English   español  
Please use this identifier to cite or link to this item: http://hdl.handle.net/10261/96657
logo share SHARE   Add this article to your Mendeley library MendeleyBASE
Visualizar otros formatos: MARC | Dublin Core | RDF | ORE | MODS | METS | DIDL
Exportar a otros formatos:

Artificial neural network approach to population dynamics of Harmful Algal Blooms in Alfacs Bay (NW Mediterranean): Case studies of Karlodinium and Pseudo-nitzschia

AuthorsGuallar, C. ; Fernández-Tejedor, Margarita; Delgado, Maximino ; Diogène, Jorge
Issue Date29-Nov-2013
CitationIntegrating New Advances in Mediterranean Oceanography and Marine Biology. Meeting program: 47 (2013)
AbstractThe dinoflagellate Karlodinium and the diatom Pseudo-nitzschia are bloom-forming genera frequently identified in Alfacs Bay. In the literature, both microalgae genera are associated with toxic events. Therefore, understanding their population dynamics and the possibility to predict their abundance is crucial for an optimal management of toxic events to the local shellfish production. Artificial neural networks have been successfully used to model the complex nonlinear dynamics of phytoplankton. In this study, this approach was applied to model Karlodinium and Pseudo nitzschia population dynamics in Alfacs Bay (NW Mediterranean). Presence-absence and prediction models were developed for both microalgae. Furthermore, with the application of Neural Interpretation Diagram (NID) and Connection Weight Approach (CWA) methodologies, ecological information was extracted from this approach, usually considered as a black box procedure. The dataset used for the model development was a long-term (1990-present) data series of environmental and phytoplankton variables from different monitoring stations established in Alfacs Bay (Ebro delta), meteorological data and Ebro River flow rates. Feed forward neural networks with momentum and flat spot elimination algorithm was used. The similarity of the different time series of each microalgae allowed performing a unique model for the whole bay for each microalgae. The variables chosen for the model development were obtained from the combination of prior knowledge of microalgae dynamics and statistical analysis of the dataset. The accuracy of the models achieved was high (Misclassification error = 6.0% and 7.7% and r2 = 0.83 and 0.71 for Karlodinium and Pseudo-nitzschia, respectively). The possibility to develop a unique model per phytoplankton species in Alfacs Bay denoted a similar microalgae dynamics in the whole ecosystem. The size of the neural nets displayed by NID showed a complex relationships between environmental and phytoplankton variables. According to the CWA results, environmental variables had stronger influence on the prediction models. These results highlighted a complex ecosystem in Alfacs Bay involving anthropogenic, climatic and hydrologic factors forcing phytoplankton dynamics
DescriptionSymposium on Integrating New Advances in Mediterranean Oceanography and Marine Biology, 26-29 November 2013, Institut de Ciències del Mar (CSIC), Barcelona, Catalunya, Spain
Publisher version (URL)http://www.icm.csic.es/bio/medocean/information.htm#schedule
Appears in Collections:(ICM) Comunicaciones congresos
Files in This Item:
There are no files associated with this item.
Show full item record
Review this work

WARNING: Items in Digital.CSIC are protected by copyright, with all rights reserved, unless otherwise indicated.