More Recent Comments

Thursday, October 03, 2013

Stephen Meyer Says that "Homology" Is a Problem in Molecular Evolution

Stephen Meyer argues (in Darwin's Doubt) that the Cambrian explosion cannot be explained by evolution but it can be explained by Intelligent Design Creationism.

His main thesis is that all the animals appeared suddenly in the Cambrian and there's no evidence that they arose from ancestors living earlier in the Precambrian. Unfortunately for him, there IS plenty of evidence in the form of molecular evolution. By comparing genes and proteins we can show that all the animal groups are related to one another and that their common ancestors are spread out over a considerable period of time as shown in the phylogenetic tree below from a paper by Dunn et al. (2008).

This evidence is a serious problem for Meyer so he has to deal with it in his book. He tries to discredit the entire field of molecular evolution by challenging the basic assumptions [Stephen Meyer Says That Constant Mutation Rates Are a "Questionable Assumption"], by setting up a strawman [Stephen Meyer Says Molecular Data Must Be Wrong Because Different Genes Evolve at Different Rates], and by pointing out that molecular dating is not precise [Stephen Meyer Says Molecular Evidence Must Be Wrong Because Scientists Disagree About the Exact Dates]. His most ridiculous argument1 against molecular evolution is that the results must be wrong because there are no transitional fossils from before the Cambrian Explosion! [The Cambrian Conundrum: Stephen Meyer Says (Lack of) Fossils Trumps Genes]

None of those arguments stand up to close scrutiny but, as I warned you last week, there are actually five arguments against the validity of molecular evolution [Darwin's Doubt: The Genes Tell the Story?].

Are you ready for the final argument showing that molecular evidence must be discounted?

It's because all those phylogenetic trees are based on the assumption than the genes being compared are homologous. But "homologous" means that the genes share a common ancestor and that's exactly what the phylogenetic trees are supposed to show. Therefore, the entire enterprise is one big circular argument and none of the molecular evidence is valid. (In fact, the entire field of molecular evolution is based on a circular argument according to Stephen Meyer.)

Here's what he says on page 110.
The assumption (of universal common descent) raises the possibility that the ancestral entities represented by divergence points in these studies are artifacts of the assumptions by which molecular data are analyzed. Indeed, the computer programs that are used to compare molecular sequences have been written to produce trees showing common ancestors and branching relationships regardless of the extent to which the genes analyzed may or may not differ. Phylogenetic studies compare two or more gene sequences and then use degrees of difference to determine divergence points and nodes on a phylogenetic tree. Inherent in that procedure is the assumption that the nodes and divergence points existed in the past.
According to Meyer, this flaw calls into question all trees constructed from genes and proteins but he's particularly concerned about those that challenge his view that God must have made the animals during the Cambrian explosion.
Thus, the deep divergence studies [molecular evidence] do not, in any rigorous sense, establish any Precambrian ancestral forms. Did a single, original metazoan or bilaterian ancestor of the Cambrian animals actually exist? The Precambrian-Cambrian fossil record taken on its face certainly doesn't document such an entity. But neither do deep-divergence studies. Instead, these studies assume the existence of such ancestors, and then merely attempt, given that assumption, to determine how long ago such ancestors might have lived. One could argue that the conflicting divergence points do at least show that some common ancestor existed in the Precambrian, since, despite their conflicting results, all divergence studies indicate at least that. But, again, to invoke molecular studies that assume the existence of a common ancestor as evidence for such an entity only begs the question. Certainly it provides no reason for using molecular evidence to trump fossil evidence. Perhaps the Precambrian rocks do not record ancestors for the Cambrian animals because none existed. To foreclose that possibility, and to resolve the mystery of the missing Precambrian ancestral fossils, evolutionary biologists cannot use studies that assume the existence of the very entity their studies are thought to establish.
In order to see the flaw in Meyer's argument you have to imagine that you are a scientist back in the early 1960s. You have sequenced a bunch of cytochrome c proteins from a variety of difference species. You notice that they are very similar in sequence and you wonder whether these similarities arise because they evolved from a common ancestors. In other words, can you use sequence similarity to demonstrate homology?

You line up and compare the similar sequences. The first thing you notice is that some sequences are almost identical whereas others differ at many positions. For example, the sequences of the human and chimpanzee proteins are practically the same but they are both very different form the yeast sequence or the bacterial sequences of the same protein. You wonder whether these differences can be plotted as a tree of similar sequences not knowing in the beginning whether it will work or not. Maybe the relationships are not treelike, in which case they may not be due to descent from a common ancestors.

Remember that fifty years ago it wasn't clear that evolution by natural selection would yield a tree of sequence similarities because different lineages will adapt to different environments at different rates. It was also remotely possible that the cytochrome c proteins in different species arose by convergence and showed no trace of common ancestry.

But the comparison turned out to be treelike, as shown by Emanual Margoliash in 1963. The tree2 resembled the evolutionary relationships deduced from the fossil record and from morphological comparisons [see The Modern Molecular Clock]. Thus, amino acid sequence similarity is a reliable indication of homology.

This conclusion is so well established that we often ignore its significance and we often forget to emphasize the difference between "similarity" and "homology." Phylogenetic trees are constructed from sequences that are similar and the fact that the relationship turns out to be treelike (it doesn't have to) means that the genes/proteins are homologous. Margoliash (1969) (and others) understood very well the difference between similarity and homology and that phylogenetic trees were constructed on evidene of similarity to reach a conclusion of homology [see Margoliash on "Homology" (1969)].

Meyer doesn't understand this. He exploits the sloppy use of "homology" by many modern molecular biologists to build a case for invalidating all molecular phylogenies that have ever been published. It's ridiculous to use an argument over semantics to rule out solid evidence of common descent.

But, that's exactly what Intelligent Design Creationists have to do because they don't have any scientific evidence to support their claims. Jonathan Wells tried the same trick back in 2000 when he wrote Icons of Evolution. Lots of people pointed out his error back then. It shows, once again, that IDiots don't learn from their mistakes.

1. It's a close call.

2. This tree is from Fitch and Margoloash (1967).

Dunn et al. (2005) Broad phylogenomic sampling improves resolution of the animal tree of life. Nature 452:745-749 [doi: 10.1038/nature06614]

Fitch, W.M. and Margoliash, E. (1967) Construction of phylogenetic trees. Science 155:279–284.

Margoliash, E. (1963) Primary structure and evolution of cytochrome c. Proc. Natl. Acad. Sci. USA 50:672-679.

Margoliash, E. (1969) Homology: A Definition. Science 163:127


Anonymous said...

So the fact that trees based on difference characters match is very compelling evidence for common descent...but I dont have a good feeling for why trees should not be expected from phylo analysis. It seems to me that if you start with cytochrome C and make random mutations, or mutations that an ID tinkerer might do, you'd still get a tree when you plugged it into a phylo analysis program. Or would you? I"m not sure. I'd like to see examples of non-trees. If trees are unlikely then getting a tree for any single sequence is evidence for common descent even before comparing it to other trees.
Meyer says that getting trees that show common ancestors is an illusion that comes from creating trees that contain the assumption of CD but this is nonsense as you show above. It seems to me this is a variation on a common theme I've seen in many ID/Creationist arguments. For example YECs will say that ones 'worldview' determines how one interprets the evidence for the age of the earth, and geologists only come to the conclusion that its billions of years old because of a prior old-earth/non-biblical bias.

Faizal Ali said...

It'd be (mildly) interesting to know whether Meyer really is this stupid, or if he just realizes that most of the readers of his book, at least those who are sympathetic to creationism, are too stupid to realize this error. But in the end that probably matters little.

Our local village IDiot Jonathan McLatchie has recently written an article that is pertinent here:

You'll notice that, in the penultimate paragraph, he contradicts Meyer by stating that homologous sequence similarities between lineages can be used to demonstrate common ancestry. Moreover, he actually states that such similarities are examples of "Complex Specified Information (CSI)". Therefore, in a single blow, he not only refutes Meyer's argument, but also contradicts the bedrock ID creationist claim that CSI can only be produced by intelligence.

These creationists really need to compare notes before mouthing off in public

Mikkel Rumraket Rasmussen said...

Yes well, one could easily imagine that the distribution of mutations in different species in some particular similar gene, like cytochrome C, did not produce a tree. Remember, the idea is that the more distantly related, the more differences. However, if there was no distance relationship, you could just as well expect that the differences between the same gene in different species would be more or less uniform. Such that, to pick an example, the number of differencse between the human-chimp gene was 20 mutations, the differences between the human-shrimp gene was also 20 mutations, the differences between the human-squid gene was also 20 mutations and so on and so forth. It would be even worse if you discovered this was true for any species you pick(I just picked humans as the source to which you compare sequences, it could be any arbitrary species) that there was no distance relationships that fall naturally into trees.

But the case against design and for common ancestry is even stronger still, because there are SOOOO many genes that fall into the same trees they massively outweigh the few genes that fall outside the consensus tree.

And this is before we compare the genetic trees to the trees we construct using comparative morphology and discover they agree too.

This could all be different, the morphologies and genes could all be a scrambled mess with no discernable pattern or relationships.

It gets even worse for design and even stronger for common ancestry still. Creationists like to claim that horizontal gene transfer is ad-hoc rationalization and that horizontal gene tranfer refutes evolutionary trees.

Well, where does horizontal gene transfer happen most frequently? Between single-celled organisms like bacteria and archaea. What does phylogenetic analysis reveal there? That's right, there's still a significantly strongly supported main tree-like pattern.

See: Seeing the Tree of Life behind the phylogenetic forest

"We set out to address the above question as objectively as possible, first of all dispensing with any pre-selected standard of tree-like evolution. The analyzed FOL consisted of 6,901 maximum likelihoodphylogenetic trees that were built for clusters of orthologous genes from a representative set of 100 diverse bacterial and archaeal genomes [1]. The complete matrix of topological distances between these trees was analyzed using the Inconsistency Score, a measure that we defined specifically for this purpose that reflects the average topological (in)consistency of a given tree with the rest of the trees in the FOL (for the details of the methods employed in this analysis, see [21]). Although the FOL includes very few trees with exactly identical topologies, we found that the topologies of the trees were far more congruent than expected by chance. The 102 Nearly Universal Trees (NUTs; that is, the trees for genes that are represented in all or nearly all archaea and bacteria), which include primarily genes for key protein components of the translation and transcription systems, showed particularly high topological similarity to the other trees in the FOL. Although the topologies of the NUTs are not identical, apparently reflecting multiple HGT events, these transfers appeared to be distributed randomly. In other words, there seem to be no prominent ‘highways’ of HGT that would preferentially connect particular groups of archaea and bacteria. Thus, although the NUTs cannot represent the FOL completely, they appear to reflect a significant central trend, an attractor in the tree space that could be equated with the STOL (Figure 1)."

This is game over for any other explanation than common descent. Where the relationships should be the hardest to work out, for single-celled life that evolves extremely fast and have much higher rates of HGT, we can still resolve a tree to an overwhelming degree.

John Harshman said...

Too bad Meyer didn't read Theobald, D. L. 2010. A formal test of the theory of universal common ancestry. Nature 465:219-222. Or he could just have noticed any of the many tests of hierarchical structure for phylogenetic data that are commonly a part of the papers he doesn't read, e.g. bootstrap, likelihood ratios, Bayesian node posteriors, etc. He could have saved a chapter. Mind you, if he had actually read the scientific literature with any other goal than quote mining, he could have saved the whole book.

un said...

But the test proposed by Theobald has been criticized by other researchers in the field, and I know that Theobald responded to some of these criticisms. But I don't think that it's well agreed upon. Do you think that the criticisms (such as this and this) are invalid or miss the point?

John Harshman said...

Yes, I do. What about you?

At any rate, Theobald's is only the most basal and explicit of such tests. I'd say that any support for a particular tree above that expected from random data is also a test of common descent for the included taxa.

Bill said...

Several misconceptions here.

First, Meyer isn't stupid. On the contrary he has studied the Cambrian stuff long enough to craftily write about his thesis and simultaneously misrepresent, distort and, most importantly, omit research and evidence that refutes his assertions. There are enough weasel-words in Doubt to feed all the weasels in the world (including lawyers) for a century. Meyer rarely states a fact, rather he implies, implicates and infers. Note that IDiot Meyer never says "x is due to intelligent design," but he says "x infers intelligence."

Second, creationists aren't interesting in learning anything. So, the first thing Buffy should teach a young creationist killer is this: never try to teach a creationist anything. They just don't care. They have their book. There's nothing more to learn. Case closed.

Third, while creationists aren't interested in learning science, they are very keen to learn a.bout politics. The evolution of Biblical creationism to scientific creationism to intelligent design creationism, and "teach the controversy" to "academic freedom" to "viewpoint discrimination" to "anti-bullying," even, demonstrates, ironically, an evolution in creationist political thought.

Finally, the root of intelligent design creationism is mostly money, although I would accept "power," too. Meyer gets paid by the Tute, not from book sales. Face it, Doubt is not a bestseller in spite of the Tute's hype. No, Meyer gets paid by people who contribute to whatever it is the DI does (anti-science lobbying?) and so long as Meyer can hold up that the Tute stands for strong, traditional family values and is opposed to the evil scientific materialist atheist machine, he brings home the bacon. I very much doubt that Meyer actually believes the IDiot BS; it's just his job.

For the creationists in Texas, however, the story is all about power: which person can tell the other person what to do. Here it's all about kicking poor people, especially if they are not white anglo-saxon non-denominational Christian, beating up women with health care restrictions, harassing immigrants and just plain yelling at kids to get off your lawn. And, boy howdy, nothing says flaming liberal more than the word evilution, by hecky, it's even got the word "evil" in it's name!

Sorry to end on such a positive note. I'll try harder next time!

un said...

I'm not an expert to judge. You know better.
I remember that you said in a comment once that phylogenetics programs can work without any assumption of trees or tree structures. Can you elaborate more on this point? Was this actually done? If yes, how did the results look like?

Mikkel Rumraket Rasmussen said...

Theobald's test is not a test in the sense that it had the ability to falsify common descent, it was a test in the sense that it was comparing different models to see which one would be best at reproducing the observed pattern.

He compared several different models and found that universal common descent was the model that best reproduced the data to an overwhelming degree. This doesn't in itself prove common descent, it just makes it more likely to be correct than his competing models chosen for the test.

There have been other similarly inspired tests of common descent, such as this:
Beyond Reasonable Doubt: Evolution from DNA Sequences


We demonstrate quantitatively that, as predicted by evolutionary theory, sequences of homologous proteins from different species converge as we go further and further back in time. The converse, a non-evolutionary model can be expressed as probabilities, and the test works for chloroplast, nuclear and mitochondrial sequences, as well as for sequences that diverged at different time depths. Even on our conservative test, the probability that chance could produce the observed levels of ancestral convergence for just one of the eight datasets of 51 proteins is ≈1×10^−19 and combined over 8 datasets is ≈1×10^−132. By comparison, there are about 10^80 protons in the universe, hence the probability that the sequences could have been produced by a process involving unrelated ancestral sequences is about 10^50 lower than picking, among all protons, the same proton at random twice in a row. A non-evolutionary control model shows no convergence, and only a small number of parameters are required to account for the observations. It is time that that researchers insisted that doubters put up testable alternatives to evolution."

Which is a nice slap in the face to ID and doubters of common descent who also don't bother proposing alternatives.

SRM said...

Its nice to see the refutation of ID claims on technical grounds but of course Bill above has it right when he highlights the actual underlying motives and machinations of the ID and creationist movements and their popularity with the masses.

But I must admit that creationist ideology is also based upon a very firm, irrefutable grounding as illustrated in the opening paragraph to this OP:

Stephen Meyer argues (in Darwin's Doubt) that the Cambrian explosion... can be explained by Intelligent Design Creationism.

Checkmate, close the books, and get thy self to a church. Is there any observation in nature that cannot be explained by Intelligent Design Creationism? The answer of course is no, and that is all that a sizeable portion of the population needs (and wants) to know. And ching-ching go the cash registers.

John Harshman said...

I remember that you said in a comment once that phylogenetics programs can work without any assumption of trees or tree structures. Can you elaborate more on this point? Was this actually done? If yes, how did the results look like?

I did? I don't recall doing such a thing, and in fact most phylogenetic analysis programs do assume that there is a single, strictly bifurcating tree. So I'm not clear on what I would have meant. I know of a few that don't make that assumption. John Alroy published one about 20 years ago, one that even allowed for the possibility of multiple, unconnected trees. And I suppose Splitstree might count. There are also gene-tree methods and even a few that try to take reticulation into account.

Alroy, J. 1995. Continuous track analysis: a new phylogenetic and biogeographic method. Syst. Biol 44:152-178.

SPARC said...

Speaking of cytochrome C: Did Meyer dismiss Denton lately? And whyt about Schwabe who developed his Genomic Potential Hypothesis from similar arguments regarding relaxins?

Unknown said...

Try producing a tree with nested hierarchies using similarities of designed objects such as cars. It won't work. This is a simple yet powerful way to demonstrate how designed objects do not show homology.

Robert Byers said...

I said this elsewhere but once again the whole flaw is the presumption that like molecular details can be used to extrapolate backwards to common origins for this or that.
Does this demand these conclusions or is it just a option.
If just a option its not evidence for common descent by this trail.
If biology was made by a creator/a guy then it also would be a option that common molecular trails would be found in biology from common laws or blueprints for biology.
Its just mere reasoning backward to see these trees.
I'm saying its not scientific evidence but only a hypothesis.
Evolutionists didn't think there could be other options and so wrongly presumed their was only one option and so that morphed into a fact or good evidence for descent by this trail.
Its not a scientific genetic conclusion to see trees as proven by extrapolation backwards.

Anonymous said...

Aha, this is what I was getting at. I vaguely remember as a 1st year undergrad, producing a cladogram from various nails, tacks and screws in a Bio lab. Perhaps Meyer's misconception is based on something like this.

christine janis said...

"There are enough weasel-words in Doubt to feed all the weasels in the world (including lawyers) for a century. "

Methinks it is like a weasel

christine janis said...

"If biology was made by a creator/a guy then it also would be a option that common molecular trails would be found in biology from common laws or blueprints for biology."

Of course. Common design can explain everything. That's why it's not science, because it's not testable and not refutable.

Robert Byers said...

Me first. I'm saying the extrapolation of present molecular evidence to determine past common origins is not scientific methodology for using molecular data for these conclusions.
Evolutionists are making a flawed case here.
Its only a option that extrapolation backwards from the present is a accurate trail to origin.
Being only a option means its only a line of reasoning.
Scientific theory is not based on options. Its based on tested facts etc.
You first!
Demonstrate molecular trees of origins using data of today is from scientific investigation of molecular data.
Betcha can't!

Unknown said...

Common design will not produce a nested hierarchy in phylogenetic trees, but common descent will. This is a straightforward way to distinguish between the two. For eg: as I mentioned above, cars made by different manufacturers may have the same parts, but you can't produce a phylogenetic tree of cars with nested hierarchies. On the contrary, living organisms clearly fall into nested groups which is strong evidence for common descent.

Robert Byers said...

Why not?
Its not just the parts but the whole body of the cars that indicates a common designer.
The slight differences in the cars is just details.

Unknown said...

The only thing, in reality that does "produce a nested hierarchy in phylogenetic trees" is the mind of the person doing the analysis and then constructing the tree, based on whatever data he or she selects, and that's why, depending on the sequences analyzed, how they are aligned, etc. the "trees" always vary and even contradict one another. It undermines the position to see these trees offered as "the" tree.