Evolving Evolvability | Evolution News and Science Today

Try to wrap your mind around the concept that evolvability evolves by natural selection. On second thought, don’t. It’s not conducive to mental health.

Valuing charity, I try to approach new evolutionary papers with dispassionate tolerance, seeking understanding before forming an opinion about them one way or another. This one was a particular challenge. It’s like trying to imagine a Mobius strip wrapping a Klein bottle in hyperspace. What on earth is meant by natural selection favoring the evolution of evolvability? Is this even a potentially useful notion for understanding how the world works?

Mentions of “evolvability” here at Evolution News can be found scattered through articles by several contributing authors, but none I searched for have treated it in detail. Now that two papers on evolvability have appeared in separate journals in February 2025, it’s a good time to examine the concept. 

The first paper, in PNAS, led by Luis Zaman from the University of Michigan, will not require much analysis, for two reasons: (1) The authors are consumed with Darwinism to the point of absurdity, and (2) Their justification is entirely built on a computer model running Avida. Even the title of the press release mentions evolution five times! “Evolution, evolution, evolution: How evolution got so good at evolving.” 

Now, a University of Michigan study shows that perhaps why evolution is so effective is that evolution is itself something that can evolve. The research is published in the Proceedings of the National Academy of Sciences.

“Life is really, really good at solving problems. If you look around, there’s so much diversity in life, and that all these things come from a common ancestor seems really surprising to me,” said Luis Zaman, an evolutionary biologist at U-M and lead author of the study. “Why is evolution so seemingly creative? It seems like maybe that ability is something that evolved itself.” [Emphasis added.]

Forms of the word “evolution” appear 38 times in this short press release, and 214 times in the paper. Such overuse of a word appears pathological, like an addiction. Worse, it contains no biological field work at all. Its conclusions are rationalized entirely by a computer model with imaginary organisms in silico that were designed to evolve or fail by natural selection. Live Science liked the paper, but because the Avida platform that supported this computer game has been debunked extensively by others at Evolution News (here, here, and here), it deserves no further serious consideration other than for the possible entertainment value, like watching clowns in a curved maze looking for a penny in the nonexistent corner.

Much Empiricism About Nothing

The second paper, published in Science, gets more into the weeds. Barnett, Meister, and Rainey titled their work “Experimental evolution of evolvability.” For a synopsis of the paper, see the Perspective by Edo Kussell (“Enabling evolvability to evolve”) in the same issue of Science, or see the press release from the Max Planck Institute for Evolutionary Biology featuring two of the authors, Michael Barnett and Paul Rainey.

A new study by researchers at the Max Planck Institute for Evolutionary Biology (MPI-EB) sheds fresh light on one of the most debated concepts in biology: evolvability. The work provides the first experimental evidence showing how natural selection can shape genetic systems to enhance future capacity for evolution, challenging traditional perspectives on evolutionary processes.

Right at the outset, we see them “challenging traditional perspectives on evolutionary processes,” leading one to proceed with caution as if handed a bottle of New Coke. Arguing that mutation and selection interact, they propose a concept called “lineage-level selection.” Here we go; just what the world needs now: not love, sweet love, but another type of natural selection. 

A caption to the opening diagram explains:

Central to this is lineage-level selection: bacterial lineages (connected nodes) were required to repeatedly evolve between two phenotypic states. Mutational transitions were initially unreliable, leading to lineage deathand replacement by more successful competitors. Final surviving lineages evolved mutation-prone sequencesin a key gene underpinning the phenotypes, enabling rapid transitions between states.

According to their concept, “natural selection optimises genetic systems for future adaptations.” Lineage selection locates the target of selection in the lineage rather than in the individual or population. In this view, your genealogy determines how natural selection will let you evolve.

Imaginary Foresight by Natural Selection

Dr. Marcos Eberlin wrote about Foresight as a sign of intelligence. In the theory of Barnett et al., however, foresight evolves (believe it or not). It’s not real foresight. It’s just imaginary foresight. They call it “evolutionary foresight.” Selection looks down through the halls of time and muses, “Which of my future lineages might win the competition for fitness?” It decides that the winner will be the most evolvable one. This is where the authors start playing mind games with your sanity. “This is not the selection you are looking for,” they say with a hypnotic gesture of the hands.

Evolution by natural selection is a blind process, but living systems can appear to possess evolutionary foresight. Mechanistically, this is conceivable. Certain configurations of gene regulatory networks, developmental systems, chromosomal architectures, and mutational processes have apparent adaptive utility in future environments. Taking advantage of such future adaptive potential requires not only memory of evolutionary history but often an ability to regenerate previously achieved phenotypic states. In this work, we show how selection on lineages can incorporate prior evolutionary history into the genetic architecture of a single cell, such that mutation appears to anticipate future environmental change.

They lost me on the assertion that “evolutionary foresight” is mechanistically conceivable. That is certainly not your grandpa’s Darwinism. At that point, I looked into their Materials and Methods to see what scientific experiments they did to support this notion. Sure enough, they ran actual lab experiments for three years on real organisms, not just computer models. 

Madness in the Methodology

They carefully studied populations of the bacterium Pseudomonas fluorescens (pictured at the top) kept in “glass microcosms” (presumably flasks or test tubes) each with billions of cells. Some of the populations were able to manufacture cellulose (CEL⁺) and some were not (CEL^–). When starved for oxygen, bacteria with the genes to make cellulose created cellulose mats on which individuals could get close to the air/liquid interface for access to oxygen. The presence of cellulose made by CEL⁺members, therefore, provided a fitness advantage (meaning, the ability to avoid dying). 

The team identified “hypermutable” loci with 10,000 times the mutation rate that they describe as similar to “contingency loci” in pathogenic bacteria. Having a contingency plan sounds like design, but they believe the ability for rapid mutation gives the bacterium “foresight” in the form of “evolutionary potential.” The press release explains,

“Our findings show that selection at the level of lineages can drive the evolution of traits that enhance evolutionary potential, offering a fascinating glimpse into how evolution can gain what appears to be ‘foresight’.” Michael Barnett, the study’s first author, added: “By demonstrating the evolution of a hyper-mutable locus, we show that adaptation is not just about surviving in the present but also about refining the ability to adapt in the future.”

The results challenge the long-held view that evolution operates without foresight. Instead, they reveal how natural selection can embed evolutionary history into genetic architecture, enabling organisms to “anticipate” environmental changes and accelerate their adaptation.

Several design words can be seen there: architecture, anticipation, embedding. Are these things that blind selectors do? In a response to the paper, David G. King, emeritus professor from Southern Illinois University, saw something different going on: neither random mutation nor directed mutagenesis:

For example, the insertions and deletions that characterize short tandem repeats (and also enable phenotypic switching in bacterial contingency genes) confer “tuning knob” or “rheostat” functionality on many, perhaps most eukaryotic genes.  Without being biased in the direction of adaptation, repeat number mutability helps assure a relatively advantageous distribution [of] mutation effects.

If so, this would indicate a function for such hypermutable loci. They act like “mutational sponges” that diffuse the harmful effects of random mutations. King explains,

This is the domain of “mutation protocols” whereby an abundant supply of unbiased mutations entails a minimal probability of harm.  Put simply, mutations produced “according to protocol” are constrained to avoid vast domains of DNA sequence space where deleterious results would be practically guaranteed.

Design is evident in concepts like a “tuning knob” or “rheostat” functionality. Another idea not discussed in the paper is the possibility that the populations of bacteria form “quasispecies” in which members of a population retain functional loci that can be shared by horizontal gene transfer. In both cases, genetic changes would not be random.

Conceptual Flaws

But since the authors wish to argue that natural selection (NS), which they admit is “a blind process,” somehow had foresight to “enhance evolutionary potential” (i.e., evolvability), their convoluted concept is subject to the critical scrutiny of NS by illustrious writers including John West (“a corrosive impact on society”), Neil Thomas (“a conceptually incoherent term”), Jonathan Wells (“cannot explain the arrival of the fittest”), and others. Have Barnett et al. twisted NS into a creative force beyond its means by its very nature as an unguided process? Here are a few considerations to keep in mind:

• No origin of species: They started with one species and ended with the same species. 
• Artificial selection: They acted like breeders, which is intelligent design, the opposite of NS.
• Investigator interference: They forced the organisms to “evolve or perish” according to criteria they had set up in advance.
• Unnatural assistance: When a population went “extinct” they transferred cells from a living population to keep it going (see the diagram in Kussell’s Perspective article).
• Limited options: They forced the organisms to exhibit only one of two phenotypic states.
• Personification: They applied terms like foresight, anticipation, and future adaptive potential to blind, mindless processes.
• Magical thinking: Only in Darwin’s Fantasyland can NS be deemed capable of “refining the ability to adapt in the future.”
• Obfuscation: Inventing concepts like “the evolution of evolvability” is no more conducive to understanding than speaking of “the phlogistification of phlogiston.”

Conclusion: Keep Your Investment on Design

Try as they might to resurrect NS from the dead, Barnett et al. and Zaman et al. are stuck with blind, unguided processes with no foresight or desire to adapt. Scientists in Darwin’s day saw through his flawed attempt to present natural selection as analogous to artificial selection, as Robert Shedinger has exposed in Darwin’s Bluff.

Design scientists, by contrast, have the tools in their toolkit to explain adaptation. It takes foresight (real foresight by a designing intelligence, not imaginary “evolutionary” foresight) to engineer a machine for robustness against potential risks. More and more, scientists are finding that life comes equipped with built-in capabilities for adapting to environmental changes. This has been the focus of lively conferences on biological engineering over the past few years. The next Conference on Engineering in Living Systems (CELS), sponsored by Discovery Institute, is coming this summer in Seattle and promises to be a fertile occasion for enlightening discussions in Adventureland and Tomorrowland instead of Fantasyland.