Generalized Additive Model

Using generalized additive models, we show evidence for nonlinear relationships between various hydrological factors and age-structured catch per unit effort of Atlantic yellowfin tuna (Thunnus albacares) for two fishing fleets. Catchability effects are distinguished from tuna environmental preference effects in the catch per unit effort variability. With respect to catchability, an important nonlinear effect of local fishing effort is highlighted for each fleet. It is interpreted as resulting from a local overfishing phenomenon of adult yellowfin tuna and from vessel fishing tactics (cooperation/spying). The environmental preferences obtained facilitate the interpretation of the hierarchical spatial distribution and age-dependent movements of the yellowfin population. We show that, on a large spatiotemporal scale (the whole ocean), low salinity is a good predictor of yellowfin habitat. Juveniles are mainly distributed in low-salinity waters (<0.035 kg·kg -1 ) when adults extend their range to waters of 0.036 kg·kg -1 . On a mesoscale, adult population annual reproductive transatlantic displacements are probably driven by temperature and salinity gradients to warm and low-salinity locations that are favorable for juveniles. North-south seasonal movements of the population are clearly related to warmwater seasonal oscillations. On a small scale, ocean thermic stability and gradients of sea surface temperature are important physical factors determining yellowfin concentration.

This paper comparatively analyses the association between urban neighborhood socioeconomic markers and ambient air pollution in Vancouver and Seattle, the two largest urban regions in the Georgia Basin -Puget Sound (GB -PS) international airshed.

Physically based models are commonly used as an integral step in landslide hazard assessment. Geomorphic principles can be applied to a broad area, resulting in first order assessment of landslide susceptibility. New techniques are now available that may result in the increased accuracy of such models. We investigate the possibility to enhance landslide susceptibility modeling by integrating two physically-based landslide models, the Factor of Safety (FS) and the Shallow Stability model (SHALSTAB), with traditional empirical-statistical methods that utilize terrain attribute information derived from a digital elevation model and land use characteristics related to forest harvesting. The model performance is measured by the area under the receiver operating characteristic curve (AUROC) and sensitivity at 90% and 80% specificity both estimated by bootstrap resampling. Our study examines 278 landslide initiation points in the Klanawa Watershed located on Vancouver Island, British Columbia, Canada. We use a generalized additive model (GAM)and a logistic regression model (GLM) combining physical landslide models, terrain attributes and land use data, and GAMs and GLMs using only subsets of these variables.In this study, all empirical and combined physical-empirical models outperformthe physically-based models, with GAMs often performing significantly better than GLMs. The strongest predictive performance is achieved by the GAMsusing terrain attributes in combination with land use data. Variables representing physically-based models do not significantly improve the empirical models, but they may allow for a better physical interpretation of empirical models. Also, based on bootstrap variable-selection frequencies,land use data, FS, slope and plan/profile curvature are relatively the most important predictor variables.

Generalized additive models are proposed for a better understanding of the underlying mechanisms for anchovy variations in abundance. Environmental variables derived from satellite imagery (surface chlorophyll, sea surface temperature and wind-mixing index), river discharge (Rhône River and Ebre River) and anchovy landings (landings per unit of effort) as proxy for abundance were used, and three fishing zones were defined along the Catalan Coast. A time shift among wind index mixing, sea surface temperature and chlorophyll was observed for these variables to be significantly correlated with anchovy. Results pointed out to processes that appear to greatly influence species abundance and affect different life stages of anchovy (conditions preceding reproduction, larvae growth and survival and recruits growth). A high proportion of anchovy LPUE variability could be explained by environmental variables. Thus, some univariate models explained deviance are more than 50%, even up to around 70% of anchovy variability. In several cases the deviance explained by a given variable was even higher at the longer time-lags. Among all univariate and bivariate models fitted, the model that best explained anchovy LPUE variability, 79% of total deviance, was a model proposed for the central zone, based on the additive effect of surface chlorophyll and Rhône River discharge, considering time lags of 15 and 18 months, respectively, for each variable.

Anchovy and round sardinella are two important small pelagic species in the Mediterranean that spawn during the summer period. This is a first attempt to model and predict the two species' potential spawning habitats in this area. Generalized additive models (GAMs) were constructed based on satellite environmental variables and presence/absence egg data, available from ichthyoplankton surveys conducted in the North Aegean Sea during early summer . These models were subsequently used to predict the probability of anchovy and round sardinella spawning in the Greek Seas as well as the entire Mediterranean and Black Sea during the same month of the year. The interaction of bottom depth and chlorophyll explained most of the deviance in the presence/absence GAMs of both species, indicating spawning over continental shelf areas with increased surface chlorophyll values. Round sardinella spawned closer to coast than anchovy. Predicted potential spawning areas for anchovy and round sardinella in unsampled areas of the Greek Seas and the entire Mediterranean and Black Sea were in good agreement with existing information on the distribution and extent of the spawning grounds, especially for anchovy. Modeling the species' reproductive activity in relation to easily accessible environmental information and applying the models in a predictive way could be an initial, low-cost step to designate potential spawning fish habitats.

We have used Landsat-5 TM and Landsat-7 ETMþ images together with simultaneous ground-truth data at sample points in the Doñana marshes to predict water turbidity and depth from band reflectance using Generalized Additive Models. We have point samples for 12 different dates simultaneous with 7 Landsat-5 and 5 Landsat-7 overpasses. The best model for water turbidity in the marsh explained 38% of variance in ground-truth data and included as predictors band 3 (630e690 nm), band 5 (1550e1750 nm) and the ratio between bands 1 (450e520 nm) and 4 (760e900 nm). Water turbidity is easier to predict for water bodies like the Guadalquivir River and artificial ponds that are deep and not affected by bottom soil reflectance and aquatic vegetation. For the latter, a simple model using band 3 reflectance explains 78.6% of the variance. Water depth is easier to predict than turbidity. The best model for water depth in the marsh explains 78% of the variance and includes as predictors band 1, band 5, the ratio between band 2 (520e600 nm) and band 4, and bottom soil reflectance in band 4 in September, when the marsh is dry. The water turbidity and water depth models have been developed in order to reconstruct historical changes in Doñana wetlands during the last 30 years using the Landsat satellite images time series.

We analyse the coupling between sea surface chlorophyll concentration (CC) and the physical environment in the Mozambique Channel (MZC) using statistical models. Seasonal and interannual patterns are studied along with the role of mesoscale dynamics on enhancement and concentration processes for phytoplankton. We use SeaWifs data for CC and two other remotely sensed data sets, TMMI for sea surface temperature (SST) and merged altimetry products for sea level anomaly and geostrophic current. Empirical Orthogonal Functions (EOF) on SSC and SST show strong seasonality and partition the MZC into three distinct sub-areas. The chlorophyll variability is mostly driven by seasonality, but more in the North (10°S–16°S) and South (24°S–30°S), and explains respectively 64% and 82% of the CC variance. In the Central part (16°S–24°S), the seasonal signal has less influence (60% variance). There, complex EOFs on Sea Level Anomaly (SLA) highlight the role of mesoscale activity (i.e. eddies and fil...

The lesser sandeel Ammodytes marinus is an important component of the North Sea ecosystem and the subject of the largest single species ®shery in this region. However, little is known about the distribution of this species outside the areas where they are ®shed. This study examines the physical characteristics of the habitat of A. marinus in an attempt to predict the distribution of this species. The characteristics and topography of sandeel habitat were described from video observations. Data on abundance, sediment characteristics and depth were collected from benthic sampling programmes around the Shetland Isles. These data were used in a general additive model framework (GAM) to examine the relative signi®cance of physical factors in in¯uencing distribution. This analysis found that A. marinus were absent from sediments with a silt/clay content of .10% and densities declined between fractions from 2 to 10%. The apparent dislike for ®ne sediments was examined further by means of sediment choice experiments. These experiments con®rmed the importance of the ®ne particle fraction in limiting distribution and indicated that sandeels would not be expected to inhabit such areas. Given the constraints of sediment requirements, densities of sandeels in benthic samples appeared to be in¯uenced by water depth. q

In 2010 the international community, under the auspices of the Convention on Biological Diversity, agreed on 20 biodiversity-related “Aichi Targets” to be achieved within a decade. We provide a comprehensive mid-term assessment of progress toward these global targets using 55 indicator data sets. We projected indicator trends to 2020 using an adaptive statistical framework that incorporated the specific properties of individual time series. On current trajectories, results suggest that despite accelerating policy and management responses to the biodiversity crisis, the impacts of these efforts are unlikely to be reflected in improved trends in the state of biodiversity by 2020. We highlight areas of societal endeavor requiring additional efforts to achieve the Aichi Targets, and provide a baseline against which to assess future progress.

Preface vii 1 Introduction 2 Overview of Supervised Learning 2.1 Introduction 9 2.2 Variable Types and Terminology 2.3 Two Simple Approaches to Prediction: Least Squares and Nearest Neighbors 2.3.1 Linear Models and Least Squares 2.3.2 Nearest-Neighbor Methods 2.3.3 From Least Squares to Nearest Neighbors 2.4 Statistical Decision Theory 2.5 Local Methods in High Dimensions 2.6 Statistical Models, Supervised Learning and Function Approximation 2.6.1 A Statistical Model for the Joint Distribution Pr(X,Y) 2.6.2 Supervised Learning 2.6.3 Function Approximation 2.7 Structured Regression Models 2.7.1 Difficulty of the Problem 2.8 Classes of Restricted Estimators 33 2.8.1 Roughness Penalty and Bayesian Methods Contents 2.8.2 Kernel Methods and Local Regression 34 2.8.3 Basis Functions and Dictionary Methods 35 2.9 Model Selection and the Bias-Variance Tradeoff Bibliographic Notes 39 Exercises 39 Linear Methods for Regression 41 3.

Summary. I discuss the production of low rank smoothers for d ≥ 1 dimensional data, which can be fitted by regression or penalized regression methods. The smoothers are constructed by a simple transformation and truncation of the basis that arises from the solution of the thin plate spline smoothing problem and are optimal in the sense that the truncation is designed to result in the minimum possible perturbation of the thin plate spline smoothing problem given the dimension of the basis used to construct the smoother. By making use of Lanczos iteration the basis change and truncation are computationally efficient. The smoothers allow the use of approximate thin plate spline models with large data sets, avoid the problems that are associated with ‘knot placement’ that usually complicate modelling with regression splines or penalized regression splines, provide a sensible way of modelling interaction terms in generalized additive models, provide low rank approximations to generalized smoothing spline models, appropriate for use with large data sets, provide a means for incorporating smooth functions of more than one variable into non-linear models and improve the computational efficiency of penalized likelihood models incorporating thin plate splines. Given that the approach produces spline-like models with a sparse basis, it also provides a natural way of incorporating unpenalized spline-like terms in linear and generalized linear models, and these can be treated just like any other model terms from the point of view of model selection, inference and diagnostics.

GAMLSS is a general framework for fitting regression type models where the distribution of the response variable does not have to belong to the exponential family and includes highly skew and kurtotic continuous and discrete distribution. GAMLSS allows all the parameters of the distribution of the response variable to be modelled as linear/non-linear or smooth functions of the explanatory variables.

The ISO 9001 (2000) define "Quality" as the ability of a set of inherent properties and characteristics of a product, system or process to fulfill requirements of customers and other interested parties. The "Plan de Calidad de Obra (PCO)" is an instrument that follows to get the necessary quality in Civil Works contracts. Graphical documentation generated during construction project elaboration or even during Civil Engineering works execution with planes, makes obligatory a meticulous treatment of this information that results transcendental in the consecution of the definitive products (projects or works) quality objective. In this work it's shown the principal graphical documentation procedures defined in the PCO attending to the edition, modification, distribution, storage, and other relevant aspects related to graphical information.

Data analysis sometimes requires the relaxation of parametric assumptions in order to gain modeling flexibility and robustness against mis-specification of the probability model. In the Bayesian context, this is accomplished by placing a prior distribution on a function space, such as the space of all probability distributions or the space of all regression functions. Unfortunately, posterior distributions ranging over function spaces are highly complex and hence sampling methods play a key role. This paper provides an introduction to a simple, yet comprehensive, set of programs for the implementation of some Bayesian nonparametric and semiparametric models in R, DPpackage. Currently, DPpackage includes models for marginal and conditional density estimation, receiver operating characteristic curve analysis, interval-censored data, binary regression data, item response data, longitudinal and clustered data using generalized linear mixed models, and regression data using generalized additive models. The package also contains functions to compute pseudo-Bayes factors for model comparison and for eliciting the precision parameter of the Dirichlet process prior, and a general purpose Metropolis sampling algorithm. To maximize computational efficiency, the actual sampling for each model is carried out using compiled C, C++ or Fortran code.

Assessment of the spatial distribution of environmental variables and of the associated uncertainty is a key issue in environmental modelling. The water content of soil plays an important role in many ecological and hydrological processes for land suitability evaluation. In this study we present a flexible procedure to interpolate soil-related variables that uses covariates to estimate the spatial trend of the variables and quantifies the uncertainty dealing with non-linear relationships. The procedure further extends approaches based on generalized additive models. The use of Gaussian simulations of the error allows the assessment of spatial uncertainty. The method was applied to available soil water capacity for three different nested extents: national, regional, and catchment. The models fitted have different significant covariates and different estimated values according to the region considered. The results suggest that the estimates from the model fitted at the appropriate extent are the most accurate. Taking into account the uncertainty of the trend, the results provided a realistic estimation of the variability and they are spatially consistent with the geomorphological patterns. Estimating the variability with the proposed procedure is useful for further environmental and land use modelling and it can be integrated with uncertainty from other variables, such as those derived from climate models.

Sophisticated statistical analyses are common in ecological research, particularly in species distribution modeling. The effects of sometimes arbitrary decisions during the modeling procedure on the final outcome are difficult to assess, and to date are largely unexplored. We conducted an analysis quantifying the contribution of uncertainty in each step during the model-building sequence to variation in model validity and climate change projection uncertainty. Our study system was the distribution of the Great Grey Shrike in the German federal state of Saxony. For each of four steps (data quality, collinearity method, model type, and variable selection), we ran three different options in a factorial experiment, leading to 81 different model approaches. Each was subjected to a fivefold cross-validation, measuring area under curve (AUC) to assess model quality. Next, we used three climate change scenarios times three precipitation realizations to project future distributions from each model, yielding 729 projections. Again, we analyzed which step introduced most variability (the four model-building steps plus the two scenario steps) into predicted species prevalences by the year 2050. Predicted prevalences ranged from a factor of 0.2 to a factor of 10 of present prevalence, with the majority of predictions between 1.1 and 4.2 (inter-quartile range). We found that model type and data quality dominated this analysis. In particular, artificial neural networks yielded low cross-validation robustness and gave very conservative climate change predictions. Generalized linear and additive models were very similar in quality and predictions, and superior to neural networks. Variations in scenarios and realizations had very little effect, due to the small spatial extent of the study region and its relatively small range of climatic conditions. We conclude that, for climate projections, model type and data quality were the most influential factors. Since comparison of model types has received good coverage in the ecological literature, effects of data quality should now come under more scrutiny.

Pollen-based climate reconstructions were performed on two high-resolution pollen marines cores from the Alboran and Aegean Seas in order to unravel the climatic variability in the coastal settings of the Mediterranean region between 15 000 and 4000 years BP (the Lateglacial, and early to mid-Holocene). The quantitative climate reconstructions for the Alboran and Aegean Sea records focus mainly on the reconstruction of the seasonality changes (temperatures and precipitation), a crucial parameter in the Mediterranean region. This study is based on a multi-method approach comprising 3 methods: the Modern Analogues Technique (MAT), the recent Non-Metric Multidimensional Scaling/Generalized Additive Model method (NMDS/GAM) and Partial Least Squares regression (PLS). The climate signal inferred from this comparative approach confirms that cold and dry conditions prevailed in the Mediterranean region during the Oldest and Younger Dryas periods, while temperate conditions prevailed during the Bølling/Allerød and the Holocene. Our records suggest a West/East gradient of decreasing precipitation across the Mediterranean region during the cooler Late-glacial and early Holocene periods, similar to present-day conditions. Winter precipitation was highest during warm intervals and lowest during cooling phases. Several short-lived cool intervals (i.e. Older Dryas, another oscillation after this one (GI-1c2), Gerzensee/Preboreal Correspondence to: I. Dormoy (isabelle.dormoy@univ-fcomte.fr) Oscillations, 8.2 ka event, Bond events) connected to the North Atlantic climate system are documented in the Alboran and Aegean Sea records indicating that the climate oscillations associated with the successive steps of the deglaciation in the North Atlantic area occurred in both the western and eastern Mediterranean regions. This observation confirms the presence of strong climatic linkages between the North Atlantic and Mediterranean regions.

The influence of synoptic-scale circulations on air quality is an area of increasing interest to air quality management in regards to future climate change. This study presents an analysis where the range of expected synoptic-scale circulation patterns over the region of Melbourne, Australia are determined and linked to regional air quality. A self-organizing map (SOM) has been applied to daily mean sea level pressure (MSLP) reanalysis to obtain twenty large-scale synoptic patterns in the Australian region. A time series of the occurrence of the synoptic archetypes was then employed within the framework of a generalized additive model (GAM) to identify links between synoptic-scale circulation and observed changes air pollutant concentrations. The GAM estimated shifts in pollutant concentration under the occurrence of each synoptic type after controlling for long-term trends, seasonality, weekly emissions, spatial variation, and temporal persistence. Results found the overall explanatory power of the synoptic archetypes in the models to be rather modest with 5.1% of the day-to-day variation in O3, 4.7% in PM10, and 7.1% in NO2 being explained. This indicates that synoptic-scale circulation features are not the primary driver of day-to-day pollutant concentrations. Nonetheless, further analysis of the partial residual plots identified that despite a modest response at the aggregate level, individual synoptic categories had differential effects on air pollutants. In particular, NO2 and O3 were 20% higher than average when synoptic conditions resulted in a northeasterly gradient wind over the Melbourne area. For PM10 maximum increases of up to 20% occurred when a strong anticyclonic system was centered directly over the Melbourne area. In sum, the unified approach of SOM and GAM provided a complementary suite of tools capable of identifying the entire range synoptic circulation patterns over a particular region and quantifying how they influence local air quality.

We had previously shown that regularization principles lead to approximation schemes which are equivalent to networks with one layer of hidden units, called Regularization Networks. In particular, standard smoothness functionals lead to a subclass of regularization networks, the well known Radial Basis Functions approximation schemes. This paper shows that regularization networks encompass a much broader range of approximation schemes, including many of the popular general additive models and some of the neural networks. In particular, we introduce new classes of smoothness functionals that lead to di erent classes of basis functions. Additive splines as well as some tensor product splines can be obtained from appropriate classes of smoothness functionals. Furthermore, the same generalization that extends Radial Basis Functions (RBF) to Hyper Basis Functions (HBF) also leads from additive models to ridge approximation models, containing as special cases Breiman's hinge functions, some forms of Projection Pursuit Regression and several types of neural networks. We propose to use the term Generalized Regularization Networks for this broad class of approximation schemes that follow from an extension of regularization. In the probabilistic interpretation of regularization, the di erent classes of basis functions correspond to di erent classes of prior probabilities on the approximating function spaces, and therefore to di erent types of smoothness assumptions.

Areas with high landslide activity and diversity were encountered in the tropical Andes of Southern Ecuador under contrasting, semi-arid to perhumid climatic conditions. The objective of this study was to determine and compare geomorphic process rates of shallow landslides along this remarkable humidity gradient and subject to different types of human-made and natural environmental changes. Geomorphic work, geomorphic power and landslide mobilization rate (LMR) were therefore calculated for shallow landslides in two study areas with two separate geological or land use-related subareas each. While landslide ages were known in the perhumid Reserva Biológica San Francisco (RBSF) area, only an approximation of the frequency of critical landslide-triggering rainfall events was available for the semi-arid Masamanaca area. Landslide volumes were estimated by volume-area scaling. Generalized additive models (GAMs) were used as landslide susceptibility models in order to analyze the relative importance of topography, and to downscale LMR values to a fine spatial resolution. LMR in the perhumid RBSF area ranged from ≈ 2 mm yr − 1 in the natural part of this area with tropical mountain rainforests to ≈5 mm yr − 1 in the human-influenced part. The semi-arid Masamanaca area, though subject to greater estimation uncertainties, displayed LMR on the order of ≈ 0.4 to 4 mm yr − 1 for shallow landslides. The results provide a basis for the spatially differentiated assessment of landscape evolution and degradation in an area with a close relation between landslide activity, natural vegetation succession and human land use.

Restoration of waterbird diversity and abundance is a key objective of river system management in Australia. Therefore, understanding the effects of climatic and hydrological variables on waterbird population dynamics is fundamental for successful river restoration programs. We investigated the population dynamics of waterbirds (total abundance) and seven functional waterbird groups in the floodplains of lower Murrumbidgee River. We found a general declining abundance trend from 1983 to 2007, except for the deep water foragers. We modelled the relative contribution of the climatic and hydrological factors to waterbird population decrease using the generalized additive model (GAM) framework after identifying the negative binomial distribution. Most of the seven functional groups were positively related to both annual rainfall and water usage, defined as the total water volume intercepted by the river reach, and the models indicated that rainfall was slightly more important. Temperature also played a role in waterbird abundance: the maximum summer temperature negatively influenced the abundance of dabbling ducks, shoreline foragers and fish eaters, while the minimum winter temperature positively affected the abundance of dabbling ducks and shoreline foragers. Overall, our results support the practice of providing environmental water for sustaining waterbird populations. However, environmental water provision is likely to be most effective when timed to coincide with antecedent rainfall.

We undertook a field study to determine whether comb cell size affects the reproductive behavior of Varroa destructor under natural conditions. We examined the effect of brood cell width on the reproductive behavior of V. destructor in honey bee colonies, under natural conditions. Drone and worker brood combs were sampled from 11 colonies of Apis mellifera. A Pearson correlation test and a Tukey test were used to determine whether mite reproduction rate varied with brood cell width. Generalized additive model analysis showed that infestation rate increased positively and linearly with the width of worker and drone cells. The reproduction rate for viable mother mites was 0.96 viable female descendants per original invading female. No significant correlation was observed between brood cell width and number of offspring of V. destructor. Infertile mother mites were more frequent in narrower brood cells.

Although common among bacteria, lateral gene transfer-the movement of genes between distantly related organisms-is thought to occur only rarely between bacteria and multicellular eukaryotes. However, the presence of endosymbionts, such as Wolbachia pipientis, within some eukaryotic germlines may facilitate bacterial gene transfers to eukaryotic host genomes. We therefore examined host genomes for evidence of gene transfer events from Wolbachia bacteria to their hosts. We found and confirmed transfers into the genomes of 4 insect and 4 nematode species that range from nearly the entire Wolbachia genome (>1 megabase) to short (<500 base pairs) insertions. Potential Wolbachia to host transfers were also detected computationally in three additional sequenced insect genomes. We also show that some of these inserted Wolbachia genes are transcribed within eukaryotic cells lacking endosymbionts. Therefore, heritable lateral gene transfer occurs into eukaryotic hosts from their prokaryote symbionts, potentially providing a mechanism for acquisition of new genes and functions.

This study examines the extent to which fish remains are affected by digestive processes and includes experimental and comparative analysis of tui chub (Cyprinidae: Gila bicolor) fish remains from modern human feeding experiments, modem coyote scats and archaeological human coprolites. Body part representation varies across the samples, but overall, bone loss resulting from digestive process is less than that documented for fish remains in previous studies. Selected elements are examined for surface modifications (pitting, rounding, adhering tissue, staining, deformation) and degree of completeness. Results show that digestive processes often modify specimens, but that many remains are unaffected. Remains that pass through human and coyote digestive tracts are similarly modified, thus distinguishing fish remains generated by the two agents is not possible using the attributes described. A sample of fish remains from western Nevada is examined for surface modification and specimen completeness to determine whether the fish ren1ains were modified by digestive agents. Results show that at least some of the fish fauna had been ingested by mammals.

Kösling, K., G. Kunter, R. H. Baayen, and I. Plag
According to the widely accepted Lexical Category Prominence Rule (LCPR), prominence assignment to triconstituent compounds depends on the branching direction. Left-branching compounds, that is, compounds with a left-hand complex constituent, are held to have highest prominence on the left-most constituent, whereas right-branching compounds have highest prominence on the second of the three constituents. The LCPR is, however, only poorly empirically supported. The present paper tests a new hypothesis concerning the prominence of triconstituent compounds and suggests a new methodology for the empirical investigation of compound prominence. According to this hypothesis, the prominence pattern of the embedded compound has a decisive influence on the prominence of the whole compound. Using a mixed-effects generalized additive model for the analysis of the pitch movements, it is shown that all triconstituent compounds have an accent on the first constituent irrespective of branching, and that the placement of a second, or even a third, accent is dependent on the prominence pattern of the embedded compound. The LCPR is wrong.

Cephalopod stocks are of increasing economic importance. Cephalopod sheries show marked inter-annual uctuations unrelated to shery landings and eOE ort. Their population dynamics, particularly recruitment, are thought to be strongly susceptible to changes in environmental conditions. This arises in part from the short life cycle, resulting in poor buOE ering of the population against changing conditions. These characteristics make traditional approaches to stock assessment and shery management inappropriate. GIS oOE ers a tool to improve understanding of spatio-temporal trends in abundance and facilitate rational management. A cephalopod shery geographical information system for Northeast Atlantic waters (CFGIS-NEA) was developed. The system covers the area from 28.0ß W to 11.0ß E, and 34.5ß N to 65.5ß N. It was designed for investigating cephalopod resource dynamics in relation to environmental variation. It is based on Unix Arc/Info, and PC ArcView, combined with the statistical software package S-PLUS and supported by a database in Microsoft Access. Environmental data (e.g. sea surface temperature and salinity, sea bottom temperature and salinity, and bathymetric data), cephalopod shery, survey and biological data, from a variety of sources, were integrated in the GIS as coverages, grids, shape les, and tables. Special functions were developed for data integration, data conversion, query, visualisation, analysis and management. User-friendly interfaces were developed allowing relatively inexperienced users to operate the system. The spatial and temporal distribution patterns of cephalopod abundance by species, the spatial and temporal relationships between cephalopod abundance and environmental factors, and the spatial and temporal patterns of cephalopod shing activity were analysed using a combination of visual (qualitative) and quantitative methods. Predictive empirical models, such as GAMs (generalized additive models), were developed for modelling cephalopod abundance using environmental variables.

This paper analyzes the impact of hedging activities of large Canadian oil and gas companies on their stock returns and firm value. Differing from the existing literature this research finds that some of these relationships are nonlinear based on the framework of nonlinear generalized additive models. The research based on this more general methodology reveals some interesting findings on oil and gas hedging activities. The large Canadian oil and gas firms are able to use hedging to protect downside risk against the unfavorable oil and gas price changes. But oil hedging appears to be more effective in protecting stock returns than gas hedging is when downside risk presents. In addition, oil and gas reserves are more likely to play a positive (negative) role when the oil and gas prices are increasing (decreasing). Finally, hedging, in particular hedging on gas, together with profitability, investment and leverage, has certain impacts on firm value.

Forestry science has a long tradition of studying the relationship between stand productivity and abiotic and biotic site characteristics, such as climate, topography, soil and vegetation. Many of the early site quality modelling studies related site index to environmental variables using basic statistical methods such as linear regression. Because most ecological variables show a typical non-linear course and a non-constant variance distribution, a large fraction of the variation remained unexplained by these linear models. More recently, the development of more advanced non-parametric and machine learning methods provided opportunities to overcome these limitations. Nevertheless, these methods also have drawbacks. Due to their increasing complexity they are not only more difficult to implement and interpret, but also more vulnerable to overfitting. Especially in a context of regionalisation, this may prove to be problematic. Although many non-parametric and machine learning methods are increasingly used in applications related to forest site quality assessment, their predictive performance has only been assessed for a limited number of methods and ecosystems.

The Peruvian anchovy or anchoveta (Engraulis ringens) forages on plankton and is a main prey for marine mammals, seabirds, fish, and fishers, and is therefore a key element of the food web in the Humboldt Current system (HCS). Here, we present results from the analysis of 21,203 anchoveta stomach contents sampled during 23 acoustic surveys over the period 1996-2003. Prey items were identified to the genus level, and the relative dietary importance of different prey was assessed by determination of their carbon content. Variability in stomach fullness was examined relative to the diel cycle, the distance from the coast, sea surface temperature, and latitude, using generalized additive models (GAMs). Whereas phytoplankton largely dominated anchoveta diets in terms of numerical abundance and comprised >99% of ingested prey items, the carbon content of prey items indicated that zooplankton was by far the most important dietary component, with euphausiids contributing 67.5% of dietary carbon followed by copepods (26.3%). Stomach fullness data showed that anchoveta feed mainly during daytime between 07h00 and 18h00, although night-time feeding also made a substantial contribution to total food consumption. Stomach fullness also varied with latitude, distance from the coast, and temperature, but with substantial variability indicating a high degree of plasticity in anchoveta feeding behaviour. The results suggest an ecological role for anchoveta that challenges current understanding of its position in the foodweb, the functioning of the HCS, and trophic models of the HCS.

Mesoscale physical features as fronts and eddies appear to play a key role in the dynamics of marine communities. In the Indian Ocean, the Mozambique Channel (MC) is a natural laboratory to investigate mesoscale eddies (100-300 km in diameter); indeed, four to seven eddies per year are know to transit through the Channel, from North to South. We studied the structuring role of the mesoscale eddies on spatial dynamics and foraging strategy of top predators using seabirds and tuna as examples. Emphasis was on the central part of the MC (16-24°S) where eddy activity is most developed. We integrated three main categories of information: (i) satellite altimetry for sea-level anomaly (SLA) and geostrophic current, remotely-sensed surface temperature (SST) and SeaWiFS data for chlorophyll concentration (CC); (ii) individual tracking of Great Frigatebirds (Fregata minor) to characterize foraging areas; and (iii) detailed catch statistics from purse-seine fisheries to describe distribution of tuna schools. Generalized Additive Models were applied to quantify the relative influence of mesoscale descriptors, SST and CC on foraging behaviour of Great Frigatebirds and location of purse-seine sets. Our results show that seabirds are more closely tied to mesoscale eddies compared to tuna. We underline the role of eddy boundaries on the response of frigatebirds and tuna. Good foraging conditions are promoted along the edge of eddies as a result of the interplay of the maturation process from cyclonic eddies and the concentration process by eddy interactions. A decrease in the number or intensity of eddies in the MC, as observed during strong El Niño events, could potentially affect the eddy-related ecosystem with putative negative repercussions on central-place foragers such as Great Frigatebirds. We discuss the importance of a better understanding of the ''eddy system", in marine conservation and tuna fisheries management in the Mozambique Channel.

Research on the climate variability during the recent decades has been immensely stimulated by the increasing manifestation of anthropogenic climate change. Such research can provide support in evaluating future climate scenarios and as such may be instrumental in extending the lead-time for adaptation. The Mediterranean region is particularly sensitive to short-term climate change due to its intermediate position between the

Hake recruitment has been examined in relation to environmental variables in two of the main reproductive areas of the central Mediterranean, the northern and central Tyrrhenian Sea. Seventeen years time series data from trawl surveys revealed high fluctuations in recruit abundance that could not be just explained by spawning biomass estimations. Generalized additive models were developed to investigate hake recruitment dynamics in the Tyrrhenian Sea in relation to spawner abundance and selected key oceanographic variables. Environmental data were explored in attempt to explain survival processes that could affect early life history stages of hake and that accounted for high fluctuations in its recruitment.

We present the results of the Air Pollution and Health: A European Approach 2 (APHEA2) project on short-term effects of ambient particles on mortality with emphasis on effect modification. We used daily measurements for particulate matter less than 10 m in aerodynamic diameter (PM 10 ) and/or black smoke from 29 European cities. We considered confounding from other pollutants as well as meteorologic and chronologic variables. We investigated several variables describing the cities' pollution, climate, population, and geography as potential effect modifiers. For the individual city analysis, generalized additive models extending Poisson regression, using a smoother to control for seasonal patterns, were applied. To provide quantitative summaries of the results and explain remaining heterogeneity, we applied second-stage regression models. The estimated increase in the daily number of deaths for all ages for a 10 g/m 3 increase in daily PM 10 or black smoke concentrations was 0.6% [95% confidence interval (CI) ϭ 0.4 -0.8%], whereas for the elderly it was slightly higher. We found important effect modification for several of the variables studied. Thus, in a city with low average NO 2 , the estimated increase in daily mortality for an increase of 10 g/m 3 in PM 10 was 0.19 (95% CI ϭ 0.00 -0.41), whereas in a city with high average NO 2 it was 0.80% (95% CI ϭ 0.67-0.93%); in a relatively cold climate the corresponding effect was 0.29% (95% CI ϭ 0.16 -0.42), whereas in a warm climate it was 0.82% (95% CI ϭ 0.69 -0.96); in a city with low standardized mortality rate it was 0.80% (95% CI ϭ 0.65-0.95%), and in one with a high rate it was 0.43% (95% CI ϭ 0.24 -0.62). Our results confirm those previously reported on the effects of ambient particles on mortality. Furthermore, they show that the heterogeneity found in the effect parameters among cities reflects real effect modification, which is explained by specific city characteristics. (EPIDEMIOLOGY 2001;12:521-531)

This paper is an attempt to develop software framework for public library based and Webenabled multilingual community information services by utilizing open standards and open source software. It uses GSDL as DL software, PERL, PHP and Java as programming environment, Unicode as character encoding standard, SeamlessUK Application Profile as metadata schema, RFC-2731 as metadata encoding standard, and an array of FLOSS based multilingual tools for designing Bengali script based user interface, processing and retrieval environment.

The Guanaco (Lama guanicoe) has suffered a progressive decline in numbers because of unregulated hunting and poaching by an assumed competition with sheep. Inadequate livestock management, including keeping sheep numbers above carrying capacity, has led to a degradation of the Patagonian steppe. Recently, interest has grown towards a reduction in sheep density and diversification of extractive activities. Guanaco populations could be potentially amenable to a number of sustainable uses. Our aim was to investigate the factors that determine guanaco distribution in southern Argentine Patagonia and to generate a predictive cartography at the regional scale. We hypothesized that guanaco distribution could be determined by primary productivity, terrain ruggedness, human disturbance and poaching, and competition with livestock. Guanaco surveys were performed from vehicles using a road survey method. To analyze the relationship between guanaco occurrence and potential predictors we built Generalized Additive Models (GAMs) using a binomial error and a logistic link. We found that guanaco occurrence increased in the less productive and remote areas, far from cities and oil camps, and decreased in regions with high sheep density. These results suggest that guanacos tend to occur where human pressure is lower. One way to promote guanaco conservation would be to highlight the economic value of guanacos under the regulations imposed by a sustainable exploitation of their populations. The predictive models developed here could be a useful tool for the implementation of conservation and management programs at the regional scale.