Statistical topography of fitness landscapes

Fitness landscapes are generalized energy landscapes that play an important conceptual role in evolutionary biology. These landscapes provide a relation between the genetic configuration of an organism and that organism’s adaptive properties. In this work, global topographical features of these fitness landscapes are investigated using theoretical models. The resulting predictions are compared to empirical landscapes. It is shown that these landscapes allow, at least with respect to the properties considered, for a rough classification into two types. In the biologically relevant limit, these two types of landscapes show very different behavior. While empirical landscapes cannot be classified on purely theoretical grounds, some of the known landscapes are classified here on a case-by-case basis. The study of theoretical models leads to consider sequences of record events of independent but non-identically distributed random variables and correlations between these record events. While these considerations appear at first to be outside the main direction of this thesis, they lead to a record-base tool of data analysis that is applied to one of the data sets considered here