The aim of this thesis was to create potential distribution models for Dicranum majus (Greater Fork Moss) and Polytrichum alpinum (Alpine Haircap) in Bohemian Switzerland National Park. In the Czech Republic these bryophyte species occur in cold climatic regions typically with higher altitudes. In Bohemian and Saxon Switzerland they can occur in really low altitudes thanks to unique microclimatic conditions of deep inversion ravines. These bryophyte species had low number of occurence records in studied area before the start of my research (4 occurence localities for Dicranum majus, 8 occurence localities for Polytrichum alpinum). Predictive habitat suitability models can be an effective tool for selecting potential new occurence localities, planning field research or management design. During field research I recorded 34 new occurence localities for Dicranum majus and 29 new occurence localities for Polytrichum alpinum in Bohemian Switzerland National Park. I used 8 topographic parameters derived from digital elevation model with 1 m resolution as environmental data. Using these data I created models of potential distribution of the most suitable habitats for both species with algorithms Artificial neural networks (ANN), Generalised linear model (GLM) and Random forest (RF). RF algorithm had the...

The larger pelagic crustacean fauna of the Gully submarine canyon is described for the first time, based on three annual summer surveys. The larger Crustacea are a significant part of the Gully fauna, and are dominated by cold temperate species, in particular the northern krill Meganyctiphanes norvegica and the decapod Sergestes arcticus. In all, at least 69 species were collected from the surface to bathypelagic depths, with seventeen being new Canadian records. With the exception of M. norvegica, inter-annual variation in the dominant species was minor. The species assemblage varied primarily with depth surveyed and diel cycle, and not year. Comparing the larger pelagic crustaceans in the Gully and over the adjacent continental slope showed that overall species number, biomass, and abundance were all greater in the canyon, the biomass of S. arcticus particularly showing a positive “Gully effect”.

L’aréthuse bulbeuse (Arethusa bulbosa L.) est une orchidée tourbicole rare au Québec de par la petite taille de ses populations et les pressions qui pèsent sur son habitat. L’aréthuse est reconnue comme étant intolérante aux changements édaphiques de son habitat, notamment en regard du drainage. Afin de déterminer si l’aréthuse est une bonne espèce indicatrice de l’intégrité écologique des tourbières, cette étude compare des parcelles où l’espèce est présente et où elle est absente. Au cours des étés 2009 et 2010, 37 tourbières du Québec méridional ont été échantillonnées. Des analyses discriminantes ont mis en évidence les facteurs naturels ou d’origine anthropique qui expliquent le mieux la différence entre les quatre types de parcelles. Aussi, la recherche d’espèces indicatrices de la présence de l’aréthuse a été réalisée grâce à la méthode INDVAL. Les résultats montrent que l’aréthuse pousse principalement dans des tourbières présentant des conditions minérotrophes, ce qui est appuyé par le pH élevé et la présence de plantes indicatrices de minérotrophie dans les parcelles contenant l’aréthuse. Cette dernière semble aussi profiter d’une certaine atténuation de la lumière par des arbres dispersés ou par les strates plus basses. Finalement, certaines perturbations de faible ampleur semblent être bénéfiques pour l’aréthuse, ce qui ne permet pas d’affirmer qu’elle est une bonne espèce indicatrice de l’intégrité écologique des tourbières. / Arethusa bulbosa L. is a peatland orchid rare in Québec because of the small size of its populations and the pressures on its habitat. It is known to be intolerant to edaphic changes of its habitat, and particularly to drainage. To determine if A. bulbosa is a good indicator species of the ecological integrity of peatlands, this study compares plots where the species is present from plots where it is absent. During the summers of 2009 and 2010, 37 peatlands were surveyed in southern Québec. Discriminant analyses were performed to find which natural or anthropogenic factors best explain the difference between plots with or without the orchid. Also, indicator species of the presence of A. bulbosa were found using INDVAL analyses. Overall, the results suggest that A. bulbosa was mostly growing in peatlands characterized by minerotrophic conditions. Two evidences supported this point, the high pH values measured and the presence of indicator species of minerotrophy in plots with the orchid. It is likely that A. bulbosa is favored by a certain interception of light by scattered trees or species growing in lower strata. Finally, despite a high floristic quality assessment index (FQAI) in A. bulbosa plots, it is likely that the species takes advantage of small-scale disturbances in it habitat and is therefore not a good indicator species of peatlands ecological integrity.

The research object. The research was made in the territory of Neris regional park (Kaisiadoriu forest enterprise, Vievis forest district) in order to evaluate the impact of recreational activities on the forest litter arthropods. The aim of the study – to evaluate different recreational digression stage impact on pinery litter arthropods: their amount and seasonal dynamics, distribution trophic groups, the abundance and variety of species in stands affected by recreation. Methods. Arthropods were collected in Barber ground traps from April through October 2005, which are plastic, 150 ml. volume, with 6.5 cm diameter receptacle. They were dug in the ground and 1/3 of trap was filled with 10 % formalin liquor, which protected insects from decaying and insectivorous birds. Ground traps in each sample plot were spread in groups by 4 (square), the distance between them was 2 m. The distance between sample plots was 20-50 meters. The results. 157 species were identified in trap catches: 96 in the control plot, 97 in the plot with low recreational damage and 110 in the plot wit high recreational damage. Entomophagous predominated and made 60-73 %, kaprophagous 8-25%, saprophagous 10% of catch. 2 extremely rare species and 7 rare species in Lithuania were found, the biggest number of rare species was found in the plot with high recreational damage. Shannon diversity index H’ was 2.5-3.31, evenness E – 0.55-0.66, Berger-Parker dominance d – 0.170-0.367. Sorensen similarity index Cs... [to full text]

Made available in DSpace on 2014-12-17T15:54:56Z (GMT). No. of bitstreams: 1 AnaCPO_DISSERT.pdf: 970919 bytes, checksum: 10654bf99e2119dae47a190c7278a096 (MD5) Previous issue date: 2011-02-21 / Studies on Brazilian biodiversity are still very few and can observe the difference of knowledge between the different regions of the country. This affirmation can be verified in the investigation to identify the rare species in Brazil and the key biodiversity areas (ACBs). In that study were identified for Brasil 2.256 rare species and 752 ACBs. The Rio Grande do Norte (RN) was the only Brazilian state that has not been identified any rare species and no ACBs, possibly due to the lack of floristic studies in this state. A particular area was selected for this study: an area of ecological tension with savanna physiognomy in Rio do Fogo, RN. This savanna community is represented in RN in a fragment and immersed in restinga and caatinga and was identified and described only through of radar imagens there are no studies to date in loco . We have prepared the following questions about this community savanna: 1) The region delimited and described by through of radar images by the RADAMBRASIL, 1976, can be associated of the Cerrado, in terms of floristic ?; 2) What is the floristic composition of this area? This area includes rare species, endemic or endangered? 3) What is the geographical and phytogeographical distribution of plant species registered in this area? 4) Those plant species registered are endemic or have affinity with other areas phytoecological Brazilian? To answer these questions we performed a floristic inventory of the August 2007 to September/2009. The results are presented in two chapters (manuscripts). The first chapter, titled "The Savannah Rio Grande do Norte: floristic links with other plant formations in the Northeast and Center-West Brazil" was submitted to the Revista Brasileira de Bot?ncia. Chapter 1 discusses the phytogeographical distribution of the species, by comparing floristic studies conducted in the Cerrado, Caatinga and Restinga in the Northeast and Cerrado of the Central Brazil. The analysis of data of this study and compilation with other studies indicated that: i) the record of 94 plant species; ii) of total species, about 64% are associated with the Cerrado, the second specialized bibliography, and about 78% as the List of Species of Flora of Brazil. However, about 73% of total species (94) are also distributed in the Caatinga, the Atlantic forest 64%, the Amazon forest 64%, the Pantanal 15% and the Pampa 12%. Floristically the data show that the community studied is influenced by other floras, has a structure where grasses dominate and also because of his appearance the same savanna then be classified as a Savana gram?neo-lenhosa do tabuleiro . Chapter 2, titled "Considerations on the flora of a savanna community in Rio Grande do Norte, northeastern Brazil: Subsidy key area for conservation" was submitted to the Revista Natureza e Conserva??o. This has the objective of improve the knowledge of flora of Rio Grande do Norte and to identify possible rare species and consequently increase the key areas of biodiversity in Brazil. The data indicated that: i) of 94 species registered in the study area, 40 were new records for the Rio Grande do Norte state; ii) These citations to unpublished state, Stylosanthes montevidensis Vogel (Fabaceae) and Aristida laevis (Nees) Kunth (Poaceae) are indicated for the first time to the Northeast of Brazil; iii) are registered in the area 24 species endemic to Brazil and 63 non-endemic; iv) Aspilia procumbens Baker (Asteraceae) registered in the area is considered a restricted species and micro endemic Rio Grande do Norte, ie rare species; v) Aspilia procumbens is also cited in the category of critically endangered species and Stilpnopappus cearensis Hubber (Asteraceae) a species vulnerable to extinction. This study shows a new area phytoecological in Rio Grande do Norte and indicates the area's potential to contribute with the sites of global significance for biodiversity conservation, either locally, regionally and nationally. This will certainly contribute to respond some targets set by the Global Strategy for Plant Conservation and the Convention on Biological Diversity such as the inventory of vegetal diversity in a region with little collection, which will provide data that contributes to questions and themes related to biodiversity. / Estudos sobre a biodiversidade brasileira s?o ainda muito escassos, sendo poss?vel observar uma grande discrep?ncia de conhecimentos entre as v?rias regi?es do pa?s. Esta afirma??o est? baseada no fato de que foram identificadas para o Brasil 2.256 esp?cies raras de plantas e 752 ?reas-chave para biodiversidade (ACBs), no entanto, o Rio Grande do Norte foi o ?nico estado brasileiro que n?o apresentou nenhuma esp?cie rara e nenhuma ?rea-chave para biodiversidade, possivelmente, devido ? escassez de levantamentos flor?sticos nesse estado. O presente estudo foi realizado em um ambiente especial: uma ?rea de tens?o ecol?gica com fisionomia sav?nica localizada no munic?pio de Rio do Fogo, RN. Essa comunidade sav?nica est? representada na forma de pequena mancha e imersa entre a Caatinga e a Restinga. Foi identificado e descrito apenas com base em imagens de radar, n?o havendo, at? o presente momento, estudos in loco. Neste contexto foram elaboradas algumas perguntas sobre essa comunidade sav?nica: 1) A regi?o delimitada e descrita por meio de imagens de radar, RADAMBRASIL de 1976, pode ser do dom?nio Cerrado, do ponto de vista flor?stico?; 2) Qual ? a composi??o flor?stica dessa ?rea? Ela inclui esp?cie rara, end?mica ou amea?ada de extin??o?; 3) Qual ? a distribui??o geogr?fica e fitogeogr?fica das esp?cies vegetais registradas nessa ?rea?; 4) As esp?cies vegetais registradas s?o end?micas ou possuem afinidade com outros dom?nios fitoecol?gicos brasileiro? Visando responder a essas indaga??es foi realizado um levantamento flor?stico, no per?odo de agosto/2007 a setembro/2009. Os resultados obtidos s?o apresentados em forma de dois cap?tulos (manuscritos). O cap?tulo 1, intitulado A Savana do Rio Grande do Norte: Rela??es flor?sticas com outras forma??es vegetais do Nordeste e Centro-Oeste Brasileiro foi submetido ? Revista Brasileira de Bot?nica. Aborda a distribui??o fitogeogr?fica das esp?cies, atrav?s da compara??o com estudos flor?sticos realizados nos cerrados, caatingas e restingas do Nordeste e cerrados do Planalto Central. As an?lises dos dados obtidos no presente trabalho e, tamb?m, atrav?s de compila??o com outros estudos apresentaram: i) o registro de 94 esp?cies vegetais; ii) desse total de esp?cies, cerca de 64 % est?o associadas ao Cerrado, segundo bibliografia especializada, e cerca de 78 % conforme a Lista de Esp?cies da Flora do Brasil. Todavia, cerca de 73 % do total das esp?cies (94) registradas tamb?m se distribuem na Caatinga, 64 % na Floresta Atl?ntica, 50 % na Floresta Amaz?nica, 15 % no Pantanal e 12 % no Pampa. Floristicamente os dados apontaram que comunidade estudada recebe influ?ncia de outras floras, possui uma estrutura onde as gram?neas dominam e tamb?m devido a sua fisionomia sav?nica a mesma, ent?o, poder ser classificada como Savana gram?neo-lenhosa do tabuleiro. O cap?tulo 2, intitulado Considera??es sobre a flor?stica de uma comunidade sav?nica no Rio Grande do Norte, Nordeste do Brasil: Subs?dio de ?rea-chave para a conserva??o foi submetido ? Revista Natureza e Conserva??o. Teve como objetivo contribuir para melhorar o conhecimento sobre a flora do Rio Grande do Norte, identificando poss?veis esp?cies raras e, assim, expandir as ?reas-chave de biodiversidade no Brasil. Os dados obtidos indicaram que: i) das 94 esp?cies encontradas na ?rea de estudo, 39 foram registradas pela primeira vez para o Rio Grande do Norte; ii) dessas cita??es in?ditas para o estado, as esp?cies Stylosanthes montevidensis Vogel (Fabaceae) e Aristida laevis (Nees) Kunth (Poaceae) s?o indicadas, tamb?m, pela primeira vez para a Regi?o Nordeste do Brasil; iii) ocorrem na ?rea 24 esp?cies end?micas do Brasil; iv) Aspilia procumbens Baker (Asteraceae) registrada na ?rea ? considerada uma esp?cie restrita e microend?mica do Rio Grande do Norte, ou seja, esp?cie rara; v) Aspilia procumbens ? tamb?m apontada na categoria das esp?cies criticamente em perigo e Stilpnopappus cearensis Hubber (Asteraceae) uma esp?cie vulner?vel a extin??o. O presente estudo aponta um novo dom?nio citoecol?gico para o Rio Grande do Norte e apresenta o potencial da ?rea em contribuir com os s?tios de signific?ncia global para conserva??o de biodiversidade, seja no ?mbito local, regional e nacional. Isso contribuir? para responder alguns dos objetivos fixados pela Estrat?gia Global para a Conserva??o de Plantas e da Conven??o sobre Diversidade Biol?gica, tal como, o invent?rio da diversidade vegetal, em uma regi?o com pouca coleta, que fornecer? dados que contribuem para quest?es e temas relativos ? biodiversidade.

No description available.

No contexto da conservação biológica, a fragmentação florestal é definida como uma separação ou desligamento não natural de áreas amplas em fragmentos espacialmente segregados, promovendo a redução dos tipos de habitats e a divisão dos habitats remanescentes em unidades menores e isoladas. A magnitude e a extensão dessas alterações na paisagem são influenciadas pelo tamanho, conectividade, forma, contexto e heterogeneidade dos fragmentos. Visando a conservação ambiental do município de Paulínia-SP, este trabalho teve como objetivo efetuar o diagnóstico da vegetação remanescente. Os fragmentos florestais identificados no município foram: Floresta Estacional Semidecidual, Floresta Estacional Semidecidual Ribeirinha, Floresta Paludosa e Cerrado. A caracterização da vegetação foi realizada através do levantamento florístico rápido baseado no tempo de amostragem. Os 104 fragmentos apresentaram riqueza de 517 espécies, sendo 447 espécies nativas regionais (região de Paulínia), 18 espécies nativas não regionais (nativas de outras regiões do Brasil) e 52 exóticas (originárias de outros países). Foram registradas espécies raras como Tabebuia insignis, Pilocarpus pennatifolius, Salix humboldtiana, Handroanthus umbellatus. A análise de similaridade florísica utilizando-se o Método de agrupamento de Sorensen foi efetiva para separar os fragmentos de Cerrado e Floresta Paludosa e mostrou tendência de agrupamento para algumas variações fisionômicas da Floresta Estacional Semidecidual. Com a finalidade de se obter dados quantitativos (abundância) das espécies, foram registradas todas as espécies diferentes encontradas em cada intervalo de tempo de amostragem no Levantamento Rápido. Obtiveram-se indicativos das 517 espécies com relação a serem raras ou comuns na paisagem, em cada formação florestal e em cada fragmento estudado. Foi aplicado um total de 801 intervalos de tempo de 15 min de amostragem e gerados 21.324 registros. A composição de espécies variou com o tamanho e grau de conservação dos fragmentos. Foram analisadas as influências do tipo de fisionomia florestal, tamanho e estado de conservação dos fragmentos na composição das espécies em relação às classes de sucessão ecológica e às síndromes de dispersão. A composição das classes de sucessão variou com o tamanho e estado de conservação dos remanescentes de Floresta Estacional Semidecidual e Floresta Estacional Semidecidual Ribeirinha. Em fragmentos menores e muito degradados houve predominância de espécies em estágios iniciais de sucessão (pioneiras e secundárias) em relação à classe de final de sucessão (clímax). Houve predominância da zoocoria, seguida de anemocoria e depois autocoria em todas as situações analisadas. / In biological conservation, forest fragmentation is defined as a separation or nonnatural disconnection of wide areas in spatially segregated fragments, generating a reduction of habitat types and the split of remnant habitat in smaller isolated units. The magnitude and extension of these landscape alterations are influenced by size, connectivity, shape, context and heterogeneity of fragments. Aiming environmental conservation of the municipality of Paulinia, SP, this study aimed to diagnose the remnant vegetation. The forest fragments identified in the municipality were: Seasonal Semidecidual Forest, Seasonal Semidecidual Riparian Forest, Swamp Forest and Savanna. The vegetation characterization was done through a quick floristic search based on sampling time. The 104 fragments presented a richness of 517 species, 447 of which are regional native species (Paulinia region) and 18 are non-regional native species (from other regions of Brazil); and 52 exotic species (from other countries). These fragments presented rare species as Tabebuia insignis, Pilocarpus pennatifolius, Salix humboldtiana, Handroanthus umbellatus. The floristic similarity analysis using the Sorensen method was particularly effective to separate Savanna and Swamp Forest fragments and showed a tendency for grouping some variations of Seasonal Semidecidual Forest. In order to obtain quantitative data (abundance) about the species, all different species found in each time interval in the quick floristic search were recorded. Indicatives for 517 species were obtained classifying then into rare or common in the landscape, in each forest formation, and in each fragment studied. 21.324 records were generated for a total of 801 intervals of 15 minutes sampling. Species composition varied with size and conservation degree of fragments. The influences of the forest physiognomy type, size and conservation status of fragments on species composition related to ecological succession classes and dispersion syndromes were analyzed. The ecological classes composition varied along with the size and conservation status for the fragments of Seasonal Semidecidual Forest and Seasonal Semidecidual Riparian Forest. In smaller and very degraded fragments there was a predominance of species from initial sucessional stages (pioneer and secondary species) in comparison to species from final stages of succession (climax). There was predominance of zoochory, followed by anemochory and autochory in all situations analyzed.

La chasse aux trophées, type de chasse récréative dont l'objectif premier est de prélever un trophée d'intérêt, est un sujet très controversé. Cette activité est potentiellement génératrice d'un effet Allee anthropogénique (EAA). Ce processus démographique stipule que la valorisation de la rareté d'une espèce peut stimuler son exploitation et mener à son extinction. L'objectif de ce travail a été de tester le risque d'EAA dans la chasse aux trophées. Nous avons mis en évidence que les espèces rares ont une forte valeur économique dans la chasse aux trophées, quelque soit la taille du trophée, ce qui témoigne d'une demande forte pour ces espèces. Nous avons également montré que le nombre d'importations/exportations de trophées et le nombre de trophées de chasse enregistrés par le Safari Club International s'accroît pour les espèces les plus rares lorsque l'indice de rareté de l'espèce augmente. La chasse aux trophées sur les espèces rares a été proposée comme outil pour financer leur conservation. Cependant, nos résultats valident le risque d'EAA pour ces espèces. De plus, les chasseurs ne semblent pas motivés en priorité par la participation à la conservation de la vie sauvage et le peu de données disponibles sur les populations, la chasse illégale, la corruption et le manque de contrôle rendent possible l'exploitation non durable de ces ressources à forte valeur économique. Néanmoins, la vie sauvage doit apporter des bénéfices économiques pour motiver sa préservation. Ainsi, la chasse aux trophées des espèces rares peut être utilisée pour financer leur conservation mais certaines mesures sont à prendre au préalable pour prévenir leur surexploitation / Trophy hunting, which is a form of recreational hunting with the main objective of collecting a trophy of interest, is a controversial subject. This activity could potentially generate an anthropogenic Allee effect (AAE). This demographic process states that the valuation of rarity could drive rare species exploitation and even lead to their extinction. Our project aims at testing the potential for an AEE in trophy hunting. We demonstrate that rare species have a high financial value, regardless of the trophy size, indicating that there is a high demand for those species. We also show that the number of trophies traded internationally and the number of recorded trophies by the Safari Club International (one of the largest clubs for international trophy hunters in the USA) rises as the degree of rarity (as measured by a rarity index) increases. Trophy hunting of rare species has been proposed as a tool to fund their conservation. However, our results indicate that there is a risk of an AAE for rare species. Furthermore, the combined effects of trophy hunting, illegal hunting, corruption as well as the lack of population knowledge and of management controls have potential to result in the unsustainable exploitation of rare species of high financial value. Nonetheless, trophy hunting has potential to generate strong financial incentives that are necessary for wildlife preservation. Such incentives are only likely to be effective if strict measures are required and enforced to prevent overexploitation of rare trophy species

Le recours à des épandages d'amendements calco-magnésiens dans les forêts acidifiées peut représenter une alternative intéressante pour corriger l'acidification des sols et la perte d'éléments nutritifs engendrés par des décennies de dépôts atmosphériques acides. Dans ce contexte, les objectifs de cette thèse sont d'évaluer les effets d'amendements calco-magnésiens sur (i) la diversité taxonomique et la structure des communautés des macroinvertébrés de sol forestier et sur (ii) deux processus fonctionnels associés notamment à l'activité des vers de terre, tels que la structuration du sol et l'évolution des humus. Au cours de nos travaux, trois sites localisés dans les Vosges (nord-est, France) ont été étudiés : deux massifs forestiers (sur grès et granite) en moyenne montagne, ayant fait l'objet d'amendements aériens en 2003 et la forêt domaniale de Humont située dans les collines sous-vosgiennes, où des amendements en 1991 et 2008 ont été réalisés. Les résultats montrent que 4 ans après amendement sur les bassins versants en moyenne montagne, l'abondance totale de la communauté de macrofaune a diminué, alors que les richesses spécifiques sont similaires. Cependant, la composition des communautés diffère avec moins de 50 % d'espèces communes aux sites amendés et témoins. Une diminution de l'abondance des prédateurs a également été observée alors que la plupart des détritivores ont été favorisés par l'amendement. Les études in vitro montrent également son effet positif sur les activités de structuration du sol par les vers de terre (production de turricules et construction de galeries). En forêt de Humont, les amendements ont eu un effet majeur à moyen terme (4 ans) et à long terme (20 ans) sur la population d'Aporrectodea velox, espèce endémique des Vosges. De part sa biomasse importante, cette augmentation de densité favorise l'amélioration des caractéristiques physico-chimiques du sol forestier tels que la structuration des sols et la morphologie des humus. / The use of liming (Ca-Mg) on acidified forest can be interesting to counteract soil acidification and loss of nutrients caused by decades of acid atmospheric deposition. In this context, the aims of this work were to assess liming effect on (i) taxonomical diversity and community structure of soil macro-invertebrates and on (ii) two associated functional processes related to earthworms: soil structuration and humus evolution. The PhD project was realised in three sites from the Vosges mountains (North-eastern, France): two forest mountain catchments (one lying on sandstone and the other on granite) limed in 2003 and the public hill forest of Humont at a lower altitude limed in 1991 and 2008. Results showed that 4 years after liming in mountain forest, the total abundance of macro-invertebrates decreased, while the species richness were similar. However, the community structure strongly differed, and less than 50% of the species were common to limed and control sites. Moreover, 43 species appeared to be indicators of liming. Predator taxa abundance decreased whereas the detritivorous were favoured by lime addition. The in vitro experiments showed a positive effect of liming on soil structuration by earthworm activities (cast production and burrowing activities). In the Humont forest, liming at medium-term (4 years) and long term (20 years) have an important effect on the Aporrectodea velox population, an endemic vosgian species. Related to its high biomass, this increase strongly improved soil physicochemical parameters such as soil structure and humus morphology.

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.