Diel vertical migration (DVM) can offer motile phytoplankton a competitive advantage over other phytoplankton species. DVM has been well documented in the harmful dinoflagellate, Karenia brevis; however, the extent to which changes in nutrient regimes and density stratification modify migration patterns has not been thoroughly investigated. A 2 m experimental column was constructed with sensors and sampling ports at four depths to obtain fluorescence and temperature measurements and discrete water samples. The environmental parameters of the experimental column represented temperature and nutrient conditions found on the West Florida Shelf (WFS), where Karenia blooms frequently occur. Results demonstrated a clear DVM pattern for K. brevis where surface aggregations occurred during light periods, and bottom aggregations occurred in dark periods. However, the rate and intensity of migration varied between experimental conditions. Injections of nutrient replete water at the surface and bottom did not appear to modify migratory behavior, in that a portion of the population resided at depth regardless of light or dark periods. Weak density stratification caused a delay in downward migration in dark periods, whereas strong density stratification caused a complete cessation of downward migration. In a weakly stratified environment with a bottom injected replete water mass, DVM patterns most closely replicate the pattern under weakly stratified conditions alone. Modifications in the natural DVM pattern of K. brevis with respect to varying environmental and nutrient conditions that occur on the WFS can have major implications current forecast model predictions, and for monitoring and mitigation strategies.

ABSTRACT In the first part of this study, a determinist mathematical approach is used to describe any kind of pollutant migration in groundwater. This theoretical background is focused on the miscible displacement and the particularities of the multiporous media are discussed. Subsequently, an objective numerical tool is developed to solve the convection-dispersion equations including immobile water effect, degradation, and adsoption. Among all the available techniques, two finite element methods in fixed meshing grids have been programmed: -the F.U.P.G. method (Full Upwind Petrov Galerkin), using a space-time upwinded weighting function with optimized coefficients, -the H.E.L.M. method (Hybrid Eulerian Lagrangian Method), using the eulerian lagrangian approach with reverse node tracking. Those two schemes are tested on a large number of reference problems. The model have been applied to study the behaviour of solutes (nitrates mainly) in the cretaceous chalk of the Hesbaye area (Belgium). Experimentations have been performed on domains of increasing size (laboratory tests, in situ tracer tests). For each interpretation, the particularities of the context have been taken into account, and miscible transport coefficients have been objectively determined. Hence, the medium is well characterized and the scale effect is quantified. This leads to previsional applications. RESUME Ce travail débute par le développement dun formalisme mathématique déterministe pour décrire, en toute généralité, la propagation de polluants dans les eaux souterraines. Cette étude théorique permet de situer le problème posé (pollution miscible diluée) dans un cadre plus large, et de souligner les particularités dun milieu à porosité multiple. Dans un second temps, un outil numérique objectif est mis au point pour résoudre les équations de convection-dispersion avec effet deau immobile, dégradation, et adsorption. Parmi le grand nombre des procédés existants, deux méthodes par éléments finis en maillage fixe ont été programmées : -la méthode F.U.P.G. (Full Upwind Petrov Galerkin) basée sur un décentrage des fonctions de pondération, optimum dans le temps et lespace, -la méthode H.E.L.M. (Hybrid Eulerian Lagrangian Method) utilisant un processus eulerien lagrangian avec traçage inverse des positions nodales. Les deux schémas sont testés sur de nombreux problèmes de référence. Ensuite, ce modèle est appliqué à des situations pratiques pour étudier le comportement de solutés (nitrates notamment) dans laquifère crayeux du Crétacé de Hesbaye (Belgique). Des domaines de taille croissante sont étudiés (essais de laboratoire, traçage in situ). A chaque étape, les coefficients de transport miscible sont déterminés de façon objective, en tenant compte de la spécificité des tests. Ainsi, leffet déchelle peut être quantifié et il est possible denvisager des scénarios prévisionnels.