In a population of altruists, all individuals thrive. But altruists are exploited by cheating individuals which do not perform altruistic acts but still benefit from those. In these conditions cooperation cannot easily evolve. This issue is resolved by kin recognition: altruists recognise each other through the use of a conspicuous tag. These altruists do well until cheaters acquire the signalling tag and disrupt the cooperation. But altruists using a different tag can then invade the population, followed by new cheaters. This mechanism can lead to a diversity of tags coexisting in the population. However it has not yet been applied in realistic biological systems. In this thesis, I formulated mathematical and simulation models to investigate the effect of diversity on the evolutionary dynamics in systems where different altruists compete with cheaters. In particular, I focused on organisms producing public goods, i.e. goods that can profit to the whole population. I considered two biological systems models: gynodioecious populations of plants, where hermaphrodites produce pollen that can be used by female-only individuals, and bacteria producing an iron-chelating molecule, called siderophore, that can be exploited by both producers and non-producers. I found that diversity in gynodioecious plants is dependent on population structure. In particular, I found that the maximal level of diversity occurs when the population structure does not favour altruists or cheaters. Next, I found a number of important results in siderophore-producing bacteria. By considering a detailed ecological model, I derived Hamilton's rule in a metapopulation and found that the level of cooperation in a population depends on the length of interaction between strains. Finally, I discovered a novel evolutionary mechanism generating and maintaining diversity and showed that it results from non-equilibrium mechanisms. These findings explain why cheaters appear readily in experiments but are rare in natural populations. My results demonstrate the importance of integrating ecological details in order to understand the mechanisms leading to cooperation and diversity, and will provide a basis and framework for future studies on the emergence and maintenance of diversity.

Chez les espèces pseudogames, les femelles utilisent les spermatozoïdes des mâles d'autres espèces pour activer leurs ovocytes et produire de nouvelles femelles sans utiliser l'ADN paternel. Ici, nous montrons un nouveau mode de reproduction pseudogame découvert chez l'espèce de nématode Mesorhabditis belari chez qui 8% de mâles conspécifiques sont présents. Ces mâles sont nécessaires pour féconder les ovocytes mais leurs gènes ne seront transmis qu'à leur fils et non à leurs filles. En effet, les femelles produisent deux types d’ovocytes, un type amphimixique où les ADN paternel et maternel constitueront le génome du zygote qui deviendra un mâle et un type gynogénétique ou l’ADN paternel déclenchera le développement du zygote mais où seul l’ADN maternel constituera le génome du zygote qui deviendra une femelle. Ainsi, la production de mâles n'a pas d'incidence sur la diversité génétique des femelles. Par une approche de théorie des jeux, nous avons montré que la production des mâles dans une faible proportion constitue une stratégie évolutivement stable seulement si les fils sont plus enclins à se reproduire avec leurs sœurs. Nous avons validé cette prédiction par une approche expérimentale en montrant une préférence entre les frères et les soeurs pour la reproduction.En parallèle, nous nous sommes intéressés aux bases moléculaires et cellulaires soutenant un tel mode de reproduction. Dans les embryons amphimixiques, la méiose femelle produit un pronoyau haploïde et la diploïdie est restaurée grâce au pronoyau paternel également haploïde. Dans les embryons gynogénétiques, l’ADN paternel ne se décondense pas, la méiose femelle est incomplète et produit un pronoyau diploïde afin de maintenir la diploïdie du zygote. Nous avons donc étudié le développement précoce des embryons de M. belari, et essayé de comprendre le type de déterminisme du sexe présent chez cette espèce / In pseudogamous species, females use the sperm of males from another species to activate their oocytes and produce females, without using the sperm DNA. Here we report a novel reproductive strategy found in the pseudogamous nematodeMesorhabditis belari, which produces its own males at low frequency. We find that the 8% of M. belari males are necessary to fertilize all oocytes but pass on their genes only to males, and never to females. Thus, the production of males has no impact on the genetic diversity of females. Using game theory, we show that the production of males at low frequency constitutes an efficient strategy only if sons are more likely to mate with their sisters. We validate this prediction experimentally by revealing a mating preference between siblings. We uncover the remarkable reproductive strategy of parthenogenetic females that pay the cost of producing males while males do not spread their genes.In parallel, we tried to understand the cellular and molecular basis at the origin of such a reproductive mode. In amphimixis embryos, female meiosis produces an haploid pronucleus and ploidy is restored with the male haploid pronucleus. In gynogenetic embryos, paternel DNA don't decondense, female meiosis is incomplete leading to a diploid pronucleus in order to maintain the diploidy of the organism. We then studied the early development of the embryos of M. belari and the type of sex determinism in this species