Myriad genetic differences distinguish the genomes of two species. What fraction of those differences arises by positive natural selection versus random genetic drift is a central question and topic of debate in evolutionary biology. A recent study in Nature Ecology & Evolution suggests positive selection could have a larger influence than previously thought.
Darwin proposed positive selection—the idea that beneficial mutations offer their bearers some advantage and so those mutations spread through a population. But genetic drift is also known to cause genome evolution through the rise of random mutations that spread by chance through the gene pool. Past lab and field studies reported conflicting evidence for the relative importance of positive selection compared to genetic drift. Lab studies tended to favor positive selection, while field studies showed stronger evidence for drift.
To sort out the conflict, researchers leading the recent study performed a series of evolution experiments in laboratory yeast (Saccharomyces cerevisiae) under conditions mimicking the changing environment of the wild. Their primary conclusion, according to coauthor and evolutionary geneticist Jianzhi “George” Zhang at the University of Michigan in Ann Arbor, is that beneficial mutations experiencing positive selection under constant laboratory conditions can switch—when the environment changes, as in the wild—to become harmful. This results in negative selection, which weeds out harmful mutations. Hence, in changing environments, the signal of past positive selection can be masked by more recent negative selection. This apparently led researchers to miss the signal of positive selection and conclude that evolution happened by drift.
The experiments are among the first to test the idea that genes which are beneficial in some environments but harmful in others could affect patterns of molecular evolution, says molecular biologist Greg Lang at Lehigh University in Bethlehem, Pennsylvania. “It’s a nice way of bridging the gap between lab evolution experiments and what we observe in natural populations,” says Lang.
To estimate whether positive selection or drift has more influence on genomes, biologists use a metric denoted by the symbol omega. An omega higher than one indicates more positive selection. An omega less than one suggests more negative selection and more drift. Omega represents the ratio of nonsynonymous mutations (those that produce different proteins due to codon changes) to synonymous (those that produce the same protein despite codon changes) across the genomes.
Past genome comparisons of wild species have found fewer nonsynonymous mutations than synonymous ones, suggesting that most natural mutations are harmful and weeded out by negative selection, or neutral and spread by drift. However, previous lab experiments indicate the opposite pattern. A 2016 study of Escherichia Coli that was allowed to grow and adapt under constant lab conditions for 50,000 generations, for example, found omega values above one, suggesting the bacteria undergo much more positive selection.
“We started thinking about what could explain this difference,” says Zhang. To test the idea that changing environments hide the signal of positive selection, he and collaborators grew S. cerevisiae in five rotating environments: in a test tube containing copper for 224 generations, followed by a different environment at pH 8 for another 224 generations, and then in a third environment containing hydrogen peroxide, and so on. Control yeasts grew in just one of the five environments, without rotating.
At the end of the experiment, the researchers compared the genomes of the yeast grown in changing environments to those grown under constant conditions. As expected, omega was lower for yeasts in changing environments than those in constant ones, indicating that beneficial mutations that were under positive selection in one environment then became harmful in the next environment and experienced negative selection; the past signal of positive selection was masked.
To better approximate differences between genomes in the wild, Lang suggests that future studies could run on timescales longer than 224 generations. One logical next step, says Zhang, is clarifying the degree of underestimation and finding alternative ways to detect positive selection. His lab is now considering follow-up studies to address these questions. “This paper is saying there’s a problem of estimating omega,” Zhang concludes. “So the real question is, How do we estimate omega from genome sequences?”