Ongoing global change has made understanding the factors that affect adaptation and survival of populations in the context of changing environments a central problem in evolutionary biology. Special focus has been given to the probability of survival through genetic adaptation to lethal environments; a process termed evolutionary rescue. Many studies of this process, both theoretical and empirical, have been carried out over the last two decades. As a result, we now understand how a number of factors may affect the probability of population survival. However, two factors that are known to affect evolutionary responses, mode of reproduction and interspecific interaction, have received limited attention. The main aim of my work was to investigate whether and how mode of reproduction and negative interspecies interactions (competition) affect the probability of evolutionary rescue. To achieve this goal, I set up a series of selection experiments, by propagating populations of unicellular alga Chlamydomonas reinhardtii in various stressful conditions, and monitored their survival and fitness. To investigate the effect of sex in these experiments, I manipulated mode of reproduction, by constructing the experimental populations allowed to reproduce either only sexually or asexually or both. To investigate the effect of competition, I manipulated the presence of the competitor(s) in the experimental populations, by cultivating them either in presence or absence of the competitor. I first tested the effect of rate of environmental deterioration and mode of reproduction on extinction dynamics and evolutionary rescue of the experimental populations. I found positive correlation between the rate of extinctions and the rate of environmental deterioration. The experiment revealed an interaction between mode of reproduction and the rate of deterioration, manifested through significantly reduced extinction rate of sexual populations relative to asexual populations in environment deteriorating at intermediate rate. I then investigated the effect of sex and competition on the probability of evolutionary rescue, by propagating the experimental populations in environment deteriorating in a simple way (the change comprising a single abiotic factor) and complex way (the change of both abiotic and biotic factors). I found the negative effect of competition on the probability of evolutionary rescue, and beneficial effect of sex in both types of environmental deterioration, reflected in higher number of rescued populations relative to asexual group. I then tested whether phylogenetic relatedness between a competitor and the focal species and the extent of their ecological similarity affect the likelihood of evolutionary rescue, by subjecting the experimental populations to the presence of 10 different competitors, isolated from two different types of habitats, and each being positioned on a different branch of the phylogenetic tree of Chlamydomonas genus. The probability of evolutionary rescue was contingent on the identity of a competitor species, but the results showed no significant effects of phylogenetic relatedness and ecological similarity. Finally, I investigated which experimental factors could potentially select for the long-term maintenance of sex, by subjecting the experimental populations to different types of selective environments (directional and fluctuating change of abiotic factors, the presence of the competitor) and monitoring the frequency of sex over the course of time. No selective environment significantly increased the rate of sex in the experimental populations. In contrast, I found reduction in frequency of sex in the populations subjected to fluctuating environmental change. My results demonstrate that both mode of reproduction and competition affect the probability of evolutionary rescue, which is generally positively affected by sex and negatively affected by competition. However, these general effects may be altered by other factors, namely mode of environmental change and the identity of the competitor species.

Évolution expérimentale aux limites de l'extinction : Le cas de Pseudomonas fluorescens soumise au stress. Les stress environnementaux : ces perturbations d'origine biotiques ou abiotiques des conditions de vie qui ont un impact sur les populations par l'altération des capacités de reproduction et/ou de survie des organismes, peuvent parfois mener une espèce à l'extinction si celle-ci n'est pas en mesure de s'adapter suffisamment rapidement. Avec l'augmentation de l'impact des activités humaines, le taux d'extinction des espèces est aujourd'hui estimé être de plusieurs ordres de grandeur au dessus du taux d'extinction basal. Il est donc important de comprendre les facteurs écologiques et évolutifs, à l'échelle de l'espèce comme de la communauté, qui déterminent la survie ou l'extinction d'une espèce soumise à un stress sévère.Au cours de cette thèse, nous utiliserons les outils de l'évolution expérimentale avec la bactérie Pseudomonas fluorescens soumise à un stress antibiotique, ainsi que des outils théoriques, pour étudier les facteurs déterminants les conditions d'un sauvetage évolutif : événement d'évolution rapide d'une population en situation d'effondrement démographique sous l'effet d'un stress létal. Nous étudierons aussi le rôle d'un stress plus modérés : une carence en ressource, sur deux aspects de la structure des communautés liés à la stabilité de ces dernières face aux perturbations environnementales : La topologie d'un réseau trophique bactérie-phage, ainsi que la distribution des tailles de populations, décrite par la loi de Taylor, pour des populations bactériennes en mono-culture ou en compétition sur un gradient de ressource. Nos travaux ont permis de mieux comprendre le sauvetage évolutif, en montrant notamment le rôle déterminant de la taille des populations et de leur diversité génétique initiale dans la capacité d'adaptation à un stress létal. Mais aussi en proposant une meilleure description des dynamiques de populations typiques de cette situation. Nous avons aussi montré la capacité d'une limitation des ressources à altérer la structure des réseaux trophiques et ainsi leur stabilité face aux perturbations. / Experimental Evolution at the Edge of Extinction, The Case of Pseudomonas fluorescens in Stressful Environments.Environmental stresses : the biotic or abiotic perturbations of life conditions which have an impact on populations through impaired organism's ability to survive and/or reproduce, can sometimes result in species extinction if rapid evolutionary adaptation doesn't occur. With the increasing impact of human activities, the rate of species extinction is now estimated to be several orders of magnitudes above the background rate. It is therefore crucial to understand the ecological and evolutionary factors, at species and community levels, which determine the survival or extinction of species exposed to severe stresses. All along this thesis, we will use experimental evolution tools with the bacterium Pseudomonas fluorescens exposed to antibiotic stress, as well as theoretical tools, to study the factors driving evolutionary rescue : the rapid evolutionary adaptation occurring in populations experiencing demographic collapse under a lethal stress. Then we will study how a moderate stress (low resource availability) drives different aspects of communities structure related to their stability against environmental perturbations : The topology of a bacteria-bacteriophage trophic network, and the distribution of population sizes along a resource gradient, described by Taylor's law, for bacterial populations either in mono-culture or in competition. Our results led to a better understanding of evolutionary rescues. In particular, we show the crucial role of initial population size and genetic diversity in the ability to evolve adaptation to an initially lethal stress. And we refined the description of populations dynamics in such cases. We show also how resource limitation can disturb trophic network structure, resulting in a lower stability against environmental perturbations.

La capacité de persistance d’une population face à un changement environnemental stressant est une question complexe à l’interface entre l’écologie et l’évolution. Le processus par lequel une population échappe à l’extinction en s’adaptant aux nouvelles conditions environnementales stressantes est nommé sauvetage évolutif. Ce cas particulier de dynamique éco-évolutive est de plus en plus étudié autant théoriquement, qu’expérimentalement, entre autres dans le contexte des changements environnementaux d’origines anthropiques. Cependant, les études modélisant ce processus négligent les interactions entre génotypes et environnements impactant le potentiel évolutif des populations faisant aux changements environnementaux. Dans le cadre de cette thèse, j’ai développé des modèles intégrant ces interactions. Pour cela, j’ai modélisé le processus de sauvetage évolutif de populations à reproduction asexuée, face à des changements environnementaux abruptes, en utilisant le paysage adaptatif de Fisher (modèle géométrique de Fisher (1930)). Ce paysage nous a permis de modéliser ces interactions génotypes-environnement et leur impact sur la proportion de mutations pouvant sauver une population. A travers deux modèles, considérant soit le sauvetage d’une population par une mutation d’effet fort, soit par un grand nombre de mutations d’effets faibles, nous avons pu dégager des prédictions pour la probabilité de sauvetage évolutif en fonction des conditions environnementales et des caractéristiques de l’organisme étudié. Ces modèles peuvent être paramétrés sur des données d’évolution expérimentale et leurs prédictions comparées à des données de traitement antibiotiques visant des pathogènes asexués. Au-delà du sauvetage évolutif, les modèles développés nous ont également permis d’établir des outils permettant de modéliser d’autres dynamiques éco-évolutives, intégrant des interactions génotype-environnement et leurs effets sur la distribution d’effets des mutations. / The persistence ability of a population facing a stressing environmental change is a complex question at the connection between ecology and evolution. The process by which a population avoid extinction by adapting to the new stressing environmental conditions is termed evolutionary rescue. This particular case of eco-evolutionary dynamic is increasingly investigated both theoretically and experimentally, among other things in the context of the environmental changes from human activity. However, the studies modelling this process neglect the interactions between genotypes and environments impacting the evolutionary potential of the populations facing environmental changes. In the context of this thesis, I developed models integrating these interactions. To this end, I modelled the process of evolutionary rescue in asexual populations, facing abrupt environmental changes, using the adaptive landscape of Fisher (Fisher’s geometric model (1930)). This landscape allowed us to model the genotypes-environments interactions and their impact on the proportion of mutations able to save a population. Using two models, considering either the rescue of a population by a mutation of strong effect, either by a large number of mutation of small effect, we derived predictions for the probability of evolutionary rescue, which depends on the environmental conditions and the characteristics of the studied organism. These models can be parametrized on data from evolutionary experiments and their predictions compared to data of antibiotic treatments aiming on asexual pathogens. Beyond evolutionary rescue, the models developed in this thesis also gave tools to model other eco-evolutionary dynamics, integrating genotype-environment interactions and their effects on the distribution of mutations effects.