The paper is focused on the Predator-Prey model modified in the case of harvesting one or both populations. Firstly there is given a short description of the basic model and the sensitivity analysis. The first essential modification is percentage harvesting. This model could be easily converted to the basic one using a substitution. The next modification is constant harvesting. Solving this system requires linearization, which was properly done and brought valuable results applicable even for the basic or the percentage harvesting model. The next chapter describes regulation models, which could be used especially in applying environmental policies. All reasonable regulation models are shown after distinguishing between discrete and continuous harvesting. The last chapter contains an algorithm for maximizing the profit of a harvester using econometrical modelling tools.

This dissertation demonstrates the construction of the Panthera Population Persistence (PPP), an individual-based model for the Sumatran tiger (Panthera tigris sumatrae) which provides proper theoretical and application frameworks for the conservation of this tiger sub-species in central Sumatra. The PPP model was developed to gain insight into tiger-preyhabitat relationships as well as the effect of human impacts on the persistence of tiger populations. The model addresses three main problems for the survival of the Sumatran tiger: tiger poaching, prey depletion, and habitat loss. The description of the PPP model serves as an in-depth study of existing literature and covers the most important factors of existing models for tiger conservation. Existing modelling approaches have been improved by the inclusion of finer description of individual-level traits and behaviours in the PPP model. The modelling approach allows a direct inter-relationships between individuals and their environment. The relationship between individual behaviours, intrinsic states, and external factors are simulated spatially explicitly in a bottom-up approach where the emergence of the population dynamics of tiger and prey can be observed under different scenarios. The integration between the PPP model and geographical information system (GIS) has provided a much more meaningful spatial data by revealing the mechanism of the response of individuals to the present land-use types. The relative importance of the parameters within the PPP model was tested using two modes of sensitivity analysis: The Morris Method and the traditional One-factor-at-a-time method. The results provided guidance for the application of reasonable sensitivity analysis during the development of individual-based models. The Morris Method suggested that the overall output of the PPP model showed a high sensitivity on the change of time required by a tigress to take care of cubs. The analysis also revealed that the number of dispersers was sensitive toward perceptual distance of individuals to detect the presence of prey. Comparison with a similar predator-prey models provided insight into the predator-prey relationship. The comparison also suggested that perceptual distance of the individual is important for any spatially explicit individual-based model involving predator-prey relationships. The parameterization of the individual perceptual distance of tigers was tested by using existing literature on prey consumption by tigers as a benchmark. The simulation results were within the range of scientific acceptance for the number of prey killed by a tiger. Thus, further use of the set of parameters for a tiger’s perceptual distance is less uncertain for the output of the PPP model. The effect of habitat quality and landscape configuration on the mortality and migration of prey were evaluated through the use of virtual habitats and landscapes. The findings suggested that a good habitat quality enables prey survival, increases the population available for predation by tigers. When a low-quality habitat is combined with a high-quality habitat, the number of migrating prey was high, reducing resources for tigers. This suggested that landscape composition should be considered when predicting population persistence of the Sumatran tiger. Optimal movement of two different prey resulted in a high density of prey in high-quality habitat, providing a concentration of prey in a tiger’s habitat, but resulted in a lower tiger predation rate than random movement and species specific movement. The PPP model has been applied to evaluate the effect of poaching, prey depletion, and their combination for the probability of extinction of a tiger population. The results from the evaluation showed that prey depletion, tiger poaching, and a combination of both, created a 100% probability of extinction within 20 years if the density and frequency of those threats at high rates. However, the duration of those threats in the system caused a 100% probability of extinction from tiger poaching. The results are able to contribute to optimize anti-poaching programs in future, to reduce significantly the probability of total extinction of Sumatran tiger. Furthermore, various landscape configurations have been tested against the probability and time of extinction for the Sumatran tiger population. The integration of spatial GIS-data in the model provides an insight into the relationship between tiger-prey-habitat. The results suggested that habitat quality surrounding a protected area plays an important role for the persistence of the Sumatran tiger population. This study also recommends agroforestry systems as reasonable land-use type in the vicinity of protected areas. They provide not only positive effects for tiger conservation purpose but they also appear as adaptable to the current land-use situation in Sumatra island.:Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Zusammenfassung . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Contents 12 1 Introduction 15 1.1 Cornerstones of Sumatran Tiger Conservation . . . . . . . . 16 1.2 Scientific Challenges to Tiger Conservation . . . . . . . . . 22 1.3 Roles of Modelling in Tiger Conservation . . . . . . . . . . 26 1.4 Individual-Based Models for Tiger Conservation . . . . . . . 30 1.5 Research Questions . . . . . . . . . . . . . . . . . . . . . . . 31 1.6 Thesis Structures . . . . . . . . . . . . . . . . . . . . . . . . 32 2 Literature Review 34 2.1 Fragmentation and Population Dynamics . . . . . . . . . . . 35 2.2 Population Extinction and its measures . . . . . . . . . . . 37 2.3 Modelling the Effect of Fragmentation on Population Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.4 Individual-Based Modelling of Population Persistence . . . . 51 2.5 Sensitivity Analysis in Individual-based Model . . . . . . . . 53 3 Methods ..........................................................................55 3.1 Study Area . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3.2 Model Description . . . . . . . . . . . . . . . . . . . . . . . 56 3.3 Land-use Map Development . . . . . . . . . . . . . . . . . . 68 3.4 Model Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 69 4 Results 73 4.1 Structure and Sensitivity Analysis of Individual-based Predator- Prey Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.2 Where to Go and How to Hide? Measuring the Relative Effect of Movement Decisions, Habitat Quality, and Landscape Configuration on theMortality andMigration of Tigers’ Prey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.3 The Extinction Potential of a Sumatran Tiger Population after the Removal of Poaching . . . . . . . . . . . . . . . . . 117 4.4 The Influence of Agroforest and Other Land-use Types on the Persistence of a Sumatran tiger (Panthera tigris suma- trae) Population: An Individual-Based Model Approach . . 135 5 General Discussion 159 5.1 Main results . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 5.2 Discussion of the results . . . . . . . . . . . . . . . . . . . . 161 6 Conclusions and Perspectives 170 6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 6.2 Perspectives for Future Research . . . . . . . . . . . . . . . 171 Bibliography 174 Appendices 191 / Die vorliegende Dissertation beschreibt die Entwicklung des Panthera Populations Persistence (PPP) Modells, eines individuenbasierten Simulationsmodells für den Sumatra-Tiger (Panthera tigris sumatrae). Dieses stellt einen geeigneten theoretischen und anwendungsbezogenen Rahmen für den Schutz dieser Tiger-Unterart in Zentralsumatra bereit. Das PPP-Modell wurde entwickelt, um Einblicke in die Tiger-Beute-Habitat-Beziehungen zu gewinnen, sowie um den Effekt anthropogener Einflüsse auf den Fortbestand von Tigerpopulationen abzuschätzen. Dabei werden die drei Hauptprobleme für das Überleben des Sumatra-Tigers analysiert: die Wilderei, der Rückgang von Beutetieren und der Verlust von geeigneten Habitaten. Die Beschreibung des PPP-Modells gibt zunächst einen umfassenden Überblick zum aktuellen Wissensstand auf dem Gebiet des Tigerschutzes und integriert die wichtigsten Faktoren bereits existierender Modellansätze. Diese konnten durch die Einbeziehung einer detaillierten Beschreibung von individuellen Merkmalen und Verhalten verbessert werden. Das PPPModell stellt somit das Individuum in einen direkten Zusammenhang mit dessen Umwelt. Die Beziehung zwischen individuellem Verhalten, intrinsischen Merkmalen und externen Faktoren werden räumlich-explizit in einem bottom-up Ansatz simuliert. Damit kann sowohl die Populationsdynamik des Tigers als auch die seiner Beutetiere unter verschiedenen Annahmen beobachtet werden. Die Verknüpfung des PPP-Modells mit Geographischen Informationssystemen (GIS) bietet die Möglichkeit, die Reaktionsmechanismen der Individuen basierend auf der gegenwärtigen Landnutzungssituation zu simulieren und somit realitätsnahe räumliche Daten zu generieren. Die relative Bedeutung der Modell-Parameter auf die Simulationsergebnisse kann durch Sensitivitätsanalysen ermittelt werden. Hier wurden zwei verschiedene Ansätze verwendet: die Morris-Methode und die herkömmliche One-factor-at-a-time Methode. Der Vergleich beider methodischen Ansätze zeigte somit beispielhaft die Eignung unterschiedlicher Sensitivitätsanalysen für individuenbasierte Modelle auf. Die Morris-Methode zeigte, dass das Gesamtergebnis des PPP-Modells eine hohe Sensitivität gegenüber der Veränderung der Zeit aufweist, die ein Tigerweibchen braucht, um ihre Jungen aufzuziehen. Die Analyse zeigt auch, dass die Anzahl an abwandernden Tigern sensitiv gegenüber der IndividuellenWahrnehmungsdistanz von Beute ist. Der Vergleich mit einem ähnlichen Räuber-Beute-Modell lässt vermuten, dass diese Wahrnehmungsdistanz eines Individuums generell als ein entscheidender Faktor für Räuber-Beute-Beziehungen in räumlich-expliziten Individuenmodellen an- gesehen werden kann. Die Parametrisierung der IndividuellenWahrnehmungsdistanz des Tigers wurde so gewahlt, dass die damit ermittelten Simulationsergebnisse den Beutekonsum des Tigers, wie in der Literatur beschrieben, weitgehen widerspiegeln. Sie ist somit für die weitere Anwendung im PPP-Modell ausreichend gut beschrieben. Simulationsszenarien, welche verschiedene Habitatqualitäten sowie Landnutzungsmuster berücksichtigen, zeigen auch deren Bedeutung für die Mortalität und Migration der Beutetiere. Eine gute Habitatqualität hat eine geringe Mortalität der Beutetiere zur Folge, welche dann wiederum für den Tiger in ausreichender Zahl zur Verfügung stehen. Treten geringe Habitatqualitäten angrenzend an ein Habitat mit hoher Qualität auf, führte dies zu einer hohen Anzahl an abwandernden Beutetieren, womit sich die Ressourcen für den Tiger verringern. Die Landschaftsmerkmale sollten also bei der Vorhersage des Populationsfortbestandes des Sumatra-Tigers berücksichtigt werden. Die optimale Bewegung von zwei verschiedenen Beutetieren ergab eine hohe Beutedichte in einem Habitat mit hoher Qualität und stellte konzentriert Beute in einem Tigerhabitat bereit. Allerdings resultierte dies auch in einer geringeren Prädationsrate des Tigers, verglichen mit zufälligen oder artenspezifischen Bewegungen. Das PPP-Modell wurde angewandt, um die Auswirkungen von Wilderei, Beutetierrückgang sowie die Kombination beider Faktoren auf die Aussterbewahrscheinlichkeit einer Tigerpopulation zu bewerten. Die Ergebnisse zeigen, dass die genannten Faktoren eine 100-prozentige Aussterbewahrscheinlichkeit innerhalb von 20 Jahren zur Folge haben, wenn die Dichte und Häufigkeit dieser Bedrohungen hoch sind. Die Dauer dieser Bedrohungen im System verursachte allerdings eine 100-prozentige Aussterbewahrscheinlichkeit nur für die Wilderei von Tigern. Betrachtet man unabhängig von Dichte und Häufigkeit einzig die Dauer der Bedrohung, führt lediglich die Wilderei zum 100%-igen Aussterben. Diese Ergebnisse können maßgeblich dazu beitragen, zukünftig Schutzprogramme gegen die Wilderei zu optimieren, um das Aussterben des Sumatra-Tigers zu verhindern. DesWeiteren wurde der Einfluss von unterschiedlichen Landnutzungsmustern auf die Aussterbewahrscheinlichkeit und -zeit einer Sumatra-Tigerpopulation aufgezeigt. Die Integration von räumlichen GIS-Daten in das Modell ermöglichte einen Einblick in die Beziehungen zwischen Tiger, Beutetieren und Habitat. Die Ergebnisse zeigen, dass die Habitatqualität um Schutzgebiete herum eine wichtige Rolle für den Fortbestand der Population spielt. Die vorliegende Arbeit empfiehlt Agroforstsysteme als eine geeignete Landnutzungsform in der Nähe von Schutzgebieten, welche sowohl positive Effekte für den Tigerschutz bietet als auch mit den gegenwärtigen Landnutzungsmustern in Sumatra vereinbar erscheint.:Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Zusammenfassung . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Contents 12 1 Introduction 15 1.1 Cornerstones of Sumatran Tiger Conservation . . . . . . . . 16 1.2 Scientific Challenges to Tiger Conservation . . . . . . . . . 22 1.3 Roles of Modelling in Tiger Conservation . . . . . . . . . . 26 1.4 Individual-Based Models for Tiger Conservation . . . . . . . 30 1.5 Research Questions . . . . . . . . . . . . . . . . . . . . . . . 31 1.6 Thesis Structures . . . . . . . . . . . . . . . . . . . . . . . . 32 2 Literature Review 34 2.1 Fragmentation and Population Dynamics . . . . . . . . . . . 35 2.2 Population Extinction and its measures . . . . . . . . . . . 37 2.3 Modelling the Effect of Fragmentation on Population Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.4 Individual-Based Modelling of Population Persistence . . . . 51 2.5 Sensitivity Analysis in Individual-based Model . . . . . . . . 53 3 Methods ..........................................................................55 3.1 Study Area . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3.2 Model Description . . . . . . . . . . . . . . . . . . . . . . . 56 3.3 Land-use Map Development . . . . . . . . . . . . . . . . . . 68 3.4 Model Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 69 4 Results 73 4.1 Structure and Sensitivity Analysis of Individual-based Predator- Prey Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.2 Where to Go and How to Hide? Measuring the Relative Effect of Movement Decisions, Habitat Quality, and Landscape Configuration on theMortality andMigration of Tigers’ Prey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.3 The Extinction Potential of a Sumatran Tiger Population after the Removal of Poaching . . . . . . . . . . . . . . . . . 117 4.4 The Influence of Agroforest and Other Land-use Types on the Persistence of a Sumatran tiger (Panthera tigris suma- trae) Population: An Individual-Based Model Approach . . 135 5 General Discussion 159 5.1 Main results . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 5.2 Discussion of the results . . . . . . . . . . . . . . . . . . . . 161 6 Conclusions and Perspectives 170 6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 6.2 Perspectives for Future Research . . . . . . . . . . . . . . . 171 Bibliography 174 Appendices 191

Killer whales are the oceans’ apex predator and are known to have important effects on ecosystems. At Subantarctic Marion Island, southern Indian Ocean, they have only been studied opportunistically, resulting in limited knowledge of their ecosystem impact here. This dissertation describes the prey and seasonal abundance, estimates the population size and assesses the predatory impact of killer whales on seals and penguins at Marion Island, using dedicated and opportunistic shore-based observations and photographic identification, from 2006 to 2009. During 823 sightings of killer whales at Marion Island (2006 to 2009) 48 predation events were recorded; in only 10 cases could prey be identified. Killer whales fed on fur seals, elephant seals and penguins. Constant effort (dedicated) observations (259 hours, 2008 to 2009) showed that killer whale abundance, which peaked in September to December with a secondary peak in April to May, is linked to the abundance of seals and penguins. Mark-recapture analyses were performed using nearly 10 000 photographs taken from 2006 to 2009. Following careful quality control criteria 37 individuals were identified and a population size of 42 (95% CI = 35-50) individuals estimated using the open population POPAN parameterization in the software program MARK. The analytical approach is more rigorous than that used in any previous population size assessment at Marion Island. Finally, the above data were integrated to assess whether top-down control of seal and penguin populations at Marion Island is generally plausible using a simple process of elimination. Based on published data I predicted the energetic ingestion requirements of adult male and female killer whales as 1 394 MJ.day-1 and 1 028 MJ.day-1, respectively. Expanding these requirements to the 37 killer whales photographically identified at Marion Island, the population requires 40 600MJ.day-1. Based on available energy density and mass data, I predicted the energy content of available seal and penguin prey and calculated the rates at which killer whales would consume these prey in various scenarios. Penguins and Subantarctic fur seals are relatively insensitive to killer whale predation owing to their large population sizes (10 000s to 100 000s), conversely, the smaller populations (100s to 1 000s) of Antarctic fur seals and southern elephant seals are sensitive to predation, particularly the latter as they have a high energy content (approximately 2 000 to 9 000 MJ). Populations of these seals are currently increasing or stable and I conclude that presently killer whale predation is not driving population declines, although they clearly have the potential for regulation of these smaller populations. Thus, if population sizes were reduced by bottom-up processes, if killer whale diet shifted, or if prey availability changed, top-down control by killer whales could become significant. This study provides baseline information for the informed management and conservation of killer whales at Marion Island, identifies avenues for further research, and provides a foundation for the continuation of structured and dedicated killer whale research at Marion Island. / Dissertation (MSc)--University of Pretoria, 2011. / Zoology and Entomology / unrestricted

В работе рассматриваются двумерная популяционная модель типа «хищник–жертва» с учетом конкуренции жертв и конкуренции хищников за отличные от жертв ресурсы, а также трехмерная популяционная модель типа «хищник–две жертвы» с учетом внутривидовой и межвидовой конкуренции жертв и конкуренции хищников за отличные от жертв ресурсы. Проводится анализ существования и устойчивости аттракторов моделей, строятся бифуркационные диаграммы и типичные фазовые портреты. Для стохастических моделей проводится анализ чувствительности аттракторов на основе теории функции стохастической чувствительности. С использованием аппарата доверительных областей: эллипсов и эллипсоидов для равновесий, а также полос и торов – для циклов, изучаются стохастические феномены: переходы между аттракторами, генерация большеамлитудных колебаний, вымирание популяций. Изучаются вероятностные механизмы вымирания популяций. / The thesis considers a two-dimensional population model of the «predator–prey» type, taking into account the competition of preys and competition of predators for resources different from the preys, and also a three-dimensional population model of the «predator–two preys» type, with intraspecific and interspecific competition of preys and competition predators for resources other than preys. An analysis is made of the existence and stability of attractors. Bifurcation diagrams and typical phase portraits are constructed. For stochastic models, an analysis of the sensitivity of attractors is carried out based on stochastic sensitivity function teqnique. Using the confidence domain method: ellipses or ellipsoids for equilibria and bands or tor for cycles, following stochastic phenomena are studied: transitions between attractors, the generation of large amplitude oscillation and the extinction of populations. The probabilistic mechanisms of extinction of populations are studied.

Biological invasions are one of the main factors responsible for the imperiled status of freshwater ecosystems, but much remains to be learned about their indirect effects on native communities. The first part of this dissertation examines community effects of long-term efforts to selectively control invasive Eurasian watermilfoil. Results of the first study show that native plants immediately recolonized treated areas and habitat complexity was unaffected. Fish community was not influenced by invasive plant control. Macroinvertebrate communities were highly variable and part of their variability could be explained by plant community attributes. Both fish and macroinvertebrates used invasive watermilfoil, which emphasizes the need for timely restoration of native macrophytes to mitigate for lost habitat. Because fish and macroinvertebrates were more affected by complexity than other attributes of plant assemblage, reestablishment of habitat complexity appears to be a promising restoration strategy. The second study, which examined species interactions after watermilfoil control, found that fish feeding activity was not correlated with invasive plants or habitat complexity and that invasive macrophyte control did not affect characteristics of fish feeding investigated. The relationship between fish and macrophytes was further explored in the context of interactions between an invasive piscivore and its native prey. First, I examined the prey naiveté hypothesis with non-native peacock bass in Paraná River, Brazil. Prey responded to visual and chemical cues of peacock bass and displayed avoidance behaviors similar to those observed with a native predator, meaning that lack of recognition was not responsible for the observed vulnerability of native species to this introduced predator. After confirming lack of naiveté, I assessed direct and indirect effects of this non-native predator on native prey. Peacock bass had no indirect effects on its prey feeding activity. Macrophyte type did not affect indirect predator-prey interactions, whereas direct predator effects slightly decreased in the presence of aquatic vegetation. I discuss implications of these findings for native biodiversity and convene other potential explanations for the observed effects of peacock bass. Both projects contribute to our understanding of the relationship between aquatic plants and their animal communities and effects of invasive species in freshwater habitats.

No description available.

No description available.

No description available.

The Eurasian spruce bark beetle Dendroctonus micans is a major pest of spruce which is expanding its range in France, Turkey, England and Wales. Its monospecific predator Rhizophagus grandis has followed naturally the bark beetle into most areas and, since the 1960s, has also been mass-produced and successfully released within newly infested locations. In this scope, the development of an effective trapping method would be very useful to assess the bark-beetle presence at previously uninfested sites, or predator establishment after release or natural spread. We demonstrated the efficiency of oxygenated monoterpenes-baited kairomone traps to monitor R. grandis in various epidemiological conditions, including areas localized behind or at the limit of the pest’s distribution, or in areas where artificial releases were performed. Because the predator is strictly species-specific, another exciting possibility offered by the kairomone trapping is the indirect monitoring of the pest itself in areas of unknown status (e.g. areas under colonization, or considered as at risk at medium- term). R. grandis is also considered as one of the most valuable natural enemies to strike aggressive North-American Dendroctonus species. In this respect, R. grandis has been recently applied in a neo-classical biological program against the red turpentine beetle D. valens, which invaded China from North America in the late 1990’s. In laboratory tests conducted on pine logs in the laboratory, or on living pine trees in the field, we demonstrated that R. grandis adults can successfully enter and reproduce into D. valens galleries. In Europe, R. grandis is the only species regularly found in the brood systems of D. micans, where adults and larvae attack the gregarious larvae of their prey. In such enclosed systems, R. grandis’ functional response is therefore influenced by various interrelated components, such as the prey density, the predator density, or the prey distribution. Measuring the predator’s success in terms of larval survival and growth rates, we demonstrated the time spent by R. grandis larvae to wound and kill their prey to be the main factor limiting their development. This factor may be considerably influenced by the proportions of diseased, wounded or molting prey rise in the brood system, for instance as a result of an increase in prey density, or due to the presence of conspecific adults (which wound their prey but do not consume them entirely). Furthermore, our tests suggest that no cannibalism or noticeable intraspecific competition occurred between R. grandis larvae, whereas some lighter mode of competition probably took place. R. grandis also exhibits a reproductive numerical response to prey density, which mainly relies on the perception of chemical stimuli and inhibitors released in the bark beetle brood system. In the current study, we developed a non-destructive approach to follow the dynamics of volatile compound production, using sequential sample collection on SPME fibers. Our tests demonstrated that the larval activity of D. micans or D. valens strongly influences the release of some oxygenated monoterpenes. However, our attempts to correlate the relative quantities of some identified chemicals to offspring production were less successful as it concerns the identification of potential oviposition stimuli and inhibitors. The problematic rose by the progression of D. micans, as well as detailed results of each of the described above studies are discussed in the two published papers and the three manuscripts forming this thesis. Bringing all these studies together, several perspectives are also presented in the general discussion. / Ravageur des épicéas, Dendroctonus micans est toujours en voie d’extension en France, en Turquie, en Angleterre et au Pays de Galles. Dans la plupart de ces zones, le dendroctone est accompagné de manière naturelle par son prédateur monospécifique, Rhizophagus grandis. Depuis les années 1960, le prédateur a également fait l’objet d’une production de masse et de programmes de lâchers dans les zones d’arrivée récente du scolyte. Dans le cadre de la lutte biologique contre D. micans, les gestionnaires forestiers doivent donc estimer au plus tôt la présence du ravageur dans des zones jusque là indemnes, mais également vérifier l’établissement du prédateur par progression naturelle ou résultant d’introductions délibérées. Dans la présente étude, nous avons démontré l’efficacité de pièges d’interception appatés à l’aide de monoterpènes oxygénés pour la capture de R. grandis. Celle-ci s’est faite dans différentes conditions épidémiologiques, incluant notamment des zones situées en arrière du front de progression du scolyte et des zones où des lâchers artificiels ont été réalisés. Comme R. grandis est strictement inféodé au dendroctone, un autre avantage de la technique est la possibilité de réaliser un dépistage indirect du ravageur dans les zones où son statut est incertain (zones en cours de colonisation, ou considérées comme à risque à moyen terme). Par ailleurs, R. grandis est également considéré comme un des meilleurs ennemis naturels potentiels pour lutter contre d’autres espèces de Dendroctonus aggressifs. Dans cette optique, R. grandis a été récemment utilisé dans un programme de lute biologique contre D. valens, ravageur invasif arrivé en Chine dans la fin des années 1990 en provenance d’Amérique du Nord. Nous avons démontré la capacité de R. grandis à s’introduire et à se reproduire dans les galeries de D. valens lors de tests de laboratoire, mais aussi sur des arbres vivants en pinèdes. En Europe, R. grandis est strictement inféodé aux galeries de D. micans, où larves et adultes du prédateur s’attaquent aux larves grégaires du scolyte. Dans ce système clos, la réponse fonctionelle de R. grandis est influencée par plusieurs facteurs étroitement corrélés, la densité de proies, la densité de prédateurs, et la distribution des proies. En mesurant l’efficacité de prédation de R. grandis en termes de survie des larves et de taux de croissance, nous avons démontré l’influence sur leur développement du temps passé par les larves à blesser et à tuer leurs proies. Ce facteur est par ailleurs fortement dépendant de la proportion de larves malades, blessées ou en cours de mue au sein du système ; une proportion qui peut augmenter en réponse à une augmentation de la densité de proies, ou lorsque des adultes sont présents (ceux-ci blessent les proies mais ne les consomment pas entièrement). Enfin, nos tests suggèrent qu’il n’existe que peu de cannibalisme ou de compétition intraspécifique entre larves de R. grandis, tandis que des modes de compétition moins importants prennent vraisemblablement place. R. grandis présente également une réponse numérique reproductive à la densité de proies disponibles, principalement basée sur la perception de stimuli et d’inhibiteurs présents dans les galeries du scolyte. Par la collecte de composés volatils présents dans ces systèmes à l’aide de fibres SPME, nous avons développé une méthode non-destructive pour suivre la dynamique de production des médiateurs chimiques. Nos tests ont démontré que l’activité des larves de D. micans ou D. valens influence fortement la dynamique de production de certains monoterpènes oxygénés. En revanche, il n’a pas été été possible de corréler les différents composés identifiés au nombre de larves de R. grandis présentes dans le système. La problématique soulevée par la progression de D. micans, de même que les résultats détaillés de chacune des études décrites ci-dessus sont discutés dans les deux papiers publiés et les trois manuscrits formant cette thèse. Les différentes perspectives apportées par ce travail sont également présentées dans la discussion générale.

South Georgia is renowned for the abundance of Antarctic krill (Euphausia superba) and a range of krill predators. Variability in krill availability at a range of scales, and the consequences of this for predator-prey interactions, mean that quantifying the spatially explicit demand for krill by those predators is essential to understanding the mechanisms underlying ecosystem changes in the region. In this thesis demand within a distinct study box to the northwest of the island has been assessed. The thesis has three sections; (1) the number of predators; (2) the distribution of predators; and (3) the demand for krill by those predators. (1) Predator densities with confidence intervals were determined from appropriately designed shipboard transect survey; counts of Antarctic fur seals (Arctocephalus gazella), macaroni penguins (Eudyptes chrysolophus), gentoo penguins (Pygoscelis papua), and Antarctic prions (Pachyptila desolata) were adjusted for sea state, distance from observer and dive behaviour. Providing the first at-sea predator density estimates for the region. (2) A comparison of the distribution of female Antarctic fur seals engaged in pup-rearing (using satellite telemetry) and the whole population that were not restricted to a single part of the population (from shipboard transect survey) was undertaken. Using two general additive models based on the relationship between seal distribution (one derived from transect and the other from telemetry) and the physical environment indicated that the spatial distribution of lactating females is representative of the general population. (3) Using the derived predator density, the local krill demand estimate was 2581 tonnes krill per day, a consumption rate of 0.45% per day of the concurrently estimated krill biomass (using shipboard acoustics). Antarctic fur seals accounted for 75% of this demand. This level of demand was less than the increase in biomass resulting from krill growth. However, based on the length-specific demand, determined from concurrent predator diet samples demand exceeded growth for krill >48mm.