京都大学 / 新制・課程博士 / 博士(医科学) / 甲第23117号 / 医科博第128号 / 新制||医科||8(附属図書館) / 京都大学大学院医学研究科医科学専攻 / (主査)教授 花川 隆, 教授 中本 裕士, 教授 溝脇 尚志 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Habitat preference of the eastern fence lizard, Sceloporus undulatus, was investigated in southwestern Virginia. Habitat features were measured at 158 lizard-centered plots and at paired random plots. Landscape-level variables, southerly aspect and mixed forest type, distinguished lizard-centered from random sites. Hatchlings were associated with relatively high temperature at perch height (23 °C), relatively high amounts (per 1- m2) of coarse woody debris (15%) and bare ground (15%), and relatively low amount of litter (34%). Adults and juveniles were associated with a relatively high number of rocks (22 per 0.01 hectare) and amount of coarse woody debris (9% per 1- m2). Habitat preferences were modeled with a Geographic Information System (GIS) using landscape-level variables and with logistic regression and Akaike's Information Criterion using site-level variables. The best-fitting site-level model for adults/juveniles included % CWD. The best-fitting model for hatchlings included % CWD and number of rocks, and the second best-fitting model also included % litter. Landscape (both classes) and site-level models (adult/juveniles only) were tested at 15 GIS-predicted "suitable" study areas and at 15 GIS-predicted "unsuitable" areas. Site-level models for hatchlings were tested with independent data collected at two study areas. Sixteen lizards were found at "suitable" areas and one at an "unsuitable" area; the GIS-based model was a good predictor of lizard presence at the landscape level. The best-fitting site-level models for adults/juveniles and hatchlings were poor predictors of lizard presence while the second best-fitting hatchling model was a good predictor of hatchling presence. / Master of Science

La capacité d’estimer les attributs de la fibre de bois (AFB) à l’échelle locale améliorerait considérablement l’information fournie par l’inventaire forestier actuel. Afin de mieux comprendre les facteurs d’influence sur les AFB, nous avons exploré la possibilité d’utiliser des estimés par télédétection de la structure forestière à l’échelle du groupement d’arbres et vérifier leur relation avec quatre AFB significatifs pour l’industrie : la densité et longueur de fibre, l’angle des microfibrilles de même que la rugosité. Notre étude a permis le développement de modèles prédictifs de ces quatre attributs en utilisant des métriques de structure en provenance de peuplements d’épinette noire (Picea mariana) et de sapin baumier (Abies Balsamea). Les métriques sélectionnées permettent de décrire quatre aspects structuraux de la forêt : la structure de canopée, la compétition dans un groupement d’arbres, la densité de végétation et la topographie locale. Des données LiDAR terrestre (Light Detection and Ranging) ont été acquises sur 53 sites répartis sur l’île de Terre-Neuve à l’aide du système Zoller+Fröhlich Imager 5006i et représentent la source d’information primaire pour l’extraction de nos métriques. Des échantillons par carottage ont été prélevés sur 10 arbres jugés représentatifs par site visité. Leur analyse par le processus SilviScan a permis de relever en détail l’état de leurs AFB. L’utilisation du critère d’information Akaike (AIC) et l’inférence multimodèle a permis d’identifier des modèles prédictifs ayant des coefficients de détermination allant de 63% à 72% dans le cas de l’épinette noire. Les métriques de structure traduisant l’aspect de compétition ont été identifiées comme étant d’importants prédicteurs. La capacité de prédiction des AFB pour le sapin baumier a toutefois été démontrée moins considérable avec des coefficients de détermination allant de 37% à 63% selon les attributs. Pour cette espèce, les métriques traduisant la structure de la canopée se sont avérées les plus importantes. Nos résultats démontrent l’importance de la structure forestière comme indicateur important de l’état des AFB et qu’ils peuvent servir dans la mise en place de modèles prédictifs pour estimer la distribution des AFB à l’échelle régionale.

Given a set of potential explanatory variables, one model selection approach is to select the best model, according to some criterion, from among the collection of models defined by all possible subsets of the explanatory variables. A popular procedure that has been used in this setting is to select the model that results in the smallest value of the Akaike information criterion (AIC). One drawback in using the AIC is that it can lead to the frequent selection of overspecified models. This can be problematic if the researcher wishes to assert, with some level of certainty, the necessity of any given variable that has been selected. This thesis develops a model selection procedure that allows the researcher to nominate, a priori, the probability at which overspecified models will be selected from among all possible subsets. The procedure seeks to determine if the inclusion of each candidate variable results in a sufficiently improved fitting term, and hence is referred to as the SIFT procedure. In order to determine whether there is sufficient evidence to retain a candidate variable or not, a set of threshold values are computed. Two procedures are proposed: a naive method based on a set of restrictive assumptions; and an empirical permutation-based method. Graphical tools have also been developed to be used in conjunction with the SIFT procedure. The graphical representation of the SIFT procedure clarifies the process being undertaken. Using these tools can also assist researchers in developing a deeper understanding of the data they are analyzing. The naive and empirical SIFT methods are investigated by way of simulation under a range of conditions within the standard linear model framework. The performance of the SIFT methodology is compared with model selection by minimum AIC; minimum Bayesian Information Criterion (BIC); and backward elimination based on p-values. The SIFT procedure is found to behave as designed—asymptotically selecting those variables that characterize the underlying data generating mechanism, while limiting the selection of false or spurious variables to the desired level. The SIFT methodology offers researchers a promising new approach to model selection, whereby they are now able to control the probability of selecting an overspecified model to a level that best suits their needs.

The spring reverse migration of landbirds (i.e., birds flying in the direction opposite to be expected), within the Great Lakes region is controversial because it is not understood if the extent or reversal of flight direction in spring is truly a change in migratory patterns or a brief anomaly. It is also not clear what the fitness and community level impacts are. My objective was to determine what and how weather influences reverse migration and to determine which species and families of birds participate the most frequently in this form of flight. I examined species which are participatory (and those that are not) as well as the impacts of specific weather covariates on the abundance of landbirds and focused explicitly on the putative reverse migration of landbirds. Field sites were located at the extreme southern tip of Fish Point, Pelee Island (2010-2012) and Point Pelee National Park (2012), where my field assistant and I visually recorded the total number of birds observed to be reverse migrating, while identifying all birds to species or family as best possible. This study was conducted over 97 days during April 26 – May 20, in 2010-2012. Information pertaining to potential reverse migration has only been formally documented twice in the Great Lakes region, most recently in 1951. I undertook a descriptive analysis to compare the numbers of individuals of bird species and families. Temperate and neotropical migrants were examined, compared, and divided into sub-sets based on their geographic ranges. I identified species at risk and vagrants which I observed during reverse migrations. Based on provincial population estimates, I determined the proportion of all reverse migrants where ≥200 individuals were observed. A descriptive analysis was undertaken to determine differences between sites (i.e., Point Pelee and Fish Point) in the final year of surveys (2012). Species and abundance were comparatively differentiated between each site and subsequently compared. While very few birds were observed flying anything but south, a total of 61,677 birds of 80 species was documented flying south. My results indicate temperate migrants vastly outnumbered neotropical migrants (as much as 4:1) and numbers of birds varied between study sites. Temperate migrants were noted to be more common (in the final study year) at Point Pelee compared to Fish Point, while neotropical migrants were more numerous at Fish Point than Point Pelee. Despite the fact that most migrant species participated in reverse migrations (i.e., of the species regularly occurring in the Pelee region at this time of year), complete absences were noted, most notably in Catharus thrushes, while species such as Rose-breasted Grosbeak and Scarlet Tanager, and families such as Tyrant Flycatchers, Vireos and Sparrows were observed to be less prevalent than anticipated. Species at risk and vagrants were noted relatively frequently during this study, suggesting that these surveys are an efficient and potentially important tool for migration monitoring in this region. Diurnal migrants, most notably Blackbirds, were observed to engage in reverse migration in higher numbers than nocturnal migrants, such as Wood Warblers. Seven weather covariates were measured and modeled with the total number of birds detected using R to determine which covariates explain the most amount of variation of the total number of birds during my surveys. I used an AICc approach to select the best model for each hypothesis. After selecting the top weather covariates with time lags according to the best (lowest) AICc values, I built general models by comparing all possible combinations of the covariates identified in the top models for each hypothesis. I included a random effect intercept for study site to discern any site difference or similarities between Point Pelee and Fish Point and specified a Poisson distribution (log-link function as implemented in the LMER package) because the data set was continuous (time-series) and count oriented. My adjusted time lag results show that most migrants tend to migrate during and ahead of inclement weather and/or periods of south winds. I also found that all identified covariates influence reverse migration to some degree. Wind direction and barometric pressure were the most significant of the covariates examined (β = 0.718 and -0.213, respectively). Specifically, wind direction is the most important covariate in explaining reverse migration, with days of south winds dramatically increasing the probability of higher numbers of birds during surveys. Low barometric pressure is also important for explaining the number of observed reverse migrants; therefore, days with lower barometric pressure have a greater likelihood of increased bird observations. Based on my observations and results I theorized that while reverse migration pertains to a distinct form of flight, it is likely not an actual form of migration. This form of flight at its simplest is likely a form of reorientation, whereby migratory birds take advantage of local weather conditions by flying south for extended distances. I anticipate that this form of flight must have serious consequences for the fitness levels and life-cycles of migratory birds. Studies looking at reverse migration provide a useful tool for migration monitoring, particularly as it is an underexplored phenomenon. Observations of thousands of birds, many of which are either species at risk or vagrants, collected in an efficient manner are vital for determining population trends related to migratory birds. Continuing this study would aid on-going monitoring programs assessing bird populations passing through the lower Great Lakes region. These studies will also help us understand the impacts of climate and climate change on migratory birds.

In educational research, students often exist in a multilevel social setting that can be identified by students within classrooms, classrooms nested in schools, schools nested in school districts, school districts nested in school counties, and school counties nested in states. These are considered hierarchical, nested, or multilevel because students are within the same community and share similar experiences which have the potential to influence an outcome. Because students within the same classrooms have similar characteristics, conclusions made on these students cannot be independent. To adapt to the hierarchical, multilevel, or nested data structure, multilevel analysis techniques such as hierarchical linear modeling (HLM) can be used to analyze the data. One purpose of HLM is to specify a model that includes appropriate random effects (Guo, 2005). One method which should be considered for inclusion or exclusion of random effects and to evaluate the goodness of fit of the final model to the data is the comparison of models with different specifications of random effects based on Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) , or Deviance Information Criterion (DIC) which corrects for bias induced by sample size and number of random effects. AIC, BIC, and DIC are information criteria that measure the statistical fit of a model. There has not been any research conducted in the multilevel literature about the impact of sample size and information criteria. This Monte Carlo Monte Carlo simulation compared the influence of sample size on the ability to select the best model in two-level hierarchical models using the information criteria Akaike Information Criterion, Bayesian Information Criterion, and Deviance Information Criterion. Results of this investigation showed that all three information criteria had very low or nonexistent success in choosing the best hierarchical linear model.

The spring reverse migration of landbirds (i.e., birds flying in the direction opposite to be expected), within the Great Lakes region is controversial because it is not understood if the extent or reversal of flight direction in spring is truly a change in migratory patterns or a brief anomaly. It is also not clear what the fitness and community level impacts are. My objective was to determine what and how weather influences reverse migration and to determine which species and families of birds participate the most frequently in this form of flight. I examined species which are participatory (and those that are not) as well as the impacts of specific weather covariates on the abundance of landbirds and focused explicitly on the putative reverse migration of landbirds. Field sites were located at the extreme southern tip of Fish Point, Pelee Island (2010-2012) and Point Pelee National Park (2012), where my field assistant and I visually recorded the total number of birds observed to be reverse migrating, while identifying all birds to species or family as best possible. This study was conducted over 97 days during April 26 – May 20, in 2010-2012. Information pertaining to potential reverse migration has only been formally documented twice in the Great Lakes region, most recently in 1951. I undertook a descriptive analysis to compare the numbers of individuals of bird species and families. Temperate and neotropical migrants were examined, compared, and divided into sub-sets based on their geographic ranges. I identified species at risk and vagrants which I observed during reverse migrations. Based on provincial population estimates, I determined the proportion of all reverse migrants where ≥200 individuals were observed. A descriptive analysis was undertaken to determine differences between sites (i.e., Point Pelee and Fish Point) in the final year of surveys (2012). Species and abundance were comparatively differentiated between each site and subsequently compared. While very few birds were observed flying anything but south, a total of 61,677 birds of 80 species was documented flying south. My results indicate temperate migrants vastly outnumbered neotropical migrants (as much as 4:1) and numbers of birds varied between study sites. Temperate migrants were noted to be more common (in the final study year) at Point Pelee compared to Fish Point, while neotropical migrants were more numerous at Fish Point than Point Pelee. Despite the fact that most migrant species participated in reverse migrations (i.e., of the species regularly occurring in the Pelee region at this time of year), complete absences were noted, most notably in Catharus thrushes, while species such as Rose-breasted Grosbeak and Scarlet Tanager, and families such as Tyrant Flycatchers, Vireos and Sparrows were observed to be less prevalent than anticipated. Species at risk and vagrants were noted relatively frequently during this study, suggesting that these surveys are an efficient and potentially important tool for migration monitoring in this region. Diurnal migrants, most notably Blackbirds, were observed to engage in reverse migration in higher numbers than nocturnal migrants, such as Wood Warblers. Seven weather covariates were measured and modeled with the total number of birds detected using R to determine which covariates explain the most amount of variation of the total number of birds during my surveys. I used an AICc approach to select the best model for each hypothesis. After selecting the top weather covariates with time lags according to the best (lowest) AICc values, I built general models by comparing all possible combinations of the covariates identified in the top models for each hypothesis. I included a random effect intercept for study site to discern any site difference or similarities between Point Pelee and Fish Point and specified a Poisson distribution (log-link function as implemented in the LMER package) because the data set was continuous (time-series) and count oriented. My adjusted time lag results show that most migrants tend to migrate during and ahead of inclement weather and/or periods of south winds. I also found that all identified covariates influence reverse migration to some degree. Wind direction and barometric pressure were the most significant of the covariates examined (β = 0.718 and -0.213, respectively). Specifically, wind direction is the most important covariate in explaining reverse migration, with days of south winds dramatically increasing the probability of higher numbers of birds during surveys. Low barometric pressure is also important for explaining the number of observed reverse migrants; therefore, days with lower barometric pressure have a greater likelihood of increased bird observations. Based on my observations and results I theorized that while reverse migration pertains to a distinct form of flight, it is likely not an actual form of migration. This form of flight at its simplest is likely a form of reorientation, whereby migratory birds take advantage of local weather conditions by flying south for extended distances. I anticipate that this form of flight must have serious consequences for the fitness levels and life-cycles of migratory birds. Studies looking at reverse migration provide a useful tool for migration monitoring, particularly as it is an underexplored phenomenon. Observations of thousands of birds, many of which are either species at risk or vagrants, collected in an efficient manner are vital for determining population trends related to migratory birds. Continuing this study would aid on-going monitoring programs assessing bird populations passing through the lower Great Lakes region. These studies will also help us understand the impacts of climate and climate change on migratory birds.

A Linear mixed-effects model (LME) is one of the possible tools for longitudinal or group--dependent data. This thesis deals with evaluating of prediction error in LME. Firstly, it is derived the mean square error of prediction (MSEP) by direct calculation. Then the covariance penalty method and crossvalidation is presented for evaluation of MSEP in LME. Further, it is shown how Akaike information criterion (AIC) can be used in mixed-effects models. Because of the model's properties two types of AIC are distinguished - marginal and conditional one. Subsequently, the procedures of AIC's calculation and its basic asymptotic properties are described. Finally, the thesis contains simulation study of behaviour of marginal and conditional AIC with the goal to choose the right variance structure of random effects. It turns out that the marginal criterion tends to select models with smaller number of random effects than conditional criterion.

FMRI datasets allow scientists to assess functionality of the brain by measuring the response of blood flow to a stimulus. Since the responses from neighboring locations within the brain are correlated, simple linear models that assume independence of measurements across locations are inadequate. Mixed models can be used to model the spatial correlation between observations, however selecting the correct covariance structure is difficult. Information criteria, such as AIC are often used to choose among covariance structures. Once the covariance structure is selected, significance tests can be used to determine if a region of interest within the brain is significantly active. Through the use of simulations, this project explores the performance of AIC in selecting the covariance structure. Type I error rates are presented for the fixed effects using the the AIC chosen covariance structure. Power of the fixed effects are also discussed.

Le travail soumis commence par un aperçu du micro-crédit tel qu’il a été introduit au Bangladesh par M. Yunus. Puis on donne un modèle stochastique des retards de versement. Comme ces retards ne donnent pas lieu à une sanction financière, ils constituent, de fait, une baisse du taux réel de crédit. Ce taux est alors, lui-même, aléatoire. On calcule un taux espéré en fonction de la probabilité de retard de remboursement hebdomadaire. On déduit que ce taux espéré est d’environ 3.5% inférieur au taux (annoncé) du cas déterministe si l’on considère que 3% des retards atteignent 4 semaines. Le travail se poursuit par une étude statistique de données du micro-crédit en Thaïlande. On commence par présenter un modèle de régression logistique du taux de remboursement par rapport aux 23 variables mesurées sur un échantillon de 219 groupes d’emprunteurs. On présente ensuite une sélection des variables les plus pertinentes selon un critère AIC ou BIC par une méthode “backward stepwise”. Finalement des expériences sur des sous-échantillons montrent une bonne stabilité du choix des variables obtenues par la sélection. / This study is inspired from a real scenario of microcredit lending introduced in Bangladesh by Yunus. A stochastic model of random delays in repayment installments is then constructed. Since delays occur without financial penalty, the interest rate is obviously lower than the exact claimed. This rate then becomes a random variable corresponding to the random repayment time, in which simulation results of its distribution are provided. The expected rate is computed as a function of in-time installment probability. It is found around 3.5% lower than the exact one in the deterministic case when considering 3% of delay occurred within four weeks in real practice. The work is extended to a statistical analysis on data of microcredit in Thailand. It is started by presenting a logistic regression model of repayment outcome containing 23 input variables measured on a sample of 219 lending groups. Applying penalized criterion, AIC or BIC together with backward stepwise elimination procedure on the full model, a more parsimonious model kept only most relevant predictors is obtained. Finally, experiments on sub-samples show a stability of the chosen predictors obtained by the selection method.