Show simple item record

dc.contributor.authorHanselman, Dana Henry
dc.date.accessioned2018-07-11T01:04:08Z
dc.date.available2018-07-11T01:04:08Z
dc.date.issued2004
dc.identifier.urihttp://hdl.handle.net/11122/8775
dc.descriptionDissertation (Ph.D.) University of Alaska Fairbanks, 2004
dc.description.abstractPacific ocean perch (Sebastes alutus) stock size in the Gulf of Alaska has been difficult to assess because of an imprecise survey biomass index. This imprecision has been attributed to low sampling effort on a species with an aggregated distribution. In this thesis, I examined the importance of estimated survey biomass in the stock assessment and ways to improve them. First, I presented the complete stock assessment for 2003, with an analysis of uncertainty. Uncertain parameters included natural mortality, recruitment, and biomass estimates. Second, I examined adaptive cluster sampling (ACS) as a method to reduce survey uncertainty. ACS results provided lower estimates of mean abundance and lower standard errors than did simple random sampling (SRS). Bootstrapping suggested that the ACS mean may be a superior measure of central tendency. ACS results were better than SRS, but not as dramatically as suggested by previous literature. I used simulations to explore why ACS did not perform optimally. These simulations showed that it would be necessary to sample over 10% of the population to obtain large gains in precision. This is impractical for a large marine population. I explored the use of hydroacoustic data recorded on survey vessels to gain precision in biomass estimation. I used the data to (1) develop a catch prediction model based on near-bottom backscatter, (2) simulate an adaptive design, (3) apply ratio estimation in double sampling using hydroacoustic data, and (4) post-stratify survey data. Using hydroacoustic data in these designs showed gains in precision over SRS and may be useful. Finally, I used the S. alutus age structured model presented above to simulate effects of five factors: survey measurement error, catchability trends, a second biomass index, data source weighting, and sensitivity of prior distributions. Simulations showed that the stock assessment model was ineffective at high measurement error and was unable to detect trends in the data. A second biomass index yielded gains in model precision. The weight given lengths measured in the fishery was most important because of its long time series, and the prior distribution on natural mortality was most influential because it was difficult to estimate.
dc.subjectAquatic sciences
dc.titleGulf Of Alaska Pacific Ocean Perch: Stock Assessment, Survey Design And Sampling
dc.typeDissertation
dc.type.degreephd
dc.identifier.departmentFisheries Division
dc.contributor.chairTerrance J. Quinn, II
refterms.dateFOA2020-03-05T16:17:42Z


Files in this item

Thumbnail
Name:
Hanselman_D_2004.pdf
Size:
3.251Mb
Format:
PDF

This item appears in the following Collection(s)

Show simple item record