A new study co-authored by SFI External Professor and Science Board co-chair Marcus Feldman demonstrates that not only is long-term evolution possible, but long-term evolutionary outcomes can be surprisingly predictable.
In recent years, some scientists have argued that natural selection occurs not just at the individual organism level, but also between lineages over the course of many generations.
The Stanford group set up a computer simulation in which 128 lineages of proteins continuously folded into new shapes, competing to bind with other molecules, called ligands, in each new configuration. The better each protein could attach itself to the ligands, the more ligands it would scoop up, and the higher its fitness – that is, its average number of "offspring" – would be. The simulation was run for 10,000 generations.
Although the chaos of 128 lineages – a total of more than 16,000 individual proteins – mutating over thousands of generations might seem unpredictable, and that it would be nearly impossible for the same thing to happen twice, it's actually the opposite, the researchers say.
"Even though things look complicated, the possible evolutionary trajectories are quite constrained," said lead author Michael Palmer, a computational biologist at Stanford. "There are only a few viable mutations at any point, which makes the dynamics predictable and repeatable, even over the long term."