Many regional and local factors can influence the distribution of native and exotic species in ecological communities. I examined the regional- and local-scale determinants of native and exotic vascular plant species richness and composition in a highly fragmented oak savanna ecosystem on southeastern Vancouver Island. In sharp contrast to most reported results, I found a negative relationship between native and exotic richness at the regional scale, and no relationship at the local scale. Two extrinsic factors, surrounding road density and climate, best explained the regional-scale relationship by each affecting natives and exotics in opposite ways. Road density and climate were also the dominant predictors of native and exotic composition at the regional scale. Patterns in the patch occupancy of individual species confirmed the importance of these factors but I found that low surrounding road densities and cool, wet conditions predicted the presence of many natives and the absence of many exotics. Environmental factors explained variation in richness and composition at the local scale, but these factors were different for natives and exotics. My results suggest that natives and exotics respond to roads and climate in fundamentally different ways. Roads increase both exotic propagule pressure and disturbance, which may facilitate exotic invasion. In contrast, disturbance from roads may increase the likelihood of local extinction for particular natives. Differing climatic preferences within the native and exotic species pools may also partially explain the observed patterns. There was no evidence that native diversity directly affects exotic diversity (or vice versa). Surprisingly, I found that connectivity was not an important predictor of richness or composition despite the high degree of habitat fragmentation in this ecosystem. / Science, Faculty of / Botany, Department of / Graduate

Historically, the Kentucky Inner Bluegrass blue ash-oak savanna-woodland was the primary ecosystem of the Inner Bluegrass Region (IBR) of Kentucky. After European settlement, the majority (>99%) of Bluegrass savanna was converted to agricultural and urban land uses. Currently remnant savanna tree species are failing to recruit. Therefore, a long-term restoration ecology project researching competition and disturbance on seedling establishment, survival, and growth has been established at Griffith Woods (the largest remaining savanna in Kentucky) in Harrison Co., KY. Fourteen native hardwood tree species (a total of 6,168 seedlings) have been experimentally planted. Light, soil, surrounding vegetation, and herbivory, factors thought to influence seedling survival, have been initially assessed. Results show that soils differed spatially in P, Ca, Mg, Zn, pH, N percent and soil organic matter percent. Light was significantly reduced by diffusive filtering through vegetation. Vegetation biomass was influenced by pH and Mg. Initial seedling survival was high, but significantly differed by species type, location, and soil pH, Mg, and Zn. This research demonstrates that under a similar range of conditions, native hardwood tree seedling establishment is possible. Therefore, the potential exists to restore Bluegrass savanna-woodland in order to return proper ecological functioning into a degraded landscape.

Le brachypode rameux (Brachypodium retusum) est une espèce herbacée pérenne qui domine les pelouses sèches de Méditerranée occidentale. Dans notre zone d’étude du Sud de la France, la recolonisation spontanée est très faible après perturbation du sol. Cette observation ne correspond pas aux résultats d'études menées dans d'autres régions montrant une capacité de colonisation élevée. L'objectif principal de cette thèse était donc de tester les différentes hypothèses pouvant expliquer les différents patrons de colonisation.Dans un premier chapitre, nous avons testé si ces différents patrons de colonisation résultent d'une différenciation génétique entre les populations. Des marqueurs AFLP ont été utilisés pour analyser la structure génétique de 17 populations de Méditerranée occidentale dont la différenciation neutre entre populations. Dans un sous-échantillon de 13 populations françaises, la différenciation de traits phénotypiques sous sélection a été testée dans un jardin commun et comparée à la différenciation neutre. Dans le deuxième chapitre, nous présentons une étude sur la différenciation adaptative des traits phénotypiques et avons testé une potentielle réponse différentielle à la manipulation de facteurs environnementaux clés. Les deux derniers chapitres de la thèse ont analysé les facteurs environnementaux qui limitent la recolonisation in situ, indépendamment du génotype ou de la population. Dans le troisième chapitre, nous avons testé l’effet du pâturage et du feu sur le recouvrement végétatif ainsi que sur la reproduction sexuée de B. retusum installé. Enfin, dans le quatrième chapitre, nous avons mesuré l’effet de l’arrosage initial et du pâturage sur l’installation de plantules transplantées initialement germées en serre et de plantules issues de semis in situ. Nos résultats ont montré que les populations de B. retusum sont génétiquement différenciées pour les marqueurs neutres mais également pour les traits phénotypiques. Cette différenciation est supérieure à la dérive seule et suggère une adaptation aux conditions environnementales, en particulier aux températures estivales et à la fréquence du gel en hiver. Une réponse différentielle à la manipulation expérimentale de facteurs environnementaux (sol, pâturage, humidité du sol) a confirmé le caractère adaptatif de la différenciation génétique. Les expériences in situ ont montré que le feu a un effet positif sur la reproduction de B. retusum et sur la communauté végétale associée, alors que son recouvrement végétatif n’était pas supérieur à la moyenne de la communauté. L'exclusion du pâturage pendant deux saisons n'a eu d'incidence sur aucun des paramètres mesurés chez les populations adultes. L'arrosage initial a affecté l’installation des plantules au cours de la première saison. Au cours de la deuxième saison, l'effet seul de l’arrosage n'était pas significatif, cependant son effet est resté positif sur la survie uniquement dans les parcelles pâturées. Le pâturage, au début du cycle de vie de B. retusum, a eu un effet négatif sur le recrutement et la croissance des plantules. En conclusion, la différenciation adaptative entre les populations peut avoir contribué aux différences régionales en termes de capacité de colonisation et doit être prise en compte lors du ciblage des populations sources pour l'introduction de plantes en restauration écologique. La réponse positive de B. retusum a indiqué que le feu était une force sélective importante dans le passé qui pourrait être utilisée pour favoriser l'espèce et la communauté végétale associée dans des opérations de conservation et de restauration. L'exclusion du pâturage à court terme est tolérée par les populations adultes de B. retusum, mais l'abandon à long terme entraîne une diminution de son recouvrement par rapport aux graminées pérennes à croissance rapide. / Ramose false brome (Brachypodium retusum) is a perennial herbaceous species that dominates dry grasslands of the Western Mediterranean. In our southern French study area, spontaneous re-colonisation is very low after soil disturbance. This observation does not correspond to the results of studies from other regions showing a high colonisation capacity. The major objective of this PhD thesis was to test different hypotheses potentially explaining the different colonisation patterns.In the first chapter, we tested whether these different colonisation patterns are the result of genetic differentiation among populations. AFLP markers were used to analyse genetic structure including neutral population differentiation in 17 Western Mediterranean populations. In a sub-sample of 13 French populations, differentiation in phenotypic traits under selection was tested in a common garden and compared to neutral differentiation. In the second chapter, we present a study on adaptive differentiation in phenotypic traits testing a potentially differential response to the manipulation of key environmental factors. The last two chapters of the PhD analysed environmental factors that limit re-colonisation in the field independent of genotype or population. In the third chapter, we tested the effect of grazing and fire on vegetative recovery as well as on sexual reproduction of established B. retusum and in the fourth chapter we measured the effect of initial watering and grazing on the establishment of transplanted seedlings pre-grown in a greenhouse and of field-sown seedlings.Our results showed that populations of B. retusum are genetically differentiated in neutral markers but also in phenotypic traits. This differentiation is superior to drift alone and suggests adaptation to environmental conditions, particularly to summer temperature and winter frost frequency. A differential response to experimental manipulation of environmental factors (soil, pasture, soil moisture) confirmed the adaptive character of genetic differentiation. The field experiments showed that fire has a positive effect on B. retusum reproduction and on the associated community whereas vegetative recovery was not higher than community average. Two seasons of grazing exclosure did not affect any of the measured parameters in adult populations. Initial watering affected seedling establishment in the first season. In the second season, the watering main effect was not significant but interestingly the effect remained positive on survival in grazed plots whereas no such effect was observed in exclosures. Grazing in early life cycle stages of B. retusum had a negative effect on seedling recruitment and growth.In conclusion, adaptive differentiation between populations may have contributed to regional differences in colonisation capacity and needs to be taken into account in targeting source populations for plant introduction in ecological restoration. The positive response of B. retusum indicated that fire was an important selective force in the past which may be used to favour the species and its associated plant community in current conservation and restoration management. Short-term grazing exclosure is tolerated by mature B. retusum populations but long-term abandonment results in a decrease of cover relative to high-growing perennial grasses. In early stages of seedling establishment grazing should be avoided to guarantee introduction success - or grazing stress needs to be compensated by watering.