eCommons

 

NUTS AND BOLTS: SPATIAL STATISTICAL MODEL DEVELOPMENT FOR FISHERIES

Other Titles

Abstract

Concerning fisheries, the question of where is as critical as the question of how many. This dissertation develops, tests, and applies two spatial models to data from one highly mobile, and one sedentary species. It (1) develops a Bayesian hidden Markov model (HMM) to estimate movement and habitat use of diadromous fish species, (2) applies this model to acoustic tracking data from the endangered New York Bight distinct population segment (DPS) of the Atlantic sturgeon (Acipenser oxyrinchus oxyrinchus), (3) develops a Bayesian extension of a regression kriging (RK) model in which a penalized spline model estimates large-scale spatial trend and a variogram estimates fine-scale spatial autocorrelation, and (4) applies a modified version of this model to continuous belt transect survey data from Atlantic sea scallops (Placopecten magellanicus). The theme of this dissertation is thus the development and thorough testing of novel Bayesian spatial model structures through simulation, followed by application. While the process is itself consistent, the methods and biology and ecology of the species thus analyzed are not. The HMM with nested detection and habitat use models performed very well on simulated data, even with approximately 70% of the data removed. Results of the application to Atlantic sturgeon suggest that while males stage downriver from spawning grounds, females migrate to and from spawning grounds with little evidence of staging. The RK model, while stable, very often gave biased estimates of the variogram parameters. The balance between the penalized spline and the variogram was tenuous at best. That said, it appears to be capable of identifying whether or not fine-scale autocorrelation did indeed exist separate from that of large-scale spatial trend. Results from application of the extended Bayesian model to Atlantic sea scallops compared to a frequentist and thus sequential application of the penalized spline and variogram models suggest that fine-scale trend may be of a much greater magnitude than originally thought. The results are themselves significant to the species in question. Furthermore, the process of testing and developing a model using biologically and ecologically informed simulations is highly effective, and constrains the infinite realm of possibilities to a manageable set of choices based in reality.

Journal / Series

Volume & Issue

Description

214 pages

Sponsorship

Date Issued

2020-08

Publisher

Keywords

Atlantic sea scallops; Atlantic sturgeon; Bayesian statistics; Hidden Markov models; Penalized splines; Regression kriging

Location

Effective Date

Expiration Date

Sector

Employer

Union

Union Local

NAICS

Number of Workers

Committee Chair

Sullivan, Patrick J.

Committee Co-Chair

Committee Member

Sethi, Suresh Andrew
Royle, Jeffrey Andrew
Hart, Deborah

Degree Discipline

Natural Resources

Degree Name

Ph. D., Natural Resources

Degree Level

Doctor of Philosophy

Related Version

Related DOI

Related To

Related Part

Based on Related Item

Has Other Format(s)

Part of Related Item

Related To

Related Publication(s)

Link(s) to Related Publication(s)

References

Link(s) to Reference(s)

Previously Published As

Government Document

ISBN

ISMN

ISSN

Other Identifiers

Rights

Attribution 4.0 International

Types

dissertation or thesis

Accessibility Feature

Accessibility Hazard

Accessibility Summary

Link(s) to Catalog Record