Peter Schuster (University of Vienna; SFI External Professor)
Abstract. Ever since Sewall Wrights publication of the landscape metaphor in 1932 evolutionary optimization is visualized as hill climbing on a rugged landscape. It took more than 50 years before the first reliable data became available that gave an idea how these evolutionary landscapes might look like and only this year the first large scale data collection on the fitness landscape of HIV virus in absence and presence of antiviral drugs was published. Almost all simplified model landscapes that were used in various applications turned out to give wrong results.
In the talk the structure of fitness landscapes combining ruggedness with neutrality – corresponding to planes – will be discussed and a mathematical model for “realistic” landscapes is introduced. Evolution on these landscapes is modeled by chemical reaction kinetics describing reproduction, mutation, and selection. The measurement of a few special fitness values is sufficient to classify the population dynamics at different mutation rates. Some applications of the model to virus evolution will be presented.