Previous Studies

(1) Evolutionary study on the floral differentiation between two varieties of Impatiens hypophylla

In various taxa of animal-pollinated plants, speciation accompanied with differentiation in flower size is widely observed. However, major selective forces causing this pattern have not been fully understood. A hopeful approach for this subject is to compare a pair of closely related taxa, which differ in flower size. It is highly desirable that such comparison could be made between intraspecific taxa with similar ecological characteristics except for floral differentiation. Impatiens hypophylla var. hypophylla and var. microhypophylla provide an excellent opportunity for this approach because they differ only in flower size and color, occur in neighboring regions, and share flowering phenology entirely overlapped. The aim of my study is to examine the factors that could have contributed to the floral differentiation between the varieties of I. hypophylla.

Following results were obtained. 1: I and T. Yahara demonstrated that a trade-off between flower number and investment per flower constrained the evolution of flower size in I. hypophylla (Sato & Yahara 1999). 2: Two varieties exhibited markedly different flowering behaviors under variation of outcross pollen availability. Based on this finding, I suggest that the evolution of autogamous self-pollination in I. hypophylla induced differences both in the selfing coefficient and in floral longevity between the varieties (Sato 2002).

Impatiens hypophylla var. hypophylla

Impatiens hypophylla var. microhypophylla

(2) Theoretical analysis of evolution of sexual expression in animal-pollinated plants

The common sexual systems in animal-pollinated plants are hermaphroditism, dioecy, and gynodioecy. I attempt to explain this pattern by extending the previous models for the evolution of sexual systems into following assumptions. (1) A portion of investment in pollinator attraction can contribute only one of female and male functions, because one sexual function of a flower can be saturated with pollinator visitations earlier than another sexual function of the same flower. (2) There is trade-offs between size and number of flowers.

The analysis was conducted to determine the each condition that females and males can increase in frequency into hermaphroditic population assuming each of two different pollinator gain-curves (relationships between investment to attractive structures of a flower and frequency of pollinator's visit to the flower). Irrespective of the gain-curves, females could exclusively increase in frequency for large regions of parameter space where there is self-fertilization and high inbreeding depression. On the other hand, gain-curves gave considerable difference for the prediction of male's. When logarithm function was assumed for pollinator gain-curve, males could not increase in frequency for any regions of parameter space that I examined. However, when sigmoidal function was assumed, males could exclusively increase for large regions of parameter space, where it is likely to be occurred in many animal-pollinated plants. The latter prediction obviously contradicts the extreme rarely of androdioecy (males coexist with hermaphrodites) in nature. Therefore, I suggested that pollinator gain-curve plays an important role for explaining the patterns of sexual systems in animal-pollinated plants.

[Figure] Area of parameter space resulting in females and males can increase in frequency when they were introduced into a hermaphroditic population with ESS allocation. Pollinator gain-curve was assumed to be a sigmoidal function of investment in attractive structures. This evaluation was conducted for each of two values of a (nonlinear constraint against increasing flower size), three values of s (selfing rate), and two values of d (inbreeding depression), each presented in a separate panel.

(3) Why flowers of selfing species die sooner?

To explain why flowers generally last for shorter durations in selfing species compared with their outcrossing relatives, I and Yahara developed models that analysed the evolution of floral longevity under different modes of selfing. When autofertilization was assumed, variation in the model parameters chiefly resulted in negative correlations between floral longevity and selfing rate. This prediction is consistent with general observations. On the other hand, when facilitated selfing was assumed, variation in the model parameter did not produce any correlation between floral longevity and selfing rate. When geitonogamous-selfing was assumed, variation in parameter chiefly produced positive correlations between floral longevity and selfing rate. From these results, we suggest that negative correlations between floral longevity and selfing rate, which have been observed in self-compatible hermaphrodite species, are mediated by autofertilization.

[Figure] Examples of correlation between floral longevity and mean selfing-rate in ESS populations, when (a) three modes of autonomous self-pollination and (b) self-pollinations mediated by pollinator visitation occur.