764 



Fishery Bulletin 88(4). 1990 



LU 

 Q_ 



CC 

 LU 

 CQ 



< 



LU 



Q 



< 



cc 



\- 



C/5 



400 



200 H 



400 



200- 



400- 

 200- 



1984 



1985 



1986 



1984 YC 1882 YC 

 AQE 2 AQE 4 



1987 



1988 



10 15 20 25 30 35 



CARAPACE LENGTH (mm) 



Figure 2 



Length-frequency distributions (stratified mean number per tow) for 

 Pandalvs borealis collected in the western Gulf of Maine during the 

 1984-88 northern shrimp surveys aboard RV Gloria Michelle. 



pairs within a given ridge tended to provide similar in- 

 terpretations of the number of modes (assumed age 

 groups) present in the observed length frequency, as 

 expected from earlier testing of the method (Shepherd 

 et al. 1987). Values of the optimum value of to corre- 

 sponding to K and Lmr parameter pairs are the deci- 

 mal fraction part of to, and are indeterminate with 

 respect to the addition or subtraction of any whole 

 number of years (Shepherd 1987). 



SRLCA results were examined in the context of 

 previously developed parameter estimates and a priori 

 assumptions, and alternative parameters were evalu- 

 ated when the highest scoring values did not agree with 

 prior interpretations of the number and position of 

 modes (assumed age groups) expected. SRLCA was 

 first applied to annual distributions independently, and 

 then to distributions from 1984 to 1988 sequentially 

 pooled in a single run. The annual distributions were 

 analyzed primarily to evaluate inter{)retations provided 

 by SRLCA given the variable patterns of growth and 

 recruitment in the data. Growth parameters and subse- 

 quently resolved age frequencies from the final pooled 

 analysis were used to estimate total mortality for com- 

 parison with rates estimated by visual resolution of the 

 length-frequency data to age. 



Shepherd (1987) noted that the number of older ages 

 determined by decomposition of the length-frequency 

 distribution according to the von Bertalanffy growth 

 equation is dependent to a large degree on the value 

 of L,nf, and suggested that parameters selected by 

 SRLCA might be most appropriate for subsequent use 

 in other length-based analyses, rather than to slice 

 length frequencies to ages, unless additional data (e.g., 

 knowledge of the expected number of cohorts) are 

 available to select the ridge containing the "correct" 

 parameter pair (see also Shepherd et al. 1987). In this 

 exercise, we elected to proceed with resolution to 

 cohorts, and subsequent age-based mortality estima- 

 tion, both because of the apparently nonequilibrium 

 nature of this northern shrimp population (thus limiting 

 the utility of length-based methods which assume 

 steady-state conditions), and to provide results com- 

 parable with those previously estimated (NSTC 1984, 

 1985, 1986, 1987, 1988). 



the growth parameters previously estimated for this 

 stock of P. borealis, in order to adequately explore the 

 high score ridges provided by the SRLCA score func- 

 tion. Preliminary evaluation of the highest scoring 

 parameter values from each ridge, along with several 

 local maxima (higher score than all eight nearest 

 neighbors) within each ridge, indicated that parameter 



Results 



Annual length frequencies 



1 984 distribution This distribution is characterized 

 by a dominant mode centered at 19.5 mm (Fig. 2). 

 A priori interpretation suggested that the dominant 

 mode at 19-20 mm should be age-group 2 shrimp (a 

 strong 1982 YC), with shrimp 22-24 mm probably age- 



