Search

The identification of relevant spatial scales in organism-environment relationships is a key step in the understanding of an ecosystem. To study this concern, the hierarchical patch dynamic theory offers a unique concept to relate the ecological processes, patterns and scales. Based on this theory, we approach the identification of relevant spatial scale in organism-environment relationships, with the intention to explicitly account for the spatial component of ecological processes. From the comparison between the PCNM method and the geostatistical approach applied to spruce's defoliation caused by spruce budworm in Ontario (Canada), the geostatistical approach is more efficient to identify relevant spatial scale inorganism-environment relationships. Even so, to answer the question of significance of identified scales, we developed a Monte-Carlo test for nested variogram models. However classical geostatistics could be difficult to use with ecological data which have positive, skewed distribution. To consider this problem, we developed a hierarchical model associated to geostatistics and applied it to modelling spatial distribution of auks (Uria spp) in the Bay of Biscay. Tree levels of patches were characterized: a very broad scale patch (200 km), broad scale patches (50 km) and fine scale patches (10 km). The spatio-temporal analysis of links between auk distribution and the oceanographic landscape discriminates between the "process variables" which feature a stable link in term of sign and correlation among time and "circumstance variables" which have unstable link. At very broad scale, the surface salinity, the mixed layer depth and the chlorophyll a, could be view as "process variables" which have been used to define the potential habitat of auks during the wintering season. At broad scale, only chlorophyll a is selected as a process variable, giving a path to model preferential habitats. These two kinds of habitats can then be connected to different levels of a hierarchical patch dynamic system., L'identification des échelles spatiales pertinentes dans l'étude des interactions espèces-environnement est indispensable pour avoir une meilleure compréhension du fonctionnement des écosystèmes. Pour aborder cette problématique la théorie des patchs hiérarchiques offre un principe unificateur permettant de relier les concepts de processus écologiques, de pattern et d'échelles. La mise en évidence des différentes échelles destructuration des organismes et des liens avec leur environnement a été abordée en se basant sur cette théorie et avec la volonté de prendre en compte de manière explicite la composante spatiale du processus étudié. A partir de la comparaison entre la PCNM et les géostatistiques, appliquées aux défoliations causées par une chenille tordeuse de bourgeons sur les épineux de l'Ontario (Canada), les géostatistiques sont apparues plus efficaces dans l'identification des relations espèces-environnement à différentes échelles. Restait la question de la significativité des échelles identifiées, pour cela un test par simulation Monte-Carlo a été développé pour les modèles de variogrammes emboîtés. Cependant, les géostatistiques classiques peuvent être difficile à utiliser avec des données écologiques présentant des distributions discrètes et fortement dissymétriques. Pour palier ces difficultés, un modèle hiérarchique couplé aux géostatistiques a été développé et appliqué à la modélisation de la distribution des guillemots (Uria spp) dans le golfe de Gascogne. Trois niveaux de structuration spatiale ont été caractérisés: un pattern à trèslarge échelle (200 km), un pattern à moyenne échelle (50 km) et un pattern à fine échelle (10 km). L'analyse spatio-temporelle des liens nous a permis de distinguer deux types de variables : les variables de processus pour lesquelles le lien est stable en terme de corrélation et de signe au cours du temps ; et les variables de circonstances pour lesquelles le lien n'est pas stable. A très large échelle, nous avons identifié trois variables de processus : la salinité de surface, la profondeur de la couche de mélange et la chlorophylle a, qui nous permettent de modéliser l'habitatpotentiel des guillemots. A moyenne échelle, seule la chlorophylle a est reconnue comme une variable de processus, donnant une piste pour modéliser les habitats préférentiels. Ces deux types d'habitats peuvent être reliés aux niveaux d'un système en patchs hiérarchiques dynamique.

International audience; Knowledge of population sizes in delimited spatial regions is crucial for most ecological research. Data from population surveys are collected with strip, line, or point transects sampling. These data are positively skewed and spatially autocorrelated, which makes estimation of uncertainty in the population size difficult. Thus, we propose a novel spatial-based estimator from a hierarchical spatial model for count data where the inhomogeneous animal density is decomposed into a deterministic trend related to potential habitat and a stationary latent field modeled by geostatistics. An empirical estimate of the latent variable is obtained including corrective terms for non-stationarity and variance resulting from a Poisson distribution of sightings. A novel block kriging estimator that takes into account both non-stationarity and the nature of count data is derived to obtain a spatial estimate of animal total abundance and variance of errors of prediction. From spatial simulated count data and real count data of common guillemot wintering in the Bay of Biscay (France), we compare mean population size and variance estimates obtained from our model-based approach to the design-based estimator (i.e., block bootstrap). The novel Poisson block kriging estimates greatly reduces uncertainty of population size estimates while block bootstrap provides larger uncertainties.

International audience; Understanding the relationships between communities of species and the environment in which they live is a challenge in the face of an increasingly variable environment. Communities of species can be composed of a remarkably large number of species and their environment can be described by multiple variables. The numerous variables composing ecological data make their analysis complex. Ecologists need to have access to statistical tools that can help them to disentangle the different patterns present within their data. Non-parametric multivariate analyses provide the theoretical, statistical basis for exploring complex relationships between multiple variables. For ecologists, the large number of multivariate methods available renders choosing the appropriate method to be difficult. After the selection of the method, arrives the question of its implementation, and again the large variety of available software does not make the task easier. Here, comes Numerical Ecology with R. In reading Numerical Ecology with R, at the beginning, it felt like a pleasant walk along a quiet river in a world of black and white. Then, feeling that I had not experienced the full color, I decided to download the R scripts http:// www.bio.umontreal.ca/ numecolR/ and use them. My reading became more active, and it was no longer simply a book. I had pleasure in discovering new way of programming different kinds of multivariate analysis by implementing the exercises of the Code It Yourself Corner. In addition to providing R code for performing a wide selection of multivariate analyses, the authors tasked us to question ourselves on the results we obtained. As a result, the authors bring the reader down the toughest paths, walking became a bit more difficult, the slope a little steeper, but I managed to reach the top without much trouble.

International audience; We tested the importance of thermal adaptations and energy efficiency in relation to the geographical distribution of two competing freshwater salmonid fish species. Presence-absence data for Arctic char and brown trout were obtained from 1502 Norwegian lakes embracing both temperature and productivity gradients. The distributions were contrasted with laboratory-derived temperature scaling models for food consumption, growth and energy efficiency. Thermal performances of the two species were almost identical. However, Arctic char exhibited double the growth efficiency (per unit of food) and appear to have out-competed brown trout from cold, low-productivity lakes, perhaps by scramble competition. Brown trout, for which previous reports have shown to be aggressive and dominant, have likely excluded the more energy-efficient Arctic char from relatively warm, productive lakes, perhaps by contest competition. Competitive interaction changing in outcome with lake productivity, rather than thermal performance, is likely a major determinant of the range distribution of the two species. Our study highlights the need for more focus on choice of relevant ecophysiological traits in ecological climate impact studies and species distribution modelling.

International symposium on “Climate change effects on fish and fisheries: Forecasting impacts, assessing ecosystem responses and evaluating management strategies” from April 26–29, 2010, in Sendai, Japan.