* , Absolute value of the t test for sample mean; , approximate critical value of t distribution; > indicates that mean values significantly differ at the 5% level.
the fact that the topology of fitness resulting from performing two tasks, mechanical stability and reproduction, contains a single phenotypic optimum (Figure 4A) that dramatically depresses the mean and inflates the SE for the mean number of optima. Nonetheless, the largest number of optima observed among all simulated landscapes is attained when fitness is quantified in terms of all three tasks, suggesting that the hypothesis would have been accepted had walks been simulated in landscapes for which fitness was defined in terms of other biological tasks in addition to light interception, mechanical stability, and reproduction.
It is instructive to compare the fitness of phenotypic maxima with the fitness of phenotypic optima. Although every simulated walk is permitted to reach all the maxima or optima in a particular landscape, the elevation of peaks (maxima or optima) differs from one landscape to another because fitness is defined in different terms in each landscape. Because phenotypic maxima and optima occupy fitness peaks, their absolute fitnesses define a landscape's elevation and, therefore, the gradient of the phenotypic transformations attending a walk. In theory, the magnitudes of the fitness of phenotypic maxima are greater than the magnitudes of the fitness of phenotypic optima. Therefore, the observation that the fitness of phenotypic maxima is greater than that of phenotypic optima is somewhat trivial. What is not unimportant, however, is that the "currency" in which fitness is measured differs among landscapes—that is, fitness was measured in different units (e.g., quanta of light absorbed, probability of mechanical failure, distance of spore dispersal). Therefore, comparisons among the elevations of different landscapes can be made only in relative, rather than absolute, terms. The ratio of the fitness of a phenotypic maximum (or optimum) and the fitness of the ancestral phenotype is useful because it normalizes the elevation of peaks and can be used to crudely compare the topologies of very different landscapes.