Anderson: Age. growth, and mortality of Pandalus borealis Kroyer 



549 



and declined to about one-sixth of its 

 former abundance by 1981. Lower ob- 

 served natural mortality for the 1975 

 year-class beyond age 4.4 could be ex- 

 plained by decreasing cod predation. 



Mortality estimates rely not only on the 

 suitable definition of the year-class modes 

 but also on the accuracy of biomass esti- 

 mates (see eq. 2). Accuracy of biomass 

 estimation is a function of the vulnerabil- 

 ity, gear selectivity, and accessibility of 

 shrimp in Pavlof Bay to the annual sur- 

 veys. Knowledge of the life history of P. borealis were 

 used to address these sources of variability in survey 

 sampling. The late summer-fall period was chosen for 

 sampling because P. borealis may be more vulnerable 

 to capture due to the formation of mating and spawn- 

 ing aggregations. Small males and the larger females 

 are not segregated, and even juvenile shrimp =6 mm 

 CL can be found in these aggregations. Sampling was 

 restricted to daylight hours to minimize the effect of 

 diel vertical migration; thus, shrimp were more suscep- 

 tible to capture by bottom trawling. Shrimp smaller 

 than 18 mm CL are known to be less vulnerable to cap- 

 ture with the 32mm mesh survey sampling gear. 

 Sampling with smaller mesh (3.2 mm) in 1986 did in- 

 dicate that more small shrimp could be captured, but 

 small mesh sampling of the entire population was not 

 deemed feasible due to increased sorting time. 



Biomass estimates are considered conservative due 

 to the lesser vulnerability of small shrimp. Likewise, 

 larger/older shrimp may be able to escape capture by 

 swimming out of the water column sampled by the 

 survey trawl, leading to an overestimate of mortality. 

 The high opening shrimp trawl used in this study 

 (Wathne 1977) was designed to sample a substantial 

 portion (3.8-4.4 m off bottom) of the water column. 

 Biomass trends tend to refute avoidance of larger 

 shrimp as a source of error. 



Biomass estimates, while they may not be an absolute 

 estimate of the population size, can represent an index. 

 Sources of variability that could affect the validity of 

 biomass estimates being used as an index were con- 

 trolled. Survey timing, gear, methods, and even vessel 

 type were the same throughout the survey series. Sur- 

 vey sampling methodology, while minimizing, may not 

 always eliminate possible changes caused by availability 

 of shrimp. Pandalus borealis may be more vulnerable 

 to capture during mating and spawning because dense 

 aggregations are probably beneficial for mating suc- 

 cess. The exact timing of this event may vary from year 

 to year. As an example, the 1972 and 1974 surveys both 

 took place during the second week in September (Table 

 1). In 1972, about 60% of females were carrying eggs 

 externally while only head roe (egg mass clearly visible 



under carapace) females were found in 1974. Changes 

 in availability probably explain the dip in the catch 

 curve for the 1971 year-class at age 4.4 and at 2.4 for 

 the 1975 year-class (Fig. 5). These apparent changes 

 in availability perhaps associated with some life-history 

 stage or abiotic factor should be studied in order to im- 

 prove both biomass and mortality estimates. 



Sex transformation 



The rate of transformation from male to female was 

 considerably different for the two dominant year- 

 classes studied (Table 6). Some members of the 1971 

 year class initiated sex transformation during their 

 fourth year; however, others transformed during either 

 their fifth or sixth year. In contrast, all members of 

 the 1975 year-class initiated and completed transfor- 

 mation during their fourth year. The age at and rate 

 of sex change for a year-class also appear to be closely 

 related to year-class size or overall population density. 

 Charnov (1979, 1981) proposed a model in which high 

 mortality rates lead to a shorter male phase. Charnov 

 et al. (1978) also considered how natural selection might 

 act to alter the sex ratio of Pandalid shrimp in response 

 to environmental influences. I feel that accelerated sex 

 transition observed for the 1975 year-class in this study 

 was influenced by short-term phenomena and is evi- 

 dence of the labile timing in sex change of this species. 

 The occurrence in 1980 and again in 1984-86 of the 

 smallest ovigerous females ever sampled (13.0 mm CL) 

 is further evidence that accelerated sex reversal may 

 be a possible mechanism by which northern shrimp 

 attempt to increase reproductive capacity in the face 

 of decreasing density (Table 1). Charnov and Ander- 

 son (1989) reported on an analysis of Pavlof Bay 

 P. borealis that demonstrated the size of shrimp at the 

 time of sex-transformation changes through time in 

 relation to changes in the breeding size distribution. 



Stock/recruit relationship and yield 



While a compensatory relationship between lower pop- 

 ulation levels or density and the occurrence of early 



