Distribution and application of a new geostatistical index standardization and habitat modeling tool for stock assessments and essential fish habitat designation in Alaska and Northwest Atlantic regions

Principal Investigator: James Thorson
Co-Principal Investigators: James Ianelli, Loretta O’Brien
NMFS Collaborators: Liz Brooks (NE), Martin Dorn (AK), Paul Spencer (AK)

Recent research has shown the value of using spatial models to generate annual estimates of stock abundance from bottom trawl survey data for West Coast groundfishes. For example, Shelton et al. (2014, CJFAS) showed that spatial variation in density for darkblotched rockfish explained a substantial portion of variation in catch rates, and Thorson et al. (In press, ICESJMS) showed that spatial models improved precision for estimated indices using simulated data relative to a nonspatial model. Spatial models can incorporate measured habitat variables (e.g., bottom substrate type) or infer habitat from biological sampling (e.g., infer rocky substrate from the presence rock-associated species), and these measured or infered habitat variables are then incorporated into the stock assessment process when the estimated index of abundance is used in a stock assessment model. Spatial models have subsequently been approved for consideration in stock assessments by the PFMC Scientific and Statistical Committee for the 2015 assessment cycle.

However, bottom trawl data are routinely collected in many other NMFS regions, including long-term data sets available for the Gulf of Alaska, Bering Sea, and Northwest Atlantic regions. This project therefore seeks to ease documentation, implementation, and exploration of spatial index standardization tools for NMFS regions outside the California Current. The project will improve documentation and examples for the publicly available R package SpatialDeltaGLMM, and will work with collaborators at the AFSC and NEFSC to explore applications of the tool for stock and habitat assessments in each region. In this way, the project will provide a proof-of-concept for sharing habitat-analysis methods between regions and centers.

FY 2015 Project 15-027

Figure 1. Density for arrowtooth flounder 2003–2012, estimated by the geostatistical delta-generalized linear mixed model (note that the white space in southern California represents the cowcod conservation area, which prohibits trawl gears including the survey design and hence is excluded when estimating spatial densities and abundance indices). From Thorson et al. (2015).

Project #


InPort ID#




Northwest Fisheries Science Center (NWFSC)


Publications & Presentations

Media & Other Products