Research Document - 2013/114
Model-based estimation of commercial- sized snow crab (Chionoecetes opilio) abundance in the southern Gulf of St. Lawrence, 1980-2012, using data from two bottom trawl surveys
By H.P. Benoît and N. Cadigan
Abstract
The assessment for snow crab in the southern Gulf of St. Lawrence (sGSL) has been founded on a dedicated fishery-independent bottom-trawl survey conducted annually since 1988 (henceforth called the crab survey, CS). The sampling frame for this survey increased gradually during much of the intervening period and there has been no accounting to date for possible changes in catchability in the survey resulting from three gear or vessel changes that have occurred since the survey’s inception. In the absence of robust methods to account for changes in the survey sampling frame and vessels/gear, long-term indices of sGSL snow crab abundance from the crab survey risk conflating true changes in abundance and apparent changes caused by modifications in the sampling protocols. Another research vessel (RV) bottom-trawl survey of the sGSL, initially focused on demersal fish (henceforth called the RV survey), has collected information on snow crab catches in survey sets since 1980. The RV survey provides synoptic estimates of snow crab abundance and distribution for a consistently sampled area of the sGSL that largely overlaps with, and is larger than, the area covered by the crab survey. Furthermore, with the exception of a single year (2003), specific data exist to estimate correction factors required to produce a standardized series for the RV survey in the face of vessel, gear and protocol changes that have occurred since 1980.
In this document, we present a model-based estimation framework that integrates data from the two surveys to produce an annual standardized index of commercial-sized sGSL snow crab abundance for 1980 to the present. The basic model is conceptually simple and assumes that crab density varies randomly with a constant mean within geographic strata and that trawl catches are basically Negative Binomial random variables. The model uses data from paired trawling at specific sites and joint fishing in common strata to estimate relative catchabilities for different RV survey vessels used since 1980 and for day vs. night to adjust for a shift from day-only to 24hr fishing that occurred in 1985. Likewise, the model exploits information from joint fishing in common strata to estimate the relative catchabilities of the vessels and gears used in the crab survey since 1988, by treating the RV survey as a standardized series. A similar logic is also applied to estimate the relative catchability of the RV survey vessel used in 2003. Because the relative catchabilities are estimated in a common model, associated uncertainties are reflected in the estimated abundance index. The model appears to fit the observations reasonably well, though discrepancies for the RV survey in certain strata and years require further investigation. Additional work aimed at simplifying the model and testing model robustness via simulation is also required. While the results presented in this document should be considered as preliminary, they indicate a strong potential for this approach which synthesizes important sources of data on snow crab abundance and distribution and which generates a standardized index whose duration is twice that of the series presently used in the assessment.
- Date modified: