Birds fly because parts (like feathers) popped into existence for no reason, but later were found useful (for flight). Is that any way to explain the complex adaptations in life?

Under the headline “Great Exaptations,” the Santa Fe Institute is celebrating a new paper in Nature by Aditya Barve and Andreas Wagner. The paper begins:

How evolutionary adaptations and innovations originate is one of the most profound questions in evolutionary biology. Previous work emphasizes the importance of exaptations, also sometimes called pre-adaptations, for this origination. These are traits whose benefits to an organism are unrelated to the reasons for their origination; they are features that originally serve one (or no) function, and become later co-opted for a different purpose. (Emphasis added.)

So we see that exaptation, pre-adaptation and co-option are essentially synonymous, supporting the counter-intuitive idea that purposelessness leads to purpose. (For a general answer to co-option, see Unlocking the Mystery of Life, the section about the bacterial flagellum.)

Barve and Wagner set out to determine whether evolution builds eyes, wings and other wonderful things primarily by natural selection or by exaptation. They also wanted to know how exaptations originate.

Exactly how new traits emerge in evolution is a question that has long puzzled evolutionary biologists. While some adaptations develop to address a specific need, others (called “exaptations”) develop as a by-product of another feature with minor or no function, and may acquire more or greater uses later. Feathers, for example, did not originate for flight but may have helped insulate or waterproof dinosaurs before helping birds fly

Their short answer is, Yes — exaptation outcompetes natural selection as evolution’s mechanism of choice.

The findings underscore the idea that traits we see now — even complex ones, like color vision — may have had neutral origins that sat latent for generations before spreading through populations, Wagner says.

“Our work shows that exaptations exceed adaptations several-fold,” he says. (Emphasis added.)

The Empirical Data

To support their claim, Barve and Wagner tinkered with the metabolism of E. coli bacteria. By altering steps in their metabolic networks in computational models, they found that some of the hypothetical bacteria were able to use other sources of carbon for energy:

Starting with the metabolism of an E. coli that can survive on glucose as its sole carbon source, they subjected the complex metabolic chemical process to a “random walk” through the set of all possible metabolisms, adding one reaction and deleting another from it with each step. They kept constant the total number of reactions and the bacterium’s ability to survive on glucose alone, but allowed everything else to change. Every few thousand steps they analyzed the altered metabolism’s reactions.

They found that most metabolisms were viable on about five other carbon sources — sugars, building blocks of DNA or RNA, or proteins — that are naturally common but chemically distinct compounds. To be certain that viability on these other carbon sources wasn’t a natural consequence of viability on glucose, they tested metabolisms starting with viability on 49 other carbon sources, and each time found that exaptations emerged allowing the metabolism to survive on any one of several other carbon sources alone.

They reasoned that neutral mutations pre-adapted them to use those energy sources, even though in the wild they fed exclusively on glucose. Since exaptation appeared to work here, it must be a universal mechanism available to evolution.

But does their limited experimentation justify the claim that all of life generates complex adaptations by exaptations?

The Problems

There are a number of problems with their conclusions, not exhausted by the following list:

1. Assume a toolkit: They assumed that all necessary nutrient transporters were present. “If this is not the case, the incidence of exaptation may be reduced,” they confessed. They felt justified, though, since “84% of E. coli transporters can transport multiple molecules, and their substrate specificity can change rapidly, thus ameliorating this constraint.” What other parts did they assume, or provide by intelligent design?

2. Search space: If lucky exaptations just “emerge” randomly, they have an excessively large search space to find a function. Stephen Meyer discussed this problem in Chapters 9-12 of Darwin’s Doubt in relation to protein folds, but the same reasoning applies here: “combinatorial inflation” guarantees that no random process will have the time to search the available functional space. Barve and Wagner admit their method relies on chance:

Briefly, this procedure involves random walks in the space of all possible networks. During any one such random walk, a metabolic network can change through the addition and deletion of reactions. Although this process resembles the biological evolution of metabolic networks through horizontal gene transfer and (recombination-driven) gene deletions, we here use it for the sole purpose of creating random samples of metabolic networks from the space of all such networks.

3. Investigator interference: To avoid combinatorial inflation, which would have guaranteed failure on grounds of probability, they applied their own intelligent design to drastically reduce their test-tube “universe” of reactions:

As we described earlier, we specifically used the REACTION and COMPOUND databases to construct our universe of reactions while excluding all reactions involving polymer metabolites of unspecified numbers of monomers, or general polymerization reactions with uncertain stoichiometry; reactions involving glycans, owing to their complex structure; reactions with unbalanced stoichiometry; and reactions involving complex metabolites without chemical information. The published E.?coli metabolic model (iAF1260) consists of 1,397 non-transport reactions. We merged all reactions in the E.?coli model with the reactions in the LIGAND database, and retained only the non-duplicate reactions. After these procedures of pruning and merging, our universe of reactions consisted of 5,906 non-transport reactions and 5,030 metabolites.

4. Wild type: They did not test the efficiency of alternate carbon sources in living bacteria, nor did they release any in the wild to see if they continued thriving on other carbon sources in real-world conditions. The whole setup was contrived in models.

5. Coding: They did not test whether the alternative lifestyles would recode the genes to preserve the change, so that it could be inherited; they just assumed standard mutation and selection would take care of it.

6. Pre-design: Living E. coli can and do use multiple carbon sources for metabolism, but Barve and Wagner cannot assume that their lab-derived hypothetical exaptations would arise without purpose in a random or neutral manner. Metabolic systems in living cells could be pre-designed to adapt. There are many examples in living cells of “backup plans” if environmental conditions present starvation of resources.

7. Designed variability: Life also uses targeted variation. Our immune system, for instance, randomly generates and varies antibodies in the thymus gland when presented with an unknown pathogen. Once a suitable fit is obtained, it is put to use — a form of “artificial selection” by a system designed to use randomness for a purpose. The researchers did not consider whether the bacteria were prepared — by design — to respond to an imposed “random walk.”

8. Extrapolation: Barve and Wagner acted recklessly in their extrapolation, taking results from an artificially small “universe” of reactions in one species of bacteria and assuming it could speak to a universal principle. From their highly contrived results, they presumed to pontificate about bird flight and eye design.

9. Bye-bye Darwin: If Barve and Wagner are right, it means 154 years of faith in natural selection to produce complex adaptations has been misguided. They appear worried about that, saying in their conclusion, “If confirmed in systematic analyses like ours, the pervasiveness of non-adaptive traits may require a rethinking of the early origins of beneficial traits.” The Santa Fe Institute puts it this way: “If exaptations are pervasive in evolution … it becomes difficult to distinguish adaptation from exaptation, and it could change the way evolutionary biologists think about selective advantage as the primary driver of natural selection.”

For these and other reasons, this paper and its accompanying news release amount to little but bluff and bluster. Their conclusions do not follow from their contrived setup. Maybe we should just ask Barve and Wagner if their own scientific work was an exaptation from hunting for meat or howling at the moon. If so, why would it have any validity? (See The Magician’s Twin, Chapters 8-9).

It’s hard to take seriously the beliefs of those who see dinosaurs finding feathers “emerging” on their arms for no purpose, or to keep warm or attract mates, then imagine that birds found them useful for flying. Time to watch Flight: The Genius of Birds again.

Image: Wikicommons.