|
|
|
|
|
|
|
result of contact between previously isolated, locally adapted populations is manifest. (See, however, the comment on page 166.) There is much to be learned about these processes, particularly with reference to schemata or coadapted sets. (Some of the most interesting work to date has been carried out by A. Brues 1972.) It is clear that the addition of migration rules to reproductive plans affords a sophisticated approach to spatially inhomogeneous environments, but we need to know a great deal more about the efficiency and robustness of such an approach (paralleling the development of chapters 5 and 7 for the homogeneous case). |
|
|
|
|
|
|
|
|
So far we have been discussing spatial inhomogeneity of payoff, but temporal inhomogeneity or nonstationarity is an even more difficult problem. There are four points at which the results of this book have a bearing on such problems. First, and most obvious, the rapid response of reproductive plans, exhibited concretely in the studies of Cavicchio (1970) and Hollstien (1971), permits "tracking" of the changing payoff function. As long as the payoff function changes at a rate comparable to the response rate, overall performance will be good. The proportions of schemata in the population will change rapidly enough to take advantage of current features of the environment. As a second point, it should be noted that the rank bestowed on a schema (its proportion in successive generations) is the geometric mean of the observed averages  (see Lemma 7.2). Thus more rapid fluctuations will favor schemata which exhibit the best (geometric) mean performance when subjected to the fluctuations. Third, if there are repetitive (not necessarily cyclic) features over time, dominance change provides a mechanism for retaining useful schemata when the features are not in force (see pages 115-16). By occasionally (say once every few generations) giving recessive status to instances of currently favored schemata, they can be reserved against adverse environmental configurations. In particular, these recessive instances have a much reduced testing rate (see page 115). As a result the recessive versions are relatively unaffected by environmental changes which quickly eliminate the dominant version. By occasionally returning an instance of a recessive schema to dominant status it can be tested against the current environmental configuration. If the dominant instance achieves above-average performance it will reproduce rapidly, producing an increasing proportion of dominant instances in the population. (If the performance is below average the newly dominant instance will quickly disappear, at no great cost to the efficiency of the adaptive plan.) Finally, by making the intrachromosomal duplication of a schema x subject to the disappearance of an environmental feature currently exploited by x, the effective mutation rate of x can be increased. For example, let the schema x be associated with a sensor (see pages 153-54) which detects the environmental feature exploited by x. Let intra- |
|
|
|
|
|