In Darwin Devolves, Michael Behe argued that the adaptations we observe in nature proceed by blunting or breaking genes, like desperate sailors in a storm trying to survive by throwing things overboard. His most memorable analogy is that tossing the seats and hood out of a car is advantageous if your primary concern at the moment is gas mileage. Recently, evolutionists argued a similar thing: loss of function can be adaptive. As Casey Luskin has noted, they did so without acknowledging Behe’s insight.

Writing in the Nature journal Heredity (open access), J. Grey Monroe et al. discuss “The population genomics of adaptive loss of function.”

Discoveries of adaptive gene knockouts and widespread losses of complete genes have in recent years led to a major rethink of the early view that loss-of-function alleles are almost always deleterious. Today, surveys of population genomic diversity are revealing extensive loss-of-function and gene content variation, yet the adaptive significance of much of this variation remains unknown. Here we examine the evolutionary dynamics of adaptive loss of function through the lens of population genomics and consider the challenges and opportunities of studying adaptive loss-of-function alleles using population genetics models. [Emphasis added.]

They examine the possibility of “positive selection” resulting from gene loss or function loss. Note that “positive selection” to a neo-Darwinist does not necessarily mean that some innovative, new thing has emerged; they just mean that a particular genotype has increased in frequency (or at least has not disappeared) and must, therefore, have been naturally “selected.” One can think of cars without seats and hoods proliferating in an environment where high gas mileage is the predominant “selection pressure.”

Old School Pop Gen

Population genomics as a discipline became standardized through the mathematical models of Haldane and Kimura and others in the 1920s through 1960s. 

Early genetics employed a functionally definitive concept of an allele. Alleles were treated as local units of inheritance based on their functional effect, observed at the phenotypic level. As such, at locus a experiencing adaptive loss of function, the (potentially multiple) variants causing the adaptive trait should act collectively as a single allele, even if due to independent mutational events. If, for example, this functionally identical set of variants experiences positive selection, it behaves like a single allele according to predictions of classic population genetic theory — increasing in frequency to fixation. Indeed, foundational models of population genetics accommodate recurrent mutation and predict that adaptation will often involve multiple independent mutational origins given realistic population sizes, selection coefficients, and mutation rates. But if we encounter such cases through analyses of DNA sequence alone, we may be troubled to find that the sequence variants only exhibit the expected evolutionary dynamics of classical alleles when considered as aggregated functional units, but not when analyzed individually. [Internal citations omitted.]

The result was that classical pop gen was nice in theory, but hard to demonstrate in lab work. A geneticist may not be aware of all the components of “classical alleles” and will be puzzled when finding a seemingly well-adapted phenotype doing fine with a missing or broken gene. Some geneticists postulated in such cases that a “soft sweep” had occurred in the lineage, eliminating individuals possessing the functional gene. 

In contrast, hard sweeps of a single adaptive variant were described during the era predominated by the view that loss-of-function mutations were necessarily deleterious, and adaptation could only proceed through mutationally rare gain-of-function alleles. Such historical dynamics speak to the interconnectedness, intentional or otherwise, between ideas about the functional molecular basis of adaptation and advances in the development of population genetic models and theories.

Unfortunately, population genetic statistics based on the expectation that adaptive alleles are mutationally rare perform poorly when this assumption is violated.

Monroe et al. continue with examples that perform poorly when violating the assumption that loss of function mutations must be necessarily deleterious. In some cases, the phenotype can appear neutral or even adaptive.

New, Improved Pop Gen

Now that whole gene sequences are becoming common, new pop gen techniques based on functional test statistics are valuable, but “can yield surprising results,” they say. Signatures of purifying selection (the weeding out of deleterious mutations) can disappear. A loss-of-function mutation can look neutral or harmless in a genetically diverse population. Geneticists need to take advantage of “functionally definitive” pop gen models and methods.

Accelerations in whole genome sequencing technologies and improved capacity to classify previously cryptic loss-of-function variants may facilitate a new generation of functionally definitive population genetic models and methods. This would not only be valuable for improving the capacity to understand the forces shaping intraspecific loss-of-function, but more generally promote a re-synthesis between studies of molecular sequence variation and the function-based conception of alleles from early population genetic theory.

So far, Monroe et al. have only suggested that loss-of-function mutations can be neutral or adaptive. They don’t give any persuasive examples of positive selection after a loss of function mutation. They only encourage use of the new models and methods to avoid missing cases of “beneficial and adaptive loss of function”:

For example, loss of function in SLC30A8 was found to be strongly associated with decreased risk of type 2 diabetes when all loss-of-function variants were collapsed into a single allele state, thus identifying its protein product as a promising therapeutic target to treat diabetes. With population whole-genome-sequence data becoming available in model and non-model species, this approach can now be readily applied by evolutionary biologists at genome wide scales to discover loss-of-function alleles contributing to phenotypic evolution in populations

The new models and methods represent “genetic diversity as a matrix of functionally relevant genetic alleles rather than a matrix of DNA sequence variants.” Examined this way, loss-of-function variants will tend to stand out. But they should not be dismissed as deleterious. They might help the car get better gas mileage.

In their “Outlook and Concluding Remarks” section, the authors raise several research questions that the new pop gen models and methods can inspire.

Loss-of-function alleles were once often held up as a paragon of deleterious genetic variation. Today a more nuanced appreciation for their potential role in adaptation has emerged. This new paradigm inspires investigations into deeper questions about the causes and consequences of adaptation by genetic loss of function. For example: Do species or populations differ in their capacity to adapt via loss of function, and if so, why? Does the high effective mutation rate of loss-of-function alleles lead to bias in the probabilities of different evolutionary outcomes? What is the contribution of adaptive loss of function to the phenomena of antagonistic pleiotropy and reproductive isolation? How does adaptation by loss of function affect long term evolutionary trajectories of populations and future evolvability? Ongoing technical breakthroughs promise to scale up the study of loss-of-function alleles experiencing positive selection for population genomic research to address these questions. More broadly, these lines of research provide paths toward advancing tools and concepts that facilitate a continued synthesis between functional molecular genomics and classic population genetic theory.

One question the authors left out must be buzzing around the heads of ID advocates. How are you going to get wings and eyes by “adaptive loss of function”? The car may be speeding down the road sportively with high gas mileage, but the seat and hood that were tossed out are not coming back by more subtractions! 

This is the latest iteration of the concept of “evolution by subtraction” discussed at Evolution News in 2013 and again in 2015, five years before Behe’s book Darwin Devolves. The conclusion then is still apropos now: “We understand subtraction, but we want to hear more about addition.”