White-tailed deer (Odocoileus virginianus) are a highly charismatic game species with considerable ecological and economic impacts across most of their range. In the Central Appalachian Mountains, deer are a keystone species in forested ecosystems. Regionally, populations vary in herd growth or decline. These fluctuations are important in that they often drive many aspects of population management and regulation, which are dependent on herd demographics. Some key population vital rates allowing better understanding of these changes in white-tailed deer herds are survival, cause-specific mortality, home-range variation, both broad and fine-scale resource selection, and ultimately population growth trends in response to changes in both population vital rates and hunter harvest regulations. In this study, I address each of these concepts within a deer population in Bath County, Virginia, that has presumably been in overall decline since the early 1990's. From June-September, 2019-2020, I monitored survival and cause specific mortality of 57 neonate white-tailed deer until 12 weeks of age. Fawn and adult female survival was 0.310 (95% CI = 0.210-0.475) and 0.871 (95%CI=0.790-0.961) respectively. During the study, I observed a total of 37 fawn mortalities and identified the cause of death using field evidence and through analyzing genetics from residual predator salvia recovered on deer carcasses. Mortalities included 28 predation events and 9 deaths from other causes (e.g., abandonment, malnutrition, or disease). Black bears accounted for 48.6% of all mortality and 64.2% of known predations within our study. My top model identified elevation as a significant predictor of fawn survival, with mortality risk increasing 20% for every 100m increase in elevation. My model using observed vital rates predicted an increasing population of λ = 1.10 (interquartile range, IQR 1.06-1.14). The population was predicted to increase by 2% with a 10% increase in doe harvest (λ = 1.02, IQR = 0.97-1.06) but declined by 7% at 20% harvest (λ = 0.93, IQR = 0.89-0.96). I found that fawning home ranges of females that successfully reared fawns to the end of the season had significantly larger home ranges than those that were unsuccessful at higher elevations. Fawning home ranges for females with fawns increased approximately 71ha in size for every 100m increase in mean home range elevation, whereas seasonal home ranges of females without fawns decreased approximately 1.5 ha for every 100m increase in mean home range elevation. Deer selected fawn-rearing areas nearer to forested edges, open habitats, and at higher elevations, while they avoided areas near disturbed and mature forests. Within the fawn rearing area, females selected locations closer to disturbed forest, open habitats, and forested edge, while avoiding mature forest habitats, and higher elevations. Females selected birth sites with higher levels of visual obstruction. Using a step-selection method for real-time resource selection across biological seasons, we found that female deer selected for open areas during the fawning, breeding, early gestational, and late gestational seasons. During the fall breeding season, females avoided forested edge, but selected for areas at higher elevations. During early gestational seasons females selected disturbed habitats and areas at higher elevations while again avoiding forested edge. Overall, my work highlights variations in population dynamics of white-tailed deer in areas of the Central Appalachian Mountains that are primarily characterized by poor habitat quality and provides novel insights into fine-scale spatial ecology of female deer across biological seasons within the region. Ultimately, while the deer population in our study was not predicted to be in decline, this work supports predation risk as being a significant factor associated with habitat quality. / Doctor of Philosophy / White-tailed deer (Odocoileus virginianus) are a game species with extensive ecological and economic impacts across most of their range. In the central Appalachian Mountains, many populations across the region vary in terms of herd size growth, stability, or decline. These fluctuations are important in that they often drive many aspects of population management and regulation, which are dependent on the status of herd demographics. Some key population vital rates that aid us in better understanding these changes in white-tailed deer herds are survival and cause-specific mortality, home-range variation in association with habitat quality and the ability to successfully reproduce, population trends under hypothetical management scenarios, and resource selection of various habitats that are available across the landscape. In this study, I address each of these concepts within a deer population in Bath County, Virginia, which has presumably been in decline since the early 1990's. As expected, fawn survival was lower than previously reported from other study areas of the Central Appalachians Mountains. Predation was the leading cause of fawn mortality, with black bears being responsible for most predation events. Fawn mortality risk was significantly associated with elevation - where fawns at higher elevations were those at greater risk. Surprisingly, the deer population in Bath County was projected to be increasing under current conditions and was also projected to be stable-to-increasing even under some hypothetical scenarios which would negatively impact population growth (i.e., 10% increase in female harvest or 17% reduction in fawn survival). Fawning home ranges of collared females which successfully reared known fawns to the end of the biological season were significantly influenced by elevation; such that females with fawns had home ranges that increased in size with increasing elevation, whereas females without fawns had home ranges which decreased slightly in size with increasing elevation. At birth sites, females selected locations characterized by higher levels of visual obstruction compared to randomly sampled areas. Of the habitat types analyzed, both selection or avoidance of specific habitats varied across both biological season and spatial scale. Ultimately, while I found that some deer populations associated with poor quality habitats in the Central Appalachians may not be in decline, deer were likely influenced greatly by habitat quality – especially pertaining to predation risk – throughout Bath County.

Em ecologia, o comportamento dos animais é freqüentemente estudado para entender melhor suas preferências por diferentes tipos de alimento e habitat. O presente trabalho esta relacionado a este tópico, dividindo-se em três capítulos. O primeiro capitulo refere-se à estimação da função da probabilidade da seleção de recurso (RSPF) comparado com um modelo de escolha discreta (DCM) com uma escolha, usando as estatísticas qui-quadrado para obter as estimativas. As melhores estimativas foram obtidas pelo método DCM com uma escolha. No entanto, os animais não fazem a sua seleção baseados apenas em uma escolha. Com RSPF, as estimativas de máxima verossimilhança, usadas pela regressão logística ainda não atingiram os objetivos, já que os animais têm mais de uma escolha. R e o software Minitab e a linguagem de programação Fortran foram usados para obter os resultados deste capítulo. No segundo capítulo discutimos mais a verossimilhança do primeiro capítulo. Uma nova verossimilhança para a RSPF é apresentada, a qual considera as unidades usadas e não usadas, e métodos de bootstrapping paramétrico e não paramétrico são usados para estudar o viés e a variância dos estimadores dos parâmetros, usando o programa FORTRAN para obter os resultados. No terceiro capítulo, a nova verossimilhança apresentada no capítulo 2 é usada com um modelo de escolha discreta, para resolver parte do problema apresentado no primeiro capítulo. A estrutura de encaixe é proposta para modelar a seleção de habitat de 28 corujas manchadas (Strix occidentalis), assim como a uma generalização do modelo logit encaixado, usando a maximização da utilidade aleatória e a RSPF aleatória. Métodos de otimização numérica, e o sistema computacional SAS, são usados para estimar os parâmetros de estrutura de encaixe. / In ecology, the behavior of animals is often studied to better understand their preferences for different types of habitat and food. The present work is concerned with this topic. It is divided into three chapters. The first concerns the estimation of a resource selection probability function (RSPF) compared with a discrete choice model (DCM) using chi-squared to obtain estimates. The best estimates were obtained by the DCM method. Nevertheless, animals were not selected based on choice alone. With RSPF, the maximum likelihood estimates used with the logistic regression still did not reach the objectives, since the animals have more than one choice. R and Minitab software and the FORTRAN programming language were used for the computations in this chapter. The second chapter discusses further the likelihood presented in the first chapter. A new likelihood for a RSPF is presented, which takes into account the units used and not used, and parametric and non-parametric bootstrapping are employed to study the bias and variance of parameter estimators, using a FORTRAN program for the calculations. In the third chapter, the new likelihood presented in chapter 2, with a discrete choice model is used to resolve a part of the problem presented in the first chapter. A nested structure is proposed for modelling selection by 28 spotted owls (Strix occidentalis) as well as a generalized nested logit model using random utility maximization and a random RSPF. Numerical optimization methods and the SAS system were employed to estimate the nested structural parameters.

Em ecologia, o comportamento dos animais é freqüentemente estudado para entender melhor suas preferências por diferentes tipos de alimento e habitat. O presente trabalho esta relacionado a este tópico, dividindo-se em três capítulos. O primeiro capitulo refere-se à estimação da função da probabilidade da seleção de recurso (RSPF) comparado com um modelo de escolha discreta (DCM) com uma escolha, usando as estatísticas qui-quadrado para obter as estimativas. As melhores estimativas foram obtidas pelo método DCM com uma escolha. No entanto, os animais não fazem a sua seleção baseados apenas em uma escolha. Com RSPF, as estimativas de máxima verossimilhança, usadas pela regressão logística ainda não atingiram os objetivos, já que os animais têm mais de uma escolha. R e o software Minitab e a linguagem de programação Fortran foram usados para obter os resultados deste capítulo. No segundo capítulo discutimos mais a verossimilhança do primeiro capítulo. Uma nova verossimilhança para a RSPF é apresentada, a qual considera as unidades usadas e não usadas, e métodos de bootstrapping paramétrico e não paramétrico são usados para estudar o viés e a variância dos estimadores dos parâmetros, usando o programa FORTRAN para obter os resultados. No terceiro capítulo, a nova verossimilhança apresentada no capítulo 2 é usada com um modelo de escolha discreta, para resolver parte do problema apresentado no primeiro capítulo. A estrutura de encaixe é proposta para modelar a seleção de habitat de 28 corujas manchadas (Strix occidentalis), assim como a uma generalização do modelo logit encaixado, usando a maximização da utilidade aleatória e a RSPF aleatória. Métodos de otimização numérica, e o sistema computacional SAS, são usados para estimar os parâmetros de estrutura de encaixe. / In ecology, the behavior of animals is often studied to better understand their preferences for different types of habitat and food. The present work is concerned with this topic. It is divided into three chapters. The first concerns the estimation of a resource selection probability function (RSPF) compared with a discrete choice model (DCM) using chi-squared to obtain estimates. The best estimates were obtained by the DCM method. Nevertheless, animals were not selected based on choice alone. With RSPF, the maximum likelihood estimates used with the logistic regression still did not reach the objectives, since the animals have more than one choice. R and Minitab software and the FORTRAN programming language were used for the computations in this chapter. The second chapter discusses further the likelihood presented in the first chapter. A new likelihood for a RSPF is presented, which takes into account the units used and not used, and parametric and non-parametric bootstrapping are employed to study the bias and variance of parameter estimators, using a FORTRAN program for the calculations. In the third chapter, the new likelihood presented in chapter 2, with a discrete choice model is used to resolve a part of the problem presented in the first chapter. A nested structure is proposed for modelling selection by 28 spotted owls (Strix occidentalis) as well as a generalized nested logit model using random utility maximization and a random RSPF. Numerical optimization methods and the SAS system were employed to estimate the nested structural parameters.

A percepção e interpretação da interação de indivíduos e populações com o ambiente e a forma como tal relação condiciona sua distribuição espacial é questão-chave e recorrente em estudos ecológicos. Padrões de uso de área observados para populações emergem em ultima análise da variabilidade entre seus indivíduos em selecionar habitats e interagir com os mesmos. Este estudo teve como foco o uso de área pela população do boto-cinza, Sotalia guianensis, e sua variabilidade individual no estuário de Cananeia, localizado na costa sudeste do Brasil (25°03\' S; 47°55\' W), durante o verão e o inverno de 2015 e o verão de 2016. Parâmetros ambientais e geográficos (distâncias da desembocadura de rios, da entrada do estuário e de áreas urbanas, profundidade, maré e autocorrelação espacial) foram testados para explicar a distribuição da população e de seus indivíduos a partir de funções de probabilidade de seleção de recursos (RSPF) em modelos aditivos generalizados (GAM). Onze indivíduos fotoidentificados com 18 ou mais recapturas foram avaliados com o uso de modelos individuais de ocupação e sua interpretação foi subsidiada por estimativas de áreas domiciliares obtidas a partir de kerneis fixos de densidade. Nas três temporadas a população apresentou densidades de grupos desiguais ao longo do estuário e todas as variáveis, com exceção da distância de áreas urbanas, explicaram as probabilidades de presença observadas. Análises individuais revelaram discrepâncias nos tamanhos e disposição geográfica de áreas domiciliares e diferenças na composição e estimativa dos parâmetros selecionados para cada indivíduo. A variabilidade individual na população deve ter papel fundamental em termos de utilização do espaço e seleção de habitat pelo boto-cinza no estuário local. / Understanding and interpreting the interaction of individuals and populations with the environment and how this relationship outlines their spatial distribution is a key question common in ecological studies. Area use patterns observed for populations are ultimately an outcome from individual variability in habitat selection and their interaction with such environments. Are use and habitat selection by the population of Guiana dolphins, Sotalia guianensis, and its individual variability were accessed in the Cananeia estuary (25°03\' S; 47°55\' W), southeastern Brazil, during the summer and winter of 2015 and the summer of 2016. Environmental and geographic parameters were estimated aiming to explain population distribution and differences within individuals. For this purpose, resource selection probability functions (RSPF) were applied in generalized additive models (GAM). Covariates tested included: distance to river mouths, distance to the estuary entrance, distance to urban areas, depth and tide. Geographic coordinates were used to model spatial autocorrelation. Eleven photo-identified individuals had their occupancy modelled and accessed in relation to their home range obtained from fixed kernel densities estimates. The population exhibited patchy group densities throughout the estuary in all seasons. Except from distance to urban areas all variables were selected in our final model for the population\'s RSPF. Individual analysis revealed discrepancies in size and location of home ranges which lead to remarkable differences in the composition and estimates of parameters selected in the models for each individual.

A percepção e interpretação da interação de indivíduos e populações com o ambiente e a forma como tal relação condiciona sua distribuição espacial é questão-chave e recorrente em estudos ecológicos. Padrões de uso de área observados para populações emergem em ultima análise da variabilidade entre seus indivíduos em selecionar habitats e interagir com os mesmos. Este estudo teve como foco o uso de área pela população do boto-cinza, Sotalia guianensis, e sua variabilidade individual no estuário de Cananeia, localizado na costa sudeste do Brasil (25°03\' S; 47°55\' W), durante o verão e o inverno de 2015 e o verão de 2016. Parâmetros ambientais e geográficos (distâncias da desembocadura de rios, da entrada do estuário e de áreas urbanas, profundidade, maré e autocorrelação espacial) foram testados para explicar a distribuição da população e de seus indivíduos a partir de funções de probabilidade de seleção de recursos (RSPF) em modelos aditivos generalizados (GAM). Onze indivíduos fotoidentificados com 18 ou mais recapturas foram avaliados com o uso de modelos individuais de ocupação e sua interpretação foi subsidiada por estimativas de áreas domiciliares obtidas a partir de kerneis fixos de densidade. Nas três temporadas a população apresentou densidades de grupos desiguais ao longo do estuário e todas as variáveis, com exceção da distância de áreas urbanas, explicaram as probabilidades de presença observadas. Análises individuais revelaram discrepâncias nos tamanhos e disposição geográfica de áreas domiciliares e diferenças na composição e estimativa dos parâmetros selecionados para cada indivíduo. A variabilidade individual na população deve ter papel fundamental em termos de utilização do espaço e seleção de habitat pelo boto-cinza no estuário local. / Understanding and interpreting the interaction of individuals and populations with the environment and how this relationship outlines their spatial distribution is a key question common in ecological studies. Area use patterns observed for populations are ultimately an outcome from individual variability in habitat selection and their interaction with such environments. Are use and habitat selection by the population of Guiana dolphins, Sotalia guianensis, and its individual variability were accessed in the Cananeia estuary (25°03\' S; 47°55\' W), southeastern Brazil, during the summer and winter of 2015 and the summer of 2016. Environmental and geographic parameters were estimated aiming to explain population distribution and differences within individuals. For this purpose, resource selection probability functions (RSPF) were applied in generalized additive models (GAM). Covariates tested included: distance to river mouths, distance to the estuary entrance, distance to urban areas, depth and tide. Geographic coordinates were used to model spatial autocorrelation. Eleven photo-identified individuals had their occupancy modelled and accessed in relation to their home range obtained from fixed kernel densities estimates. The population exhibited patchy group densities throughout the estuary in all seasons. Except from distance to urban areas all variables were selected in our final model for the population\'s RSPF. Individual analysis revealed discrepancies in size and location of home ranges which lead to remarkable differences in the composition and estimates of parameters selected in the models for each individual.

No description available.

The execution of the scientific applications on the Cloud comes with great flexibility, scalability, cost-effectiveness, and substantial computing power. Market-leading Cloud service providers such as Amazon Web service (AWS), Azure, Google Cloud Platform (GCP) offer various general purposes, memory-intensive, and compute-intensive Cloud instances for the execution of scientific applications. The scientific community, especially small research institutions and undergraduate universities, face many hurdles while conducting high-performance computing research in the absence of large dedicated clusters. The Cloud provides a lucrative alternative to dedicated clusters, however a wide range of Cloud computing choices makes the instance selection for the end-users. This thesis aims to simplify Cloud instance selection for end-users by proposing a probabilistic machine learning framework to allow to users select a suitable Cloud instance for their scientific applications. This research builds on the previously proposed A2Cloud-RF framework that recommends high-performing Cloud instances by profiling the application and the selected Cloud instances. The framework produces a set of objective scores called the A2Cloud scores, which denote the compatibility level between the application and the selected Cloud instances. When used alone, the A2Cloud scores become increasingly unwieldy with an increasing number of tested Cloud instances. Additionally, the framework only examines the raw application performance and does not consider the execution cost to guide resource selection. To improve the usability of the framework and assist with economical instance selection, this research adds two Naïve Bayes (NB) classifiers that consider both the application’s performance and execution cost. These NB classifiers include: 1) NB with a Random Forest Classifier (RFC) and 2) a standalone NB module. Naïve Bayes with a Random Forest Classifier (RFC) augments the A2Cloud-RF framework's final instance ratings with the execution cost metric. In the training phase, the classifier builds the frequency and probability tables. The classifier recommends a Cloud instance based on the highest posterior probability for the selected application. The standalone NB classifier uses the generated A2Cloud score (an intermediate result from the A2Cloud-RF framework) and execution cost metric to construct an NB classifier. The NB classifier forms a frequency table and probability (prior and likelihood) tables. For recommending a Cloud instance for a test application, the classifier calculates the highest posterior probability for all of the Cloud instances. The classifier recommends a Cloud instance with the highest posterior probability. This study performs the execution of eight real-world applications on 20 Cloud instances from AWS, Azure, GCP, and Linode. We train the NB classifiers using 80% of this dataset and employ the remaining 20% for testing. The testing yields more than 90% recommendation accuracy for the chosen applications and Cloud instances. Because of the imbalanced nature of the dataset and multi-class nature of classification, we consider the confusion matrix (true positive, false positive, true negative, and false negative) and F1 score with above 0.9 scores to describe the model performance. The final goal of this research is to make Cloud computing an accessible resource for conducting high-performance scientific executions by enabling users to select an effective Cloud instance from across multiple providers.

La sélection des habitats est un comportement important reliant des individus aux conditions environnementales de leur habitat. Elle est généralement étudiée pour faire des inférences sur les patrons de distribution des populations. Or, la sélection des habitats peut varier entre individus d’une même population et cette variation peut excéder la variation observée entre les populations. D’une part, si la sélection des habitats est adaptative, on peut supposer que les individus sélectionneront des habitats leur permettant de maximiser leur performance. D’autre part, les conditions environnementales dans les habitats peuvent affecter les performances individuelles, impliquant ainsi que la sélection des habitats peut avoir des conséquences physiologiques. Par ailleurs, l’environnement social peut influencer la performance physiologique des individus. L’objectif général de la thèse est l’étude des déterminants et des conséquences physiologiques de la sélection des habitats chez les poissons. Dans un premier temps, nous avons créé et comparé la capacité prédictive de modèles de sélection des habitats pour l’achigan à petite bouche Micropterus dolomieu intégrant la variabilité individuelle. Nos résultats ont démontré que l’intégration de la variabilité individuelle permettait d’identifier les variables influençant la sélection des habitats au niveau individuel, des groupes et de la population. Les modèles incluant les variables représentant la présence de refuges dans les habitats avaient un meilleur pouvoir prédictif que ceux qui ne les incluaient pas. Par ailleurs, des groupements d’individus présentant des similitudes dans leur sélection d’habitats ont été identifiés. Malgré tout, la variabilité dans la sélection des habitats entre les individus était nettement plus grande que la variabilité entre les groupes. Nous avons démontré que la présence de refuge était la variable la plus importante à considérer dans les modèles de sélection d’habitats pour les achigans à petite bouche. Nous avons ensuite investigué si la présence de refuge pouvait influencer différents traits métaboliques des achigans à petite bouche grâce à des expériences de respirométrie en laboratoire. La présence de refuge a diminué les taux métaboliques au repos (RMR) des achigans provenant d’un lac alors qu’il n’y a pas eu d’effet sur les achigans provenant d’une rivière. En considérant la position hiérarchique des individus, nous avons noté que les individus dominants avaient un temps de récupération plus court en présence de refuge alors que la présence de refuge n’a rien changé pour les individus soumis. Finalement, nous avons étudié si l’environnement social, en particulier la taille du groupe social, pouvait influencer l’estimation des taux métaboliques des poissons en présence ou en absence de refuge. Nous avons cette fois mené des expériences sur des vairons Phoxinus phoxinus, des poissons très sociaux. Les vairons gardés en petits groupes avaient des taux métaboliques plus élevés que ceux gardés en grands groupes. La présence de refuge a diminué les taux métaboliques indépendamment de la taille des groupes. Nos résultats ont démontré que la taille des groupes peut influencer les dépenses énergétiques des individus, ce qui souligne l'importance de comprendre le rôle des dynamiques sociales sur les variations dans les traits métaboliques. Les résultats de la thèse démontrent l’importance de tenir compte de l’environnement physique et de l’environnement social pour mieux comprendre les conséquences physiologiques de la sélection des habitats. / Habitat selection is an important behaviour that relates individuals to the environmental conditions in their habitat, and is generally studied to infer population-level patterns of distributions. Habitat selection varies among individuals and there is growing evidence that individual differences often exceed population differences in habitat selection. On the one hand, if habitat selection is adaptive, it could be hypothesized that individuals would select habitats that would maximize their fitness. On the other hand, environmental conditions in habitats can have physiological consequences, which can be amplified or masked by the social environment. Therefore, the general objective of this thesis was to better understand the determinants and physiological consequences of habitat selection. We created and compared the predictive capacity of habitat selection models for smallmouth bass Micropterus dolomieu integrating individual variability. Our results show that by integrating individual variability, we could identify variables influencing individual-, group-, and population-level habitat selection. Models that included variables referring to presence of shelter had the best predictive capacity. Further, we identified groups of individuals defined by their habitat selection. Nevertheless, variation in habitat selection among individuals was higher than that among groups. Presence of shelter was the main correlate of habitat selection for smallmouth bass. We then we tested whether presence of shelter could influence smallmouth bass metabolic traits estimated during respirometry trials. In presence of shelter, resting metabolic rates (RMR) were lower than in absence of shelter for smallmouth bass from a lake population. There was no difference in RMR for smallmouth bass from a river population. Further, dominant individuals showed reduced recovery time (RT) in presence of shelter, while no difference was observed in subordinate individuals. We investigated how social group size and availability of shelter could influence metabolic rate. This project was conducted on Eurasian minnow Phoxinus phoxinus, a highly social fish. Fish held in smaller groups had higher standard metabolic rate as compared to that of fish held in larger groups. Presence of shelter during respirometry trials was associated with reduced metabolic rates, regardless of group size fish were held in. Our results suggest that social group size may directly influence energy demands of individuals, highlighting the importance of understanding the role of group size on variations in physiological traits associated with energy expenditure. Our results highlight the importance of considering the physical and social environment to better understand the physiological consequences of habitat selection.

Management and conservation of populations of animals requires information on where they are, why they are there, and where else they could be. These objectives are typically approached by collecting data on the animals’ use of space, relating these to prevailing environmental conditions and employing these relations to predict usage at other geographical regions. Technical advances in wildlife telemetry have accomplished manifold increases in the amount and quality of available data, creating the need for a statistical framework that can use them to make population-level inferences for habitat preference and space-use. This has been slow-in-coming because wildlife telemetry data are, by definition, spatio-temporally autocorrelated, unbalanced, presence-only observations of behaviorally complex animals, responding to a multitude of cross-correlated environmental variables. I review the evolution of techniques for the analysis of space-use and habitat preference, from simple hypothesis tests to modern modeling techniques and outline the essential features of a framework that emerges naturally from these foundations. Within this framework, I discuss eight challenges, inherent in the spatial analysis of telemetry data and, for each, I propose solutions that can work in tandem. Specifically, I propose a logistic, mixed-effects approach that uses generalized additive transformations of the environmental covariates and is fitted to a response data-set comprising the telemetry and simulated observations, under a case-control design. I apply this framework to non-trivial case-studies using data from satellite-tagged grey seals (Halichoerus grypus) foraging off the east and west coast of Scotland, and northern gannets (Morus Bassanus) from Bass Rock. I find that sea bottom depth and sediment type explain little of the variation in gannet usage, but grey seals from different regions strongly prefer coarse sediment types, the ideal burrowing habitat of sandeels, their preferred prey. The results also suggest that prey aggregation within the water column might be as important as horizontal heterogeneity. More importantly, I conclude that, despite the complex behavior of the study species, flexible empirical models can capture the environmental relationships that shape population distributions.