As I said, any mutation rate over 1pipg seems to be forcing the population from the optimal peak
It's difficult to visualize a realistic "fitness landscape" over which evolution operates. There's a dimension for every range of variation, an unimaginable number of dimensions. We tend to think of the fitness landscape as three-dimesional, two dimensions of variation that the third of fitness, with a surface with hills and valleys showing the fitness of each combination of variation. THis is misleading in one importatn way. In the three-dimensional visualization, there are peaks from which it's impossible to go uphill and valleys from which it's impossible to go downhill.
In the real multi-dimensional case,
many "peaks" are the higher-dimensional equivalent of saddle points. That is, an organism may be as fit as possible considering variation of one characteristic but ther may be higher fitness possible by varying some other characteristic.
Another factor is that
the fitness landscape changes with time. Even in our three-dimensional visualization, peaks and valleys move so that what was optimum yesterday may no longer be optimum tomorrow.
Finally,
the peak may be local. There is often a nearby peak of higher fitness, and a sufficiently large jump can cross the vally between peaks.