Research Index
Consortium for Estuarine Ecoindicator Research for the Gulf of Mexico
|
Data Analysis Core
![[Dr. Chet Rakocinski]](../images/rakocinskic.jpg)
Dr. Chet Rakocinski chet.rakocinski@usm.edu
The University of Southern Mississippi Gulf Coast Research Laboratory, Ocean Springs, MS
Four primary steps required to achieve integration of the ecoindicator data include: (1) Relating macrobenthic indicators and other indicators provided such as primary production, microbial diversity, macrobenthic production, and nutrient metabolism; (2) Establishing directional linkages between population, community, and ecosystem levels; (3) Developing a conceptual framework for understanding concerted changes in ecoindicators at multiple scales, and how those changes reflect shifts in ecosystem integrity, function, and resilience; and (4) Extrapolating ecoindicators to the ecosystem scale through GIS and Remote Sensing.
Neural Networks: NN can be used to identify the simplest possible associations among ecoindicators in a way that incorporates inherent non-linear relationships among such the variables.
Remote Sensing and GIS: RS and GIS can be used both to acquire process-ecoindicator information (such as chlorophyll) on the ecosystem level, to spatially integrate multiple sources of information, and to extrapolate ecoindicator information over larger spatial scales.
Dr. Rakocinski will serve as Data Manager for the Data Analysis Core.
![[Dr. Kenneth Rose]](../images/rosek.jpg)
Dr. Kenneth Rose karose@lsu.edu
Louisiana State University Coastal Fisheries Institution, Baton Rouge, LA
Keywords: mathematical and simulation modeling; fish population dynamics; food web theory; natural resource management; quantitative ecology
Dr. Rose will use the population recruitment metrics of fecundity, fertility, hatching success and survival measured in these experiments, and in the field for population modeling studies.
Dr. Peter Noble panoble@washington.edu
University of Washington, Seattle, WA
Keywords: integration; neural networks; sensitivity analysis; multivariate analysis; ecosystem dynamics and structure; complexity and dimensionality analysis; biofilms
Dr. Noble will establish a web-driven database for integration and distribution of project data. The database will be constructed using MySQL, which allows project participants to execute web-based interactions including: submit data, annotate, sort, filter, and find specific data within a given experiment, and extract data from multiple experiments. The database will integrate data from all aspects of the study and include such information as collection data, water quality, biogeochemistry, species/taxonomy, T-RFLP patterns, DNA sequences, sequence annotation, DNA macroarray data, etc.
Dr. Peter Noble will determine how relevant linkages change with scale and with the dynamics of the system by using artificial neural networks (NNs), with cross-validation through conventional multivariate statistics. These combined statistical approaches should yield similar but not identical results since NNs are better at dealing with the nonlinear nature of biological data than conventional statistics.
