Urban areas occupy a large and increasing proportion of global land area. To date, urbanization generally produces disastrous consequences for native species. Yet some developed areas appear to support native species better than others, suggesting that humans need not harm nature with their mere presence. Indeed, our cities--if managed appropriately--may play a crucial role in sustaining the world's biological diversity. Focusing on birds, I here present four investigations into the causes and consequences of, and potential solutions to, the problem of reduced biodiversity in urban areas. In the first, I develop and implement a community-based monitoring project (the Tucson Bird Count) to acquire previously unavailable data on the distribution of birds throughout Tucson and its many habitats. In the second investigation, I focus on one suite of birds--those that require desertscrub habitats--and develop a model to understand better the relationship between these species and the composition of their habitats in and around Tucson. In the third investigation, I evaluate the ability of 62 alternative future scenarios to restore and sustain Tucson's desert birds, assessing particular strategies that can be used to reduce the impact of development on these species. In the fourth investigation, I compare the displacement of humans relative to bird diversity in Tucson and 4 other cities. This analysis reveals a systematic pattern of urban humans concentrated in neighborhoods of impoverished diversity. This pattern likely applies to many other cities worldwide, and has tragic implications both for human quality of life and for the conservation of nature everywhere. Fixing this situation will require innovative approaches--based on sound biology--to sustaining nature nearer the places we spend our lives. This dissertation is a first step in that direction.

How does variability in nectar secretion rates affect the interaction between hummingbirds and the plants that: they pollinate? Other researchers have suggested that variability may influence pollinators to leave a plant earlier, thus increasing the potential for cross-pollination. My dissertation asks the following questions: Does variability in nectar secretion rates influence hummingbird visitation at a flower patch? Does it do so in a manner that is likely to increase; cross-pollination? And, does variability in these rates benefit the hummingbird by improving their foraging efficiency? By asking if variability benefits both plants and hummingbirds, I implicitly ask the question: Can variability in nectar secretion rates be a mechanism that helps keep this potentially antagonistic interaction positive? I developed an artificial flower that simulates nectar secretion. Using patches of these artificial flowers, I varied renewal rates of flowers found within a patch. The appendices describe the results of my studies. The aviary experiment (Appendix A) investigates how Broad-billed hummingbirds (Cynanthus latirostris) forage given variability in nectar renewal-rates. The field experiment (Appendix B) investigates how hummingbird visitation to patches with different renewal-rates might affect cross-pollination. Appendix C describes the hummingbird community at the study site. Appendix D identifies the plants visited by these hummingbirds. In the aviary experiment, Broad-billed hummingbirds changed their foraging when exposed to high variability and limited nectar. They visited a higher percentage of rewarding flowers, foraged more systematically, and significantly decreased their foraging time. These changes made their foraging more efficient. In the field experiment, increased renewal-rate variability was associated with hummingbirds visiting fewer artificial flowers per foraging bout. These results suggest cross-pollination may be increased. However, levels of renewal-rate variability did not affect the visitation rates to the flower patch or the distribution of nectar within the patch. Hummingbirds spent significantly more time probing the last flower in a foraging bout than other flowers, and preferentially ended foraging bouts on a rapidly renewing flower. I suggest that the energetic cost of hovering flight likely influenced this behavior. My results support the hypothesis that variability in nectar secretion rates may benefit both partners in this pollination mutualism.

Two major conceptual approaches for understanding the evolutionary ecology of insect-plant interactions, the plant-insect paradigm and the tri-trophic paradigm, have focused primarily upon dietary specialists and their host-plants. Here, I attempted to evaluate the utility of both paradigms for explaining the maintenance of food-mixing by the individually polyphagous caterpillars of Grammia geneura (Strecker) (Lepidoptera: Arctiidae). First, I conducted three experiments testing the hypothesis that individual G. geneura caterpillars perform better on mixed-plant diets than on single-plant diets due to improved physiological efficiency of food utilization. However, caterpillar performance was not always superior on mixed-plant diets. In the one case in which food-mixing improved performance, the host-plant species included in the mixture were individually of low suitability. Behavioral observations of individual caterpillars both in the above experiments, in nature, and in two further laboratory experiments with chemically-manipulated, synthetic diets supported the idea that such dietary benefits resulted from dilution of plant secondary metabolites, achieved behaviorally via the physiological mechanisms of neophilia and post-ingestive feedbacks on feeding. I also investigated the possibility that food-mixing was maintained by the unpredictable availability of high-quality host-plant species. A field survey of caterpillar feeding preference, frequency of parasitism, and host-plant availability suggested that this variation in food availability combined with the increased risk of parasitism incurred by individuals experiencing prolonged development (e.g. by searching excessively for a rare, preferred host-plant species) should favor polyphagy, and reinforce opportunistic food-mixing. However, because individual caterpillars showed a tendency to leave nutritionally superior host species for nutritionally inferior ones, I tested the idea that individuals ate some host-plant species for defense against parasitoids. Two experiments showed that diet modified the survival of parasitized caterpillars, and that at least one pair of host-plant species revealed a trade-off between their nutritive and defensive value to caterpillars. Taken together, the experiments in this study underscore the importance of the tri-trophic approach toward understanding the pattern and process of foraging in generalist as well as specialist herbivores.

Consideration of scale and ecosystem resilience is integral to any conceptual model of the effects of disturbance on ecosystems. Organisms, populations, communities, and ecosystems are differentially affected by disturbance based on the scale at which they occupy the landscape. Scale of observation influences perceptions about ecosystem resilience. There is no single correct scale at which ecological phenomena should be studied, and management decisions require the interfacing of phenomena that occur on very different scales of space and time. Fire disturbance affects a variety of ecosystem factors including nutrient cycling, species diversity, and population and community dynamics. My experimental research on fire in shortgrass steppe examined the effects of fire and season of fire on various components of shortgrass steppe at multiple spatial and temporal scales and organizational units. My experimental design was completely randomized, with 3 treatments, and 4 replicates per treatment. Treatments were dormant-season fire, growing-season fire, and unburned. Response variables were (1) ground cover; (2) microbiotic crust nitrogen fixation, and chlorophyll a content; and (3) species richness, abundance, and relative abundance of small mammals and arthropods. Microbiotic crust cover never differed significantly among treatments for all periods, however, acetylene reduction and chlorophyll a content of crusts differed significantly among treatments. Dormant-season fire-treated crusts had significantly lower rates of acetylene reduction than unburned crusts, while growing-season fire-treated crusts did not differ significantly from unburned or dormant-season fire-treated crusts. Dormant-season fire-treated crusts had significantly lower chlorophyll content than unburned crusts, while growing-season fire-treated crusts did not significantly differ from unburned or dormant-season fire treated-crusts. Initially, growing-season fire significantly reduced grass cover compared to unburned and dormant-season fire. Approximately 30 months later there were no significant differences in grass cover among treatments. Bare ground response was basically the inverse of grass cover response. The only significant differences in litter cover between treatments occurred immediately after the growing-season fire. Arthropod species richness differed significantly among treatments; growing-season fire plots had a significantly higher number of beetle species. However, overall beetle abundance did not significantly differ among treatments. Significant differences were never detected in overall rodent species richness or abundance among treatments.

Animals vary in their response to the distribution of resources in time and space. Opportunistic foraging is evident in many species and has indirectly been shown to be linked to measures of cognition such as innovation and problem solving. However, in the field of cognitive ecology, the operationalization and empirical use of opportunism is problematic. In chapter 1, I review the concept of opportunism in the zoological literature and propose an operational definition. The review suggests that many definitions of the concept are not useful, in particular those that equate it with generalism or use it to describe random choice in foraging. With the operational definition I propose ('latency to switch to a new, abundant, food source'), the relationship between ecological flexibility and cognition is then addressed through a small-scale comparative study in chapter 2. Here, the purpose is to determine if an opportunistic species will perform better at problem solving, and have lower neophobic tendencies, than a less opportunistic species. The study compares two sister species of Thraupidae with different foraging strategies: the Barbados bullfinch (Loxigilla barbadensis), an opportunistic forager, and the black-faced grassquit (Tiaris bicolor), a conservative forager. In the field, I carried out focal observations along with opportunism and neophobia experiments. In captivity, wild-caught individuals were run through a set of behavioural and cognitive tests, which included a neophobia test and a problem-solving obstacle removal task. Results show that although both species share overlapping foraging modes, territorial habits and neophobic tendencies, the Barbados bullfinch is much more opportunistic, bolder and better at problem-solving than the black-faced grassquit. / Les animaux diffèrent dans leurs réponses à la distribution spatiale et temporelle des ressources. Plusieurs espèces manifestent un mode opportuniste de quête alimentaire et des preuves indirectes suggèrent que l'opportunisme est associé à des mesures de cognition telles que l'innovation et la résolution de problèmes. Toutefois, dans le domaine de l'écologie cognitive, la définition et l'opérationalisation de l'opportunisme pose problème. Dans le premier chapitre de ce mémoire, je fais une revue de littérature du concept d'opportunisme et j'en propose une définition opérationelle. La revue suggère que plusieurs acceptions du concept sont peu utiles, en particulier celles qui le confondent avec le concept de 'genéralisme' et celles qui lui donnent le sens de 'capture au hasard' de proies. A partir de la définition opérationelle que je propose ('la latence d'exploitation d'une nouvelle et abondate source de nourriture'), la relation entre l'opportunisme et la cognition est testée au chapitre 2 dans une étude comparative à petite échelle. Je prédis qu'une espèce opportuniste sera plus rapide à résoudre un problème alimentaire et sera moins néophobe qu'une espèce conservatrice. L'étude compare deux espèces génétiquement très proches, le sporophile de la Barbade (Loxigilla barbadensis), une espèce opportuniste, et le sporophile cici (Tiaris bicolor), une espèce conservatrice. J'ai effectué sur le terrain des observations focales et des expériences sur l'opportunsime et la néophobie. En captivité, j'ai soumis des individus piégés sur le terrain à des tests de néophobie et d'enlèvement d'obstacle pour atteindre de la nourriture. Les résultats révèlent que le sporophile de la Barbade est plus opportuniste, moins néophobe et meilleur à résoudre le problème que le sporophile cici, mais que ni sa territorialité ni son mode d'alimentation sur le terrain ne diffèrent suffisamment de celui du sporophile cici pour expliquer les différences de cognition.

This study examines Bobolink (Dolichonyx oryzivorus) nest success, nest-site selection, and nestling growth and survival with the goal of establishing conservation and management plans for hayfields of Québec and Ontario. Bobolink nest survival increased with age of the nest and was higher in managed hayfields than at less-managed sites in close proximity to wetlands. Female bobolinks selected nests with higher forb composition and higher overhead concealment. If haying in the region is optimally delayed until 15 July and minimally until 1 July, this allows time for Bobolink young to fledge, as well as harvesting of hay late to use as animal fodder. For late-cut fields, increased forb composition provides high-quality nesting habitat. For early-cut fields, which ultimately lead to nest and nestling destruction, nesting may be deterred by a lack of forbs. Basic Bobolink nest ecology and nestling growth is described to serve as a baseline for further research on this understudied species. / Cette étude porte sur la réussite de la nidification, la sélection du site de nidification et l'écologie des oisillons du Goglu des Prés (Dolichonyx oryzivorus), pour objectif d'établir des plans de conservation et de gestion des champs de foin du Québec et de l'Ontario. La survie des nichées de Goglu des Prés a augmenté avec l'âge des nids et était supérieure dans un champ de foin géré que dans un site moins géré a proximité d'un milieu humide. Les femelles ont choisi des nids avec une plus grande composition en herbes non-graminéennes et un camouflage aérien plus important. Si la récolte dans la région est rapporte de façon optimale jusqu’au 15 juillet et au minimum jusqu’au 1er juillet, il y aurait suffisamment de temps pour que les jeunes Goglus atteignent la phase d’envol et aussi si une récolte de foin tardif soit faite pour le fourrage. Pour les champs tardifs, une plus grande composition en herbes non-graminéennes fournit du habitat de haute qualité. Pour les champs coupées tôt, qui finalement cause la destruction des nids et l’oisillon, la nidification des Goglus peu être dissuadé par une absence des herbes non-graminées. Finalement, l’écologie du nid et la croissance de l’oisillon chez le Goglu des Prés ont été décrites afin de servir de référence pour de futures recherches sur cette espèce qui n’est pas suffisamment étudiée.

The introduced small Asian mongoose (Herpestes javanicus) is a widespread predator of sea turtle eggs and hatchlings in the Caribbean. I studied the behavioural ecology of mongoose predation on the nests of critically endangered hawksbill sea turtles (Eretmochelys imbricata) in Barbados. Combining short-term field experiments with seven years of hawksbill nesting data, I investigated how mongoose foraging behaviour, antipredator behaviour and landscape use explain the spatial and temporal patterns of sea turtle nest predation. An experiment combining artificial nests and predator tracking revealed a direct relationship between fine-scale variation in mongoose activity and nest predation risk. The combination of mongoose avoidance of open areas and the spatial distribution of hawksbill nests relative to patches of beach vegetation accurately predicted the observed peak in nest predation near the vegetation edge. Egg-burial depth by nesting hawksbills also affected predation risk, but this was primarily due to the increased digging effort required rather than any increase in nest concealment with depth. A second experiment with artificial nests confirmed the causal relationship between burial depth and predation risk and showed that substrate disturbance is a primary cue for nest detection by mongooses. At the landscape scale, mongooses tracked local nest abundance but showed a fine-scale negative response to human beach use, suggesting that human activity on nesting beaches may improve nest survival by deterring predators. Finally, an analysis of nest survival times showed that nests were most vulnerable to predation in first days following oviposition and that predation risk increased over the nesting season, providing a general framework for planning where and when predation reduction methods should be applied. I conclude that predation risk for sea turtle nests is likely to depend on: i) how predator nest-finding behaviour is modulated by nest characteristics s / La petite mangouste indienne (Herpestes javanicus) est une espèce introduite dans de nombreuses îles des Caraïbes et est un prédateur important des oeufs de tortues marines. J'ai étudié l'écologie comportementale de la prédation par les mangoustes sur les nids de tortue imbriquée (Eretmochelys imbricata), une espèce en danger critique d'extinction, à la Barbade. En combinant des études expérimentales de courte durée avec des données de prédation des nids de tortues imbriquées s'échelonnant sur sept années, j'ai investigué comment le comportement d'approvisionnement, le comportement anti-prédateur, et l'utilisation du paysage par les mangoustes expliquent les patrons spatiaux et temporels de la prédation des nids. Une étude expérimentale combinant des nids artificiels et une mesure passive de l'activité des prédateurs a démontré une relation directe entre la variation spatiale de l'activité des mangoustes et le risque de prédation des nids. Conjointement, l'évitement des espaces dépourvus de végétation sur la plage par les mangoustes et la distribution spatiale des nids de tortues imbriquées en fonction de la végétation ont prédit de manière précise le patron de prédation élevé observé dans la zone bordée de végétation. La profondeur des nids affectait également le risque de prédation mais avait peu d'influence sur la détection des nids par les mangoustes. L'effet relié à la profondeur était surtout dû à l'effort d'excavation supplémentaire. Une deuxième étude expérimentale utilisant des nids artificiels a confirmé la relation directe entre la profondeur des nids et la prédation, démontrant que la mangouste utilise la perturbation du sable créée lors de la ponte comme principal indice de détection des nids. A l'échelle du paysage, l'activité des mangoustes suivait la disponibilité des nids de tortues sur la plage. Cependant, il y avait une relation négative entre l'activité des mangoustes et l'u

The process by which one species becomes many (speciation) was considered by Darwin to be the mystery of mysteries. Speciation is no longer a complete mystery, but we still have a lot to discover about its mechanisms. Of particular interest is the process where reproductive isolation evolves as a byproduct of local adaptation, called ecological speciation. Ecological speciation is of interest to theoretician who can look at the condition that allow local adaptation in face of gene flow, but it is also interesting to field biologist because they can study the process as it is happening. Many factors, such as natural selection, sexual selection, environmental difference between the environments, can influence the evolution of reproductive isolation. In this thesis, I used individual-based numerical simulation to address the relative contribution of natural selection, sexual selection, environmental difference between the environments on the progression of the ecological speciation process. I then evaluated the effect of phenotypic plasticity on a similar system. To inform field biologist about a method to infer ecological speciation using neutral markers, I ran a power analysis to determine in what conditions the method was accurate. Finally, I looked at the relative role competition, sexual selection and the shape of the resource distribution in a sympatric scenario. I found that natural selection greatly influence progress toward ecological speciation, but without the added contribution of sexual selection, speciation could not be achieved. Phenotypic plasticity can either promote or constrain progress toward ecological speciation, depending on the timing of migration relative to the expression of the plasticity. Using neutral markers to infer ecological speciation is reliable if migration rate is intermediate, i.e. about one migrant per generation per population. In sympatry, sexual selection was the most important promoter of speciation, but could not complete / Le processus par lequel une espèce devient plusieurs (spéciation) était considéré par Darwin comme le mystère des mystères. La spéciation n'est plus un mystère complet, mais nous avons encore beaucoup à expliquer à propos de ses mécanismes. En particulier, comment l'isolement reproductif peut évoluer en tant que sous-produit de l'adaptation locale. Ce processus est connu sous le nom de spéciation écologique. La spéciation écologique est intéressante pour les théoriciens qui peuvent étudier les conditions dans lesquelles l'adaptation locale peut évoluer face au flux génique. Les biologistes de terrain s'intéressent aussi à la spéciation écologique, car ils peuvent l'étudier pendant qu'elle progresse. Plusieurs facteurs peuvent influencer l'évolution de l'isolement reproductif : la sélection naturelle, la sélection sexuelle et l'environnement, entre autres. Dans cette thèse, j'utilise des simulations numériques orientées-individus afin de déterminer la contribution relative de la sélection naturelle, de la sélection sexuelle et des différences entre les environnements dans la progression du processus de spéciation écologique. J'ai ensuite étudié l'effet de la plasticité phénotypique sur un système similaire. Afin d'assister les biologistes de terrain, j'ai fait une étude sur la puissance statistique d'une méthode largement utilisée afin d'inférer la spéciation écologique à partir de marqueurs génétiques neutres. Finalement, j'ai examiné le rôle relatif de la compétition, de la sélection sexuelle et de la forme de la distribution des ressources sur la spéciation sympatrique. J'ai trouvé que la sélection naturelle influence grandement le progrès vers la spéciation écologique, mais sans l'ajout de la sélection sexuelle, la spéciation ne pouvait pas être complète. La plasticité phénotypique peut soit promouvoir, soit contraindre la progression vers la spéciation écologique, tout dépend si l'express

Natural ecosystems are open to flows of energy, materials, and organisms. These subsidies are ubiquitous and influence ecosystem structure and functioning at local and regional extents. Subsidized consumers can attain higher biomass and abundance than unsubsidized consumers but the indirect, cascading effects generated from subsidized consumers are not well understood. I derive ecosystem models to investigate the relationship between subsidies and trophic cascades. I show that the ratio of subsidy to equivalent in situ prey may not be the best predictor of consumer response to subsidies, particularly when subsidies are temporally variable. I predict strong generalist consumer responses to subsidies in ecosystems with high in situ prey and relatively frequent subsidies. Next, I use response ratios to quantify the relative effect of predator regulation of herbivores and consumer-mediated recycling on producers stocks. I observe that predator regulation of herbivores has a larger, positive, effect on producer stocks than consumer-mediated recycling, however, consumer-mediated recycling can influence producer biomass in many cases. The relative contribution of both mechanisms to cascading trophic interactions depends on feeding relationships between predator and prey, nutrient turnover rates and the rate of external nutrient loading. Trophic cascade theory and experiments to date, have been conducted in closed ecosystems but recent evidence suggests that predators can have cascading effects across ecosystem boundaries. I derive a model of a recipient ecosystem and demonstrate that ecosystems with higher rates of constant subsidies experience stronger trophic cascades because these subsidies facilitate higher secondary production and consumption. I extend this previous model to meta-ecosystem extents in order to consider the effects of reciprocal pulsed subsidies on ecosystem functioning. I show that reciprocal pulsed subsidies can be reinforcing, particularly when th / Les écosystèmes naturels sont ouverts aux flux d'énergie, de matière, et d'organismes. Omniprésents, ces apports allochtones ont un impact sur la structure et le fonctionnement des écosystèmes. S'il est clair que des consommateurs bénéficient directement de ces flux, nous n'avons qu'une connaissance rudimentaire de leurs effets indirects sur les réseaux trophiques. Je développe des modèles d'écosystème afin d'étudier la relation existante entre flux de matière et d'organismes et cascades trophiques. Je démontre ainsi que le ratio de la biomasse des apports sur la biomasse de proie locale ne prédit pas l'effet des flux sur les consommateurs, notamment lorsque ces apports sont variables dans le temps. L'impact des flux sur la biomasse de consommateur se révèle d'autant plus grand quand la biomasse de proie locale est importante et que les flux sont fréquent. J'examine ensuite l'impact relatif du contrôle des herbivores et du recyclage de nutriments par les prédateurs sur la population de plantes et montre que, bien que les deux soient positifs, l'effet du contrôle des herbivores est plus fort. La contribution relative de ces deux mécanismes pour les cascades trophiques dépend des interactions entres proies et prédateurs, des taux de recyclages et des flux de nutriments provenant de l'extérieur. Les cascades trophiques sont généralement étudiées et conceptualisées dans des écosystèmes fermés. Cependant, il a récemment été mis en évidence que les prédateurs peuvent avoir des effets indirects dépassant les frontières d'un écosystème. A l'aide d'un modèle d'écosystème bénéficiant de flux allochtones, je démontre en effet que les cascades trophiques sont d'autant plus fortes quand les écosystèmes reçoivent des flux de matière et d'organismes fréquents. Ce modèle est ensuite étendu à l'échelle de méta-écosystèmes afin d'examiner l'effet des flux réciproques et variables dans le temps sur la force des cascade

Despite the important role spatial processes play in natural communities, far too little theoretical work has been devoted to exploring how complex food web communities may be assembled in space, and how the spatial structure of trophic interactions may provide a stabilizing mechanism for complex food web networks. In this thesis I develop a food web metacommunity model based on a classic Levins-type patch-dynamic model which views trophic interactions between species as occurring in a spatially subdivided habitat. I then use this model to explore both simple and complex trophic networks in an ecological and evolutionary context. I first review and evaluate previous attempts at defining a patch-dynamic metacommunity model of trophic interactions. After correcting the flaws in previously published models I develop a corrected model and apply it to simple trophic configurations. I show how the stability of simple trophic interactions, like omnivory loops, depends upon the interacting effects of space and network configuration. I then use the model to study the evolution of dispersal in a simple predator-prey system. Specifically, I investigate how both predator and prey dispersal rates will evolve in response to increasing patch extinction rates caused by locally strong top-down predator effects. I show how the predator's evolutionarily stable (ESS) dispersal rate will increase, as expected, in response to increasing local extinction, while the prey's ESS dispersal rate exhibits a counterintuitive nonmonotonic response – actually decreasing for some ranges of extinction. I explain how the prey's counterintuitive response arises because of the way trophic interactions between species play out at different spatial scales. After applying the model to simple networks I then explore the assembly of complex food webs. I show that, under very simple assumptions, diverse and complex food web networks can be assembled through the creation of network branches which provide opportunities for the build-up species and multiple food chain paths in the food web. I also show how these network branches can emerge simply as the result of the spatial distribution of trophic interactions, and the structural support provided by omnivory and generalist feeding links. I then attempt to determine if natural food webs show a relationship between biodiversity and network branching. To this end I examine a set of empirical food webs and observe a striking linear scaling relationship between food web size and the degree of branching in the minimum spanning tree of a food web. This empirical corroboration of the theory suggests that the theory reported here may be of value as a guide to how space and dispersal interact to structure natural food webs at large scales. / Malgré l'importance des processus spatiaux dans les communautés écologiques naturelles, peu de théories examinent le rôle de l'espace dans l'assemblage et la stabilisation des réseaux trophiques complexes. Dans cette thèse, je développe un modèle de réseau trophique spatial (métacommunauté) fondé sur un model dynamique de métapopulation du type Levins, où les interactions trophiques entres les espèces ont lieu au sein d'une série de populations locales. Ce modèle de métacommunauté me permet d'examiner les réseaux trophiques simples et complexes dans un contexte écologique et évolutif.Dans le premier chapitre, je résume et critique les modèles actuels de métacommunauté du type Levins incorporant les interactions trophiques dans un contexte spatial. Après avoir identifié les erreurs de ces modèles, je développe un modèle corrigé afin d'examiner des réseaux trophiques simples. Je montre que la stabilité des interactions trophiques simples (telles que les boucles omnivores) dépend de l'interaction entre la structure spatiale et la topologie du réseau trophique. Dans le deuxième chapitre, j'utilise ce modèle afin de déterminer l'évolution du taux de dispersion du prédateur et de sa proie lorsque la prédation favorise l'extinction des populations locales. Je montre que face à une augmentation du taux d'extinction, le taux de dispersion évolutivement stable du prédateur augmente de façon monotone alors que celui de la proie varie de façon non-monotone et diminue pour certains niveaux d'extinction. Je démontre que cette réponse contre-intuitive de la proie est due à la structure spatiale des interactions trophiques entre les espèces.Dans le troisième chapitre, j'utilise le modèle afin d'étudier l'assemblage de réseaux trophiques complexes dans l'espace. Je montre que l'addition de branches dans le réseau trophique (ramification) permet l'accumulation d'espèces dans des chaînes alimentaires distinctes et la création de réseaux trophiques complexes. Je démontre que cette ramification du réseau trophique est due à la distribution spatiale des interactions trophiques ainsi que le support structurel apporté par les boucles omnivores et généralistes.Dans le quatrième chapitre, j'essaye de déterminer si la relation entre la biodiversité et la ramification des réseaux trophiques observée dans le modèle est applicable aux réseaux trophiques naturels. Je montre qu'il existe une forte relation linéaire entre la taille des réseaux trophiques naturels et le nombre de branches qui caractérise leur arbre couvrant minimum. Cette vérification empirique du modèle indique que la théorie développée dans cette thèse pourrait permettre de mieux comprendre les rôles que jouent l'espace et de la dispersion dans l'assemblage et la structure des réseaux trophiques naturels à grandes échelles.