Ibba, "New Roles for Codon Usage," Science, Vol. 329:1473-1474 (September 17, 2010) reporting on Fangliang Zhang, Sougata Saha, Svetlana A. Shabalina, Anna Kashina, "Differential Arginylation of Actin Isoforms Is Regulated by Coding Sequence-Dependent Degradation," Science, Vol. 329:1534-1537 (September 17, 2010). See also Gene-Wei Li, Eugene Oh, and Jonathan S. Weissman, "The anti-Shine-Dalgarno sequence drives translational pausing and codon choice in bacteria," Nature (2012) doi:10.1038/nature10965 (published online March 28, 2012); Gina Cannarozzi, Nicol N. Schraudolph, Mahamadou Faty, Peter von Rohr, Markus T. Friberg, Alexander C. Roth, Pedro Gonnet, Gaston Gonnet, and Yves Barral, "A Role for Codon Order in Translation Dynamics," Cell, Vol. 141: 344-354 (April 16, 2010); Tamir Tuller, Asaf Carmi, Kalin Vestsigian, Sivan Navon, Yuval Dorfan, John Zaborske, Tao Pan, Orna Dahan, Itay Furman, and Yitzhak Pilpel, "An Evolutionarily Conserved Mechanism for Controlling the Efficiency of Protein Translation," Cell, Vol. 141: 344-354 (April 16, 2010).

[7.] Austin L. Hughes, Masatoshi Nei, "Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection," Nature, Vol. 335: 167-170 (September 8, 1988).

[8.] See for example Manyuan Long and Charles H. Langley, "Natural selection and the origin of jingwei, a chimeric processed functional gene in Drosophila," Science, Vol. 260: 91-95 (April 2, 1993); Dmitry I. Nurminsky, Maria V. Nurminskaya, Daniel De Aguiar, and Daniel L. Hartl, "Selective sweep of a newly evolved sperm-specic gene in Drosophila," Nature, Vol. 396: 572-575 (December 10, 1998); David J. Begun, "Origin and Evolution of a New Gene Descended From alcohol dehydrogenase in Drosophila," Genetics, Vol. 145: 375-382 (February, 1997); Jianzhi Zhang, David M. Webb and Ondrej Podlaha, "Accelerated Protein Evolution and Origins of Human-Specific Features: FOXP2 as an Example," Genetics, Vol. 162: 1825-1835 (December, 2002); Esther Betran and Manyuan Long, "Dntf-2r, a Young Drosophila Retroposed Gene With Specific Male Expression Under Positive Darwinian Selection," Genetics, Vol. 164: 977-988 (July, 2003).

[9.] Austin L. Hughes, "Looking for Darwin in all the wrong places: the misguided quest for positive selection at the nucleotide sequence level," Heredity, Vol. 99: 364-373 (2007).

[10.] Ibid.

[11.] Ibid. (emphasis added).

[12.] Ibid.

[13.] Chihiro Hiramatsu, Amanda D. Melin, Filippo Aureli, Colleen M. Schaffner, Misha Vorobyev, Yoshifumi Matsumoto, Shoji Kawamura, "Importance of Achromatic Contrast in Short-Range Fruit Foraging of Primates," PloS One, Vol. 3: e3356 (October, 2008).

[14.] Shozo Yokoyama, Takashi Tada, Huan Zhang, and Lyle Britt, "Elucidation of phenotypic adaptations: Molecular analyses of dim-light vision proteins in vertebrates," Proceedings of the National Academy of Sciences USA, Vol. 105: 13480-13584 (September 9, 2008).

[15.] Shozo Yokoyama, Takashi Tada, Huan Zhang, and Lyle Britt, "Elucidation of phenotypic adaptations: Molecular analyses of dim-light vision proteins in vertebrates," Proceedings of the National Academy of Sciences USA, Vol. 105: 13480-13584 (September 9, 2008).

[16.] "Hundreds of Natural-Selection Studies Could be Wrong, Study Demonstrates," Phys.org, at http://phys.org/news157648673.html (March 30, 2009).

[17.] Masafumi Nozawa, Yoshiyuki Suzuki, and Masatoshi Nei, "Reliabilities of identifying positive selection by the branch-site and the site-prediction methods," Proceedings of the National Academy of Sciences USA, Vol. 106: 6700-6705 (April 21, 2009) (emphasis in original) (internal citations removed).

[18.] Masatoshi Nei, Mutation-Driven Evolution (Oxford University Press, 2013), p. 81.

[19.] Ibid., pp. 81-82.

[20.] Masatoshi Nei, Yoshiyuki Suzuki, and Masafumi Nozawa, "The Neutral Theory of Molecular Evolution in the Genomic Era," Annual Reviews of Genomics and Human Genetics, Vol. 11: 265-289 (2010) (emphasis added).

[21.] Ibid. (emphases added).

[22.] Ibid. (emphasis added.

[23.] Austin L. Hughes and Masatoshi Nei, "Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection," Nature, Vol. 335: 167-170 (September 8, 1988). See also Austin L. Hughes and Masatoshi Nei, "Nucleotide substitution at major histocompatibility complex class II loci: Evidence for overdominant selection," Proceedings of the National Academy of Sciences USA, Vol. 86: 958-962 (February, 1989).

[24.] Michael J. Behe, Darwin's Black Box: The Biochemical Challenge to Evolution, pp. 175-176 (Free Press, 1996).

[25.] Michael J. Behe, The Edge of Evolution: The Search for the Limits of Darwinism, p. 95 (Free Press, 2007).