Faculté des sciences

Evolutionary conflicts and ecological constraints at reproduction in the dioecious plant Silene latifolia

Burkhardt, Anne ; Bernasconi, Giorgina (Dir.) ; Benrey, Betty (Codir.)

Thèse de doctorat : Université de Neuchâtel, 2009.

Les fleurs sont d’une étonnante diversité de formes et de couleurs. Les plantes ayant des sexes séparés (dioécie) sont d’un intérêt particulier car différentes pressions de sélection peuvent s’exercer sur les deux sexes, amenant à l’évolution d’un dimorphisme sexuel des fleurs. De nombreuses études ont montré que les pollinisateurs favorisaient les plantes à grandes ou... Plus

Ajouter à la liste personnelle
    Résumé
    Les fleurs sont d’une étonnante diversité de formes et de couleurs. Les plantes ayant des sexes séparés (dioécie) sont d’un intérêt particulier car différentes pressions de sélection peuvent s’exercer sur les deux sexes, amenant à l’évolution d’un dimorphisme sexuel des fleurs. De nombreuses études ont montré que les pollinisateurs favorisaient les plantes à grandes ou nombreuses fleurs, exerçant ainsi une sélection sur les fleurs. Le succès reproducteur d’une plante ne dépend pas seulement de l’attraction des pollinisateurs, mais aussi de la dynamique réelle du dépôt de pollen et de la réceptivité des structures femelles. De plus, les ennemis des fleurs affectent le succès reproducteur des plantes, et pourraient aussi influencer l’évolution des traits floraux et leur dimorphisme. J’examinai les effets de la taille des fleurs de Silene latifolia (lignées de plantes sélectionnées pour de larges fleurs (LF) ou de petites fleurs (SF)) et de l’avortement des fruits (mécanisme de défense de la plante) sur le succès reproducteur du pollinisateur et prédateur de graines Hadena bicruris. Les fruits des plantes LF contenaient significativement plus de graines, et de plus grandes larves en émergèrent comparativement aux plantes SF. L’avortement réduisit significativement la taille des larves et le temps passé dans le fruit. Les larves s’établirent plus souvent chez les plantes LF, lesquelles avaient aussi moins tendance à avorter que les plantes SF. J’étudiai la sélection exercée sur les fleurs et l’effet du nombre de fleurs sur le succès reproducteur de la plante. Les plantes ayant un plus grand nombre de fleurs étaient favorisées, mais la prédation des graines ne contribua pas à cette sélection. La production de graines, le nombre d’oeufs pondus par l’insecte, et les proportions de fruits attaqués ou avortés ne différèrent pas de manière significative entre les lignées. Les plantes SF élevèrent significativement plus de larves, et perdirent plus de fruits par prédation et avortement que les plantes LF. L’avortement était contre sélectionné, et le prédateur de graines y contribua significativement chez les plantes SF mais pas LF. J’examinai l’effet des traits au niveau d’une fleur sur la diversité génétique des graines à l’intérieur d’un fruit. Je montrai que le pollen du mâle déposé en premier obtint une proportion significativement plus grande de paternité que celui d’un mâle déposé plus tard, et cet effet était visible bien avant l’arrivée des tubes polliniques à l’ovaire. Les stigmates se fanèrent simultanément avec la croissance des tubes polliniques. Un partage de paternité plus équilibré impliqua une réduction de la masse de chaque graine. Ni la surface du stigmate ni la taille de l’ovaire n’étaient corrélées avec le succès relatif de paternité des deux mâles. En conclusion, l’évolution du dimorphisme sexuel chez les fleurs de Silene latifolia pourrait être due à une sélection par les pollinisateurs pour un plus grand nombre de fleurs chez les deux sexes, et à une sélection par le prédateur de graines pour un plus petit nombre de fleurs chez les femelles en raison des coûts de l’avortement. Ce dernier, bien qu’étant préjudiciable tant au développement de la larve que pour le succès reproducteur de la plante, pourrait diminuer la perte de ressources pour la plante s’il empêche les larves d’attaquer d’autres fruits. Au niveau de la fleur, seule une petite fenêtre temporelle permet le succès de paternité, et le partage de paternité que l’on observe sur le terrain est probablement le résultat du dépôt simultané de pollen de plusieurs mâles ou de la visite de plusieurs pollinisateurs durant un temps court. Les prochaines études sur l’évolution des traits floraux des plantes ne devraient pas se limiter aux pollinisateurs mais devraient simultanément considérer les ennemis naturels ainsi que les changements plus subtiles qui ont lieu au niveau des fleurs.
    Summary
    Flowering plants show an astonishing diversity of flower shapes and colours. Among plant reproductive systems, dioecy is of particular interest because each sex can be subjected to independent selection pressures that may lead to the evolution of different floral traits in the two sexes (sexual dimorphism; chapter 1). On the one hand, many studies have confirmed the role of pollinators as selective agents on floral traits. Pollinators are attracted by plants with large flowers and with many open flowers (large floral display), and have been shown to contribute to selection on these traits. Plant fitness via pollination will not only depend on the probability of attracting pollinators but also on the actual dynamics of pollen deposition and reception, which are influenced by the floral phenology and characteristics of the receptive surfaces. On the other hand, natural enemies of flowers can directly affect plant fitness, and may also influence the evolution of floral traits and their dimorphism. Some plants are pollinated by insects that also act as natural enemies by consuming part of the seeds produced (nursery pollination). To reduce the cost of seed predation, female plants may use fruit abortion as a defence mechanism. However defences can be costly themselves, and defence costs may contribute to selection on floral traits. Thus pollinators and seed predators may select for opposite optima in floral traits in male and female plants. In this thesis I investigate the effect of the pollinating-seed predator Hadena bicruris on the fitness of the dioecious plant Silene latifolia, the contributions of pollinators and seed predators to selection on sexually dimorphic floral traits, the role of fruit abortion in mediating floral trait selection, the effect of flower size and fruit abortion on the insect performance, and at the flower level, the effect of floral phenology and characteristics of the receptive surfaces on the fitness of male and female plants. In chapter 2, I investigated the effects of flower size, seed provisioning and fruit abortion on the fitness of larvae of the pollinating-seed predator. Using selection lines that varied in flower number and size (large- vs. small-flowered plants), I infested two flowers per plant with eggs of H. bicruris, measured the growth of the resulting larvae, fruit abortion, and seed provisioning of one non-attacked fruit. Fruits from large-flowered plants (LF) contained significantly more seeds and more nutrients, and gave rise to significantly larger larvae than small-flowered plants (SF). Fruit abortion had a dramatic effect on larval growth, reducing significantly the mass and time at emergence of larvae from the fruit. Also the success of larval establishment was higher on LF plants and these plants had a lower probability of fruit abortion (significant for the second fruit). Thus fruit abortion is detrimental for larval development- and likely reduces the amount of resources lost by the plant, and therefore may help to stabilize this plant-nursery pollinator mutualism. In chapter 3, I investigated the effect of varying floral display and flower size on plant fitness, and pollinator and seed predator selection on floral traits. I conducted two common garden experiments. In, the first experiment, I exposed plants from SF and LF selection lines to naturally occurring pollinators and seed predators, and recorded fruit predation, abortion and seed production, and selection on floral traits. In the second experiment, I measured the risk of infestation by eggs of H. bicruris on plants that displayed only flowers. I found positive total selection but no seed predator selection on flower number. Total seed production, number of eggs received, and proportions of predated or aborted fruits did not differ significantly between lines. SF plants reared significantly more larvae, tended to have a larger parasite load, and lost more fruits due to predation and abortion than LF plants. Interestingly, fruit abortion was negatively selected, and seed predator selected against abortion in the SF but not in the LF plants. Fruit abortion was significantly more common in plants with high parasite load. While pollinators or non-ovipositing H. bicruris may select for large floral display in males, the seed predator may contribute indirectly to selection for small floral display in females by imposing higher abortion costs to plants with a large floral display. In chapter 4, I investigated how traits at the flower level affect the within fruit genetic diversity, a component of male and female fitness. As shown in a field survey fruits of S. latifolia are usually sired by multiple fathers but the mechanisms were not known. In greenhouse experiments I studied the effect of timing between hand-pollination with the pollen of two males on paternity and seed mass, the effect of pollen load on seed set, the time needed for pollen tubes to reach the ovary, and stigma wilting after pollen deposition. The first-arriving pollen sired significantly more seeds than later-arriving pollen, and this advantage was seen several hours before the pollen tubes could reach the ovary. The stigma papillae wilted simultaneously with pollen tube growth. A more even share of paternity between the two males resulted in a significantly lower individual seed mass. Thus multiply sired fruits of S. latifolia as usually found in the field are likely to result from simultaneous deposition of pollen from several male plants (pollen carry over) or from multiple pollinator visits within a short time interval. Male and female function may have conflicting interest over the duration of stigma receptivity, forward wilting would benefit the first-arriving pollen donor by increasing its paternity share compared to later-arriving donors, while wilting may be costly to the female function by reducing the genetic diversity of offspring and/or the number of pollen grains captured by the stigma. In chapter 5, together with Sara Teixeira, I investigated whether the size of the receptive stigma surface or the size of the ovary affect the shares of paternity between two competing pollen donors. Sara hand-pollinated the flowers of two sisters and of one unrelated female with the pollen of two males, and determined paternity. I counted the number of stigma lobes, measured stigma surface and ovary size on unpollinated flowers of these females. Flowers had four to seven stigma lobes, but most flowers had five lobes. I found significant variation between the two populations in stigmatic surface and ovary size. Those two traits were significantly positively correlated with each other, but did not correlate with the relative siring success of the two males. Thus a larger stigmatic surface apparently does not increase competition between two pollen donors, and results in similar shares of paternity. However, it may influence pollen capture under natural pollination. In conclusion (chapter 6) my results suggest that the evolution of sexually dimorphic traits in S. latifolia flowers may be due to opposing selection on male and female plants. While pollinators may select for larger display in both sexes, females would pay a larger cost to fruit abortion with increasing display size. Thus its natural enemy is likely to contribute to selection for small floral display in females. Fruit abortion seem to play an important role in this plant-insect interaction. This resistance trait is likely to be beneficial to the plant or to its offspring on the long term because it is expected to reduce the seed predator population. However on the short term, it is costly to the female plant. The nature of this cost remains to be elucidated. At the flower level, my results suggest that only a small window of time is available for a male’s pollen to obtain siring success. The role of stigma surface and wilting for pollen capture and paternity, and thus for the fitness of male and female plants under natural conditions need further investigation. Stigma surface may increase pollen capture, and wilting may be used by male or female to manipulate the outcome of pollen competition. Alternatively, wilting may serve as a defence mechanism against venereal diseases or may render the flower unattractive to ovipositing females of seed predators. Future studies on the evolution of floral traits in plants should not be limited to pollinators, but should simultaneously consider natural enemies as well as the more subtle changes at the flower level.