Дисертації з теми "Species distribution modelling (SDM)"

Xylosandrus compactus et X. crassiusculus sont deux scolytes originaires d’Asie du Sud-Est et invasifs sur plusieurs continents, dont la biologie et l’écologie atypiques favorisent l’invasion. Une approche pluridisciplinaire a été utilisée au cours de cette thèse afin (i) d’identifier l’origine des populations invasives et leurs routes d’invasion, et (ii) de déterminer les zones dans lesquelles elles pourraient s'étendre et s’établir. Les routes d’invasion ont été retracées à l’aide d’un marqueur mitochondrial et de marqueurs génomiques, et les zones favorables à l’établissement de chaque espèce ont été déterminées à l’aide de modèles de distribution d’espèces (SDM). Malgré leur proximité écologique et phylogénétique, les deux espèces ont une histoire d’invasion différente. Deux lignées ont été identifiées chez X. compactus, l’une originaire d’Inde ou du Vietnam ayant envahi l’Afrique et l’autre originaire de la région de Shanghai et ayant envahi indépendamment les Amériques et les îles du Pacifique, puis l’Europe. X. crassiusculus est composé de deux clusters très divergents, majoritairement allopatriques et possédant des niches écologiques différentes. Le cluster 1 a envahi indépendamment les îles du Pacifique et l’Afrique. Le cluster 2 est responsable de l’invasion en Amérique, en Europe, en Afrique et en Océanie, avec plusieurs invasions indépendantes de multiples origines (dont des événements dits "tête de pont") suivies de dispersion intra-continentale. Les SDM ont montré pour les deux espèces l’existence de zones favorables où elles ne sont pas encore présentes et qui sont donc susceptibles d’être envahies secondairement. Nous anticipons également un impact du changement climatique sur leurs potentielles distributions futures. A l'inverse, l'évolution récente du climat n'est pas responsable de l'invasion récente de l'Europe, ce continent étant favorable depuis plusieurs décennies
Xylosandrus compactus and X. crassiusculus are two ambrosia beetles originating from Southeastern Asia and invasive on several continents, whose atypical biology and ecology favour invasion. During this thesis, a multidisciplinary approach was used to (i) identify the origin of invasive populations and their invasion routes and (ii) determine the areas in which they could spread and establish. Invasion routes were traced using a mitochondrial marker and genomic markers, and the suitable areas for each species were identified using species distribution models (SDM).Despite their ecological and phylogenetic proximity, the two species have different invasion histories. Two lineages were identified in X. compactus, one originating from India or Vietnam, who invaded Africa, and the other from the Shanghai area, who independently invaded the American-Pacific and Europe. X. crassiusculus comprises two very diverging clusters, mostly allopatric and with different ecological niches. Cluster 1 independently invaded Pacific islands and Africa. Cluster 2 is responsible for the invasion in the Americas, Europe, Africa and Oceania, with several independent introductions from multiple sources (including bridgehead, where an invasion occurs from an area already invaded) followed by intra-continental dispersion. For both species, SDM showed suitable areas where the pests are not present yet and which could be invaded. We expect an impact of climate change on their future potential distributions. Conversely, the recent evolution of climate is not responsible for their recent invasion in Europe, which has already been suitable for decades

Les invasions biologiques, deuxième cause de perte de biodiversité à l’échelle mondiale, représentent un risque majeur auquel nos sociétés doivent faire face. On parle d’invasion biologique lorsque des activités humaines permettent à une espèce de franchir des barrières qui jusqu’alors limitaient sa dispersion ou sa multiplication, entrainant une explosion géographique et démographique de l’espèce dans un nouvel écosystème et s’accompagnant éventuellement d’impacts économiques, sociétaux ou écologiques. La façon la plus efficace et la moins coûteuse de limiter les impacts causés par les espèces envahissante et de les prévoir en amont afin de mettre en place des mesures de prévention ciblées et efficaces et d’essayer de les éviter.L’objectif de cette thèse est de démontrer qu’il est possible d’améliorer la prédiction des invasions biologiques en développant et combinant différentes approches de modélisation de façon innovante. Les questions posées sont de savoir s’il est possible de prévoir quelles pourraient-être ces espèces, où elles pourraient devenir envahissantes et quels impacts elles pourraient avoir. Toutes les méthodes développées dans cette thèse ont été appliquées à des hyménoptères sociaux ; nombreuses espèces de fourmis (famille des Formicidés) ou frelon asiatique (famille des Vespidés, vespa Velutina nigrithorax), mais elles sont généralisables et réutilisables pour tout autre taxa.Je montre dans cette thèse qu’il est possible de développer un outil statistique de détection des espèces risquant de devenir envahissantes. J’applique cet outil aux fourmis, je fournis une liste de 15 espèces de fourmis risquant de devenir envahissantes à travers le monde et je cartographie les zones du globe risquant d’être envahies par ces espèces. Chacun des continents est menacée par au moins une de ces invasions potentielles. Dans un deuxième temps, je mets au point d’un cadre méthodologique permettant d’améliorer les prédictions d’aires de distribution des espèces en combinant leurs exigences climatiques et d’habitat, tout en respectant l’échelle géographique à laquelle ces facteurs agissent sur la distribution des espèces. J’applique cette méthode au frelon asiatique, ce qui me permet d’identifier les habitats qui lui sont les plus favorables et d’utiliser ces informations pour raffiner la prédiction de son aire favorable. En combinant le climat et l’habitat, je prédis une aire potentielle de distribution 56% plus restreinte par rapport aux estimations basées sur le climat uniquement. Enfin, dans un troisième temps, je m’intéresse au développement d’une méthode permettant de prédire spatialement les impacts causés par une invasion biologique. Pour ce faire, je prédis d’abord l’abondance potentielle du frelon asiatique en France. Je couple ensuite cette prédiction avec des données de présence de ruches et un modèle présidant l’impact du frelon asiatique sur la survie des colonies d’abeilles. J’estime enfin que cette invasion peut conduire à l’effondrement de 41% des colonies d’abeilles domestiques en France.Cette thèse met en lumière l’utilité d’intégrer la modélisation dans la construction du savoir autour des invasions biologiques, approche relativement nouvelle dans ce champ disciplinaire. De plus, elle illustre comment la modélisation et l’élaboration de prédictions peuvent aider à objectiver la prise de décision concernant la gestion des espèces envahissantes et optimiser leur efficacité en ciblant les habitats, les régions et les espèces d’action prioritaires
Biologicals invasions, the second cause of biodiversity loss worldwide, represent a major threat that our societies have to face. Invasive species correspond to species that, due to human activities, cross geographic and reproduction barriers and expand into new areas in large numbers. This spread into new ecosystems may have severe socio-economic or ecological impacts. The most efficient way to limit these impacts is to predict and avoid biological invasions before they occur by setting up appropriate management plans.The aim of this PhD thesis is to demonstrate that existing predictive models can be further developed and combined together to improve biological invasion predictions. All of the methods developed in this thesis have been applied to social Hymenoptera: ant species (Formicidae) and the Asian hornet (Vespa velutina nigrithorax), but they are generalizable to any other taxa. The questions asked are: can we predict future invader species? Can we improve the spatial predictions of their distribution? Can we predict invasive species impact?First, I show in this thesis that it is possible to develop a model that detects future invasive species, even before they have had the chance to be moved outside their native range. I apply this screening tool to more than 2000 ant species, provide a list of the 15 ant species that are highly likely to become invasive and map their global suitability to highlights the area the most at risk from these invasions. All continents are threatened by at least one of these potential invasions. Second, I set up a methodological framework to improve species distribution predictions by combining multi-scale drivers. I apply this method to the invasive Asian hornet, identify its high affinity habitats, and use this information to refine suitability maps. I show that integrating multiple drivers, while still respecting their scale of effect, produced a potential range 55.9% smaller than that predicted using a climatic model alone. Finally, I propose a method to predict invasive species impacts in a spatially explicit way and I apply it to the estimate the Asian hornet’s impact on honeybee colonies in France. To do so, I estimate the Asian hornet nest density across France and combine it with an agent-based hive model to estimate honeybee mortality risk. I show that up to 41% of the honeybee colonies are likely to collapse due to the Asian hornet.Overall, these studies demonstrate how modelling techniques can provide valuable inputs to improve invasive species management decision by offering tools to optimize prevention strategies and target areas, species or habitats where action is needed in priority. Biological invasions involve our scientific, political and cultural perceptions in an intricate way; this PhD thesis highlights the usefulness of bringing together modelling techniques and the rest of biological invasion knowledge to better grasp invasion science complexity

Main goal of forest diseases’ management is to reduce economic, biological and aesthetic damages and biodiversity loss caused by plant parasites. The many strategies used can be grouped under two main actions, prevention (prophylaxis in some early writings) and therapy (treatment or cure). Prevention is limited primarily by the lack of knowledge of the organisms involved, including host plants. Mathematical models have been used to extend the understanding of plant disease epidemiology on a number of fronts, providing an opportunity for a more rational use of resources on expensive field trials and representing a step towards more sustainable control measures. From a curative point of view, current efforts by scientists have focused on developing diseases management (Pest Management = PM) concepts in order to balance the benefits of pesticides with the ecological concerns of their residues contaminating the environment. In this thesis, the two PM principles were applied from an innovative point of view on two case studies: ash dieback caused by Hymenoscyphus fraxineus, which can be considered the most serious disease for Fraxinus genus in Europe, and chestnut ink disease, caused by Phytophthora cambivora and P. cinnamomi. In the first part of the thesis, the two diseases are introduced, in order to permit the evaluation of similarities and differences (chapter I). Subsequently, from chapter II to chapter V, the experimental trials performed are described. In particular, in chapter II a study of the ecological niche of H. fraxineus, with the characterization of the environmental variables associated with naturally infected zones, is reported. This procedure was realized with Species Distribution Models (SDM), widely utilized in the ecological field and only recently applied to plant pathology. The presence of the pathogen was highly correlated to three summer predictors: abundant precipitation, high soil moisture and low air temperature, in comparison with the averages of the study area. The ensemble forecasting technique was then applied to obtain a prediction of the potential distribution of the pathogen at European scale, considering the distribution maps of Fraxinus excelsior and Fraxinus angustifolia, susceptible to the parasite. At last, an innovative method of network analysis permitted to identify the suitable areas that are not reachable by the pathogen with a natural spread. Chapter III reports a study conducted to evaluate six fungicides for their potential to control ash dieback. Initially, in vitro tests of the active ingredients against five different strains of the pathogen indicated thiabendazole, propiconazole and allicin as the most effective fungicides, with lower median lethal doses than procloraz. In contrast, copper sulphate and potassium phosphite were totally ineffective. Subsequently, the antifungal activities of the best three compounds were investigated in planta against H. fraxineus by trunk injection on European ashes inoculated with an indigenous strain. The test was preceded by preliminary trials to maximize the efficacy of injections; in the experimental conditions highest speed was reached with the addition of 1.2 % acetic acid to the aqueous solution and making treatments in early morning or late afternoon. Considering the results of in planta trial, thiabendazole and allicin significantly slowed down the growth of the necroses in the growing season, in contrast propiconazole injections were impracticable. The studies in chapters IV and V recall the methodologies applied to ash dieback, with application to chestnut ink disease complex. In particular, in chapter IV fuzzy logic theory was applied considering the environmental variables, such as minimum winter temperature, summer drought, slope's aspect, streams' distance and soil's permeability, that mainly can influence the development of the disease. The model was validated with a broad field survey conducted in a chestnut area in Treviso province. Moreover, uncertainty maps (regarding model structure, inputs and parameters) were produced for the correct interpretation of the prediction. Great part of the chestnut area in the study zone resulted as suitable for the development of ink disease, whereas only the 18.8 %, corresponding to higher elevation zones, presented inferior risks. In a second study (chapter V), a comparative efficacy trial on four potassium phosphite formulations by means of endotherapy against chestnut ink disease is performed. P. cinnamomi was isolated with baiting technique from symptomatic chestnuts and was inoculated on 50 asymptomatic trees. As a result of endotherapic treatments, the unique solution that significantly slowed down necroses' growth was potassium phosphite (35 %) with an addition of 0.1 % micronutrient solution. An additional endotherapic trial was conducted in a preliminary way in the chestnut where P. cinnamomi was isolated, with the main aim to evaluate growth stimulation of active growing callus next to the shape flame necroses by the injected solution of potassium phosphite 70 %. In this case, results did not highlight a significant difference between treated trees and water control ones, probably for the need of longer times for older trees. On the base of the achieved results, epidemiological modelling and endotherapic treatments, applied both to ash dieback and chestnut ink disease, can represent fundamental tools in the management of these important diseases and should be applied in an Integrated Pest Management (IPM) approach, together with appropriate cultural techniques to maximize benefits.
Lo scopo principale della gestione delle malattie forestali è la riduzione dei danni economici, biologici ed estetici e delle perdite di biodiversità dovute alle malattie delle piante. Le molteplici strategie usate nella gestione delle malattie possono essere raggruppate in due azioni principali, la prevenzione (anche detta profilassi) e la terapia (trattamento o cura). La prevenzione è principalmente limitata dalla mancanza di conoscenza in merito all'organismo in oggetto e i suoi ospiti. I modelli matematici sono stati utilizzati per approfondire la conoscenza delle malattie delle piante con vari obiettivi. Essi offrono l'opportunità di affrontare un uso razionale delle risorse riguardo ai costosi monitoraggi e rappresentano un passo fondamentale verso misure di controllo più sostenibili. Da un punto di vista curativo, oggigiorno gli sforzi sono focalizzati allo sviluppo di concetti di gestione delle malattie che bilancino i benefici dei pesticidi con le preoccupazioni in merito ai residui che possono contaminare l'ambiente. In questa tesi, i due principi della gestione della malattia sono stati affrontati con due casi studio: il dissecamento del frassino, causata da Hymenoscyphus fraxineus, che può essere considerata la più grave malattia del genere Fraxinus in Europa, e il mal dell'inchiostro del castagno, causata da Phytophthora cambivora (Petri) Buism. and P. cinnamomi Rands. Nella prima parte della tesi sono state introdotte le due malattie, in modo da poterne appurare somiglianze e differenze (Capitolo I). Successivamente, dal capitolo II al capitolo V sono descritte le prove sperimentali effettuate. In particolare, nel capitolo II è stato approntato uno studio della nicchia ecologica di H. fraxineus, con la caratterizzazione di variabili ecologiche e ambientali associate a zone naturalmente infette. Tale procedura è stata effettuata tramite Species Distribution Models (SDM), ampiamente utilizzati in ambito ecologico e da poco tempo anche nell'ambito della patologia vegetale. La presenza del patogeno è risultata fortemente correlata a tre variabili ambientali estive, in particolare abbondanti precipitazioni, alta umidità del suolo e basse temperature, in comparazione con la media dell'area di studio. Successivamente la tecnica dell'ensemble forecasting è stata applicata per ottenere una predizione della distribuzione potenziale del patogeno a scala europea, considerando la distribuzione di F. excelsior e F. angustifolia, ospiti della malattia. Infine, un innovativo metodo di network analysis ha permesso di individuare le aree ecologicamente adatte al patogeno ma non raggiungibili con una diffusione naturale. Nel capitolo III viene descritto uno studio condotto per valutare sei diversi fungicidi contro H. fraxineus. Inizialmente è stata effettuata una prova in vitro dei prodotti commerciali contro cinque ceppi del patogeno. Tiabendazolo, propiconazolo e allicina sono risultati i fungicidi più efficaci, con dose letale mediana più bassa, rispetto, per esempio, al principio attivo procloraz. Al contrario, il solfato di rame e i fosfiti di potassio si sono rilevati completamente inefficaci. Successivamente, i tre migliori fungicidi sono stati applicati in planta tramite trattamenti endoterapici su frassini maggiori inoculati al tronco con un ceppo autoctono. Tale test è stato anticipato da prove preliminari per massimizzare l'efficienza delle iniezioni; nelle condizioni stazionali e climatiche delle prove, maggiori velocità sono state raggiunte con soluzione acquosa addizionata con 1.2 % di acido acetico, effettuando i trattamenti la mattina presto o nel pomeriggio tardo. Considerando i risultati della prova in planta, tiabendazolo e allicina hanno rallentato in maniera significativa la crescita delle necrosi, al contrario non si è riusciti a iniettare la soluzione a base di propiconazolo. I capitoli IV e V riprendono le metodologie applicate contro la patologia del dissecamento del frassino, applicandole al mal dell'inchiostro del castagno. In particolare nel capitolo IV, la teoria fuzzy è stata adottata nello studio del complesso del mal dell'inchiostro, includendo nella costruzione del modello variabili ambientali quali temperatura minima invernale, siccità estiva, esposizione, distanza da corsi d'acqua e permeabilità del suolo, che più possono influire sullo sviluppo della malattia. Il modello è stato validato con un'ampia ricerca sul campo condotta nei castagneti nell'area di Treviso. Inoltre, sono state prodotte delle mappe dell'incertezza (inerenti a struttura, input e parametri del modello) per la corretta interpretazione della previsione. Buona parte dell'area a castagneto nella zona di studio si è rivelata adatta allo sviluppo del mal dell'inchiostro, mentre solo il 18.8 %, corrispondente alle aree più elevate, presentava rischi inferiori. Un secondo studio (capitolo V) ha riguardato una prova comparativa di efficacia di quattro formulazioni di fosfiti di potassio tramite endoterapia. P. cinnamomi è stata isolata con la tecnica del baiting in un castagneto affetto da mal dell'inchiostro ed è stata inoculata su 50 castagni asintomatici. In seguito ai trattamenti endoterapici, l'unica soluzione che ha significativamente rallentato la crescita delle necrosi è stata quella a base di fosfiti di potassio (35 %) addizionata con 0.1 % di soluzione di micronutrienti. Un'ulteriore prova di endoterapia è stata condotta in via preliminare nel castagneto abbandonato in cui era stata isolata P. cinnamomi, al fine di valutare la stimolazione alla crescita del callo cicatriziale da parte della soluzione iniettata fosfiti di potassio 70 %. I risultati ottenuti in questo caso non hanno evidenziato una differenza significativa rispetto ai controlli trattati con acqua, probabilmente per una necessità di tempi più lunghi considerando piante di età maggiore. In base ai risultati raggiunti, la modellistica epidemiologica e i trattamenti endoterapici sperimentati in merito alle patologie del dissecamento del frassino e al mal dell'inchiostro del castagno possono rappresentare degli strumenti fondamentali nella gestione integrata delle malattie considerate, da applicare insieme ad appropriate tecniche colturali per massimizzarne i benefici.

Submitted by Cláudia Bueno (claudiamoura18@gmail.com) on 2016-05-19T19:46:17Z No. of bitstreams: 2 Dissertação - André Felipe Alves de Andrade - 2016.pdf: 2007118 bytes, checksum: 69341ce5bc6c459857d1feff5d92366e (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5)
Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2016-05-20T13:54:17Z (GMT) No. of bitstreams: 2 Dissertação - André Felipe Alves de Andrade - 2016.pdf: 2007118 bytes, checksum: 69341ce5bc6c459857d1feff5d92366e (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5)
Made available in DSpace on 2016-05-20T13:54:17Z (GMT). No. of bitstreams: 2 Dissertação - André Felipe Alves de Andrade - 2016.pdf: 2007118 bytes, checksum: 69341ce5bc6c459857d1feff5d92366e (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2016-02-26
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES
Coral reefs are of extreme importance to both nature and society, due to being responsible for several services and harbouring hundreds of species. Despite such critical importance, reef corals current suffered heavy losses since the Anthropocene, with 20% of world´s corals damaged beyond recovery due to human pressure and coastal development. This scenario is even worse, since corals are especially vulnerable to climate change and the entire ecosystem could go extinct by 2050. In this study we focus on comparing the already established impacts from human development and the yet happen losses from climate change on Brazilian corals, a unique fauna that still have gaps in knowledge. We created environmental suitability models for 24 species and quantified individual losses from both climate change and human activities. From the individual results we derived an overall pattern, in which we found out that future losses from climate alteration are equivalent to current losses from human activities. We then used the spatial distribution of those activities and key areas for conservation, determined with software Zonation, to select six areas in the Brazilian exclusive economic zone where proactive and reactive conservation strategies should be implanted, given its importance to biodiversity and concentrated anthropogenic impacts. Overall suitability losses were of approximately 30% for both sources and 60% of the areas will continue to be suitable in the future. Therefore, Brazilian corals will experience heavy losses from climate, especially the loss of highly suitable areas, which are compared to effects from human economic activities. Coral situation is likely to be even worse, if we were to consider bleaching, ocean acidification and diseases, events expected to increase with the rising temperature.
Ecological Niche Modelling (ENM) is widely used for conservation purposes, predicting species invasion, evolutionary aspects and a whole array of applications. However, for most cases, evaluating the efficiency of those models poses as problematic, as commonly used methods (i.e. random methods) do not assure the required independence between data used to create the model and data used to evaluate the model. We developed a new transferabilitybased framework that ensures the much-needed independence between subsets. We created an alternate approach that geographically splits occurrence datasets, while intrinsically controls issues related to previous transferability approaches, such as overfitting, extrapolation and sampling bias. We used 26 Atlantic coral species to perform three different geographical divisions quantifying the effect of different splits on model predictive efficiency. We demonstrate that transferability should be used as an effective method to evaluate ENMs. Geographical split of the area in deciles proved as a reliable evaluation method, assuring independence between datasets and being less prone to common transferability issues. Our odds-and-evens framework provides improvements to the ongoing debate of ENMs evaluating by its transferability. This new method corrects the issue of artificiality causing sampling bias and overfitting, common in previous methodologies, while also is less prone to extrapolation issues, a common problem in transferability approaches. Moreover, the framework appears as a feasible and useful alternative to the problematic and commonly used random partition of datasets evaluation.

This thesis aims to explore distributional patterns of tree species in the neotropical lowland rain forest biome based on diversity analyses, dated phylogenies and species distribution models, using the family Sapotaceae as a case study. Sapotaceae is an abundant and diverse group in the neotropical lowland rain forest and its distributional patterns are representative of other tree clades in this biome. These characteristics make this family a good model to test ecological and biogeographic hypothesis in neotropical rain forests. An analysis of beta-diversity measured by the number of shared species was used as a test of biotic homogeneity of Morrone's (2001) widely used system of neotropical biogeographic units. Biotic homogeneity was generally low, and Morrone's (2001) biogeographic regionalisation was found not to coincide with the distributional patterns of Sapotaceae species. Divergence times of Sapotaceae species were estimated using a dated phylogeny based on DNA sequences of the nuclear ribosomal internal transcribed spacer (ITS) to explore the effects of Andean uplift, closure of the Isthmus of Panama and Pleistocene climatic changes on the evolutionary history of lowland rain forests in northern South America. The Andean uplift was found to have affected patterns of distribution by creating new habitats and altering hydrologic systems in northern South America, and in some cases by isolating lineages to the east and west of the Eastern Cordillera of the Andes. The closure of the Panama Isthmus and Pleistocene climatic changes do not seem to have strongly affected patterns of distribution or diversification in Sapotaceae. In general, the lack of congruent dates for many repeated biogeographic splits in the phylogeny (e.g., Amazon-Choco) suggests that idiosyncratic dispersal events have had a substantial effects on Sapotaceae's biogeography. Finally, species distribution models generated for Sapotaceae in the Neotropics were used to identify areas of high predicted species richness in Colombia. The highest diversity of Sapotaceae species was predicted for the inter-Andean valleys and northern Amazon. These results were compared to the current system of Protected Areas in this country, demonstrating that areas of high conservation value based on predicted species richness have a low coverage of Protected Areas. Such gaps highlight the potential need for new systems for the delimitation of basic units for conservation at national levels in Colombia.

Models that are used for predicting species' potential distributions are important tools that have found applications in a number of areas of applied ecology. The majority of these models can be classified as correlative, as they rely on strong, often indirect, links between species distribution records and environmental predictor variables to make predictions. Correlative models are an alternative to more complex mechanistic models that attempt to simulate the mechanisms considered to underlie the observed correlations with environmental attributes. This study explores the influence of the type and quality of the data used to calibrate correlative models. In terms of data type, the most popular techniques in use are group discrimination techniques, those that use both presence and absence locality data to make predictions. However, for many organisms absence data are either not available or are considered to be unreliable. As the available range of profile techniques (those using presence only data) appeared to be limited, new profile techniques were investigated and evaluated. A new profile modelling technique based on fuzzy classification (the Fuzzy Envelope Model) was developed and implemented. A second profile technique based on Principal Components Analysis was implemented and evaluated. Based on quantitative model evaluation tests, both of these techniques performed well and show considerable promise. In terms of data quality, the effects on model performance of false absence records, the number of locality records (sample size) and the proportion of localities representing species presence (prevalence) in samples were investigated for logistic regression distribution models. Sample size and prevalence both had a significant effect on model performance. False absence records had a significant influence on model performance, which was affected by sample size. A quantitative comparison of the performance of selected profile models and group discrimination modelling techniques suggests that different techniques may be more successful for predicting distributions for particular species or types of organism than others. The results also suggest that several different model design! sample size combinations are capable of making predictions that will on average not differ significantly in performance for a particular species. A further quantitative comparison among modelling techniques suggests that correlative techniques can perform as well as simple mechanistic techniques for predicting potential distributions.

This thesis explores some important practical considerations concerning the use of species distribution models in marine conservation planning. Using geo-referenced gorgonian distribution data, together with explanatory environmental variables, predictive models have been used to map the spatial distribution of suitable gorgonian (sea fan) habitat in two study sites; Hatton Bank, in the Northeast Atlantic, and Lyme Bay on the south coast of Devon. Generalized Linear Models (GLMs), Generalized Additive Models (GAMs) and a Maximum Entropy (Maxent) model have been used to support critical investigation into important model considerations that have received inadequate attention in the marine environment. The influence of environmental data resolution on model performance has been explored with specific reference to available datasets in the nearshore and offshore environments. The transferability of deep-sea models has been similarly appraised, with recommendations as to the appropriate use of transferred models. Investigating these practical issues will allow managers to make informed decisions with respect to the best and most appropriate use of existing data. This study has also used novel approaches and investigated their suitability for marine conservation planning, including the use of model classification error in the spatial prioritisation of monitoring sites, and the adaptation of an existing presence-only modelling method to include absence data. Together, these studies contribute both practical recommendations for marine conservation planning and novel applications within the wider species distribution modelling discipline, and consider the implications of these developments for managers, to ensure the ongoing improvement and development of models to support conservation planning.

A variety of species distribution models (SDMs) were fit to data collected by a 15,000km road-side visual survey of Aloidendron dichotomum populations in the Northern Cape region of South Africa, and Namibia. We fit traditional presence/absence SDMs as well as SDMs on how proportions are distributed across three species stage classes (juvenile, adult, dead). Using five candidate machine learning methods and an ensemble model, we compared a number of approaches, including the role of balanced class (presence/absence) datasets in species distribution modelling. Secondary to this was whether or not the addition of species’ absences, generated where the species is known not to exist have an impact on findings. The goal of the analysis was to map the distribution of Aloidendron dichotomum under different scenarios. Precipitation-based variables were generally more deterministic of species presence or lack thereof. Visual interpretation of the estimated Aloidendron dichotomum population under current climate conditions, suggested a reasonably well fit model, having a large overlap with the sampled area. There however were some conditions estimated to be suitable for species incidence outside of the sampled range, where Aloidendron dichotomum are not known to occur. Habitat suitability for juvenile individuals was largely decreasing in concentration towards Windhoek. The largest proportion of dead individuals was estimated to be on the northern edge of the Riemvasmaak Conservancy, along the South African/Namibian boarder, reaching up to a 60% composition of the population. The adult stage class maintained overall proportional dominance. Under future climate scenarios, despite maintaining a bulk of the currently habitable conditions, a noticeable negative shift in habitat suitability for the species was observed. A temporal analysis of Aloidendron dichotomum’s latitudinal and longitudinal range revealed a potential south-easterly shift in suitable species conditions. Results were however met with some uncertainty as SDMs were uncovered to be extrapolating into a substantial amount of the study area. We found that balancing response class frequencies within the data proved not to be an effective error reduction technique overall, having no considerable impact on species detection accuracy. Balancing the classes however did improve the accuracy on the presence class, at the cost of accuracy of the observed absence class. Furthermore, overall model accuracy increased as more absences from outside the study area were added, only because these generated absences were predicted well. The resulting models had lower estimated suitability outside of the survey area and noticeably different suitability distributions within the survey area. This made the addition of the generated absences undesirable. Results highlighted the potential vulnerability of Aloidendron dichotomum given the pessimistic, yet likely future climate scenarios.

This thesis presents the first insights into inshore dolphin distribution of the western Pilbara. The region is undergoing rapid coastal development, which has the potential to threaten Indo-Pacific bottlenose dolphin (Tursiops aduncus) and Australian humpback dolphin (Sousa sahulensis) populations. Understanding the distribution of these species is essential for their conservation. Species distribution models (SDMs) were developed using dolphin sightings data that were opportunistically collected during dugong aerial surveys. A geographical information system (GIS) was used to generate the training data, which consisted of the binomial presence-absence of dolphins, distances from mainland and islands, sea surface temperature (SST), ocean fronts and bathymetric derivatives. Preliminary models were developed using generalised additive model (GAM) and component-wise boosting techniques. Models could not be fit to the data using either technique. It was unclear whether this was a result of relatively few dolphin sightings across a large study area, pseudo-absences, weak environmental variables, or a combination of all of these factors. Maximum Entropy (MaxEnt) software was subsequently used as an alternative modelling technique to model the presence of dolphins, along with automatically generated background data, in order to avoid problems associated with unreliable absence data. Bottlenose and humpback dolphins were sympatric, with overlap in occurrence across the study area. Bottlenose dolphin presence was associated with the slope at the 20 m contour and waters around the Muiron Islands. This is likely to be a productive area that could be important for foraging. Humpback dolphin presence was associated with intertidal areas, including shallow coastal waters near the mainland and surrounding islands. The presence of numerous offshore islands would thus explain why humpback dolphins were recorded more than 50 km from the coastline. MaxEnt models were limited in their predictive power. Dedicated aerial surveys for inshore dolphins, using standardised techniques, are required to obtain reliable species data. In addition to increasing the sample size available for modelling, greater certainty in group size and composition could allow count, calf and mixed species group data to be modelled. Adequate species conservation needs to incorporate various ecological processes that occur at different spatial and temporal scales. Guidance is provided for undertaking boat-based studies and biopsy sampling, gathering opportunistic sightings data, and undertaking satellite telemetry research in addition to dedicated aerial surveys. Distribution modelling for inshore dolphins of northern WA is an intrinsically challenging research project, due to limited habitat data and the elusive nature of the subject species, particularly humpback dolphins. This research has obtained the first insights into the distribution of inshore dolphins in northern WA. Through lessons learnt, this research has paved the way for the development of future models to have a greater predictive ability, which will be useful for the conservation of threatened inshore dolphin species.

La SSD (Species Sensitivity Distribution) est une méthode utilisée par les scientifiques et les régulateurs de tous les pays pour fixer la concentration sans danger de divers contaminants sources de stress pour l'environnement. Bien que fort répandue, cette approche souffre de diverses faiblesses sur le plan méthodologique, notamment parce qu'elle repose sur une utilisation partielle des données expérimentales. Cette thèse revisite la SSD actuelle en tentant de pallier ce défaut. Dans une première partie, nous présentons une méthodologie pour la prise en compte des données censurées dans la SSD et un outil web permettant d'appliquer cette méthode simplement. Dans une deuxième partie, nous proposons de modéliser l'ensemble de l'information présente dans les données expérimentales pour décrire la réponse d'une communauté exposée à un contaminant. A cet effet, nous développons une approche hiérarchique dans un paradigme bayésien. A partir d'un jeu de données décrivant l'effet de pesticides sur la croissance de diatomées, nous montrons l'intérêt de la méthode dans le cadre de l'appréciation des risques, de par sa prise en compte de la variabilité et de l'incertitude. Dans une troisième partie, nous proposons d'étendre cette approche hiérarchique pour la prise en compte de la dimension temporelle de la réponse. L'objectif de ce développement est d'affranchir autant que possible l'appréciation des risques de sa dépendance à la date de la dernière observation afin d'arriver à une description fine de son évolution et permettre une extrapolation. Cette approche est mise en œuvre à partir d'un modèle toxico-dynamique pour décrire des données d'effet de la salinité sur la survie d'espèces d'eau douce
Species Sensitivity Distribution (SSD) is a method used by scientists and regulators from all over the world to determine the safe concentration for various contaminants stressing the environment. Although ubiquitous, this approach suffers from numerous methodological flaws, notably because it is based on incomplete use of experimental data. This thesis revisits classical SSD, attempting to overcome this shortcoming. First, we present a methodology to include censored data in SSD with a web-tool to apply it easily. Second, we propose to model all the information present in the experimental data to describe the response of a community exposed to a contaminant. To this aim, we develop a hierarchical model within a Bayesian framework. On a dataset describing the effect of pesticides on diatom growth, we illustrate how this method, accounting for variability as well as uncertainty, provides benefits to risk assessment. Third, we extend this hierarchical approach to include the temporal dimension of the community response. The objective of that development is to remove the dependence of risk assessment on the date of the last experimental observation in order to build a precise description of its time evolution and to extrapolate to longer times. This approach is build on a toxico-dynamic model and illustrated on a dataset describing the salinity tolerance of freshwater species

Subsistence fishing is a vital component of Alaska’s North Slope borough economy and culture that is being threatened by human disturbance. These threats mean the fish must be protected, but the size of the region makes conservation planning difficult. Fortunately, advances in species distribution models (SDMs), environmental DNA (eDNA), and remote sensing technologies provide potential to better understand species’ needs and guide management. The objectives of my study were to: (1) map the current habitat suitability for twelve fish species, occurring in Alaska’s North Slope,(2) determine if SDMs based on eDNA data performed similarly to, or improved, models based on traditional sampling data, and (3) predict how species distributions will shift in the future in response to climate change. I was able to produce robust models for 8 of 12 species that relate environmental characteristics to a species’ presence or absence and identify stream reaches where species are likely to occur. Unfortunately, the use of eDNA data did not produce useful models in Northern Alaskan rivers. However, I was able to generate predictions of species distributions into the future that should help inform management for years to come.

Context: Understanding spatial patterns of biodiversity and species’ distributions is important for scientific theory, and for conservation and management of the natural world. Climatic variables are widely recognised as strong correlates of species richness over large spatial extents. Correlates of species richness at smaller extents (regional and landscape scales) are less well established, but environmental heterogeneity is widely thought to be important. A large number of environmental heterogeneity measures have been used, but in particular there is a growing interest in ‘geodiversity’, which I define here as the diversity of abiotic terrestrial and hydrological nature, comprising earth surface materials and landforms. Recent research has emphasised both geodiversity’s inherent value and its potential as a correlate and predictor of spatial biodiversity and species’ distribution patterns. However, despite this clear potential of geodiversity for improving our understanding of how patterns of life relate to environmental heterogeneity, its incorporation into biodiversity and species’ distribution modelling is substantially underdeveloped. In this thesis, using a macroecological approach I begin to address some of these knowledge gaps by analysing the relationships between geodiversity data, and its constituent ‘geofeatures’, and species richness and distributions for multiple taxa and across several scales (grain size and extent) and geographic locations. My main aims in this thesis are to more fully evaluate geodiversity itself, and improve our understanding of its role with respect to the spatial patterning of biodiversity, both conceptually and empirically. Locations and Spatial Scales Analyses were carried out within and across Great Britain (England, Scotland, and Wales) and Finland. The order of the four quantitative papers generally reflects the largest spatial extent (i.e. size of the study area) at which they were conducted, from national (PAPERS II and III) through landscape (PAPER IV), to the local scale (vegetation plots within a small upland river catchment; PAPER V). PAPER II is a study across several spatial extents (from landscape to national) and uses two grain sizes (1 km2 and 100 km2). PAPER I is a review paper that considers multiple scales and geographic locations conceptually. Time period Present day: data were from between 1995 and 2016 across all of the quantitative studies. Taxa Multiple: alien and native vascular plants across Great Britain (PAPER II); threatened bryophytes, beetles, fungi, lepidoptera, lichens, mammals, molluscs, and vascular plants across Finland (PAPER III); common and rare vascular plants across the Cairngorms, Scotland (PAPER IV); angiosperms, conifers, fungi, lichens, liverworts, lycophytes, mosses, and pteridophytes (and productivity) across an upland river catchment within the Cairngorms (PAPER V); and conceptual consideration of multiple taxa (PAPER I). Methods: For studies in Great Britain, plant data were provided by the Botanical Society of Britain and Ireland (BSBI) for PAPERS II and IV, and by the Centre for Ecology and Hydrology (CEH) for PAPER V. The threatened species data in Finland were from Finnish Environment Institute (PAPER II). Species richness (PAPERS II, III, and IV), rarity-weighted richness (RWR; PAPER III), species’ distributions (PAPERS IV and V), and productivity (measured using NDVI from colour infrared aerial imagery; PAPER V) were all analysed using Boosted Regression Tree (BRT) modelling, allowing comparisons between studies. For geodiversity data in the British studies, I compiled geodiversity data on landforms, soils, hydrological and geological features using existing national datasets (e.g. British Geological Survey), and used a geomorphometric method to extract landform coverage data (landforms included: hollows, ridges, valleys, and peaks). These data were analysed alongside environmental data, which varied between papers, relating to climate, standard topography (e.g. slope; elevation), land use, and human population. The sources of other geodiversity data in Finland, and environmental data on topography and climate, came from a variety of sources, which are detailed within each paper. Results: Geodiversity improved biodiversity and species’ distribution models throughout all of the quantitative analyses and generally declined in importance as spatial scale coarsened beyond the landscape scale. At most spatial scales and in most places, the roles of climate and/or coarse topography dominated, and geodiversity played a relatively small role, as was expected. Geodiversity, however, made consistent positive contributions to the models independently of traditionally used topographic metrics such as standard deviation of elevation and slope. Taxonomically, geodiversity: (i) was slightly more relevant for native vascular plants than alien in Great Britain (PAPER II); (ii) of similar relevance to common and rare vascular plants in the Scottish Highlands, except that the coverage of soil parent material was especially important for rare species’ distributions (PAPER IV); of similar relevance to most sessile taxa (angiosperms, fungi, mosses, liverworts, lichens, pteridophytes, and lycophytes; conifers were not related to geodiversity) in an upland Scottish river catchment (PAPER V); and more important for threatened vascular plants and bryophytes over other studied taxa in Finland (PAPER II). Geodiversity also improved models of productivity, and the variability in productivity, in PAPER V. Main conclusions and Future Directions: Geodiversity improves our understanding of, and ability to model, the relationship between biodiversity and environmental heterogeneity at multiple spatial scales, by allowing us to get closer to the real-world conditions and processes that affect life. I found that the greatest benefit comes from measuring ‘geofeatures’, which describe the constituent parts of geodiversity separately, rather than as one combined variable. Automatically extracted landform data, the use of which is novel in ecology, biogeography and macroecology, proved particularly valuable throughout this body of work, and as too did data from expert geological and hydrological maps. The idea of ‘Conserving Nature’s Stage’ (CNS), and identifying areas that are most capable of supporting high biodiversity into the future, the benefits and caveats of which are discussed in this thesis, has recently emerged. It requires a sound empirical and conceptual basis, to which my research contributes. In this thesis, I have gone some way towards demonstrating the conceptual and empirical value of incorporating geodiversity into ecological analyses across multiple spatial scales, paving the way for this recent approach to be more extensively used for theoretical and applied purposes. I accomplished this by carrying out an assessment of existing geodiversity literature and, importantly, looking forwards to consider the prospects of geodiversity within ecology (PAPER I), supported by four quantitative studies. The conservation significance is emphasised in PAPER III. Much remains to be done, however, and future research directions are detailed in PAPER I. We need to develop predictive models to test the role of geodiversity across an array of geographical and taxonomic domains, as well as to assess metrics beyond species richness and species’ distributions. One example may involve beta diversity: does spatial turnover in species relate to spatial turnover in geofeatures? Fully analysing the role of geodiversity through time will also be important, including in relation to refugia, given predicted environmental changes in climate. In progressing with this line of enquiry, we will improve our knowledge and understanding of patterns of life on Earth and, specifically, how the geophysical landscape helps shape them.

It is thought that climate change will have a major impact on species distributions by changing the habitat suitability for species. Species distribution modelling is a modern approach to assess the potential effect of climate change on biodiversity. We used 11 environmental variables with the MaxEnt algorithm to model the distributions of 114 Egyptian medicinal plant species under current conditions, then projecting them into three different future times (2020, 2050, and 2080) under two different climate-change emission scenarios (A2a and B2a), under two hypotheses about the capability of the species for dispersal (unlimited and no dispersal). Species richness maps for current and future times were produced. We tested the value of Egypt’s Protected Areas under climate change by estimating the species richness inside and outside under each scenario. We assessed Egyptian medicinal plants based on IUCN Red List categories and criteria, and then used the SDMs for conservation planning with and without consideration of socioeconomic factors using Zonation software. The A2 emission scenario was more harmful than B2 under all assumptions. Species richness inside Protected Areas was significantly higher than outside for all models. Based just on the records, between 75% and 90% of species could be classified as Least Concern, according to the assumptions made. Similarly, based on SDMs all species could be classified as LC at the current time, whilst in the future under climate change, up to 18% of species face the risk of extinction, depending on assumptions and based on the absolute time gap between the two future times. Based on 10 years, most species were assigned as Least Concern. Areas within PAs were no better in conservation prioritization value than area outside when socioeconomic costs (especially the Human Influence Index) were taken into account. Species distribution models appear to be extremely useful for conservation planning under climate change, particularly when only sparse data are available. Socioeconomic information adds a new dimension to conservation planning, which is actually misleading and incomplete without it.

Species distribution models are used increasingly in biogeography and conservation biology to predict how species are distributed and to understand attributes of species' environmental requirements. Whilst previous work has established that standard statistical approaches, such as General Linear Models anc General Additive Models, can be useful in modelling species' distribution classification trees are both more flexible and perform better in a range of scenarios and are thus being adopted as 'state-of-the-art'. In this thesis I investigate the relative performance of a range of four different classification tree modelling techniques namely CART, Bagging, Random Forests and Boosted Regression Trees.

This dissertation focused in the comprehensive analysis of the capabilities of some non-tested types of Artificial Neural Networks, specifically: the Probabilistic Neural Networks (PNN) and the Multi-Layer Perceptron (MLP) Ensembles. The analysis of the capabilities of these techniques was performed using the native brown trout (Salmo trutta; Linnaeus, 1758), the bermejuela (Achondrostoma arcasii; Robalo, Almada, Levy & Doadrio, 2006) and the redfin barbel (Barbus haasi; Mertens, 1925) as target species. The analyses focused in the predictive capabilities, the interpretability of the models and the effect of the excess of zeros in the training datasets, which for presence-absence models is directly related to the concept of data prevalence (i.e. proportion of presence instances in the training dataset). Finally, the effect of the spatial scale (i.e. micro-scale or microhabitat scale and meso-scale) in the habitat suitability models and consequently in the e-flow assessment was studied in the last chapter.
Esta tesis se centra en el análisis comprensivo de las capacidades de algunos tipos de Red Neuronal Artificial aún no testados: las Redes Neuronales Probabilísticas (PNN) y los Conjuntos de Perceptrones Multicapa (MLP Ensembles). Los análisis sobre las capacidades de estas técnicas se desarrollaron utilizando la trucha común (Salmo trutta; Linnaeus, 1758), la bermejuela (Achondrostoma arcasii; Robalo, Almada, Levy & Doadrio, 2006) y el barbo colirrojo (Barbus haasi; Mertens, 1925) como especies nativas objetivo. Los análisis se centraron en la capacidad de predicción, la interpretabilidad de los modelos y el efecto del exceso de ceros en las bases de datos de entrenamiento, la así llamada prevalencia de los datos (i.e. la proporción de casos de presencia sobre el conjunto total). Finalmente, el efecto de la escala (micro-escala o escala de microhábitat y meso-escala) en los modelos de idoneidad del hábitat y consecuentemente en la evaluación de caudales ambientales se estudió en el último capítulo.
Aquesta tesis se centra en l'anàlisi comprensiu de les capacitats d'alguns tipus de Xarxa Neuronal Artificial que encara no han estat testats: les Xarxes Neuronal Probabilístiques (PNN) i els Conjunts de Perceptrons Multicapa (MLP Ensembles). Les anàlisis sobre les capacitats d'aquestes tècniques es varen desenvolupar emprant la truita comuna (Salmo trutta; Linnaeus, 1758), la madrilla roja (Achondrostoma arcasii; Robalo, Almada, Levy & Doadrio, 2006) i el barb cua-roig (Barbus haasi; Mertens, 1925) com a especies objecte d'estudi. Les anàlisi se centraren en la capacitat predictiva, interpretabilitat dels models i en l'efecte de l'excés de zeros a la base de dades d'entrenament, l'anomenada prevalença de les dades (i.e. la proporció de casos de presència sobre el conjunt total). Finalment, l'efecte de la escala (micro-escala o microhàbitat i meso-escala) en els models d'idoneïtat de l'hàbitat i conseqüentment en l'avaluació de cabals ambientals es va estudiar a l'últim capítol.
Muñoz Mas, R. (2016). Multivariate approaches in species distribution modelling: Application to native fish species in Mediterranean Rivers [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/76168
TESIS

Thesis (MA)--Stellenbosch University, 2004.
ENGLISH ABSTRACT: At a social, ecological and biological level it is important tha t we gain a better understanding of species distribution and the constraints to species distribution. Various modelling tools and approaches are available to provide this type of functionality. The GARP (Genetic Algorithm for Rule set Production) Modelling System (GMS) was selected because of its strong predictive modelling abilities and its ability to represent the results of model iterations in both a tabular and cartographic manner. A shortcoming in this system was identified in tha t it requires strong information technology skills in order to carry out the modelling process. This can be attributed to the lack of a user-friendly interface to the system. In order to address this a loosely coupled system was developed that provides an easy to use web-based front end to the GMS. This Integrated Modelling System extends the core functionality of the GMS by providing a system that provides detailed history for each analysis, allows fine tuning of the modelling process, integrates directly with a biodiversity database containing specimen observations, and provides a simple ‘wizard’ interface to the modelling process.
AFRIKAANSE OPSOMMING: Van ’n sosiale, ekologiese en biologiese standpunt is dit belangrik dat ons spesies verspreiding en die beperkings daarvan verstaan. ’n Verskeidenheid sagteware pakkette en metodologiee is beskikbaar om spesies verspreiding te modelleer. Die GARP (Genetic Algorithm for Rule set Production) sagteware was gebruik vir sy sterk voorspellingsvermoe, en sy kapasiteit vir kartografiese en tubulere tentoonstelling van model resultate. ’n Tekortkoming met hierdie stelsel was gei'dentifiseer - dit is nie gebruikersvriendelik nie en gebruikers het sterk informasie tegnologie vermoens nodig. Om hierdie tekortkominge aan te spreek was ’n sagteware program ontwerp wat van GARP gebruik maak deur middel van ’n webblaaier. Hierdie ge'integreerde stelsel bou op die basiese funksionaliteit van GARP om ’n werk omgewing te skep wat ’n gedetailleerde geskiedenis van elke model stoor, fyn beheer oor die model toelaat, direk met ’n bio diver siteits databasis koppel, en van ’n eenvoudige ’wizard’ stelsel gebruik maak om gebruikers opsies te bepaal.

Statistical modelling is often used to relate the presence or abundance of species to environmental predictors, thereby providing a basis for predictive mapping of species or biodiversity. The variables included must thus be relevant and reflect actual changes in the environment. Therefore, the quantification of species–environment relationships is an important aspect of predictive modelling. This thesis examines how phytobenthic species or communities in the Baltic Sea relate to environmental gradients, and if different aspects of phytobenthic species distribution in the Baltic Sea could be explained by spatial or temporal variation in environmental factors. Predictive distribution modelling usually focuses on how environmental variables control the distribution of species or communities. Thus the relative weight of the predictor variables on different scales is of importance. In this thesis, I show that the relative importance of environmental variables depends both on geographic scale and location, and that it also differs between species or species groups. There are no simple explanations to the temporal variability in species occurrence. I here show that the temporal changes in species distribution within the phytobentic zone varies in a spatial context. I also try to find temporal and spatio-temporal patterns in species distribution that could be related to changes in climate or anthropogenic disturbance. However, the findings in this thesis suggest that single factor explanations are insufficient for explaining large-scale changes in species distribution. A greater understanding of the relationship between species and their environment will lead to the development of more sensitive models of species distributions. The predictions can be used to visualise spatial changes in the distribution of plant and animal communities over time.
At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 2: Submitted. Paper 3: Manuscript. Paper 4: Manuscript. Paper 5: Manuscript.

Understanding the spatial distribution of species in relationship to climatic and environmental variables is key to conservation and management of important species, as their distribution might change under climate change and variability. Based on presence absence data from scientific trawl surveys, this study used Generalized Additive Model (GAM) and Krigging with External Drift (KED) statistical techniques to determine the spatial distribution of three marine fish species of commercial interest: Merluccius capensis, Merluccius paradoxus, and Thyrsites atun, on the West and South coasts of South Africa. The modelled distributions reflect the previously determined range and habitats of the two species of hake and are in accordance with the common knowledge on the biology of the two species. Presence-absence modelling found depth to be the main factor for explaining hake distribution on both coasts. For the West coast an interaction between sea surface temperature and chlorophyll-a combined with depth as a factor was found to provide the best model. On the South coast depth was the only factor retained. The models for M. capensis and M. paradoxus are potentially useful in mapping and determining future distributions based on environmental factors. The model obtained for the spatial distribution of T. atun has a lower explanatory power than those of the two hake species.

Species Distribution Modelling (SDM) plays a key role in a number of biological applications: assessment of temporal trends in distribution, environmental impact assessment and spatial conservation planning. From a statistical perspective, this thesis develops two methods for increasing the accuracy and reliability of maps of density surfaces and provides a solution to the problem of how to collate multiple density maps of the same region, obtained from differing sources. From a biological perspective, these statistical methods are used to analyse two marine mammal datasets to produce accurate maps for use in spatial conservation planning and temporal trend assessment. The first new method, Complex Region Spatial Smoother [CReSS; Scott-Hayward et al., 2013], improves smoothing in areas where the real distance an animal must travel (`as the animal swims') between two points may be greater than the straight line distance between them, a problem that occurs in complex domains with coastline or islands. CReSS uses estimates of the geodesic distance between points, model averaging and local radial smoothing. Simulation is used to compare its performance with other traditional and recently-developed smoothing techniques: Thin Plate Splines (TPS, Harder and Desmarais [1972]), Geodesic Low rank TPS (GLTPS; Wang and Ranalli [2007]) and the Soap lm smoother (SOAP; Wood et al. [2008]). GLTPS cannot be used in areas with islands and SOAP can be very hard to parametrise. CReSS outperforms all of the other methods on a range of simulations, based on their fit to the underlying function as measured by mean squared error, particularly for sparse data sets. Smoothing functions need to be flexible when they are used to model density surfaces that are highly heterogeneous, in order to avoid biases due to under- or over-fitting. This issue was addressed using an adaptation of a Spatially Adaptive Local Smoothing Algorithm (SALSA, Walker et al. [2010]) in combination with the CReSS method (CReSS-SALSA2D). Unlike traditional methods, such as Generalised Additive Modelling, the adaptive knot selection approach used in SALSA2D naturally accommodates local changes in the smoothness of the density surface that is being modelled. At the time of writing, there are no other methods available to deal with this issue in topographically complex regions. Simulation results show that CReSS-SALSA2D performs better than CReSS (based on MSE scores), except at very high noise levels where there is an issue with over-fitting. There is an increasing need for a facility to combine multiple density surface maps of individual species in order to make best use of meta-databases, to maintain existing maps, and to extend their geographical coverage. This thesis develops a framework and methods for combining species distribution maps as new information becomes available. The methods use Bayes Theorem to combine density surfaces, taking account of the levels of precision associated with the different sets of estimates, and kernel smoothing to alleviate artefacts that may be created where pairs of surfaces join. The methods were used as part of an algorithm (the Dynamic Cetacean Abundance Predictor) designed for BAE Systems to aid in risk mitigation for naval exercises. Two case studies show the capabilities of CReSS and CReSS-SALSA2D when applied to real ecological data. In the first case study, CReSS was used in a Generalised Estimating Equation framework to identify a candidate Marine Protected Area for the Southern Resident Killer Whale population to the south of San Juan Island, off the Pacific coast of the United States. In the second case study, changes in the spatial and temporal distribution of harbour porpoise and minke whale in north-western European waters over a period of 17 years (1994-2010) were modelled. CReSS and CReSS-SALSA2D performed well in a large, topographically complex study area. Based on simulation results, maps produced using these methods are more accurate than if a traditional GAM-based method is used. The resulting maps identified particularly high densities of both harbour porpoise and minke whale in an area off the west coast of Scotland in 2010, that might be a candidate for inclusion into the Scottish network of Nature Conservation Marine Protected Areas.

The Agulhas Bank supports a speciose fish community, many of which are commercially important. Despite substantial research being conducted on aspects of their biology spatial aspects of their distribution and abundance in relation to environment parameters has been ignored. This study, therefore, addressed aspects related to the distribution and abundance of representative species on the Agulhas Bank within a Geographic Information System (GIS). Four candidate species were chosen due to their importance either in numbers or unit mass to the South African demersal trawl fishery. The species also shared morphological and taxonomic similarities. The candidate species chosen were the two Cape hake species, shallow-water hake Meluccius capensis, and deep-water hake Merluccius paradoxus, and the two pleuronectiform species being Agulhas sole Austroglossus pectoralis and redspotted tonguesole Cynoglossus zanzibarensis. The use of a GIS was appropriate and allowed for hidden spatial patterns be exposed and illustrated visually, while also facilitating the quantification of the relationships between distribution/abundance and certain environmental predictors using statistical methods The Department of Marine and Coastal Management, Cape Town, supplied biological data in the form of length frequency and biomass information from spring (AprillMay) and autumn (September/October) cruises conducted between 1986 and 1993 on the R. V. Africana. The Council for National Geoscience, Cape Town, supplied sediment data for the entire southern African coastline. Initial exploratory data analysis highlighted potential relationships between environmental variables and abundance for each specie's life-history stanzas. Variations in spatial distribution were found to be significantly different between each life-history stanzas within species. Fish density as a function of the additive effects of the various environmental parameters, including temperature, depth and sediment type, was assessed using a Poisson Generalized Additive Model (GAM), while distribution was analysed with a logistic GAM. A predictive logistic model was then created, taking into consideration the importance of the predictor variables for each species, allowing for predictive estimates to be made for each species by inputting environmental information within the study area. The importance of certain environmental variables influencing distribution and abundance were noted. General patterns indicated that sediment was the most important to both the distribution and abundance of the two pleuronectiform species and juvenile life-history stanzas, while the adult gadoids' distribution and abundance appeared to be depth dependent.

Invasive weeds inflict significant harm on native species, ecosystem processes, and natural disturbance regimes. When managing these weed threats, some of the most useful tools are the outputs of predictive distribution models. As they supplement existing distribution data to assess where in the landscape is most susceptible to weed invasion, they allow for more efficient weed management because the areas most suited to weed species may be targeted for control. This research develops a habitat suitability model for the weed lantana (Lantana camara L. sensu lato) in a portion of the Greater Blue Mountains World Heritage Area at present and under forecast warmer climates. A generalised additive model (GAM) is used, which fits the regression curve used for prediction to the calibration data themselves and allows for an exploration of which environmental conditions favour lantana as well as where in the landscape is most suitable for the weed. Temperature was positively correlated with suitable habitat and explained over 90% of the variation in lantana presence predicted by the model. 15% of the study area was found to be suitable for lantana at present, with this figure reaching 58% after a simulated 3??C rise in temperature. Mapping habitat suitability across the study area allowed for the identification of five distinct pathways for lantana to further invade the Blue Mountains. Responding to calls for the integration of weed management with biodiversity conservation, the research also integrates the habitat suitability model with information regarding the distribution of vegetation communities and endangered species in the Blue Mountains. Thirteen native vegetation communities were found to have more than 20% suitable habitat for lantana at present, and an additional three contained more than 80% suitable habitat after a simulated 3??C rise in temperature. Five of these communities are listed as threatened under relevant legislation and harbour at least 27 endangered species, placing additional urgency on their conservation. This research has successfully used modelling techniques to identify areas for targeted weed management integrated with biodiversity conservation. The methods are easily adaptable to other weeds and regions and could thus be used to illustrate the comprehensive threat weeds pose to Australia???s biodiversity.

Otters (Family Mustelidae; Subfamily Lutrinae) are considered bioindicators of the wetlands they live in. Because they are engaged in a variety of ecosystem processes, any disruption to these ecosystems, such as hunting or climate change, is first reflected in the otter populations in the area. For species conservation and management, a basic knowledge of species ecology is required. Feeding and habitat ecology are critical for species survival because they allow for informed decisions regarding the protection and management of a species' preferred habitats. Additionally, the Himalayan mountain ranges are anticipated to warm faster than the worldwide average. As a result, any animal found in this region, such as the otter, is of special significance. Future climatic changes are expected to have a disastrous effect on both the ecosystem of this area and the animals that occupy the environment. Himalaya is home to three otter species: Eurasiasn otter (Lutra lutra), Smooth-coated otter (Lutrogale perspicillata), and Small-clawed otter (Aonyx cinereus). Otter research in the Himalayan area has received less academic attention than other parts of the world, leading to a lack of scientific knowledge about the species and its ecosystem. In recent years, the conservation and scientific significance of otters in the Himalaya has grown significantly. However, several unknown aspects of this species have not been thoroughly investigated. Hence, this research aimed to address a knowledge gap on current distribution and future scenarios for otter species found in Himalaya. Determining the occurrence of rare and endangered species in freshwater habitats is critical for ecological research and conservation concerns, but it may be time consuming and difficult. In the present research we also tested an environmental DNA approach to detect Lutra lutra in the South Italian and Himalayan rivers. We used the recently proposed Climate Niche Factor Analysis (CNFA) framework combined with Species Distribution Models to assess the vulnerability of three otter species found in the Himalayan region, namely Aonyx cinereus, Lutra lutra, and Lutrogale perspicillata, to 2050 climate and land-use change. Our study revealed that future climate and land-use change will shrink (6–15%) and shift (10–18%) the geographic ranges of the three Himalayan species, with land-use changes having much more severe consequences than climate change. Among the vulnerability factors, sensitivity was more important than exposure in determining the otters' vulnerability. In contrast to the most generalist, L. lutra, the most specialised species, L. perspicillata, exhibited the greatest vulnerability. Our findings showed that combining climate and land-use change components in CCVAs may result in divergent estimates of species vulnerability when compared to methods that solely include climate change. Furthermore, intrinsic factors like species sensitivity were shown to be much more relevant in predicting vulnerability than extrinsic factors like habitat exposure. We also examined the ecological niche overlap, similarity, and potential distribution of the three otter species—L. lutra, A. cinereus, and L. perspicillata— throughout their ranges. We created potential distribution maps for three of four Asian otter species (Lutra lutra, Aonyx cinereus, and Lutrogale perspicillata) using SDM, as well as a comparison of their geographic and climatic niche overlap and similarity. According to our findings, L. lutra, L. perspicillata, and A. cinereus are appropriate for 77.48 percent (SD=10.19), 77.32 percent (SD=9.31), and 75.26 percent (SD=10.68) of the respective IUCN ranges. The L. lutra niche was the largest of the three species, with a core centred on high NDVI (of July), NPP, and Driest Quarter Precipitation values. The niches of the other two species are distributed along a gradient of NDVI (February), Annual Mean Temperature, Mean Temperature of Wettest Quarter, and Precipitation of Wettest Month, with the A. cinereus niche lying toward greater values of these predictors than the L. perspicillata niche. Significant niche similarities were found between A. cinereus and both L. lutra (D=0.40, p0.05) and L. perspicillata (D=0.56, p0.05) in niche similarity tests. Furthermore, substantial similarity was found between L. perspicillata and A. cinereus (D=0.56, p0.05), but not between L. perspicillata and L. lutra (D=0.31, p0.05). Finally, niche similarities between L. lutra and the other two species were not significant (with A. cinereus, D= 0.40, p0.05; with L. perspicillata, D= 0.31, p0.05). Two of the three species are now classified as Vulnerable on the IUCN red list due to an estimated population decrease of more than 30% in the past 30 years, while L. lutra's East Asian range is extremely fragmented and rare, as is the situation with many terrestrial mammals in South Asia. Despite being well adapted to human-threatened environments, each species' appropriate habitat only covers around 75% of its present range. Wetland loss, poaching, pollution, and a decrease in prey biomass are dangers to all the species. Our results indicate that stringent protection and freshwater habitat restoration may help them expand their existing range and expand their present regions of residence in Asia. In this thesis, we also evaluated the potential of utilising environmental DNA (eDNA) to detect the presence of Eurasian otters from South Italy. Environmental DNA (eDNA) detection may offer novel and reliable methods for monitoring and conservation of rare and elusive species like the Eurasian otter, Lutra lutra. We used an experimental approach based on a target qPCR assay to detect L. lutra eDNA from water samples as a rapid monitoring tool to complement fine-scale conventional field surveys on a broad scale. This is the first time in Italy that an eDNA-based method has been used to monitor the presence of L. lutra. We compared the eDNA-based findings to conventional survey observations and found that this novel method to large-scale monitoring of such secretive aquatic species is reliable. We also tested an environmental DNA method to detect Lutra lutra in Himalayan rivers. Little is known about the Eurasian otter's habitat preferences throughout the Himalaya, particularly in the Trans-Himalayan region. This indicates a major knowledge gap in the region's poorly-known species biology and conservation. Future climate change is expected to have devastating consequences for the ecosystem of Himalayan region as well as the species that occupy the terrain. The findings of eDNA monitoring methods revealed the presence of L. lutra eDNA in 11 out of 15 Himalaya samples tested positive, a rate of 73 percent compared to 53 percent using the conventional monitoring technique. The PCA biplot overlaid on the scatter plot revealed that absent locations had a large Channel width and low Nitrate, Acidity, Surface velocity, and relative humidity values. We used glmulti to test 2100 models by autoruning various combinations of variables. The variables that contributed to the top five GLM models explaining the occurrence of eDNA at the 15 sample locations were acidity, channel width, and surface velocity. The concentration of acidity (pH) was shown to be the most important factor in the presence of otters (AIC = 10.45). In this context, environmental DNA (eDNA) surveys may help to overcome the expenses and constraints of conventional monitoring techniques in terms of species detection. For the first time, our results demonstrated that non-invasive molecular monitoring methods may effectively detect the presence of Lutra lutra in freshwater bodies, particularly on a medium-large scale. However, when evaluating an area with just a few sampling sites or a limited number of repetitions, the use of the eDNA method to detect otter presence in water samples should be approached with care. Future research should focus on testing eDNA surveys in various seasons and environments, since eDNA detection may be affected by temperature and other environmental variables that contribute to DNA degradation.

Species distribution models (SDMs) represent nowadays an essential tool in the research fields of ecology and conservation biology. By combining observations of species occurrence or abundance with information on the environmental characteristic of the observation sites, they can provide information on the ecology of species, predict their distributions across the landscape or extrapolate them to other spatial or time frames. The advent of SDMs, supported by geographic information systems (GIS), new developments in statistical models and constantly increasing computational capacities, has revolutionized the way ecologists can comprehend species distributions in their environment. SDMs have brought the tool that allows describing species realized niches across a multivariate environmental space and predict their spatial distribution. Predictions, in the form of probabilistic maps showing the potential distribution of the species, are an irreplaceable mean to inform every single unit of a territory about its biodiversity potential. SDMs and the corresponding spatial predictions can be used to plan conservation actions for particular species, to design field surveys, to assess the risks related to the spread of invasive species, to select reserve locations and design reserve networks, and ultimately, to forecast distributional changes according to scenarios of climate and/or land use change. By assessing the effect of several factors on model performance and on the accuracy of spatial predictions, this thesis aims at improving techniques and data available for distribution modelling and at providing the best possible information to conservation managers to support their decisions and action plans for the conservation of biodiversity in Switzerland and beyond. Several monitoring programs have been put in place from the national to the global scale, and different sources of data now exist and start to be available to researchers who want to model species distribution. However, because of the lack of means, data are often not gathered at an appropriate resolution, are sampled only over limited areas, are not spatially explicit or do not provide a sound biological information. A typical example of this is data on ‘habitat’ (sensu biota). Even though this is essential information for an effective conservation planning, it often has to be approximated from land use, the closest available information. Moreover, data are often not sampled according to an established sampling design, which can lead to biased samples and consequently to spurious modelling results. Understanding the sources of variability linked to the different phases of the modelling process and their importance is crucial in order to evaluate the final distribution maps that are to be used for conservation purposes. The research presented in this thesis was essentially conducted within the framework of the Landspot Project, a project supported by the Swiss National Science Foundation. The main goal of the project was to assess the possible contribution of pre-modelled ‘habitat’ units to model the distribution of animal species, in particular butterfly species, across Switzerland. While pursuing this goal, different aspects of data quality, sampling design and modelling process were addressed and improved, and implications for conservation discussed. The main ‘habitat’ units considered in this thesis are grassland and forest communities of natural and anthropogenic origin as defined in the typology of habitats for Switzerland. These communities are mainly defined at the phytosociological level of the alliance. For the time being, no comprehensive map of such communities is available at the national scale and at fine resolution. As a first step, it was therefore necessary to create distribution models and maps for these communities across Switzerland and thus to gather and collect the necessary data. In order to reach this first objective, several new developments were necessary such as the definition of expert models, the classification of the Swiss t rritory in environmental domains, the design of an environmentally stratified sampling of the target vegetation units across Switzerland, the development of a database integrating a decision-support system assisting in the classification of the relevés, and the downscaling of the land use/cover data from 100 m to 25 m resolution. The main contributions of this thesis to the discipline of species distribution modelling (SDM) are assembled in four main scientific papers. In the first, published in Journal of Biogeography different issues related to the modelling process itself are investigated. First is assessed the effect of five different stepwise selection methods on model performance, stability and parsimony, using data of the forest inventory of State of Vaud. In the same paper are also assessed: the effect of weighting absences to ensure a prevalence of 0.5 prior to model calibration; the effect of limiting absences beyond the environmental envelope defined by presences; four different methods for incorporating spatial autocorrelation; and finally, the effect of integrating predictor interactions. Results allowed to specifically enhance the GRASP tool (Generalized Regression Analysis and Spatial Predictions) that now incorporates new selection methods and the possibility of dealing with interactions among predictors as well as spatial autocorrelation. The contribution of different sources of remotely sensed information to species distribution models was also assessed. The second paper (to be submitted) explores the combined effects of sample size and data post-stratification on the accuracy of models using data on grassland distribution across Switzerland collected within the framework of the Landspot project and supplemented with other important vegetation databases. For the stratification of the data, different spatial frameworks were compared. In particular, environmental stratification by Swiss Environmental Domains was compared to geographical stratification either by biogeographic regions or political states (cantons). The third paper (to be submitted) assesses the contribution of premodelled vegetation communities to the modelling of fauna. It is a two-steps approach that combines the disciplines of community ecology and spatial ecology and integrates their corresponding concepts of habitat. First are modelled vegetation communities per se and then these ‘habitat’ units are used in order to model animal species habitat. A case study is presented with grassland communities and butterfly species. Different ways of integrating vegetation information in the models of butterfly distribution were also evaluated. Finally, a glimpse to climate change is given in the fourth paper, recently published in Ecological Modelling. This paper proposes a conceptual framework for analysing range shifts, namely a catalogue of the possible patterns of change in the distribution of a species along elevational or other environmental gradients and an improved quantitative methodology to identify and objectively describe these patterns. The methodology was developed using data from the Swiss national common breeding bird survey and the article presents results concerning the observed shifts in the elevational distribution of breeding birds in Switzerland. The overall objective of this thesis is to improve species distribution models as potential inputs for different conservation tools (e.g. red lists, ecological networks, risk assessment of the spread of invasive species, vulnerability assessment in the context of climate change). While no conservation issues or tools are directly tested in this thesis, the importance of the proposed improvements made in species distribution modelling is discussed in the context of the selection of reserve networks.

Medical cartography can provide valuable insights into the epidemiology and ecology of infectious diseases, providing a quantitative representation of the distribution of these pathogens. Such methods therefore provide a key step in informing public health policy decisions ranging from prioritising sites for further investigation to identifying targets for interventions. By increasing the resolution at which risk is defined, policymakers are provided with an increasingly informed approach for considering next steps as well as evaluating past progress. In spite of their benefits however, global maps of infectious disease are lacking in both quality and comprehensiveness. This thesis sets out to investigate the next steps for medical cartography and details the use of species distribution models in evaluating global distributions of a variety of zoonotic diseases of public health importance. Chapter 2 defines a methodology by which global targets for infectious disease mapping can be quantitatively assessed by comparing the global burden of each disease with the demand from national policymakers, non-governmental organisations and academic communities for global assessments of disease distribution. Chapter 3 introduces the use of boosted regression trees for mapping the distribution of a group of vector-borne diseases identified as being a high priority target, the leishmaniases. Chapter 4 adapts these approaches to consider Ebola virus disease. This technique shows that the West African outbreak was ecologically consistent with past infections and suggests a much wider area of risk than previously considered. Chapter 5 investigates Marburg virus disease and considers the variety of different factors relating to all aspects of the transmission cycle that must be considered in these analyses. Chapters 6 and 7 complete the mapping of the suite of viral haemorrhagic fevers by assessing the distribution of Crimean-Congo haemorrhagic fever and Lassa fever. Finally, Chapter 8 considers the risk that these viral haemorrhagic fevers present to the wider African continent, quantifying potential risk of spillover infections, local outbreaks and more widespread infection. This thesis addresses important information gaps in global knowledge of a number of pathogens of public health importance. In doing so, this work provides a template for considering the global distribution of a number of other zoonotic diseases.

A three-dimensional (3-D) chemical transport model (CTM) has been used to study the tropospheric chemistry and troposphere-stratosphere transport of biogenic very short-lived species (VSLS). Calculations have been performed in order to quantify their contribution to stratospheric bromine loading and also the relative importance of the source gas injection (SGI) and product gas injection (PGI) pathways. Simulations with a chemistry-climate model (CCM) have also been performed to assess the impact of VSLS on stratospheric ozone (03) and how their transport to the stratosphere may respond to climate change. Five brominated VSLS were considered in this work; bromoform (CHBr3), dibromomethane (CH2Br2), dibromochloromethane (CHBr2Cl), bromodichloromethane (CHBrC12) and bromochloromethane (CH2BrCl). The CTM shows these gases contribute ~5 parts per trillion (pptv) of inorganic bromine (BryvSLS) to the stratosphere, representing ~25% of total stratospheric bromine in 2009. SGI accounts for between ~69-75% of this supply, with the remainder from PG!. The longest lived organic product gas (PG) from CHBr3 and CH2Br2 degradation is carbonyl dibromide (CBr20). Its tropospheric lifetime with respect to photolysis is ~7 days. It is unlikely that organic PGs make a significant contribution to the total PGI from VSLS. The CTM performs reasonably well in reproducing tropospheric VSLS observations from a number of aircraft campaigns. In the tropical tropopause layer (TTL), CH2Br2 is overestimated when vertical transport is diagnosed from the meteoro- logical analyses. An improved agreement is obtained when vertical transport is calculated using diabatic heating rates, resulting in slower transport through the TTL. CCM runs show during periods of background aerosol loading, the impact of BryvSLS on mid-latitude 03 is small. The impact is larger in polar regions, where BryvSLS enhances the BrO-ClO loss cycle, resulting in a reduction of up to ~15 Dobson units (DU) of the total 03 column over Antarctica. The CCM has also been used to diagnose the potential response of SGI to climate change. For 2000, the modelled SGI is ~1.7 pptv, lower than the CTM estimate but still in good agreement with observations. For 2100, the CCM estimate increases to ~1.9 and ~2.7 pptv when the model is forced with Intergovernmental Panel on Climate Change (IPCC) representative concentration pathways (RC Ps) 4.5 and 8.5, respectively. The increase is due to enhanced tropical deep convection, reducing transport time-scales in the upper troposphere, particularly over the tropical Western Pacific region.

Biodiversity, the variety of life on the planet, is declining due to climate change, population and species interactions and as the result f demographic and landscape dynamics. Integrated model-based assessments play a key role in understanding and exploring these complex dynamics and have proven use in conservation planning. Model-based assessments using Species Distribution Models constitute an efficient means of translating limited point data to distribution probability maps for current and future scenarios in support of conservation decision making. The aims of this doctoral study were to investigate; (1) the use of a hybrid genetic programming to build high quality models that handle noisy real-world presence and absence data, (2) the extension of this solution to exploit the parallelism inherent to genetic programming for fast scenario based decision making tasks, and (3) a conceptual framework to share models in the hope of enabling research synthesis. Subsequent to this, the quality of the method, evaluated with the true skill statistic, was examined with two case studies. The first with a dataset obtained by defining a virtual species, and the second with data extracted from the North American Breeding Bird Survey relating to mourning dove (Zenaida macroura). In these studies, the produced models effectively predicted the species distribution up to 30% of error rate both presence and absence samples. The parallel implementation based on a twenty-node c3.xlarge Amazon EC2 StarCluster showed a linear speedup due to the multiple-deme coarse-grained design. The hybrid fuzzy genetic programming algorithm generated under certain consitions during the case studies significantly better transferable models.
Biodiversidade, a variedade de vida no planeta, está em declínio às alterações climáticas, mudanças nas interações das populações e espécies, bem como nas alterações demográficas e na dinâmica de paisagens. Avaliações integradas baseadas em modelo desempenham um papel fundamental na compreensão e na exploração destas dinâmicas complexas e tem o seu uso comprovado no planejamento de conservação da biodiversidade. Os objetivos deste estudo de doutorado foram investigar; (1) o uso de técnicas de programação genética e fuzzy para construir modelos de alta qualidade que lida com presença e ausência de dados ruidosos do mundo real, (2) a extensão desta solução para explorar o paralelismo inerente à programação genética para acelerar tomadas de decisão e (3) um framework conceitual para compartilhar modelos, na expectativa de permitir a síntese de pesquisa. Subsequentemente, a qualidade do método, avaliada com a true skill statistic, foi examinado com dois estudos de caso. O primeiro utilizou um conjunto de dados fictícios obtidos a partir da definição de uma espécie virtual, e o segundo utilizou dados de uma espécie de pomba (Zenaida macroura) obtidos do North American Breeding Bird Survey. Nestes estudos, os modelos foram capazes de predizer a distribuição das espécies maneira correta mesmo utilizando bases de dados com até 30% de erros nas amostras de presença e de ausência. A implementação paralela utilizando um cluster de vinte nós c3.xlarge Amazon EC2 StarCluster, mostrou uma aceleração linear devido ao arquitetura de múltiplos deme de granulação grossa. O algoritmo de programação genética e fuzzy gerada em determinadas condições durante os estudos de caso, foram significativamente melhores na transferência do que os algoritmos do BIOMOD.

Among tools that are used to fill knowledge gaps on natural systems, ecological modelling has been widely applied during the last two decades. Ecological models are simple representations of a complex reality. They allow to highlight environmental drivers of species ecological niche and better understand species responses to environmental changes. However, applying models to Southern Ocean benthic organisms raises several methodological challenges. Species presence datasets are often aggregated in time and space nearby research stations or along main sailing routes. Data are often limited in number to correctly describe species occupied space and physiology. Finally, environmental datasets are not precise enough to accurately represent the complexity of marine habitats. Can we thus generate performant and accurate models at the scale of the Southern Ocean ?What are the limits of such approaches ?How could we improve methods to build more relevant models ?In this PhD thesis, three different model categories have been studied and their performance evaluated. (1) Mechanistic physiological models (Dynamic Energy Budget models, DEB) simulate how the abiotic environment influences individual metabolism and represent the species fundamental niche. (2) Species distribution models (SDMs) predict species distribution probability by studying the relationship between species presences and the environment. They represent the species realised niche. (3) Dispersal lagrangian models predict the drift of propagules in water masses. Results show that physiological models can be developed for marine Southern Ocean species to simulate the metabolic variations in link with the environment and predict population dynamics. However, more data are necessary to highlight detailed physiological contrasts between populations and to accurately evaluate models. Results obtained for SDMs suggest that models generated at the scale of the Southern Ocean and future simulations are not relevant, given the lack of data available to characterise species occupied space, the lack of precision and accuracy of future climate scenarios and the impossibility to evaluate models. Moreover, model extrapolate on a large proportion of the projected area. Adding information on species physiological limits (observations, results from experiments, physiological model outputs) was shown to reduce extrapolation and to improve the capacity of models to estimate the species realised niche. Spatial aggregation of occurrence data, which influenced model predictions and evaluation was also succefully corrected. Finally, dispersal models showed an interesting potential to highlight the role of geographic barriers or conversely of spatial connectivity and also the link between species distribution, physiology and phylogeny history. This PhD thesis provides several methodological advice, annoted codes and tutorials to help implement future modelling works applied to Southern Ocean marine species.
Doctorat en Sciences
info:eu-repo/semantics/nonPublished

A conservação da biodiversidade é uma das questões mais urgentes e importantes da atualidade. Pesquisadores da biodiversidade usam modelos de distribuição de espécies para fazer inferências sobre ocorrência e localização de espécies. Estes modelos são úteis para políticas de conservação de biodiversidade. Ferramentas para modelagem de distribuição de espécies usam grandes conjuntos de dados de biodiversidade que estão globalmente distribuídos, podendo estar em diferentes plataformas computacionais, o que dificulta seu acesso e manipulação. A comunidade científica precisa de infraestrutura onde pesquisadores de biodiversidade possam colaborar e compartilhar modelos, dados e resultados, e estejam aptos a reproduzir experimentos de outros pesquisadores. Neste contexto uma Arquitetura Orientada a Serviços (SOA) que suporta a colaboração em uma rede de modelagem de distribuição de espécies na Web é apresentada. Este ambiente computacional baseia-se em um catálogo de experimentos de modelagem e em um conjunto de serviços web geoespaciais, o Web Biodiversity Collaborative Modelling Services WBCMS.
Biodiversity conservation has become a most urgent and important theme at present. Biodiversity researchers use species distribution models to make inferences about species occurrences and locations. These models are useful for biodiversity conservation policies. Species distribution modelling tools use large biodiversity datasets which are globally distributed, sometimes in different computational platforms, and are hard to access and manipulate. The scientific community needs infrastructure in which biodiversity researchers can collaborate and share models, data, results, as well as reproduce experiments from others researchers. In this context, we present a Service-Oriented Architecture (SOA) that supports the collaboration in species distribution modelling network on the Web. This computational environment is based on a Catalogue of modelling experiments and on a set of Geoweb services, the Web Biodiversity Collaborative Modelling Services WBCMS.

Parmi les outils qui permettent de mieux comprendre les systèmes naturels, la modélisation écologique a connu un essor particulièrement important depuis une vingtaine d’années. Les modèles écologiques, représentation simplifiée d’une réalité complexe, permettent de mettre en avant les facteurs environnementaux qui déterminent la niche écologique des espèces et de mieux comprendre leur réponse aux changements de l’environnement. Dans le cas des faunes marines antarctiques, la modélisation écologique fait face à plusieurs défis méthodologiques. Les jeux de données de présence des espèces sont très souvent agrégés dans le temps et dans l’espace, à proximité des stations de recherche. Ces données sont souvent trop peu nombreuses pour caractériser l’espace environnemental occupé par les espèces ainsi que leur physiologie. Enfin, les jeux de données environnementales manquent encore de précision pour finement représenter la complexité des habitats marins. Dans ces conditions, est-il possible de générer des modèles performants et justes à l’échelle de l’océan Austral ? Quelles sont les approches possibles et leurs limites ? Comment améliorer les méthodes afin de générer de meilleurs modèles ? Au cours de ce travail de thèse, trois types de modèles ont été étudiés et leurs performances évaluées. (1) Les modèles physiologiques de type DEB (Dynamic Energy Budget) simulent la manière dont l’environnement abiotique influe sur le métabolisme des individus et proposent une représentation de la niche fondamentale des espèces. (2) Les modèles de distribution d’espèces (SDMs pour Species Distribution Models) prédisent la probabilité de distribution des espèces en étudiant la relation spatiale entre données de présence et environnement. Ils proposent une représentation de la niche réalisée des espèces. Enfin (3), les modèles de dispersion de type lagrangien prédisent le mouvement de propagules dans les masses d’eau. Les résultats montrent que les modèles physiologiques réussissent à simuler les variations métaboliques des espèces antarctiques en fonction de l’environnement et à prédire les dynamiques de populations. Cependant, davantage de données sont nécessaires pour pouvoir caractériser finement les différences physiologiques entre populations et évaluer correctement les modèles. Les résultats obtenus pour les SDMs montrent que les modèles générés à l’échelle de l’océan Austral et leurs prédictions futures ne sont pas fiables du fait du manque de données disponibles pour caractériser l’espace occupé par les espèces, du manque de précision des scénarios climatiques futurs et de l’impossibilité d’évaluer les modèles. De plus, les modèles extrapolent sur une très grande proportion de l’espace projeté. L’apport d’information complémentaire sur les limites physiologiques des espèces (observations, résultats d’expériences, sorties de modèles physiologiques) permet de réduire l’extrapolation et d’augmenter la capacité des modèles à décrire la niche réalisée des espèces. L’agrégation spatiale des données, qui influençait les prédictions et l’évaluation des modèles a également pu être corrigée. Enfin, les modèles de dispersion ont montré un potentiel intéressant pour révéler le rôle des barrières géographiques ou à l’inverse, la connectivité spatiale, mais également le lien existant entre distribution, physiologie et histoire phylogénétique des espèces. Ce travail de thèse propose de nombreux conseils et fournit des codes annotés parfois sous forme de tutoriels, afin de constituer une aide utile aux futurs travaux de modélisation sur les espèces marines antarctiques
Among tools that are used to fill knowledge gaps on natural systems, ecological modelling has been widely applied during the last two decades. Ecological models are simple representations of a complex reality. They allow to highlight environmental drivers of species ecological niche and better understand species responses to environmental changes. However, applying models to Southern Ocean benthic organisms raises several methodological challenges. Species presence datasets are often aggregated in time and space nearby research stations or along main sailing routes. Data are often limited in number to correctly describe species occupied space and physiology. Finally, environmental datasets are not precise enough to accurately represent the complexity of marine habitats. Can we thus generate performant and accurate models at the scale of the Southern Ocean ? What are the limits of such approaches ? How could we improve methods to build more relevant models ? In this PhD thesis, three different model categories have been studied and their performance evaluated. (1) Mechanistic physiological models (Dynamic Energy Budget models, DEB) simulate how the abiotic environment influences individual metabolism and represent the species fundamental niche. (2) Species distribution models (SDMs) predict species distribution probability by studying the relationship between species presences and the environment. They represent the species realised niche. (3) Dispersal lagrangian models predict the drift of propagules in water masses. Results show that physiological models can be developed for marine Southern Ocean species to simulate the metabolic variations in link with the environment and predict population dynamics. However, more data are necessary to highlight detailed physiological contrasts between populations and to accurately evaluate models. Results obtained for SDMs suggest that models generated at the scale of the Southern Ocean and future simulations are not relevant, given the lack of data available to characterise species occupied space, the lack of precision and accuracy of future climate scenarios and the impossibility to evaluate models. Moreover, model extrapolate on a large proportion of the projected area. Adding information on species physiological limits (observations, results from experiments, physiological model outputs) was shown to reduce extrapolation and to improve the capacity of models to estimate the species realised niche. Spatial aggregation of occurrence data, which influenced model predictions and evaluation was also succefully corrected. Finally, dispersal models showed an interesting potential to highlight the role of geographic barriers or conversely of spatial connectivity and also the link between species distribution, physiology and phylogeny history. This PhD thesis provides methodological advices, annoted codes and tutorials to help implement future modelling works applied to Southern Ocean marine species

O interesse em valorizar os serviços ecossistêmicos fornecidos pela vegetação natural aumentou em um esforço para mitigar os efeitos da mudança do uso da terra. Nesta linha de pensamento, desenvolvemos um índice para valorar as comunidades de árvores - do ponto de vista antropocêntrico - da savana brasileira (Cerrado). O índice e a cartografia produzida servirão como ferramenta para a priorização da conservação, bem como para revelar como a colonização e a expansão da agricultura tem ocorrido. Para desenvolver o índice, foram produzidas novas camadas ambientais com resolução de 90m; A distribuição das 93 espécies mais comuns foi modelada; e a cartografia da distribuição de cada uso humano das árvores (alimentos, aromáticos, fibras, cosméticos, cortiça, etc., totalizando 20 usos) e um índice de valor total fo desenvolvido. O novo índice de valor, nomenado a Soma de Usos (SoU, Sum of Uses), representa o número esperado de usos para a montagem de espécies potenciais que poderia estar ocorrendo no lugar em condições ideais. O impacto da agricultura foi avaliado pela contabilização da área que foi convertida em lavouras. Nossos resultados indicam fortemente que a colonização humana e a expansão de terras cultivadas eliminaram as árvores de áreas que antes eram melhores prestadores de serviços ambientais. Por outro lado, observamos também que as áreas protegidas no Cerrado estão localizadas onde esperamos encontrar valor marginal para as espécies ótimas. Esses resultados nos levam a pensar que a estratégia de conservação pode estar longe de ser ideal para o maior remanescente arável do mundo.
The interest in valuing the ecosystem services provided by the natural vegetation has increased in an effort to mitigate the effects of land use change. In this line of thinking, we developed an index to value the tree communities -from an anthropocentric point of view- of the Brazilian savannah (Cerrado). The index and the cartography produced will serve as a tool for prioritization of conservation, has well as to unveil how colonization and agriculture expansion has taken place. In order to develop the index: new environmental layers at 90m resolution were produced; the most common 93 species’ distribution was modelled; and cartography for each use humans derive from the trees (food, aromatic, fiber, cosmetic, cork, etc., totaling 20 uses) and a total value index were developed. The new index of value, namely the Sum of Uses (SoU), represent the expected number of uses for the potential species assemblage that could be taking place under optimal conditions. The impact of agriculture was assessed by accounting for the area that has been converted to croplands. Our results strongly indicate that human settlement and cropland expansion have cleared the trees of areas that once were better than average ecosystem service providers. On the other hand, we also observe that protected areas in the Cerrado are located where we expect to find marginal value for the optimal communities. These results lead us to think that the conservation strategy might be far from optimal for the largest remaining arable patch in the world.

Conservation of biodiversity and sustainable resource use are central aims within ecology. This thesis focuses on the current data and environmental frameworks used to support these aims across different states in Europe. In particular, it examines the impact of geo-political boundaries on data-use, funding and planning for temporal movement of species in response to climate change. It also examines the current environmental framework agreements in Europe and their capacity to deal with trans-boundary aspects of biodiversity change. Through examination of European biodiversity datasets, undertaking species distribution modelling of forest taxa, examining economic data, palaeo-ecological data, and assessing international environmental framework agreements, this thesis identifies a number of important knowledge gaps. Probably unsurprisingly, the distribution of biodiversity in Europe mostly does not match political entities, all of which have individual aims, financial resources, and biodiversity management regimes in place. All have a significant impact on biodiversity conservation planning because i) the use of geo-politically truncated data influences modelling predictions, ii) financial commitment to biodiversity conservation varies between countries influencing success outcomes, iii) biodiversity persistence in current and future climate change does not recognise geo-political boundaries, and iv) many of the key environmental frameworks are implemented within countries and do not considering trans-boundary issues. Overall these findings significantly improve the understanding of conservation and resource management in Europe and fill a number of important knowledge gaps. They highlight the importance of appropriate trans-boundary ecological datasets and the need for more consistency across Europe in financial resources for biodiversity conservation. They also highlight the need for appreciation of areas of high-persistent biodiversity regardless of geo-political boundaries and environmental framework agreements that support cross-border conservation measures.

Conservation frequently requires immediate responses to prevent further declines of imperilled populations, often in the absence of detailed information. Consequently, population distribution patterns are often used to guide conservation decisions. However, distribution patterns may be misleading if threats have restricted species to low quality habitat. This issue means it is not always apparent where management efforts should be concentrated for maximum conservation gain. My aim was to improve the effectiveness of threatened species conservation by investigating this issue in whio (blue duck - Hymenolaimus malacorhynchos), a New Zealand riverine duck that has undergone serious declines. I used population and spatial modelling to answer three questions: (1) what are the threats to whio, (2) how can these threats be managed, and (3) managing which whio habitats will give the greatest conservation gain? A spatial analysis of contemporary whio habitat using boosted regression trees revealed whio are only secure in 1 % of their historical range, with predation likely causing significantly greater range contraction (83 %) than habitat modification (29 %). In that analysis, I identified 39,000 km of occupiable whio habitat, providing extensive opportunities to expand their contemporary range through management. Intensive monitoring identified stoats (Mustela erminea) as the primary cause of whio population declines, with stoat predation severely reducing whio nest survival (10 % and 54 % in the absence and presence of stoat control, respectively). Population viability analyses indicated whio populations in the absence of stoat control were at high risk of extinction (λ = 0.74) but large-scale, low-intensity predator control was useful for short-term whio conservation. However, whio populations with stoat control still had a declining population growth rate (λ = 0.95) and further intervention may be required to prevent whio extinctions. Such management needs to target high quality habitat to ensure the greatest conservation value. Analyses of habitat quality revealed whio fitness was highest in warm, low gradient rivers, although fitness gradients differed between North and South Islands. Comparisons of fitness relationships with spatial model predictions showed that South Island whio occurred more frequently in poorer habitat, indicating they may occupy a relict distribution. Limited resources for conservation mean identifying effective management techniques is critical for species persistence. My modelling approach enabled the effectiveness of whio management to be assessed and areas of high quality habitat where such management should provide the greatest benefit to be identified. These tools are directly applicable to the conservation management of many threatened species by quickly informing managers in situations where distributions may not follow habitat quality.

Conservation strategies are mainly focused on species existing in an environment shaped by natural and anthropogenic pressures. Yet, evidence shows that climate is changing faster than ever and expected to continue to change in the near future, which can be devastating for plants with restricted ranges. Turkey harbors many endemic species that might be affected from these changes. However, available data is scarce and biased, complicating the anticipation of future changes. Aim of this study is to improve our understanding of endemic species distributions and forecasting effects of climate change via species distribution modelling (SDM). The study is based on two Anatolian (Crocus ancyrensis and Crataegus tanacetifolia) and two Ankara (Salvia aytachii and Centaurea tchihatcheffii) endemics. Independent presence and absence data (ranging between 19-68 and 38-61, respectively) for each species was collected through fieldwork in and around the Upper Sakarya Basin in 2008 and 2009. With the software Maxent, SDMs were performed by using 8 least correlated environmental features and random presence records (of which 25% were used for confusion matrix). SDMs for current distributions of C. ancyrensis, C. tchihatcheffii and C. tanacetifolia were reliable enough for future extrapolations despite errors originating from scale, non-equilibrium status and biotic interactions, respectively. The model for S. aytachii failed due to absence of limiting factor (soil type) in the model. Future projections of those three species modelled using CCCMA-CGCM2 and HADCM3 climate models indicated three possible responses to climate change: (1) Extinction, especially for habitat specialists
(2) Range expansion, especially for generalist species
and (3) Range contradiction, especially for Euro-Siberian mountainous species. Species modelling can be used to understand possible responses of plant species to climate change in Turkey. Modelling techniques should to be improved, however, especially by integrating other parameters such as biotic interactions and through a better understanding of uncertainties.

Les cétacés grands plongeurs, cachalots Physeteridae et Kogiidae, et baleines à bec Zipiidae, sont des espèces marines rares. Leur faible densité, aire de distribution étendue et faible disponibilité en surface génèrent de faibles taux d’observations. Cette particularité constitue un défi pour la modélisation d’habitat de ces espèces, préalable à leur conservation. Les modèles doivent gérer l’abondance de zéros qui limitent leur capacité à inférer des résultats écologiquement cohérents. Cette thèse vise donc à trouver une méthodologie adaptée aux jeux de données abondants en zéros, à déterminer comment les variables environnementales influencent la distribution des grands plongeurs et à prédire les zones potentielles qu’ils utilisent. Tester la capacité de prédiction de différents modèles d'habitat confrontés à un nombre décroissant d’observations a permis de souligner la pertinence d’un modèle, même si un minimum de 50 observations est nécessaire pour fournir des prédictions fiables. Des données issues de différentes campagnes visuelles ont été assemblées afin de produire les premières cartes de densités de grands plongeurs à l’échelle de l’océan Atlantique Nord et la mer Méditerranée. Les densités les plus élevées sont prédites dans les eaux entre 1500 et 4000 m de profondeur et près des fronts thermiques, particulièrement le long des pentes continentales et à l'ouest de l'océan Atlantique Nord. Par ailleurs, l’analyse de la transférabilité des modèles a montré une variation des habitats préférentiels en fonction des écosystèmes. Finalement, cette thèse permet de discuter les défis de la modélisation statistique appliquée aux espèces rares et les applications de gestion associées
Deep-diving cetaceans, sperm- and beaked whales Physeteridae, Kogiidae and Ziphiidae, are rare marine species. Due to their low densities, wide distribution ranges and limited presence at the water surface, visual surveys usually result in low sighting rates. This paucity of data challenges the modelling of their habitat, prerequisite for their conservation. Models have to cope with a great number of zeros that weakens the ability to make sound ecological inferences. Consequently, this thesis aimed at finding a methodology suitable for datasets with a large number of zeros, determining how environmental variables influence deep-diver distributions and predicting areas preferentially used by these species. By testing the predictive performance of various habitat models fitted to decreasing numbers of sightings, I selected the most suitable model and determined that at least 50 sightings were needed to provide reliable predictions. However, individual surveys can rarely provide sufficient deep-diver sightings thus I merged many visual survey datasets to produce the first basin-wide deep-diver density maps in the North Atlantic Ocean and the Mediterranean Sea. Highest densities were predicted in waters from 1500-4000 m deep and close to thermal fronts ; hotspots were predicted along the continental slopes, particularly in the western North Atlantic Ocean. In addition, a model transferability analysis highlighted that habitat drivers selected by the models varied between contrasted large ecosystems. Finally, I discussed challenges related to statistical modelling applied to rare species and the management applications of this thesis

To characterise the habitat preferences of ring ouzel (Turdus torquatus) and blackbird (T. merula) in Switzerland, we adopt species distribution modelling and predict the species’ spatial distribution. We model on two different scales to analyse in how far downscaling leads to a different set of predictors to describe the realised habitat best. While the models on macroscale (grid of one square kilometre) cover the entire country, we select a set of smaller plots for modelling on territory scale. Whereas ring ouzels occur in altitudes above 1’000 m a.s.l. only, blackbirds occur from the lowlands up to the timber line. The altitudinal range overlap of the two species is up to 400 m. Despite both species coexist on macroscale, a direct niche overlap on territory scale is rare. Small-scale differences in vegetation cover and structure seem to play a dominant role for habitat selection. On macroscale however, we observe a high dependency on climatic variables mainly representing the altitudinal range and the related forest structure preferred by the two species. Applying the models for climate change scenarios, we predict a decline of suitable habitat for the ring ouzel with a simultaneous median altitudinal shift of +440 m until 2070. In contrast, the blackbird is predicted to benefit from higher temperatures and expand its range to higher elevations.
Unter Verwendung von Habitatmodellen beschreiben wir die Habitatpräferenz von Amsel (Turdus merula) und Ringdrossel (T. torquatus) in der Schweiz. Mit Hilfe verschiedener Klimaszenarien prognostizieren wir zudem die künftige potentielle Verbreitung beider Arten. Zur Beschreibung eines eventuell auftretenden Skalensprungs, d.h. einer Änderung in der Beschreibungskraft der Variablen auf verschiedenen räumlichen Ebenen, erstellten wir Modelle auf zwei unterschiedlichen Skalen. Während das Modell auf Makroskala mit einer Maschenweite von einem Quadratkilometer die gesamte Schweiz abdeckt, erstellten wir zudem eine Auswahl an Untersuchungsgebieten auf Revierebene. Ringdrosseln zeigen ihren Verbreitungsschwerpunkt in der subalpinen Lage, während Amseln vornehmlich das Tiefland und die Tallagen besiedeln und nur vereinzelt in hohe Lagen vordringen. In einem Gürtel von ungefähr 400 Höhenmetern siedeln beide Arten parallel.Trotz dieses auf der Makroskala erkennbaren Überschneidungsbereiches konnten wir in unserer Untersuchung auf Revierebene, von einer Ausnahme abgesehen, keine Koexistenz beobachten. Kleinräumige Unterschiede in der Habitatstruktur, insbesondere in der Vegetationsbedeckung scheinen demnach für die Habitatselektion von maßgeblicher Bedeutung zu sein. Auf Makroebene hingegen wurde der Einfluss klimatischer Variablen deutlich, die neben der Höhenlage auch dort typische Vegetationsstrukturen widerspiegeln. Wie die Klimaszenarien zeigen, nehmen geeignete Ringdrosselhabitate bei steigenden Temperaturen ab und die Art weicht im Mittel um 440 m in höhere Lagen zurück. Für Amseln scheint sich eine zunehmende Erwärmung jedoch positiv auszuwirken, während das Verbreitungsgebiet im Tiefland beibehalten wird, dringt sie von den Tälern ausgehend zunehmend in höhere Lagen vor.

The amazon ecosystems have been compromised by historical forms of occupation and land-use causing habitat loss and forest fragmentation. These anthropogenic disturbances associated to climate changes have direct consequences on the distribution of species and their in situ persistence. Currently, 76 of 14.003 plants taxonomically identified in the Amazon have been listed by the Brasilian Ministry of the Environment as threatened species, though we believe this number to be much bigger in the reality. Among them, Amazon-nut (Bertholletia excelsa), a native tree species, national and internationally known for its cultural, social-economic and nutritional value has been classified as vulnerable. For developping of public policy turned to its management and conservation is fundamental to know the percentage of habitat available, as well as the nature and scale of threats to this environments. Species distribution modelling is an increasingly important tool for predicting habitat suitability and for understanding species environmental tolerances, but has been rarely used in Brazil, especially for Amazonian species. This study aimed to model the potential distribution of B. excelsa in the Amazon biome and to know the factors that control its distribution. To enhance our analysis, case studies were carried out with stakeholders aiming to know their perceptions about the main threats to the species and potential solutions.This research project was based on two hypotheses: (i) There is a suitable habitat to Amazon-nut which require different objectives for conservation and planting; (ii) If the local people are aware of the species vulnerability, they are able to point out the factors that cause this condition. In the chapter 1, habitat was investigated using MAXENT algoritm. We collected 3,325 Amazon-nut records and organized one hundred-and-two environmental variables into climatic, edaphic and geophysical categories at a spatial resolution of 30 arcs-second (~1km). Multi-colinearity between variables was dealt with multivariate statistics associated to expert\'s knowledge, and presence data biased with the spatial filtering. The best model was selected adopting quantitative metrics and visual examination. The most importante biophysic variables we identified were: altitude (m), coarse soil fragments (<2mm) and clay (%). Finaly, the best model indicated 2.3 million km2 i.e., 32% of the Amazon basin has potential for B. excelsa to grow. In the chapter 2, the factors that affect Amazon-nut conservation and planting were discussed with local communities, public managers and researchers, totalyzing 203 participants. Focus groups, individual interviews and questionaire techniques were used to gather information. Data were categorized and the perceptions among stakeholders compared using quali-quantitative analyses. We found that there are currently 36 problems responsible for the species vulnerability and 72% of them belong to environmental and political contexts. Deforestation was the main problem mentioned, followed by fruit depreciation, control failures and lack of organization in the communities. For three groups of stakeholders, the main solutions were related to political context. The results obtained in this study contribute to increase ecological knowledge on the species, to demonstrate the complexity of sustainable use in the Amazon and to guide decisions makers in the selection of priority areas for conservation and potential planting.
Os ecossistemas amazônicos vêm sendo impactados ao longo dos anos por diversos processos de uso e ocupação do território, os quais têm resultado em perdas de habitats e na fragmentação da paisagem nativa. Essas perturbações antrópicas, associadas às mudanças climáticas, têm consequências diretas sobre a distribuição e persistência das espécies in situ. Das 14.003 plantas da Amazônia reconhecidas taxonomicamente, somente 76 estão atualmente listadas pelo Ministério do Meio Ambiente brasileiro como espécies ameaçadas, embora acredita-se que esse número seja muito maior. Entre elas, a Castanha-da-Amazônia (Bertholletia excelsa), uma espécie de árvore nativa, reconhecida nacional e internacionalmente pela sua importância cultural, socioeconômica e nutricional, encontra-se classificada como vulnerável. Para nortear políticas públicas na conservação e no plantio dessa espécie, um profundo entendimento sobre o habitat disponível para ela, bem como a origem e escala das ameaças à esse ambiente, é necessário. A modelagem de distribuição de espécies é uma ferramenta que oferece predições espaciais robustas sobre a adequabilidade de habitat e tolerância das espécies, mas tem sido pouco utilizada no Brasil, sobretudo para espécies Amazônicas. Nesse contexto, esse estudo objetivou modelar a distribuição potencial da B. excelsa no bioma Amazônia, bem como conhecer os fatores que controlam sua distribuição. Para aprofundar essas análises, estudos de caso foram realizados com o objetivo de conhecer a percepção de atores sociais envolvidos com a espécie sobre as principais ameaças e potenciais soluções. Essa tese baseou-se em duas hipóteses: (i) existem áreas com maior adequabilidade para a ocorrência da Castanha-da-Amazônia que demandam diferentes objetivos, para conservação e para o plantio; (ii) se a população local é conciente da vulnerabilidade da espécie, ela pode indicar os fatores que geram essa condição. No capítulo 1, o habitat foi investigado por meio de simulações usando o algoritmo MAXENT. Um total de 3.325 ocorrências e 102 variáveis ambientais foram obtidas, e posteriormente organizadas por categorias climática, edáfica e geofísica. A resolução espacial escolhida foi de 30 arc-segundo (~1km). A multi-colinearidade entre as variáveis foi reduzida por meio da estatística multivariada associada ao conhecimento de especialistas, e as tendências nas ocorrência foram tratadas através da filtragem espacial. O melhor modelo foi selecionado usando métricas quantitativas e examinações visuais. As variáveis biofísicas mais importantes encontradas foram altitude (m), solos com fragmentos grosseiros (<2mm) e argila (%). Por fim, o modelo indicou que 2.3 million km2 i.e., 32% da região amazônica é apropriado para B. excelsa crescer. No capítulo 2, os fatores que afetam a conservação e o plantio da espécie foram discutidos com comunidades, gestores e pesquisadores locais, totalizando 203 participantes. As técnicas de discussão em grupo focal, entrevistas individuais e questionários foram utilizadas para a coleta das informações. Os dados foram categorizados e as opiniões entre os diferentes grupos comparadas utilizando análises quali-quantitativas. Concluiu-se que atualmente existem 36 problemas responsáveis pela vulnerabilidade da espécie, dos quais 72% encontram-se no contexto ambiental e político. O desmatamento foi a principal forçante apontada, seguida pela desvalorização do fruto, falhas na fiscalização e falta de organização nas comunidades. Para os três grupos, as principais soluções foram voltadas para o contexto político. Os resultados obtidos nesse estudo contribuiem para aumentar o conhecimento ecológico da espécie, para demostrar a complexidade do uso sustentável na Amazônia, e orientar tomadores de decisão na seleção de áreas prioritárias para conservação e potenciais para o plantio.

This thesis explores and proposes methods to improve species distribution models. Throughout this thesis, a rich class of statistical modelling techniques has been developed to address crucial and interesting issues related to the data input into these models. The overall contribution of this research is the advancement of knowledge on species distribution modelling through an increased understanding of extraneous zeros, quality of the ecological data, variable selection that incorporates ecological theory and evaluating performance of the fitted models. Though motivated by the challenge of species distribution modelling from ecology, this research is broadly relevant to many fields, including bio-security and medicine. Specifically, this research is of potential significance to researchers seeking to: identify and explain extraneous zeros; assess the quality of their data; or employ expert-informed variable selection.

Knowing where species are and understanding why is paramount for developing relevant and sustainable conservation and resource management strategies. The need for this information is becoming urgent as fishing activity, resource extraction and the impacts of coastal developments continue to put marine resources under increasing pressure. As logistical and financial constraints can restrict our ability to collect data in the marine environment, the ability to predict distributions based on known associations with different environmental variables would enhance our capacity to manage these resources. Before attempting to predict the distribution of species and groups of species, the underlying species-environment relationships must be examined to determine whether associations between species and the environment can: (i) be identified, (ii) be used to develop models that can accurately predict distributions, and (iii) are general enough to allow accurate predictions beyond the sampled area. Most studies to date have compared the composition of fish assemblages between sites to determine how different environmental variables influence distribution. While widely applied, these methods do not consider how individual species respond to multiple environmental gradients and they lack the ability to predict distributions across different combinations of variables along those gradients. This lack of prediction also limits our capacity to assess what marine biodiversity is presently threatened by global, regional, and local human pressures on marine ecosystems. '...' Thus, summarising and modelling species data at higher levels would result in models with poorer predictive accuracy and a loss of ecological information. The generality of the species-environment relationships defined by the models were assessed by evaluating the transferability of models between different areas. Models developed from data collected over a wider geographic extent could more accurately predict the distribution of species across a smaller spatial extent than vice versa. This indicated that while general theories of the ecology of temperate demersal fish can be defined, the actual patterns of distribution may vary from site to site, suggesting caution when using predictions beyond the sampled area for management purposes. Overall, species distribution modelling identified how different species and groups of species responded to the combined influence of multiple environmental gradients and was able to accurately predict distributions based on the defined associations. Their application has led to a greater understanding of the species environment relationships and will help to identify those areas that may be important for conservation. Their predictive ability will allow general predictions of distribution of fish species across unsurveyed areas and provides the ability to assess the potential impact from implementing different policy and management strategies.

Bacillus anthracis is the causal agent of anthrax which primarily affects ungulates, occasionally carnivores and less frequently humans. The endospores of this soil-borne bacterium are highly resistant to extreme conditions, and under ideal conditions, anthrax spores can survive for many years in the soil. The bacterium is generally found in soil at sites where infected animals have died. When these spores are exposed, they have the potential to be ingested by a mammalian species which could lead to an anthrax outbreak. Anthrax is almost never transmitted directly from host to host, but is rather ingested by herbivores while drinking, grazing or browsing in a contaminated environment, with the exception of scavengers and carnivores consuming infected prey. Anthrax is known to be endemic in the northern part of Kruger National Park (KNP) in South Africa (SA), with occasional epidemics spreading southward into the non-endemic areas. The aim of this study is to identify and map areas that are ecologically suitable for the harbouring of B. anthracis spores within the KNP. Anthrax surveillance data and selected environmental variables were used as inputs to the maximum entropy (Maxent) species distribution modelling method. Five-hundred and ninety-seven anthrax occurrence records, dating from the year 1988 to 2011, were extracted from the Skukuza State Veterinary Office’s database. A total of 40 environmental variables were used and their relative contribution to predicting suitability for anthrax occurrence was evaluated using Maxent software (version 3.3.3k). Variables showing the highest gain were then used for subsequent, refined model iterations until the final model parameters were established. The environmental variables that contributed the most to the occurrence of anthrax were soil type, normalized difference vegetation index (NDVI), land type and precipitation. A map was created using a geographic information system (GIS) that illustrates the sites where anthrax spores are most likely to occur throughout the Park. This included the known endemic Pafuri region as well as the low lying soils along the Shingwedzi-Phugwane-Bubube rivers and the Letaba-Olifants river drainage area. The outputs of this study could guide future targeted surveillance efforts to focus on areas predicted to be highly suitable for anthrax, especially since the KNP uses passive surveillance to detect anthrax outbreaks. Knowing where to look can improve sampling efficiency and lead to increased understanding of the ecology of anthrax within the KNP.
Dissertation (MMedVet)--University of Pretoria, 2013.
Production Animal Studies
unrestricted