NetLogo has become a standard platform for agent-based simulation, yet there appears to be widespread belief that it is not suitable for large and complex models due to slow execution. Our experience does not support that belief. NetLogo programs often do run very slowly when written to minimize code length and maximize clarity, but relatively simple and easily tested changes can almost always produce major increases in execution speed. We recommend a five-step process for quantifying execution speed, identifying slow parts of code, and writing faster code. Avoiding or improving agent filtering statements can often produce dramatic speed improvements. For models with extensive initialization methods, reorganizing the setup procedure can reduce the initialization effort in simulation experiments. Programming the same behavior in a different way can sometimes provide order-of-magnitude speed increases. For models in which most agents do nothing on most time steps, discrete event simulation—facilitated by the time extension to NetLogo—can dramatically increase speed. NetLogo’s BehaviorSpace tool makes it very easy to conduct multiple-model-run experiments in parallel on either desktop or high performance cluster computers, so even quite slow models can be executed thousands of times. NetLogo also is supported by efficient analysis tools, such as BehaviorSearch and RNetLogo, that can reduce the number of model runs and the effort to set them up for (e.g.) parameterization and sensitivity analysis.

Le modèle Aevol est un modèle d'évolution expérimentale in silico développé par Carole Knibbe et Guillaume Beslon pour étudier l'évolution de la structure des génomes. Aevol a permis d'identifier une très forte pression de sélection indirecte vers un certain niveau de variabilité mutationnelle du phénotype : la survie à long terme d'une lignée étant conditionnée à sa capacité à produire des mutations avantageuses sans pour autant produire trop de mutations délétères, un certain compromis entre robustesse et évolvabilité est indirectement sélectionné. Une conséquence de cette pression de sélection indirecte est le rôle central joué par le taux spontané de réarrangements chromosomiques dans la détermination de la structure du génome. Dans ce travail, nous avons modifié le modèle Aevol pour introduire d'une part un processus explicite de régulation de l'expression des gènes et d'autre part, une sensibilité aux similarités entre séquences dans les événements de recombinaison de l'ADN. Nous avons ainsi pu étudier l'effet de ces variations sur la sélection de second-ordre. Nous avons en particulier observé que celle-ci est extrêmement robuste aux choix de modélisation : les effets liés aux réarrangements sont en effet observés de la même façon lorsque les organismes possèdent un réseau de régulation (qui plus est, ces effets sont visibles sur le réseau lui-même), lorsque les réarrangements se produisent préférentiellement entre séquences similaires et lorsque les transferts horizontaux sont possibles. De plus, les effets de cette pression de sélection de second-ordre ne sont pas limités au niveau génomique : de forts taux de réarrangements tendent à donner lieu à des génomes présentant beaucoup d'opérons, très peu d'ARNs non-codants et des réseaux de régulation très simples. Au contraire, chez les organismes ayant évolué avec de faibles taux de réarrangement, la plupart des gènes sont transcrits sur des ARNs monocistroniques. Ces organismes possèdent un grand nombre d'ARNs non-codants et présentent des réseaux de régulation très complexes. Ces effets observés dans le modèle à différents niveaux d'organisation peuvent s'apparenter à de nombreuses caractéristiques observées chez les organismes réels. Ainsi les pressions sélectives indirectes observées grâce au model Aevol permettent de reproduire un large spectre de propriétés biologiques connues en ne modifiant que le seul taux de réarrangements dans le modèle. Ces mécanismes de sélection indirecte apparaissent donc comme de bons candidats pour expliquer ces mêmes observations sur les organismes réels. / The Aevol model is an in silico experimental evolution model that was specifically developped by Carole Knibbe to study the evolution of the structure of the genome. Using Aevol, a very strong second-order selective pressure towards a specific level of mutational variability of the phenotype was revealed: it was shown that since the survival of a lineage on the long term is conditionned to its ability to produce beneficial mutations while not loosing those previously found, a specific trade-off between robustness and evolvability is indirectly selected. A consequence of this indirect selective pressure is the central role played by the spontaneous rate of chromosomal rearrangements in determining the structure of the genome. More specifically, it was shown that because some rearrangements (large duplications and large deletions) have an impact not only arround their breakpoints but on the whole sequence between them, non-coding sequences are actually mutagenic for the coding sequences they surround. The consequence is a clear trend for organisms having evolved under high rearrangement rates to have very short genomes with hardly any non-coding sequences while organisms evolving in the context of low rearrangement rates have huge, mostly non-coding genomes. Here, we modified the Aevol model to introduce an explicit regulation of gene expression as well as a sensitivity to sequence similarity in DNA recombination events. We observed that the effects of the second-order pressure mentioned above are very robust to modelling choices: they are similarly observed when gene regulation is made available, when rearrangements occur preferentially between similar sequences and even when a biologically plausible process of horizontal transfer is allowed. Moreover, the effects of this second-order selective pressure are not limited to the genomic level: high rearrangement rates usually lead to genomes that have many polycistronic RNAs, almost no non-coding RNAs and very simple regulation networks. On the contrary, at low rearrangement rates organisms have most of their genes transcribed on monocistronic RNAs, they own a huge number of coding RNAs and present very complex and intricate regulation networks. These astounding effects at different levels of organization can account for many features found on real organisms. Thus, the indirect selective pressure that was identified thanks to the Aevol model allows to reproduce a large panel of known biological properties by changing the sole spontaneous rearrangement rate, making this pressure a good candidate for explaining these observations on real organisms.

Le déplacement animal joue un rôle déterminant dans la structuration spatiale et la dynamique des populations biologiques, en particulier des espèces fortement mobiles. L’espace et l’environnement font ainsi partie intégrante du cycle de vie des tortues marines. Ce travail de thèse propose de caractériser l’écologie spatiale des tortues marines, du stade juvénile au stade adulte, dans le Sud-ouest de l’océan Indien, principalement par l’usage de deux méthodes : la télémétrie satellitaire et la modélisation individu-centrée. Il montre en premier lieu que la phénologie de la reproduction de la tortue verte à travers la région est principalement liée à la température de surface de la mer au voisinage des sites de reproduction. Sont ensuite étudiés les patrons de dérive des nouveau-nés générés par les courants océaniques qui impacteraient inégalement leurs traits d’histoire de vie selon l’emplacement du site de naissance. Concernant le stade immature, les résultats suggèrent un cycle de développement trans-équatorial pour la tortue caouanne dans l’océan Indien. Pour le stade adulte, cette étude caractérise les couloirs et la connectivité migratoires de la tortue verte dans la région. Enfin, l’intégration de ces résultats permet de comprendre la structuration des patrons migratoires régionaux et leur influence sur la dynamique des populations. L’ensemble des connaissances acquises fournit un support concret d’aide à la décision pour la mise en place de plans de gestion et de conservation des tortues marines dans le Sud-ouest de l’océan Indien. Cela souligne l’importance d’une approche à grande échelle pour la protection d’un patrimoine biologique partagé par plusieurs nations. / Animal movement is crucial to the ecology of spatially structured population, particularly for highly mobile species. Marine turtles’ life cycle is indeed closely related to spatial and environmental factors. This work analyses the spatial ecology of marine turtles, from early juvenile to adult stages, in the Southwest Indian Ocean, primarily through the use of two methods: satellite tracking and individual-based modeling. Firstly, this analysis argues that green turtle’s reproductive phenology across the region is mainly related to the sea surface temperature in the vicinity of the nesting site. Then, it shows how drifting trajectories of hatchlings in oceanic currents unevenly influence their life history traits depending on the position of the natal site. By tracking late juvenile stage, this work also suggests a trans-equatorial developmental cycle for loggerhead turtle in the Indian Ocean. At adult stage, it describes migratory corridors and connectivity for green turtle across the region. Finally, an integrative approach considering all these results allows for an understanding of the regional migratory patterns and their influence on population dynamics. The results of this work provide a practical policy decision tool for management and conservation of marine turtles in the Southwest Indian Ocean and highlight the need for a large-scale approach in the protection of biological resources and heritage shared by multiple nations.

Despite their important ecosystem benefits for terrestrial and marine flora and fauna and the human livelihood mangrove forests suffer a high loss rate mainly due to human activity. Aside from these impacts, natural forest disturbances exist more commonly in mangroves compared to other forests as a direct consequence of their exposed coastal location. Within this thesis I investigate the influence of natural disturbance regimes on the mangrove forest dynamics focusing in particular on the ecological role of disturbances, disturbance patterns, forest structure, succession behavior and long-term vulnerability evaluation. The study areas were set in the Indian River Lagoon in Florida (USA) and in Can Gio an UNESCO Biosphere Reserve (Vietnam). In addition, theoretical simulation studies were carried out to complement the field studies. Thereby, in our study at the Indian River Lagoon site I investigated the ecosystem response to hurricane events of an artificially impounded mangrove forest. In Can Gio, the suitability of lightning strike – caused gaps for setting a homogenous plantation into more natural-like state according to species composition and forest structure was analyzed. Finally, a theoretical simulation study was carried out to compare lightning strike and hurricane events regarding their homogenization and heterogenization effects on the spatio-temporal forest structure. The findings of the field study in the Indian River Lagoon indicate that hurricane events had a severe impact on forest areas in higher successional stages by creating open patches, whereas areas in lower successional stages remained largely undisturbed. Furthermore, the impoundment determines the species selection of the post-hurricane succession by favoring flooding-tolerant species. However, regeneration was found to be impaired by the artificially high inundation regime at some disturbed patches. The lightning-strike disturbances enhance the species composition in the monospecific plantation in Can Gio by providing a sufficient light regime for entering seeds to establish. In addition, lightning-strike gaps increased the plantation structure complexity. Regenerating lightning-strike gaps remained as “green islands” within windthrow sites in the plantation due to their low stature and provided seeds for surrounding disturbed areas thereby accelerating their recolonization. The results of the simulation analysis of a theoretical landscape showed that in the simulated highly complex natural mature forests all disturbance regimes entail homogenization on the spatial structure compared to an undisturbed scenario. The hurricane scenario showed an increased temporal variation of the forest dynamics whereas lightning-strike gaps were not able to contribute to additional heterogeneity in the simulated area, despite of having the same tree mortality probability during disturbances. The interaction of the large-scale impoundment in the Indian River Lagoon and medium-sized hurricane events is characterized by partially impeded post-hurricane regeneration. In contrast, small-scaled lightning strikes influenced the regeneration of medium-sized windthrow sites positively within the homogenous plantation. We therefore suggest management activities aimed at creating small clearances within the plantation in Can Gio to simulate additional small-scale disturbances in order to facilitate heterogenization of the plantation structure. Natural disturbances are found to be able to enhance the species diversity and the interactions of ecological processes. In particular, where sustainable management strategies focused on maintaining ecosystem services especially in restored sites or plantations act as a supportive part. Natural disturbances are an integral component of mangrove forests and fulfill specific ecological functions. However, our findings indicate that these disturbances, on top of altered environmental conditions associated with climate change and direct human impacts, might jeopardize the natural development in unnatural forest structures as on plantations or restored sites. This thesis gives an extensive overview about the effect of various disturbances in different mangrove forest systems, including semi-natural forests and strongly modified plantations, on species composition and forest structure. Field studies and simulation analyses contribute in equal parts to the results of the thesis.

Research Doctorate - Doctor of Philosophy (PhD) / The original contribution of this thesis is to demonstrate the use of a generalised and parsimonious approach to building individual-based computer models of ecological systems with the objective of advancing our mechanistic understanding of these systems. Two models are presented; the first, a model of predator-prey interaction, produces the expected non-linear dynamics and illustrates the importance of the timing of variable updating and individual variation for the persistence of the populations. This model is applied to two near-exclusive, cycling predator-prey systems, those of the Canadian lynx and snowshoe hare and the Fennoscandian mustelid predators and their microtine prey. The reproduction of the patterns found in the empirical data of these systems by the model suggests that the underlying mechanism of these predator-prey systems may be more simple than is suggested by other more complex models reported in the literature. The second model describes a system similar to that of a grazing herbivore in a two-dimensional space. The emergence of complex behaviour resulting from the use of space in the model, including metapopulation-like local extinction and re-population and the effects of corridors and edge qualities on the species are demonstrated. The inclusion of a graphical display of the two-dimensional space in the computer interface to the model reveals important details of system behaviour not evident in the population means, including herding behaviour. The latter is dependent on herbivore mobility and the re-growth of resources in an heterogeneous environment, and emerges in the absence of social behaviour. The problem of detecting herding behaviour automatically is addressed, including the development of qualitative and quantitative definitions of herding in the model.

La faible représentation de la biodiversité dans les modèles de végétation a longtemps été un obstacle à la compréhension et à la projection des processus écosystémiques. La forte biodiversité des forêts tropicales, leur rôle clé dans les cycles biogéochimiques globaux, ainsi que leur vulnérabilité aux perturbations anthropiques directes et indirectes, amplifient les difficultés et enjeux de ces questions de recherche. En particulier, l'augmentation prédite de la fréquence et de l'intensité des sécheresses pourrait impacter la structure et composition floristique de ces forêts, comme dors et déjà observé au cours d'expériences naturelles et artificielles. Cette thèse explore ces questions de recherche à travers deux approches complémentaires, de modélisation et de mesures écophysiologiques. Dans le premier chapitre, je décris un simulateur de croissance forestière individu-centré et spatialement-explicite, TROLL, qui intègre les progrès récents en physiologie des plantes. Les processus sont paramétrés à l'aide de traits fonctionnels espèce-spécifiques, pour une forêt tropicale amazonienne. Une régénération forestière est simulée, et validée par des observations faites en Guyane française. La sensibilité du modèle à plusieurs paramètres globaux clés est évaluée. Enfin, l'influence de la variation de la richesse et composition spécifiques sur les propriétés écosystémiques est explorée. La réponse des forêts tropicales à la sécheresse est mal connue, empêchant la représentation pertinente des processus en jeu dans les modèles de végétation. Les chapitres 2 à 5 de cette thèse ont ainsi pour but de documenter la tolérance à la sécheresse et sa diversité dans une forêt amazonienne. Une méthode récente et rapide de détermination d'un trait de tolérance des feuilles à la sécheresse, le potentiel hydrique des feuilles au point de perte de turgescence (ptlp), est validée et utilisée, permettant de quantifier pour la première fois un tel trait de tolérance à la sécheresse dans une forêt amazonienne à l'échelle de la communauté. Ce jeu de données permet l'exploration des déterminants de la tolérance à la sécheresse des feuilles, à travers les espèces d'arbres, les tailles des individus, les stades de succession, les expositions à la lumière, ainsi que les lianes. La variabilité de ptlp observée suggère une large diversité de réponses à la sécheresse au sein des communautés de plantes amazoniennes. Ceci est confirmé par le suivi direct du flux de sève au cours d'une saison sèche sur divers arbres de canopée. Enfin, je discute les implications de ces résultats pour le développement des futurs modèles de végétation. / A great part of uncertainties in our current understanding and projections of the carbon cycle lies in the vegetation compartment. The problem of biodiversity representation in vegetation models has long been an impediment to a detailed understanding of ecosystem processes. The high biodiversity of tropical forests, their disproportionate role in global biogeochemical cycles, together with their vulnerability to direct and indirect anthropogenic perturbations, amplify the relevance of this research challenge. In particular, the predicted increase in drought intensity and frequency in the tropics may impact forest structure and composition, as already observed in natural and artificial experiments. This thesis explores how new advances in modelling and ecophysiology should help improve our understanding of these processes in the future. In the first chapter, I describe an individual-based and spatially-explicit forest growth simulator, TROLL, that integrates recent advances in plant physiology. Processes are linked to species-specific functional traits parameterized for an Amazonian tropical rainforest. This model is used to simulate a forest regeneration, which is validated against observations in French Guiana. Model sensitivity is assessed for a number of key global parameters. Finally, we test the influence of varying the species richness and composition on ecosystem properties. Tropical forest response to drought is not well understood, and this hampers attempts to model these processes. In chapters 2 to 5 I aimed at documenting drought-tolerance and its diversity in an Amazonian forest. A rapid method of determination of a leaf drought tolerance trait, the leaf water potential at turgor loss point (ptlp), was validated and applied to a range of plant species. We established the first community-wide assessment of drought tolerance in an Amazonian forest. These results inform on the drivers and determinants of leaf drought tolerance, across tree species and lianas, tree size, successional stages, light exposition, and seasons. Variability in ptlp among species indicates the potential for a range of species responses to drought within Amazonian forest communities. This is further confirmed by direct monitoring of whole-plant water use on diverse canopy trees during a marked dry season. Finally, I discuss the implications of these results to increase the dialogue between the vegetation modeling community and ecology, to enhance model's predictive ability, and to inform policy choices.

Les changements climatiques ont un impact certain sur les écosystèmes marins. Un déplacement vers le sud des principaux systèmes de front servant de sites de nourrissage pour de nombreuses espèces de prédateurs supérieurs est susceptible de se produire dans les zones subantarctiques. Les « central place foragers », tels que les pinnipèdes, sont susceptibles de faire face à une augmentation de la distance entre leurs lieux d'alimentation et leurs colonies d'élevage. Nous avons étudié l’impact des changements climatiques sur le succès d’élevage et la dynamique de population des otaries à fourrure (Arctocephalus gazella) des Îles Kerguelen par le biais du développement du modèle individu centré, MarCPFS (Marine Central Place Foragers Simulator) qui montre que la survie du couple femelle-jeune est particulièrement sensible à la répartition des proies (abondance et structure), à la capacité de mémorisation des meilleurs sites de nourrissages trouvés par la femelle pendant la période d'élevage, à la taille des femelles et à la distance qu'il faut parcourir pour trouver la ressource. Les résultats suggèrent qu’au cours des trois prochaines décennies un déplacement vers le sud supérieur à 2km an 1 pourrait compromettre la survie et la durabilité des populations. Un couplage avec un modèle de simulation de la dynamique océanique et de la ressource (SEAPODYM) a permis des projections jusqu’à la fin de ce siècle basées sur le scénarios RCP8.5 du GIEC (2014), confirmant les tendances mises à jour précédemment. Ces travaux montrent que les perspectives de maintien des populations sont pessimistes y compris en envisageant une adaptation de la taille des individus. / Climate change has certain impact on the marine ecosystems. A southward shift in productive frontal systems serving as the main foraging sites for many top predator species is likely to occur in the Subantarctic areas. Central place foragers, as seabirds and pinnipeds, are thus likely to cope with an increase in the distance between foraging locations and their land-based breeding colonies. We studied the impact of climate change on the breeding success and population dynamics of Antarctic fur seals (Arctocephalus gazella) at Kerguelen Islands by means of an individual based model we developped, MarCPFS (Marine Central Place Foragers Simulator) which showed that the survival of the female-pup pair is particularly sensitive to the distribution of preys (abundance and structure), to the memorization abilities of the best resource sites found by the female during the rearing period, to the female size and to the foraging distance which it is necessary to cover at each trip. The results suggest that during the next three decades a southward shift greater than 2 km year-1 could compromise the survival and the sustainability of the populations. A coupling with a model of simulation of the oceanic dynamics and the resource (SEAPODYM) allowed projections till the end of this century based on scenarios RCP8.5 of the IPCC (2014), confirming the trends previously obtained. Globally, these works lead us to conclude with pessimistic perspectives about the sustainability of populations even when considering an evolution through time towards individuals of greater sizes.

Despite their important ecosystem benefits for terrestrial and marine flora and fauna and the human livelihood mangrove forests suffer a high loss rate mainly due to human activity. Aside from these impacts, natural forest disturbances exist more commonly in mangroves compared to other forests as a direct consequence of their exposed coastal location. Within this thesis I investigate the influence of natural disturbance regimes on the mangrove forest dynamics focusing in particular on the ecological role of disturbances, disturbance patterns, forest structure, succession behavior and long-term vulnerability evaluation. The study areas were set in the Indian River Lagoon in Florida (USA) and in Can Gio an UNESCO Biosphere Reserve (Vietnam). In addition, theoretical simulation studies were carried out to complement the field studies. Thereby, in our study at the Indian River Lagoon site I investigated the ecosystem response to hurricane events of an artificially impounded mangrove forest. In Can Gio, the suitability of lightning strike – caused gaps for setting a homogenous plantation into more natural-like state according to species composition and forest structure was analyzed. Finally, a theoretical simulation study was carried out to compare lightning strike and hurricane events regarding their homogenization and heterogenization effects on the spatio-temporal forest structure. The findings of the field study in the Indian River Lagoon indicate that hurricane events had a severe impact on forest areas in higher successional stages by creating open patches, whereas areas in lower successional stages remained largely undisturbed. Furthermore, the impoundment determines the species selection of the post-hurricane succession by favoring flooding-tolerant species. However, regeneration was found to be impaired by the artificially high inundation regime at some disturbed patches. The lightning-strike disturbances enhance the species composition in the monospecific plantation in Can Gio by providing a sufficient light regime for entering seeds to establish. In addition, lightning-strike gaps increased the plantation structure complexity. Regenerating lightning-strike gaps remained as “green islands” within windthrow sites in the plantation due to their low stature and provided seeds for surrounding disturbed areas thereby accelerating their recolonization. The results of the simulation analysis of a theoretical landscape showed that in the simulated highly complex natural mature forests all disturbance regimes entail homogenization on the spatial structure compared to an undisturbed scenario. The hurricane scenario showed an increased temporal variation of the forest dynamics whereas lightning-strike gaps were not able to contribute to additional heterogeneity in the simulated area, despite of having the same tree mortality probability during disturbances. The interaction of the large-scale impoundment in the Indian River Lagoon and medium-sized hurricane events is characterized by partially impeded post-hurricane regeneration. In contrast, small-scaled lightning strikes influenced the regeneration of medium-sized windthrow sites positively within the homogenous plantation. We therefore suggest management activities aimed at creating small clearances within the plantation in Can Gio to simulate additional small-scale disturbances in order to facilitate heterogenization of the plantation structure. Natural disturbances are found to be able to enhance the species diversity and the interactions of ecological processes. In particular, where sustainable management strategies focused on maintaining ecosystem services especially in restored sites or plantations act as a supportive part. Natural disturbances are an integral component of mangrove forests and fulfill specific ecological functions. However, our findings indicate that these disturbances, on top of altered environmental conditions associated with climate change and direct human impacts, might jeopardize the natural development in unnatural forest structures as on plantations or restored sites. This thesis gives an extensive overview about the effect of various disturbances in different mangrove forest systems, including semi-natural forests and strongly modified plantations, on species composition and forest structure. Field studies and simulation analyses contribute in equal parts to the results of the thesis.

NetLogo has become a standard platform for agent-based simulation, yet there appears to be widespread belief that it is not suitable for large and complex models due to slow execution. Our experience does not support that belief. NetLogo programs often do run very slowly when written to minimize code length and maximize clarity, but relatively simple and easily tested changes can almost always produce major increases in execution speed. We recommend a five-step process for quantifying execution speed, identifying slow parts of code, and writing faster code. Avoiding or improving agent filtering statements can often produce dramatic speed improvements. For models with extensive initialization methods, reorganizing the setup procedure can reduce the initialization effort in simulation experiments. Programming the same behavior in a different way can sometimes provide order-of-magnitude speed increases. For models in which most agents do nothing on most time steps, discrete event simulation—facilitated by the time extension to NetLogo—can dramatically increase speed. NetLogo’s BehaviorSpace tool makes it very easy to conduct multiple-model-run experiments in parallel on either desktop or high performance cluster computers, so even quite slow models can be executed thousands of times. NetLogo also is supported by efficient analysis tools, such as BehaviorSearch and RNetLogo, that can reduce the number of model runs and the effort to set them up for (e.g.) parameterization and sensitivity analysis.

No description available.