Habitat fragmentation can have an impact on a wide variety of biological processes including abundance, life history strategies, mating system, inbreeding and genetic diversity levels of individual species. Although fragmented populations have received much attention, ecological and genetic responses of species to fragmentation have still not been fully resolved. The current study investigated the ecological factors that may influence the demographic and genetic structure of the giant white-tailed rat (Uromys caudimaculatus) within fragmented tropical rainforests. It is the first study to examine relationships between food resources, vegetation attributes and Uromys demography in a quantitative manner. Giant white-tailed rat densities were strongly correlated with specific suites of food resources rather than forest structure or other factors linked to fragmentation (i.e. fragment size). Several demographic parameters including the density of resident adults and juvenile recruitment showed similar patterns. Although data were limited, high quality food resources appear to initiate breeding in female Uromys. Where data were sufficient, influx of juveniles was significantly related to the density of high quality food resources that had fallen in the previous three months. Thus, availability of high quality food resources appear to be more important than either vegetation structure or fragment size in influencing giant white-tailed rat demography. These results support the suggestion that a species’ response to fragmentation can be related to their specific habitat requirements and can vary in response to local ecological conditions. In contrast to demographic data, genetic data revealed a significant negative effect of habitat fragmentation on genetic diversity and effective population size in U. caudimaculatus. All three fragments showed lower levels of allelic richness, number of private alleles and expected heterozygosity compared with the unfragmented continuous rainforest site. Populations at all sites were significantly differentiated, suggesting restricted among population gene flow. The combined effects of reduced genetic diversity, lower effective population size and restricted gene flow suggest that long-term viability of small fragmented populations may be at risk, unless effective management is employed in the future. A diverse range of genetic reproductive behaviours and sex-biased dispersal patterns were evident within U. caudimaculatus populations. Genetic paternity analyses revealed that the major mating system in U. caudimaculatus appeared to be polygyny at sites P1, P3 and C1. Evidence of genetic monogamy, however, was also found in the three fragmented sites, and was the dominant mating system in the remaining low density, small fragment (P2). High variability in reproductive skew and reproductive success was also found but was less pronounced when only resident Uromys were considered. Male body condition predicted which males sired offspring, however, neither body condition nor heterozygosity levels were accurate predictors of the number of offspring assigned to individual males or females. Genetic spatial autocorrelation analyses provided evidence for increased philopatry among females at site P1, but increased philopatry among males at site P3. This suggests that male-biased dispersal occurs at site P1 and female-biased dispersal at site P3, implying that in addition to mating systems, Uromys may also be able to adjust their dispersal behaviour to suit local ecological conditions. This study highlights the importance of examining the mechanisms that underlie population-level responses to habitat fragmentation using a combined ecological and genetic approach. The ecological data suggested that habitat quality (i.e. high quality food resources) rather than habitat quantity (i.e. fragment size) was relatively more important in influencing giant white-tailed rat demographics, at least for the populations studied here . Conversely, genetic data showed strong evidence that Uromys populations were affected adversely by habitat fragmentation and that management of isolated populations may be required for long-term viability of populations within isolated rainforest fragments.

Ce travail de thèse porte sur la connectivité des populations de coraux du genre Pocillopora dans le Sud-Ouest de l'océan Indien et l'océan Pacifique tropical. Ces coraux sont répartis sur toute la frange tropicale des océans Indien et Pacifique. Traditionnellement, les espèces étaient identifiées sur la base critères morphologies [17 espèces décrites dans Veron (2000)]. Différentes études utilisant des données génétiques ont révélé que la délimitation des espèces était parfois floue chez ces coraux. Ainsi, au cours de ce travail, l'utilisation de méthodes de délimitation d'espèces à partir d'ADN mitochondrial (ABGD, GMYC, PTP) et nucléaire (haplowebs) 16 hypothèses primaires d'espèces (PSH) ont été identifiées. Ces PSH ont ensuite été confrontées à des tests d'assignement à partir de marqueurs microsatellites, révélant un minimum de 18 hypothèses d'espèces secondaires (SSH). Une fois que les hypothèses d'espèces sont définies, il est possible de réaliser des études de connectivité. Au cours de ce travail, deux hypothèses d'espèces présentant des écologies différentes ont été choisies pour mener ces analyses. La première, Pocillopora damicornis type β (SSH05) a été échantillonnée dans les lagons et la seconde, Pocillopora eydouxi (SSH09) a, quant à elle, été échantillonnée sur la pente externe. L'estimation de la structure génétique des populations a permis d'estimer les modes de reproduction (sexuée ou asexuée) chez ces deux hypothèses d'espèces et les analyses de connectivité ont révélé des patterns de structuration complexes pour chacune des SSHs. / This work focuses on the connectivity among populations of the coral genus Pocillopora in the Southwestern Indian Ocean and the Southwestern Pacific Ocean. These corals are widely distributed throughout the tropical fringe of the Indian and Pacific oceans. Traditionally, species were identified on the basis of morphological criteria [17 species described in Veron (2000)]. Different studies using genetic data revealed that the delimitation of species was sometimes blurred in these corals. Thus, in this work, the use of species delineation methods from mitochondrial (ABGD, GMYC, PTP) and nuclear (haplowebs) DNA, 16 primary species hypotheses (PSH) were identified. These PSHs were then confronted to assignment tests from microsatellite loci, revealing a minimum of 18 secondary species hypotheses (SSH). Once the species hypotheses are defined, it is possible to conduct connectivity studies. In this work, two SSHs with different ecologies were chosen to carry out these analyses. The first, Pocillopora damicornis type β (SSH05) was sampled in the lagoons and the second, Pocillopora eydouxi (SSH09) was sampled on the outer slope. The estimation of the genetic structure of the populations made possible to estimate the reproductive modes (sexual or asexual) in these two SSHs and the connectivity analyzes revealed complex structuring patterns for each of the SSHs.