<p>I combined comparative and experimental studies in the field and in the greenhouse to examine factors influencing reproductive success in two insect-pollinated herbs, the tristylous, selfincompatible perennial <i>Lythrum salicaria</i> and the self-compatible, biennial <i>Pedicularis palustris</i>. More specifically, I explored (i) the effects of plant population characteristics on the intensity and outcome of interactions with pollinators and seed predators, (ii) whether flower morphology affects duration of stigma receptivity, and (iii) whether damage-induced reduction in floral display reduces pollinator visitation and increases pollen limitation.</p><p>As predicted, the rate of pollinator visitation tended to increase and the number of flowers probed per plant tended to decrease with increasing population size in <i>L. salicaria</i>, but these relationships only approached statistical significance. By taking advantage of the pollen size polymorphism that is typical of many heterostylous plants, I could show that the number of compatible pollen grains received increased with population size, and that this was associated with a reduction in pollen limitation and increased seed output per flower. The deposition of high numbers of incompatible conspecific and heterospecific pollen grains did not appear to reduce seed set.</p><p>In <i>P. palustris</i>, fruit set and seed predation varied markedly among populations and years, but this variation could only partly be explained by variation in population size. Fruit set was positively related to population size, and seed predation was negatively related to population size, in one of three years. Similarly, the level of pollen limitation, which was quantified in two years, varied among populations, but was not related to population size, density or isolation.</p><p>In <i>L. salicaria</i>, both the duration of stigma receptivity and the effect of prior self-pollination on seed output varied among style morphs. These differences may contribute to morph-specific differences in pollen limitation and seed production documented in the field.</p><p>The results of a field experiment demonstrated that damage to the shoot apex may markedly reduce fruit production in <i>L. salicaria</i>, and suggested that this is mainly because damage reduces flower number. I found no evidence that a reduced floral display increased the severity of pollen limitation.</p><p>Taken together, the results show that interactions with both pollinators and herbivores may markedly affect reproductive output in the two plant species, and that the intensity of both mutualistic and antagonistic interactions vary considerably in space and time. Moreover, they are consistent with the hypothesis that pollination success should depend less on population size in selfcompatible than in self-incompatible plants.</p>

I combined comparative and experimental studies in the field and in the greenhouse to examine factors influencing reproductive success in two insect-pollinated herbs, the tristylous, selfincompatible perennial Lythrum salicaria and the self-compatible, biennial Pedicularis palustris. More specifically, I explored (i) the effects of plant population characteristics on the intensity and outcome of interactions with pollinators and seed predators, (ii) whether flower morphology affects duration of stigma receptivity, and (iii) whether damage-induced reduction in floral display reduces pollinator visitation and increases pollen limitation. As predicted, the rate of pollinator visitation tended to increase and the number of flowers probed per plant tended to decrease with increasing population size in L. salicaria, but these relationships only approached statistical significance. By taking advantage of the pollen size polymorphism that is typical of many heterostylous plants, I could show that the number of compatible pollen grains received increased with population size, and that this was associated with a reduction in pollen limitation and increased seed output per flower. The deposition of high numbers of incompatible conspecific and heterospecific pollen grains did not appear to reduce seed set. In P. palustris, fruit set and seed predation varied markedly among populations and years, but this variation could only partly be explained by variation in population size. Fruit set was positively related to population size, and seed predation was negatively related to population size, in one of three years. Similarly, the level of pollen limitation, which was quantified in two years, varied among populations, but was not related to population size, density or isolation. In L. salicaria, both the duration of stigma receptivity and the effect of prior self-pollination on seed output varied among style morphs. These differences may contribute to morph-specific differences in pollen limitation and seed production documented in the field. The results of a field experiment demonstrated that damage to the shoot apex may markedly reduce fruit production in L. salicaria, and suggested that this is mainly because damage reduces flower number. I found no evidence that a reduced floral display increased the severity of pollen limitation. Taken together, the results show that interactions with both pollinators and herbivores may markedly affect reproductive output in the two plant species, and that the intensity of both mutualistic and antagonistic interactions vary considerably in space and time. Moreover, they are consistent with the hypothesis that pollination success should depend less on population size in selfcompatible than in self-incompatible plants.