You are here

Predicting the distribution and ecological niche of unexploited snow crab (Chionoecetes opilio) populations in Alaskan waters: A first open-access ensemble model

TitlePredicting the distribution and ecological niche of unexploited snow crab (Chionoecetes opilio) populations in Alaskan waters: A first open-access ensemble model
Publication TypeJournal Article
Year of Publication2011
AuthorsHardy, SM, Lindgren, M, Konakanchi, H, Huettmann, F
JournalAmerican Journal of Physiology – Regulatory, Integrative and Comparative Physiology and Comparative Biology
Pagination1–15
Abstract

Synopsis Populations of the snow crab (Chionoecetes opilio) are widely distributed on high-latitude continental shelves ofthe North Pacific and North Atlantic, and represent a valuable resource in both the United States and Canada. In USwaters, snow crabs are found throughout the Arctic and sub-Arctic seas surrounding Alaska, north of the AleutianIslands, yet commercial harvest currently focuses on the more southerly population in the Bering Sea. Populationdynamics are well-monitored in exploited areas, but few data exist for populations further north where climate trendsin the Arctic appear to be affecting species’ distributions and community structure on multiple trophic levels. Moreover,increased shipping traffic, as well as fisheries and petroleum resource development, may add additional pressures innorthern portions of the range as seasonal ice cover continues to decline. In the face of these pressures, we examined theecological niche and population distribution of snow crabs in Alaskan waters using a GIS-based spatial modelingapproach. We present the first quantitative open-access model predictions of snow-crab distribution, abundance, andbiomass in the Chukchi and Beaufort Seas. Multi-variate analysis of environmental drivers of species’ distribution andcommunity structure commonly rely on multiple linear regression methods. The spatial modeling approach employedhere improves upon linear regression methods in allowing for exploration of nonlinear relationships and interactionsbetween variables. Three machine-learning algorithms were used to evaluate relationships between snow-crab distributionand environmental parameters, including TreeNet, Random Forests, and MARS. An ensemble model was then generatedby combining output from these three models to generate consensus predictions for presence–absence, abundance, andbiomass of snow crabs. Each algorithm identified a suite of variables most important in predicting snow-crab distribution,including nutrient and chlorophyll-a concentrations in overlying waters, temperature, salinity, and annual sea-icecover; this information may be used to develop and test hypotheses regarding the ecology of this species. This is the firstsuch quantitative model for snow crabs, and all GIS-data layers compiled for this project are freely available from theauthors, upon request, for public use and improvement.

DOI10.1093/icb/icr102
Username Tag: