2-D Adaptive Landscape - Background
The adaptive landscape, invisioned by Sewell Wright, considers simultaneously the value of a trait and the fitness of individuals possesing that trait value. The function describing the fitness of the trait for a given value can be thought of as a landscape, with peaks being trait values of high fitness and valleys being trait values of low fitness. The trait value with the highest fitness is located at the global peak. All peaks other than the global peak are termed local peaks.
Since selection tends to increase the average fitness in a population, the mean trait value will tend to move towards the nearest local peak on the adaptive landscape.
Selection alone will not drive a population across a low fitness valley in order to reach a peak with a higher maximum. However, stochastic processes, such as mutation or genetic drift, may allow a population to cross a low fitness valley. Selection may then drive the population mean towards the nearest peak.
Another force may also allow a population to explore new areas of the adaptive landscape: a changing environment. When the adaptive landscape is constantly changing due to a changing environment, selection will still drive the population towards the nearest local peak, but that nearest peak may be changing through time. In this type of system, the population may be constantly moving, without ever reaching a fitness peak.