Record Details
Habitat suitability and uncertainty : a Bayesian approach to mapping benthic invertebrate distributions
ScholarsArchive at Oregon State University
| Field | Value |
|---|---|
| Title | Habitat suitability and uncertainty : a Bayesian approach to mapping benthic invertebrate distributions |
| Names |
Havron, Andrea M.
(creator) Goldfinger, Chris (advisor) |
| Date Issued | 2015-05-13 (iso8601) |
| Note | Graduation date: 2015 |
| Abstract | Mitigating for increased human impact to the seafloor associated with resource extraction activities and renewable energy development can benefit from an understanding of the distribution of sensitive marine benthic species. Habitat suitability predictive modeling is a cost effective statistical tool to infer species distribution patterns from constrained sampling locations. However, uncertainties related to the data, collection methods, and the statistical process can carry over into final maps. Challenges remain in accurately displaying the spatial uncertainty of habitat suitability models in a way that is transparent and easy to interpret. This thesis uses Bayesian networks to develop habitat suitability maps for several species of benthic invertebrates along the shelf and slope of the continental US west coast. In addition to predictive maps, methods were developed to create two complementary maps communicating model prediction uncertainty and experience, a measure of equivalent sample size. Species modeled include three species of benthic macrofauna: a marine bivalve Axinopsida serricata, a marine gastropod Aystris gausapata, and a marine polychaete, Sternaspis fossor; and three species of benthic megafauna of the dictyonine glass sponge assemblage: Aphrocallistes vastus, Heterochone calyx, and Farrea occa. Benthic macrofauna models were learned from benthic sampling data collected from sites along the Pacific Northwest shelf, spanning from Northern California to Washington State, south of the Olympic Coast National Marine Sanctuary. Benthic megafauna models were learned from NOAA’s historical sponge and coral observations dataset. Data were collected from bottom trawls conducted between 100m -1300m water depth along the US west coast continental shelf and upper slope. Netica® software was used to implement the design and analysis of the statistical models. Final macrofauna models were selected using a cross-validation technique. A generalized benthic macrofauna model structure for invertebrates living within marine sediment was developed for reusability and update capacity. With additional maps of uncertainty, marine resource managers and decision makers are better equipped to interpret habitat suitability maps in light of the best available science, and improve map usability for marine spatial planning purposes. Methods developed and presented here are broadly applicable to a wide range of other species and ecosystems, particularly in settings with small sampling effort. |
| Genre | Thesis/Dissertation |
| Topic | Bayesian networks |
| Identifier | http://hdl.handle.net/1957/55893 |
