A recent paper in Nature finds that epistasis (interactions between genetic changes) is much more pervasive than previously assumed. This strongly limits the ability of beneficial mutations to confer fitness on organisms.
In classic neo-Darwinism, mutations can be considered independent alterations of a local gene. Mutations could be neutral, deleterious, or beneficial. As Darwin personified it, “natural selection is daily and hourly scrutinizing, throughout the world, the slightest variations; rejecting those that are bad, preserving and adding up all that are good.” This simplistic tale has been complicated by epistasis.
Epistasis might be compared to changes in software as opposed to a dictionary. In the dictionary, a single-letter change might not cause a drastic effect on the message. In software, though, routines frequently are dependent on other routines. Software routines have inputs and outputs; they form networks of alliances. Changing a subroutine could ripple throughout the software, causing multiple effects, most probably harmful ones. That’s why software engineers routinely run whole-system tests after making changes.
Epistasis is like that; a mutation in one gene can cause harm in distant genes. Neo-Darwinism has to be modified to incorporate the effects of epistasis. It has to postulate that neutral genes are only neutral in the whole, and that beneficial mutations are only beneficial in the whole. Point mutations can no longer be considered in isolation; what’s beneficial in one context could be deleterious in another.
A team of geneticists in Spain searched for the degree of epistasis in published genomes and found it far more pervasive than thought. In “Epistasis as the primary factor in molecular evolution,” Breen et al. found the following:

… the measured rate of amino-acid substitution in recent evolution is 20 times lower than the rate of neutral evolution and an order of magnitude lower than that expected in the absence of epistasis. These data indicate that epistasis is pervasive throughout protein evolution: about 90 per cent of all amino-acid substitutions have a neutral or beneficial impact only in the genetic backgrounds in which they occur, and must therefore be deleterious in a different background of other species. Our findings show that most amino-acid substitutions have different fitness effects in different species and that epistasis provides the primary conceptual framework to describe the tempo and mode of long-term protein evolution. (Emphasis added.)

It would be hard to improve on their explanation of how epistasis affects evolution, so here it is:

An amino-acid substitution that is neutral or beneficial in one genetic context may be deleterious in another. Such a situation, when the fitness effect of one allele state depends on the allele states at other loci, is called epistasis. Both the neutral and selective theories of protein evolution provide an accurate framework for understanding long-term protein evolution only if amino-acid states in different genetic contexts have the same effect on fitness, that is, if epistasis is rare. In the absence of epistasis, when the fitness effects of all amino-acid states are independent of one another, substitutions in different species are expected to have similar effects on fitness except in cases where these substitutions enable differences in adaptation to environmental conditions. In that case, if an amino-acid state were found in one species in a protein sequence that is not directly involved in environmental adaptation, such as a housekeeping protein, then the same amino-acid state should be acceptable in an orthologous site in a different species. However, if epistasis is common then amino-acid substitutions that were beneficial or neutral in one species should often be deleterious in another. Therefore, unravelling the extent and basis of epistasis may be crucial to understanding differences in protein sequences between species and long-term protein evolution. At present, studies of the differences in the fitness of substitutions in different genetic contexts consider specific genes or events, and it is unknown what fraction of amino-acid substitutions that occur in one species would also be acceptable in another species if they were to occur in orthologous sites (but see ref. 11). Here we develop an approach to quantifying the impact of epistasis in protein evolution and show that the fitness effects of most amino-acid substitutions must depend on the genetic context in which they occur. (References deleted.)

They found that epistasis is not only much more prevalent than previously thought, it is the “primary conceptual framework to describe the tempo and mode of long-term protein evolution.” It should be intuitively obvious that altering a gene coupled to other genes makes neo-Darwinian progress much more improbable. A beneficial mutation has to be beneficial in more contexts. Similarly, neutral or nearly neutral mutations will be less frequent, since there is a larger probability that they will have deleterious effects on other genes. This sheds light on why the authors found that “positive selection was not common in the evolution of the proteins in our data set,” according to a test they used to search for it.
Even if a beneficial mutation survives to improve fitness on one species in one environment, there is no guarantee the same mutation will improve another species. “Thus, an amino acid that was beneficial to one species because of a specific environmental adaptation may be detrimental to a species that does not live in the same environment,” they said. Evolutionists cannot avoid this problem, because “epistatic interactions are the norm and not the exception when we consider amino-acid substitutions in protein sequences.”
With all this bad news for neo-Darwinism, could the authors rescue evolutionary progress? Too bad, no. Their last paragraph consisted only of hard questions raised by their findings:

We identify epistasis as a powerful factor affecting long-term protein evolution and one that must necessarily be invoked to explain why the vast majority of amino-acid substitutions that occur in one species cannot occur in another regardless of whether or not positive selection plays the dominant role in the course of fixation of amino-acid substitutions in specific genetic contexts. An epistatic perspective of molecular evolution leads to the formulation of several fundamental questions, in addition to the largely unanswered questions posed by John Maynard Smith in 1970 (ref. 12). First, given a specific site, substitutions in how many other sites in the same gene or in the entire genome could change the strength of selection associated with substitutions at this site? Second, out of the entire network of pairwise epistatic interactions between sites across the genome, are there many non-overlapping epistatic subnetworks or are most sites interconnected within the entire network of epistatic interactions? Third, what is the ratio of intergenic to intragenic epistatic interactions? Fourth, what is the molecular basis of epistatic interactions within the genome? Finally, pervasive epistasis in long-term protein evolution raises the possibility that similar epistatic interactions may be prevalent in short-term evolution and that situations when a polymorphism is benign or beneficial to one individual but deleterious to another individual within the same population may be more common than is thought at present.

This is potentially devastating news for neo-Darwinians. Just as they were still struggling to deal with old unanswered questions John Maynard Smith raised 42 years ago, these five new questions threaten to erode the raw material for Darwin’s sieve of positive selection in profound and fundamental ways.
Will this paper have an immediate (we might say “epistatic”) influence on evolutionary theory? Probably not. Science is like a large ship that turns slowly. Thousands of scientific papers are published each week. It’s doubtful that many scientists will read or even notice this paper; those who do may shrug it off as a puzzle to be solved later, since neo-Darwinism has long been the accepted paradigm. Remember that it took decades (some say seventy years) for Mendel’s paper to be noticed and considered seriously by evolutionary theorists; even then, they didn’t abandon the paradigm — they just incorporated Mendelian inheritance in it. The Darwinian web of belief is so strong that its proponents will simply come up with new models to incorporate epistasis into the web. These geneticists were just raising new puzzles to be solved within the paradigm, not seeking to overthrow it.
It takes an outsider to read this paper and see how disturbing it should be to the consensus neo-Darwinian theory. All that Darwin skeptics can do is continue to point to papers like this as severe challenges to the consensus view. Perhaps a few will listen and take it seriously.
More likely, if Thomas Kuhn’s view of the incommensurability of paradigms is even partially correct, the Darwinians and the Darwin skeptics will just talk past each other. It will require a new generation of young, open-minded scientists not yet wedded to the Darwinian paradigm to lead the long-overdue scientific revolution.