How Geitonogamous Selfing Affects Sex Allocation in Hermaphrodite Plants
Total Page:16
File Type:pdf, Size:1020Kb
How geitonogamous sel®ng affects sex allocation in hermaphrodite plants T. J. DE JONG, P. G. L. KLINKHAMER & M. C. J. RADEMAKER University of Leiden, Institute of Evolutionary and Ecological Sciences, PO Box 9516, 2300RA Leiden, The Netherlands Keywords: Abstract dioecy; Does the mode of self-pollination affect the evolutionarily stable allocation to geitonogamy; male vs. female function? We distinguish the following scenarios. (1) An pollination; `autogamous' species, in which sel®ng occurs within the ¯ower prior to sel®ng; opening. The pollen used in sel®ng is a constant fraction of all pollen grains sex allocation. produced. (2) A species with `abiotic pollination', in which sel®ng occurs when pollen dispersed in one ¯ower lands on the stigma of a nearby ¯ower on the same plant (geitonogamy). The sel®ng rate increases with male allocation but a higher sel®ng rate does not mean a reduced export of pollen. (3) An `animal-pollinated' species with geitonogamous sel®ng. Here the sel®ng rate also increases with male allocation, but pollen export to other plants in the population is a decelerating function of the number of simultaneously open ¯owers. In all three models sel®ng selects for increased female allocation. For model 3 this contradicts the general opinion that geitonogamous sel®ng does not affect evolutionarily stable allocations. In all models, the parent bene®ts more from a female-biased allocation than any other individual in the population. In addition, in models 2 and 3, greater male allocation results in more local mate competition. In model 3 and in model 2 with low levels of inbreeding depression, hermaphroditism is evolutionarily stable. In model 2 with high inbreeding depression, the population converges to a ®tness minimum for the relative allocation to male function. In this case the ®tness set is bowed inwards, corresponding with accelerating ®tness gain curves. If the sel®ng rate increases with plant size, this is a suf®cient condition for size-dependent sex allocation (more allocation towards seeds in large plants) to evolve. We discuss our results in relation to size-dependent sex allocation in plants and in relation to the evolution of dioecy. Introduction reduces the evolutionarily stable (ES) allocation to sons below 50%. By analogy, Charlesworth & Charlesworth In his paper on sex ratios of insects, Hamilton (1967) (1978, 1981) and Lloyd (1987a) modelled the effects of showed that, if the breeding structure of a population sel®ng on the allocation to male and female function in forces sons into competition with one another, this cosexual plants. They showed that sel®ng reduces the ES allocation to male function. These conclusions were used to explain the facts that species with high sel®ng rates show low pollen±ovule ratios (Cruden, 1977) and, within Correspondence: Dr T. J. de Jong, University of Leiden, Institute of species, genotypes with high sel®ng allocate less to pollen Evolutionary and Ecological Sciences, PO Box 9516, 2300RA Leiden, The Netherlands. production (Schoen, 1982; Lloyd, 1984; Charnov, 1987; Tel: 31±71±5275118; fax: 31±71±5274900; Parker, 1995). However, different modes of sel®ng exist e-mail: [email protected] (Lloyd, 1979; Lloyd & Schoen, 1992; Schoen & Lloyd, 166 J. EVOL. BIOL. 12 (1999) 166±176 Ó 1999 BLACKWELL SCIENCE LTD Geitonogamy and sex allocation 167 1992) and Lloyd (1987a) pointed out that the mode of self-pollination in¯uences the ES allocations. We distinguish between three modes of sel®ng. First, in Charlesworth & Charlesworth's (1978, 1981) model it is assumed that sel®ng occurs before the outcross pollen arrives and that the few pollen grains used in sel®ng do not signi®cantly reduce the pollen export of the plant. This model (model 1 in what follows) seems appropriate for an autogamous species, when the anthers and stigma touch brie¯y before the ¯ower opens. Second, geitonogamous (between-¯ower) sel®ng may occur in species with wind or water pollination. The self pollen then competes with outcross pollen. In wind- pollinated plants the architecture of the plant and physical factors such as wind speed or turbulence may determine what fraction of the pollen is dispersed (Freeman et al., 1997). Some pollen grains may land on the stigmas of neighbouring ¯owers but their numbers are probably low compared with the total pollen released. Therefore, the sel®ng rate is expected to increase with the number of open ¯owers, but the fraction of all pollen dispersed is hardly affected by the number of open ¯owers (model 2). Third, in animal-pollinated species, geitonogamy oc- curs when the pollinator moves between ¯owers. Here outcross pollen and the self pollen from neighbouring Fig. 1 Outline of the three models. In model 1 (`autogamy') the ¯owers are applied simultaneously to the stigma. Bumble sel®ng rate (S) does not depend on the number of ¯owers (n) and bees, honey bees and hummingbirds will usually tend to pollen export is a constant fraction of the pollen produced (solid line) visit more ¯owers in succession when more open ¯owers or no pollen is used for sel®ng (broken line). In model 2 (`abiotic are available (Klinkhamer et al., 1989; Snow et al., 1995). pollination') the sel®ng rate increases with the number of ¯owers, Higher sel®ng rates on plants with many ¯owers have but not at the expense of pollen export. In model 3 (`animal pollination') sel®ng also increases with the number of ¯owers. Pollen been shown for a number of animal-pollinated plants export is a saturating function of the number of open ¯owers. As a (DommeÂe, 1981; Crawford, 1984; Dudash, 1991; Schoen result of the pollen discounting an inverse relation exists between & Lloyd, 1992; Harder & Barrett, 1995a; Snow et al., the sel®ng rate and the fraction of the pollen that is dispersed to 1995; Vrieling et al., 1997 and unpublished results). other plants in the population. When a bee moves between successive ¯owers on the same plant, it constantly loses pollen. This loss occurs partly on the stigma, but more important losses are made explain this result. Next we examine how a shift to self- in ¯ight, on the corolla and through grooming (Thom- incompatibility would affect ES allocation to male func- son, 1986; Rademaker et al., 1997). Thus if the plant tion in the three models. Our calculations have assumed presents more simultaneously open ¯owers, this induces that all plants in the population are of equal size. This is pollinators to visit more ¯owers in succession and this clearly unrealistic. Plants differ greatly in size within gives more opportunities for the pollen of the ®rst visited populations, and genotypes may be phenotypically plas- ¯ower to be lost before the pollinator leaves the plant. tic, i.e. adjust sex allocation to their size. This problem Pollen export is then a decelerating function of the has been discussed for outcrossing species (Klinkhamer number of simultaneously open ¯owers (Hessing, 1988 et al., 1997). For sel®ng species, the factors affecting for Geranium caespitosum). Because of the pollen dis- ®tness gain curves have been listed (de Jong & counting (Holsinger, 1991) in model 3, a negative Klinkhamer, 1994), without reaching a de®nite conclu- correlation is expected between the sel®ng rate and the sion. Here we sketch the simplest model without fraction of the pollen exported, as shown by Harder & inbreeding depression and with the sel®ng rate increas- Barrett (1995a). ing with the number of ¯owers. Is it then adaptive to be The different modes of sel®ng and their consequences phenotypically plastic and to adjust gender to size, i.e. for the relation between the number of ¯owers, the allocate more to female function when large and more to sel®ng rate and the pollen export per ¯ower are depicted male function when small? How do wind- and animal- in Fig. 1. In this paper we use the differences between pollinated species compare in this respect? Finally, we wind- and insect-pollinated plants to show how geitono- brie¯y discuss in which model dioecy is most likely to gamy reduces male allocation to levels below 50% and to evolve. J . E V O L . B I O L . 1 2 ( 1 9 9 9 ) 1 6 6 ± 1 7 6 Ó 1 9 9 9 B L A C K W E L L S C I E N C E L T D 168 T. J. DE JONG ET AL. The general model wm 1 Sm T m 1 dSm T m Consider a hermaphrodite plant species in which seed 1 dSm T m Em=En 1±Sn T n: production is not pollen limited. Each individual plant 1b has T units of resource. Plants may allocate this resource freely between seeds and ¯owers, so that there exists a wn can be scaled to 2 and represents the number of copies complete trade off or full `compensation' (Charnov, of the (haploid)genome that an individual passes on to 1982; Lloyd, 1987a; Seger & Eckhart, 1996). For the next generation through seeds. The ®rst term in simplicity, assume that one unit of resource is required eqn 1b denotes the female ®tness contribution through to construct a ¯ower with pollen, but without seed, and outcrossing. The second term is the female ®tness that a single seed also costs one unit of resource. We contribution through sel®ng. The third term is the male assume that plants can vary the ratio of seeds to ¯owers, ®tness contribution through sel®ng. We write the second but that each ¯ower contains the same amount of and third terms separately to be able to discriminate pollen. The alternative is, of course, to assume that between the female and male contribution to ®tness later pollen production per ¯ower varies but this poses further on in this paper.