Terceiro and Idoine 



SRLCA application to Pandalus boreslis survey data 



763 



Figure 1 



Western Gulf of Maine region with major bathymetric features and 

 50-fathom (100-m) isobath. Northern shrimp survey area extends 

 from 68°W longitude to the 50-fathom isobath, except for stratum 

 2, where the sampled area extends to the 30-fathom (60-m) isobath. 



Length-frequency data from identical sample strata 

 sets (strata 1, 3, and 5-8) are available for 1984-88 

 (NSTC 1984, 1985, 1986. 1987, 1988). Trawl gear con- 

 sists of a modified 4-seam conmiercial shrimp trawl 

 with 35 mm (1.4 inch) stretched mesh in the body of 

 the net and 32 inm (1.3 inch) stretched mesh in the ex- 

 tension and codend. with "rockhopper" ground gear 

 to allow sampling over rough bottom (Mclnnes 1986). 

 Samples (2 kg) of each tow are retained for length 

 measurement and sex determination. All shrimp in the 

 sample are measured, with mid-dorsal carapace lengths 

 aggregated by 0.5-mm intervals (nearest 0.5 mm below 

 that measured). A 1-kg subsample is retained for deter- 

 mination of sex and spawning stage. 



To date, abundance and biomass indices from the RV 

 Gloria Michelle survey (stratified mean number per 

 tow, stratified mean weight [kg] per tow) have proven 

 to be accurate predictors of year-class size and com- 

 mercial fishery performance (total catch and catch-per- 

 unit-effort; NSTC 1988). Coefficients of variation for 

 stratified mean number and weight (kg) per tow, ag- 

 gregated over all sample strata, have averaged (1984- 

 88) about 13% and 12%, respectively, indicating rela- 

 tively high precision. 



A priori assumptions 



Information is available from previous analyses of the 

 Gulf of Maine northern shrimp length-frequency dis- 

 tributions to provide a baseline interpretation for 

 evaluation of SRLCA results. The von Bertalanffy 

 growth equation has been the accepted means of 

 describing the growth of P. horealis (Frechette and 

 Parsons 1983). For the Gulf of Maine stock, growth 

 parameters were expected to be in the range of those 

 previously estimated for this stock. These include (1) 

 parameters derived here by nonlinear least-squares 

 regression analysis of mean carapace length (CL) at age 

 data summarized by Haynes and Wigley (1969; Lj^f 

 = 32 mm, K = 0.46, t„ = -0.12), and (2) parameters 

 derived from NEFC groundfish bottomtrawl survey 

 data for northern shrimp (NEFC unpubl. data in 

 Mclnnes 1986; L,,,,- = 35.2 mm, K = 0.36, t,, = 0.06). In 

 both of these studies, age was estimated by visual in- 

 spection of length-frequency distributions. 



For Gulf of Maine northern shrimp survey data, a 

 1 March birthday was assumed (Apollonio et al. 1986), 

 with an average catch date of 1 August (NSTC 1984, 

 1985, 1986, 1987, 1988); thus t, was set equal to 0.42 

 for all SRLCA runs in this exercise. The usual presence 

 of four age groups (ages 1-4) in the survey catch was 

 suspected (Mclnnes 1986), with a maximum age of 5 

 years (Haynes and Wigley 1969, Apollonio et al. 1986), 

 based on modes in length frequencies and sexual 

 characteristics (Allen 1959, McCrary 1971). Northern 

 shrimp in a length range of 13-18 mm CL are generally 

 assumed to be of age-group 1, and mainly immature 

 and mature males, while shrimp 18-22 inm CL are 

 assumed to be age 2 and mostly mature males. Animals 

 in the 22-25 mm CL interval are assumed age 3, usual- 

 ly females with no previous spawning history. Shrimp 

 larger than 25 mm CL are assumed to be females of 

 age-group 4, with possible age-group 5 female shrimp 

 at CL greater than 29 mm. The survey samples north- 

 ern shrimp in a CL range of 10-32 mm (Fig. 2), with 

 animals assumed fully recruited to the gear at about 

 19-20 mm CL, or age 2 and older (Blott et al. 1983). 

 Previous interpretations of survey results and commer- 

 cial fishery performance suggest that the 1982 and 

 1987 year-classes (YC) of Gulf of Maine northern 

 shrimp were strong, and the 1983 YC very weak, with 

 the remaining cohorts (1984-86) of about equal 

 strength (NSTC 1987, 1988). 



Model evaluation 



Exploratory runs were performed using L,„f values 

 ranging from 20 to 50 mm, in 1-mm steps, and K values 

 ranging from 0.20 to 0.50, in 0.01 steps, encompass- 

 ing a very wide range of values (about ± 50%) around 



