|
Activity Number:
|
53
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #307026 |
|
Title:
|
Using Mixture Models To Estimate Abundance of Patchy Species
|
|
Author(s):
|
Elizabeth Conners and Loveday Conquest*+
|
|
Companies:
|
National Oceanic & Atmospheric Administration and University of Washington
|
|
Address:
|
Campus Box 355020, 1122 NE Boat Street, Seattle , WA, 98105,
|
|
Keywords:
|
mixture models ; delta distribution ; fisheries
|
|
Abstract:
|
A common problem in fisheries and many other areas of environmental science is estimating the overall abundance of a resource with an aggregated or patchy spatial distribution. Typically, the majority of survey observations will have zero or very low density, with a few high-density observations representing the majority of the biomass. This research looks at model-based estimates of abundance based on the assumption that samples are drawn from a mixture of distributions representing distinct biological processes. The behavior of mixture model estimators is examined using both simulated and real data. GLMs are used to examine spatial pattern in component classification for Atka mackerel data from the Aleutian Islands, Alaska, and model-predicted classes are used as the basis for both mixture model and restratified estimators.
|
