In this paper we endeavor to test the controversial ideas that exist about the role of fragmentation in a conservation context. In line with earlier understanding, we find that habitat fragmentation alone results in a strong detrimental effect (especially for the predator population). Connecting the fragmented habitats facilitates predator survival and hence prey survival as compared to the unconnected fragmented case. Our main result is counterintuitive: in the presence of a high quality predator, connected fragmented habitats ensure a better chance for coexistence than does even the unfragmented case. We explain why a connected fragmented habitat might thus be beneficial for the stabilization of the system, and how connections between sub-habitats are able to protect prey population from over-exploitation. In the model, habitat fragmentation is separated from the effects of habitat destruction, in order to better understand how populations react to habitat transformation.

La flore adventice est à la base des chaînes trophiques dans les agroécosystèmes. Leur diminution dans les parcelles cultivées est une des causes de la baisse de la biodiversité. Par conséquent, des alternatives sont étudiées pour diminuer l’utilisation des herbicides dans le cadre du plan ECOPHYTO. Par exemple, la gestion de la flore adventice via les consommateurs de graines. En effet, les agroécosystèmes sont des milieux où des espèces granivores ont les capacités de consommer les graines d’adventices. Parmi les vertébrés, nous pouvons y trouver des oiseaux (alouette des champs) ainsi que des micromammifères (mulot). Parmi les invertébrés, les carabes sont identifiés comme principaux granivores dans les agroécosystèmes tempérés. L’objectif général de cette thèse est d’étudier les interactions entre carabes et graines d’adventices afin d’apporter des connaissances sur les interactions interspécifiques et d’identifier les traits physiologiques des espèces influençant la force des interactions entre les espèces. Un protocole a été établi afin de standardiser toutes les expériences et de garder comme variables uniquement l’espèce de graines et celle du carabe. Ensuite, nous avons étudié les interactions entre les graines d’adventices et les carabes selon deux axes de recherche. Le premier a permis de comparer les niveaux de consommation de la communauté de carabes sur une même espèce de graines. Nous avons pu ainsi observer que la masse corporelle des carabes n’était pas en relation avec la variation des niveaux de consommation. Mais le ratio entre la longueur des mandibules et la largeur du labrum (force mandibulaire) a un effet significatif lorsque seules les espèces consommant plus de cinq graines sont prises en compte. Pour le deuxième axe, nous avons mesuré la consommation des carabes sur 42 espèces de graines pour identifier les capacités de consommation et explorer la stratégie d’alimentation selon deux caractéristiques : la masse des graines (taille) et leur concentration lipidique. Les graines de faibles masses avec un taux lipidique élevé sont mangées en plus grand nombre. Enfin, nous avons déterminé les types de réponse fonctionnelle pour évaluer le potentiel de régulation de 4 espèces de carabes sur deux espèces de graines d’adventices. Pour chaque espèce (mâle et femelle) nous avons trouvé une réponse fonctionnelle de type II, sauf pour les femelles de P.rufipes. Cependant nos résultats doivent être interprétés avec précaution car en milieu naturel d’autres facteurs interviennent dans les interactions, telle que la présence de proies alternatives ou d’autres prédateurs. En conclusion, nous montrons que les interactions entre les graines d’adventices et les carabes sont nombreuses et complexes. Les caractéristiques physiologiques intervenant dans les interactions sont très nombreuses et très variables selon les espèces. Des études supplémentaires sont nécessaires pour évaluer l’intérêt agronomique des carabes dans la gestion des adventices. / Weed flora limits crop yields therefore herbicides inputs are important in order to manage weeds. Weeds are at the bottom of the trophic pyramid in agroecosystems, and this explains partly the decrease of biodiversity. Alternatives to manage weeds are studied to reduce negative effect of herbicides on the environment in order to preserve biodiversity. For example, weeds could be managed with granivorous species. Indeed, weed seeds are the origin of the weed community, thus seed consumption could limit weed abundance in cultivated fields. In agroecosystems, several taxa are known to be granivorous such as vertebrates (birds and rodents) and invertebrates (ground beetles). Ground beetles are considered as the main granivorous taxa in agroecosystems and are abundant in temperate agroecosystems. The main goal of this PhD was to study trophic interactions between communities, weed seed and ground beetles, with a prey-predator system in order to estimate the role of ground beetles through their seed consumption in weed seeds management. First, we established a protocol with standardize steps in order to restrain variables to weed seed species at ground beetles’ species. A similar protocol has been adapted to replicate experimentation with the same individuals to measure consumption at different weed seed densities. We then studied interactions between weed seeds and ground beetles with two research axes. We first explored interspecific variation of consumption on the same weed seeds species, Viola arvensis, by ground beetles. Body mass and ratios between mandible length and labrum width had no relation with seed consumption by ground beetles. Thus, other factors were suggested to explain consumption variation such as gut symbionts of ground beetles or preferences for specific weed seed species. Secondly, we measured consumption of 42 weed seed species by four ground beetles to identify abilities in weed seed consumption for these beetles and to explore feeding strategy according to two seed characteristics, seed mass (size) and seed lipid content. Generalism degrees are different according to ground beetles’ species and seem affected by physiological characteristics in both communities. Finally, to assess the potential in weed seed management of ground beetles we determined functional response to four ground beetles on two weed seeds species. We determined type II response for all species (male and female) tested, except for females of P.rufipes. Our results should be interpreted cautiously as, under natural conditions, generalist predators meet many alternatives prey and it could influence functional response type. We showed that there are many complex interactions between weed seed and ground beetles communities, especially due to a wide range of diversity in physiological characteristics in both communities. The potential of ground beetles to regulate seeds must be more precisely explored to evaluate their agronomic interest in weed management.

Made available in DSpace on 2015-03-26T13:35:24Z (GMT). No. of bitstreams: 1 texto completo.pdf: 1414332 bytes, checksum: 6e2f42018f3e3dcdf9e8cbccab567e7a (MD5) Previous issue date: 2006-03-20 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Pattern, structure and emergent collective properties are ubiquitous in systems with many units (alive or inanimated) coupled through nonlinear interactions. Within this context, the study of cooperative phenomena in population dynamics of ecological interest has attracted the attention of the mathematicians and physicists since Lotka and Volterra in the 1920s. Thenceforth, in addition to differential equations, theoretical ecology has continuously incorporated powerful and well-established techniques of contacts processes, cellular automata models and others, developed in the fields of condensed matter physics, statistical physics and computational physics. In the present work, a predator-prey model with pursuit and escape triggered by alarm pheromones is proposed and studied through analytical methods and computer simulations. Such models can show oscillatory behavior of the population density, phase transitions that belong to distinct universality classes and rich stationary phase diagrams. Two distinct levels of description were used. In a first approach, we consider a model of cellular automata in which predators and preys walk on a square lattice, according specific rules for each species, in a homogeneous environment and with periodic boundary conditions. The second part of our study is based on the analysis of partial differential equations that also describes the dynamics of a prey-predator system with the same characteristics above. Both, spatially uniform or mean field like and explicit spatio-temporal partial differential equations were considered. These models can represent relevant tools to design better strategies of biological control of pests by predators. In successful cases, the pests and its predators must persist in stable interactions at a low level of pest density. / Padrões, estruturas, propriedades coletivas emergentes são ubíquas em sistemas com muitas unidades (vivas ou inanimadas) acopladas por meio de interações não-lineares. Dentro desse contexto, o estudo de fenômenos cooperativos em dinâmica de populações de interesse ecológico tem atraído a atenção de físicos e matemáticos desde os anos de 1920 com Lotka e Volterra. Portanto, além de equações diferenciais, a teoria ecológica tem continuamente incorporado poderosas e bem-estabelecidas técnicas dos processos de contatos, modelos de autômatos celulares e outros, desenvolvidos no campo de física da matéria condensada, física estatística e física computacional. No presente trabalho, um modelo presa-predador com perseguição e fuga mediada por um feromônio de alarme é proposto e estudado através de métodos analíticos e simulações computacionais. Tais modelos podem exibir comportamentos oscilatórios da densidade de população, transições de fases que pertencem a classes de universalidade distintas e um diagrama de fases rico. Duas abordagens distintas de descrição foram usadas. Numa primeira abordagem, propomos um modelo de Autômato Celular (AC) onde predadores e presas se movimentam, segundo regras específicas para cada espécie, num ambiente homogêneo e com condições de contorno periódicas. A outra parte do nosso estudo é baseado na análise de EDP s que também descrevem a dinâmica de um sistema presa-predador com as mesmas características citadas acima. É feito um estudo considerando as equações sem termos espaciais, isto é, tipo campo médio e depois considerando esses termos. Esses modelos podem representar ferramentas relevantes para o estudo das melhores estratégias para o controle biológico de pragas por predadores. Em casos bem sucedidos, as pestes e seus predadores devem persistir em interações estáveis e com uma baixa densidade da população de pragas.