POPULATION STRUCTURE AND MORPH-SPECIFIC FITNESS DIFFERENCES IN TRISTYLOUS LYTHRUM SALICARIA.

In tristylous plant populations, style-morph frequencies are governed by an interaction between frequency-dependent selection due to disassortative mating and stochastic processes. Provided that there are no inherent fitness differences among morphs, frequency-dependent selection should result in equal morph frequencies at equilibrium. Stochastic models indicate that the short-styled morph has the highest and the long-styled morph the lowest probability of being lost from local populations as a result of random processes. We surveyed the morph composition of 82 populations of the tristylous, self-incompatible herb Lythrum salicaria in two archipelagos, one in central and one in northern Sweden, located close to the range-margin of the species. To examine whether deviations from even morph frequencies can be explained by among-morph differences in reproductive success, we quantified flower and seed production in six and three populations in the northern and southern archipelago, respectively, and we recorded segregation ratios in offspring produced in six trimorphic populations in the northern area. Seed germination and offspring growth were studied in the greenhouse. Ninety percent of the populations in the southern archipelago (N = 31) and 69% of the populations in the northern archipelago (N = 35) were trimorphic; the remaining populations were dimorphic (only populations consisting of at least three flowering plants considered). Dimorphic populations were smaller than trimorphic populations, as predicted by stochastic models. There was a striking difference in the morph composition of L. salicaria populations between the two archipelagos. In the southern archipelago, there was a slight excess of the long-styled morph and a corresponding deficiency of the short-styled morph. In contrast, the northern populations were characterized by a marked deficiency of the mid-styled morph: the average frequency of the mid-styled morph in trimorphic populations was 0.21, and nine of eleven dimorphic populations lacked the mid-styled morph. In both archipelagos, the long-styled morph (the most common morph) produced about 20% fewer seeds per fruit than the other morphs. The long-styled morph also tended to produce fewer seeds per plant. A hand-pollination experiment performed in two of the northern populations indicated that seed production per flower was pollen-limited in the long-styled morph but not in the other two morphs. Seed germination and offspring size after 24 weeks of growth did not differ among morphs. The mid-styled morph tended to have a higher representation in the offspring than in the parental generation in all six trimorphic populations studied further indicating that the deficiency of the mid-styled morph in the northern archipelago does not represent an equilibrium. Taken together, the results do not support the hypothesis that morph-specific differences in reproductive success can account for deviations from even morph frequencies in L. salicaria. It is suggested that among-morph differences in other components of fitness and historical factors may contribute to the current morph structure.