A revival of the landscape paradigm: Large scale data harvesting provides access to fitness landscapes

T he first large-scale fitness analysis of HIV in presence and absence of antiviral drugs has been presented in a recent publication [1]. This study confirms expectations, in particular ruggedness and neutrality of the HIV fitness landscape, but it provides at the same time also unexpected results like long-range correlations and surprisingly little influence of antiviral agents on the landscape structure. Here, we make use of this occasion to present a brief account on the development of the landscape concept from a pure metaphor to an experimentally assisted tool for modeling and understanding evolution. Sewall Wright [2], one of the three great scholars of population genetics conceived the idea of a fitness landscape as a metaphor for the evolutionary process: species or subspecies are occupying local fitness optima in a rugged landscape (Fig. 1). Fitness is plotted upon genotype space, which is a high-dimensional space with individual genotypes differing in the recombination pattern of genes as elements. The metaphor was introduced as a visualization of his theory of shifting balance as a model of evolution that in a nutshell may be formulated as a process in three logical steps [3, 4]: (i) Random drift leads to semi-isolated subpopulations or demes within the global population, which are losing fitness because of accidental loss of fittest genotypes known as Muller’s ratchet [5], (ii) mass selection acts on complex genetic interaction networks and raises the fitness of subpopulations, and (iii) interdemic selection raises the fitness of the global population. Environmental change shifts the adaptive peaks on the landscape and drives evolutionary dynamics. Clearly, the landscape concept is suggestive for the Wright model and facilitates its understanding. Wright’s model of evolution has been heavily criticized by Ronald Fisher and others. Apart from the practical impossibility to measure fitness in the first half of the 20th century and the apparent lack of knowledge on the nature of genes and genotypes that rendered void any attempt to quantify fitness on landscapes, Fisher’s and Wright’s views on evolution differed in many aspects. Fisher’s concept called large population size theory assumes that (i) the ecological context of evolution is large panmictic populations rather than small subpopulations, (ii) the PETER SCHUSTER