The beneficial mutations that helped yeast survive were losses:
An interesting paper that strongly reinforces the lessons of Darwin Devolves was recently published in Nature Ecology and Evolution.1 University of Michigan biologists Piaopiao Chen and Jianzhi Zhang looked at the effect of changing environments on the evolution of laboratory yeast Saccharomyces cerevisiae…
The most interesting point of their fine work to me is that all of the beneficial mutations almost certainly are loss- or degradation-of-function. That is, the mutations in the various conditions benefit the yeast by destroying pre-existing genes or diminishing their activity. Chen and Zhang followed two different categories of mutations: 1) mutations that substitute single nucleotide residues; and 2) mutations that delete chunks of DNA or cause a stop codon to appear in a gene. The latter category is highly likely to outright destroy the activity of the protein that the mutated gene codes for. Nonetheless, this category is actually the more frequently found of the two. The former category — substitution mutations — does not necessarily destroy a protein’s activity, but that’s certainly the way to bet here. The reason is that most of the selected genes that have substitution mutations (where the normal amino acid residue in the protein the gene codes for is swapped out for a different one) actually have multiple positions that can be beneficially substituted. That’s the signature of a mutation that is helping by degrading or destroying a protein’s activity, simply because there are many more positions where substitution will degrade activity than ones that will improve activity.Michael Behe, “Helpful Devolutionary Mutations Are Rapid and Unavoidable: Paper Reinforces Darwin Devolves” at Evolution News and Science Today
Paper: Chen, P. and Zhang, J. 2020. Antagonistic pleiotropy conceals molecular adaptations in changing environments. Nature Ecology & Evolution 4:461-469. (paywall)
The trouble with Darwin Devolves is that it is likely to be both quite right and a big problem for schoolbook Darwinism. Just as it is much easier to—without thinking much—throw something out than fix or adapt it, life forms will far more likely randomly mutate by dumping complex equipment than by reengineering it. It’s not that life forms can’t develop complex new equipment. But such changes probably aren’t an instance of natural selection acting on random mutation. And in these times, that’s the controversial part: design in nature.
See also: Michael Behe and the broken wolves. In Darwin Devolves, he explains how much evolution depends on breaking genes. He picks up the theme in this video series. In the case of wolves, we call the broken ones dogs.