L'intensification de l'agriculture a amélioré de façon considérable la production alimentaire ces dernières cinquante années mais elle s'est accompagnée d'un impact croissant sur l'environnement. En particulier, la modernisation de l'agriculture a impliqué une simplification de la structure des paysages agricoles rendant nos agro-ecosystèmes plus sensibles au risque épidémique. L'utilisation de la diversité génétique des cultures est une solution prometteuse pour réduire le risque d'occurrence et de propagation des maladies des cultures. Elle nécessite cependant une gestion collective des espaces agricoles. En conséquences, l'échelle d'étude ne doit plus se focaliser sur la parcelle mais sur le paysage. Dans cette thèse, nous nous interessons aux processus se déroulant à l'échelle du paysage et au rôle de la diversité des plantes cultivées pour le contrôle des épidémies. Nous avons identifié trois questions: comment les populations pathogènes se propagent-elles dans un paysage d'hôtes hétérogène ? Comment les différents génotypes composant la population pathogène entrent-ils en compétition au sein d'une population hôte diversifiée ? et, à plus long terme, comment les populations pathogènes évoluent-elles en réponse à la structure des populations hôtes ? Chacune de ces questions a été approfondie grâce à l'analyse de données obtenues en condition de production mais aussi par des approches théoriques. Nous avons montré que la composition et la structure spatiale des populations hôtes influence fortement la population pathogène. Cependant, les recommandations que peut fournir ce travail pour gérer la diversité génétique dépendent de l'objectif visé. / Agriculture intensification has improved food production impressively in the past 50 years but this came with an increasing impact on the environment. In particular, modern agriculture has led to the simplification of the environmental structure over vast areas. As a consequence, agro-ecosystems are particularly susceptible to epidemics. The increase of crop genetic diversity is a promising way for reducing the risk of occurrence and development of diseases in crops but the technical and organisational conditions required to manage the genetic resources at this scale have not been established yet. This will require shifting the scale of crop protection investigations from the field to the agricultural landscape. In this PhD thesis we focus on landscape-scale processes and on the potential role of functional diversity in cultivated landscapes to better control plant diseases. We identified three questions: how does a pathogen population spread over a heterogeneous host landscape? How do pathogen genotypes compete in a diversified host population? And, in a longer term, how do pathogen populations evolve in response to host landscape structure? Each of these questions is investigated through the analysis of real data and the development of theoretical approaches. We demonstrate that the composition and the spatial structure of the host landscape greatly influence the pathogen population dynamics and evolution. The recommendations that this work could provide in order to practically manage the genetic resources will depend on the desired aim and will request further collaborative work withthe professional operators.

AN ABSTRACT OF THE DISSERTATION OF GRETCHEN ELIZABETH PADGETT-FLOHR for the Doctor of Philosophy degree in Zoology, presented on March 5, 2009 at Southern Illinois University of Carbondale. TITLE: BATRACHOCHYTRIUM DENDROBATIDIS IN CENTRAL CALIFORNIA AMPHIBIANS MAJOR PROFESSOR: Eric Schauber Amphibian chytridiomycosis has been identified as a disease responsible for the decline and extinction of many amphibian taxa world wide, but little research has been conducted on the disease in Mediterranean climates. To address this gap in the data I studied the amphibian assemblage present across a ~6,475 ha site in central California and investigated the occurrence of the etiological agent, Batrachochytrium dendrobatidis (BD) from organismal, community, landscape and historical perspectives. I initially tested the accuracy and reliability of a proposed diagnostic screening test for BD in four larval species that occur on the site. The screening test proposed by Fellers et al. (2001) and Vredenburg and Summers (2001) consisted of examining larval amphibian mouthparts for abnormalities and or defects, based on their hypothesis that mouthpart defects are clinical signs of BD infection. Sensitivity and specificity of the diagnostic screening test were 76% and 58%, respectively, indicating that the proposed screening test was not a reliable diagnostic test for BD infection for the four species I examined. I conducted controlled laboratory experiments to examine the consequences of BD infection in the two threatened California species that occur on my study site: Rana draytonii and Ambystoma californiense. Both species were susceptible to infection, but all infected animals survived the 18-month study. Infected A. californiense sloughed skin at three times the rate of uninfected salamanders, a pattern that may have long-term energetic costs potentially leading to population-level consequences of sublethal infection by BD. I conducted a retrospective survey of the California Academy of Sciences' (San Francisco, California, USA) amphibian collection, testing for BD in four amphibian species collected from central California between 1897 and 2005 to assess whether the pathogen is novel versus endemic. The earliest detection of BD was in two Rana catesbeiana collected in 1961, and the data support the hypothesis that BD was a novel pathogen introduced into central California prior to 1961 that spread geographically and taxonomically from at least one point of introduction and is now endemic throughout most of central California. I analyzed how environmental factors, amphibian community composition, land use practices, and landscape structure affect the dynamics of the pathogen's distribution on my study site in central California. The distribution of BD in ponds within the landscape varied markedly between years and increases were associated with precipitation, mean minimum and maximum temperatures, and presence of particular species. Pseudacris regilla infection patterns were highly indicative of overall patterns of pond BD status. Fourteen ponds were identified as BD hotpots (BD-positive three of four years). Occurrences of the pathogen within the landscape were spatially autocorrelated and ponds in close proximity to BD hotspots were more likely to test positive. Local land use, (presence/absence of grazing or recreational activity and developed lands), apparently did not influence BD status of a pond. My studies show that BD was likely a novel pathogen introduced into California ca. 1961 that has since become established as an endemic pathogen throughout most of central California. The listed amphibian species that occur in central California can be infected with BD but appear to be resistant to manifesting amphibian chytridiomycosis, and the data from the studies herein could support one of two hypotheses: that natural selection acting over the past 48 years has selected for those individuals that were resistant to the disease; or that the species on my site have always been resistant to BD. The research I conducted further supports the hypothesis that BD is locally vectored by native amphibians (e.g. P. regilla) moving between ponds and that local ecological constraints likely limit vectoring of BD by non-native species. These findings contribute substantially to elucidating and understanding the responses of amphibian populations to disease/pathogen introduction and lay groundwork for future investigations into the host-pathogen-environment relationship as it relates to declining amphibian populations.