A TIME COMMITMENT HYPOTHESIS FOR SIZE-DEPENDENT GENDER ALLOCATION.

Abstract

In many species of plants, gender expression is correlated with size. Usually large plants are than small plants (Policansky 1981, 1987; Bierzychudek 1982, 1984a,b; Lovett Doust and Cavers 1982; Schlessman 1987, 1988, 1991; Pickering and Ash 1993), but the reverse is sometimes seen in wind pollinated species (see Burd and Allen 1988; de Jong and Klinkhamer 1994; Lundholm and Aarssen 1994). Models that predict evolutionarily stable sex ratios have been developed to explain this phenomenon (Freeman et al. 1980; Charnov 1982, ch. 4, 16; Lloyd 1983; Lloyd and Bawa 1984; Charnov and Bull 1985; Schlessman 1988; de Jong and Klinkhamer 1989, 1994; Bulmer 1994, ch. 10), and the majority of these are derived from Ghiselin's (1969) size-advantage hypothesis for sequential hermaphroditism in animals. This hypothesis postulates that the sexes each have different reproductive successes (RS) at different sizes, and that the sex whose relative rate of gain of RS with size is greatest should be expressed by large individuals. For example, if females gain RS more quickly with size than males, then large individuals should be female. Many of these models predict an abrupt transition from all male to all female allocation at a threshold size (i.e., diphasy; e.g., see Charnov 1982, ch. 4, and Schlessman 1988 for a review). When applying this hypothesis to flowering plants it is usually suggested that females have a higher energetic cost of reproduction relative to males as a result of the resource requirements for seed and fruit development (Bierzychudek 1984b; Lloyd and Bawa 1984; Schlessman 1988, 1991; de Jong and Klinkhamer 1994). Larger plants are believed to have greater resources or resource gathering potential and consequently, the increase in RS with size is greatest for females. Here we discuss a simple alternative evolutionary explanation that is frequently overlooked. Much available evidence suggests that mortality rate (per unit time) in plants is strongly correlated with size (Harper 1977; Gross 1981; Mithen et al. 1984; Silvertown and Lovett Doust 1993). The smallest plants in a population inevitably sustain the highest mortality rates, with mortality often declining substantially as size increases. Additionally, the female function of plants usually requires a longer reproductive time commitment relative to the male function as a result of the time required for fruit production. These two factors combine to create a selective regime promoting the evolution of a positive correlation between femaleness and size. The reasoning is simple. Suppose the smallest plants have the highest mortality rate. Then because the probability of dying early in the season is high when small, often there will not be sufficient time for female reproduction. The likelihood of reproductive success when small will be greater for males than females because pollen production and dispersal occur relatively fast. Thus, the female function will gain the most from an increase in size. Under this time-commitment hypothesis one would expect natural selection to favor the gender allocation rule, if small be male, if large be female. Although this prediction is the same qualitative one made by models based on energetic costs, it is not widely appreciated that this alternative explanation exists. Given that it is based on demographic and reproductive characteristics that appear to be quite common, we suggest that it deserves further attention. Additionally, we suggest that this time-commitment hypothesis can also explain instances of environmental sex determination (Charnov and Bull 1977), whereby plants in good quality sites are more female than plants in poor quality sites. From the above verbal argument however, it is difficult to determine whether diphasy or a graded gender transition is expected. Consequently, we present a simple mathematical model to address this question, and to clarify the conditions under which our verbal argument is valid. The model also allows a more detailed prediction of the expected pattern of gender expression and therefore suggests potential ways in which this hypothesis might be tested empirically.