One of the primary problems with the adaptive landscape
concept is that only three
variables can be graphed - two
alleles and
fitness. This ignores
linkage disequilibrium, which will be another factor in determining the way in which the
population will
evolve. Another problem is that it
assumes that
mating is
random, which is often not the case.
Overall, the adaptive landscape is useful for telling us how fit a population is at a given time point corresponding to a given allelic combination. It's pretty useless when it comes to telling us what the population is going to do next. It's a shame, really, because it's not possible to work out what a population is going to do without using iterative models and continuing until equilibrium is reached. Being able to start from a point on a landscape and simply draw a line to the top of the nearest peak would make life significantly easier.
(Of course, there's the other problem that given a constant selection pressure populations will not necessarily reach the point of maximum fitness given the alleles available to them. But that's another issue.)