Selection caused by self‐fertilization

The generally held view that increased self‐fertilization should be advantageous in the absence of counteracting selective forces reducing viability or fertility is reexamined. It is pointed out that the models on which this view is based all imply a gain in male (pollen) fertility with increased selfing. Hence, the postulated advantage may equally well be due to increased fertility, a fact which reopens the discussion on the selective significance of differential selfing. A new model for differential self‐fertilization is presented which avoids built‐in fertility selection by explicitly considering pollen available for self‐ and for cross‐pollination. Plant types are distinguished with respect to their amounts r i of pollen available for self‐pollination, and these types are assumed to be identical with respect to their pollen and ovule fertilities. Moreover, the efficiency of cross‐pollination is allowed to depend, for example, on population density, thus giving rise to a parameter b called “crossing potential”, while the efficiency of self‐pollination is described by a parameter a called “selfing potential”. These parameters may be conceived of as ecological parameters. Increasing r i produces a simultaneous increase in each of the four measures of self‐fertilization (introduced in Part I of the present series) irrespective of the values of the ecological parameters. It is then shown that increased selfing can be both advantageous and disadvantageous in terms of fitness, dependig on the ecological parameters as well as on the mode of self‐fertilization (i.e. where selfing occurs before or after outcrossing). The main result is, roughly, that for both selfing modes high crossing and low selfing potential favour increased cross‐fertilization, while the reverse favours increased self‐fertilization. However, the regions for a and b in which this holds true differ substantially for the two selfing modes. In the complements of these regions strange conditions for the evolution of increased selfing or outcrossing, respectively, exist. The significance of these results for explaining experimental observations is discussed.