In March 2010, Douglas Theobald published a paper in Nature purporting to demonstrate “A formal test of the theory of universal common ancestry.” According to his cheering squad at National Geographic, the paper “supports the widely held ‘universal common ancestor’ theory first proposed by Charles Darwin more than 150 years ago.” National Geographic is mistaken on one obvious point: Darwin wasn’t the first to propose universal common ancestry. But never mind that. The paper makes no official claim to be a response to scientific skeptics of universal common ancestry, but given Theobald’s notoriety as the author of the widely criticized “Talk Origins Common Ancestry FAQ,” his motivation is clear. If there were no doubts about universal common ancestry (“UCA”), his paper would be unnecessary. This becomes especially clear when you see the trivially obvious point his paper actually establishes as part of his “test” of universal common ancestry.
Before going any further, I must make it clear that intelligent design (ID) is certainly not incompatible with common ancestry. ID refers to the mechanism of change, and does not claim that species are necessarily unrelated. So ID grants that it’s possible that all living species shared a common ancestor, but ID doesn’t require it. In fact, ID leaves multiple options open, which will be discussed in my next post.
In contrast, neo-Darwinism is inextricably wedded to common ancestry and requires a common ancestor (or common gene pool) for all living organisms. That’s why neo-Darwinists must defend UCA at all costs. Testing UCA allows us to evaluate neo-Darwinian predictions, and understand what the evidence shows.
Your Theory Is Only as Good as Your Null Hypothesis
Whenever someone purports to test a theory, you must ask, “What’s the null hypothesis?” In other words, the test of a theory is only as good as the hypothesis that you’re trying to test it against. To see what I mean, consider an extreme example.
Say that I hypothesize that plants need milk to grow. “Well, there’s an easy way to test my hypothesis,” I say. “I’ll test the hypothesis that plants will grow when fed gasoline as my null hypothesis. If they don’t grow using gasoline, then that will lend support to my hypothesis that they grow via milk.” So I feed my plants gasoline, and they promptly die; I feed them milk and they don’t die. I conclude that this shows plants require milk to live.
Obviously, there are many other ways to explain my experimental observation (that plants die when fed gasoline) apart from the conclusion that plants live on milk. So even if my hypothesis is correct, testing it against a preposterous hypothesis doesn’t make a very compelling test. While plants might survive on milk for a time, my test ignores the possibility that there may be better options for keeping plants hydrated — such as water.
Returning to Theobald’s paper, he repeatedly claims that his test provides a “formal, fundamental test of UCA, without assuming that sequence similarity implies genetic kinship.” Again he writes, “I report tests of the theory of UCA using model selection theory, without assuming that sequence similarity indicates a genealogical relationship.” But given that his null hypothesis is that certain genes found ubiquitously throughout living cells arrived at their highly similar DNA sequences via independent convergent evolution, he’s testing universal common ancestry by comparing it to a preposterous null hypothesis.
All Theobald is really testing is whether unguided common ancestry for similar genes is more likely than unguided independent ancestry for similar genes — e.g., convergent evolution. As Koonin and Wolf put it), Theobald uses a “general null hypothesis of independent ancestry.” If you don’t believe me, read this passage from Theobald’s paper:
Sequence similarity is an empirical observation, whereas the conclusion of homology is a hypothesis proposed to explain the similarity. Statistically significant sequence similarity can arise from factors other than common ancestry, such as convergent evolution due to selection, structural constraints on sequence identity, mutation bias, chance, or artifact manufacture. For these reasons, a skeptic who rejects the common ancestry of all life might nevertheless accept that universally conserved proteins have similar sequences and are “homologous” in the original pre-Darwinian sense of the term (homology here being similarity of structure due to ”fidelity to archetype”). Consequently, it would be advantageous to have a method that is able to objectively quantify the support from sequence data for common-ancestry versus competing multiple-ancestry hypotheses.
In other words, he finds: (1) the odds are very slim of highly similar DNA sequences arising via independent evolutionary processes (“multiple-ancestry”), and (2) therefore inheritance of these sequences from some universal common ancestor must be the correct explanation. The problem is that everyone knows that (1) is obviously true, and finding (1) doesn’t necessarily demonstrate (2). The National Geographic article explains this logic:
The “best competing multiple ancestry hypothesis” has one species giving rise to bacteria and one giving rise to Archaea and eukaryotes, said Theobald, a biochemist at Brandeis University in Waltham, Massachusetts.
But, based on the new analysis, the odds of that are “just astronomically enormous,” he said. “The number’s so big, it’s kind of silly to say it” — 1 in 10 to the 2,680th power, or 1 followed by 2,680 zeros.
Biologists call the independent development of similar traits in different lineages “convergent evolution.” The wings of bats, birds, and insects are prime examples: They perform similar functions but evolved independently of one another.
But it’s highly unlikely that the protein groups would have independently evolved into such similar DNA sequences, according to the new study, to be published tomorrow in the journal Nature.
“I asked, What’s the probability that I would see a human DNA polymerase [protein] sequence and another protein with an E. coli DNA polymerase sequence?” he explained.
“It turns out that probability is much higher if you use the hypothesis that [humans and E. coli] are actually related.”
National Geographic notes in a subheadline: “Creationism called ‘absolutely horrible hypothesis’ — statistically speaking.” The problem is that Theobald didn’t test universal common ancestry against “creationism.” He tested universal common ancestry against the impossibly unlikely hypothesis that these genes independently arrived at highly similar sequences via blind, unguided convergent evolution. Given his outlandish null hypothesis, no wonder common descent came out looking so good.
Again, if you don’t believe me, consider what reviewers of a critique of Theobald’s paper had to say (approving the critique!):
Cogniscenti cringed when they saw the Theobald paper, knowing that “it is trivial”. It is trivial because the straw man that Theobald attacks in a text largely formulated in convoluted legalese, is that significant sequence similarity might arise by chance as opposed to descent with modification. Ignoring the strength of the universality of the genetic code and the commonality of central intermediary metabolism among cells as evidence, Theobald construed a non-issue that the referees of his paper, whoever they may have been, found convincing and novel (!).
(Comments by William Martin of the University of Duesseldorf in review of Eugene V Koonin and Yuri I Wolf, “The common ancestry of life,” Biology Direct, Vol. 5:64 (2010).)
As we’ll see in the next post, the problem with Theobald’s test is that he didn’t consider the possibility that a multiple-ancestry model might involve common design–a game-changing possibility for the whole field of systematics that he hoped this paper would scuttle through technical argument.