Hannah and Jones: Fishery-induced population changes in Pandalus jordani 



43 



Area 22- August, Area 21- July, Area 

 26- August, and Area 19- August. The 

 most complete time-series for age-2 

 shrimp were found to be Area 19- 

 August, Area 22- April, Area 21-June, 

 and Area 26-May. Age-2 length com- 

 parisons are not independent of the 

 age-1 results. The age-2 analysis was 

 employed to help rule out any appar- 

 ent changes in growth of age-1 shrimp 

 caused by changes in fishery or gear 

 selectivity over time. 



To further investigate factors influ- 

 encing age-1 shrimp growth, we con- 

 ducted some exploratory correlation 

 analysis. Since the time-series for the 

 most complete index areas still con- 

 tained some missing samples, and 

 since many of the environmental fac- 

 tors we wished to test are not area- 

 specific, it was desirable to combine 

 our four age-1 growth indices into one, 

 more complete, time-series. Carapace 

 length at age in pink shrimp exhibits 

 a gradient effect increasing from north 

 to south along the coast and also in- 

 creasing through the season. To re- 

 move these effects and yet preserve the interannual 

 variations in size, one of the four indices was chosen 

 as a standard and the other three were adjusted by an 

 additive factor equal to the difference between the 

 mean of the chosen standard and that of the individual 

 index area-month. Subsequently, the adjusted index 

 area-months were averaged into one time-series. Area 

 22 was chosen as the standard, and the resultant time- 

 series for age-1 was without gaps and most points were 

 based on two or more adjusted means. 



Linear regression was used to examine factors in- 

 fluencing variation in this age-1 shrimp growth in- 

 dex. The independent variables tested included sea sur- 

 face temperature at Charleston, Oregon (Oreg. Dep. 

 Fish. Wildl., unpubl. data) upwelling at 45°N, 125°W 

 (Bakun 1973; NMFS Pacific Environ. Group, Mon- 

 terey, CA, unpubl. data), inverse-barometer corrected 

 sea level at Newport, Oregon, (Pittock et al. 1982; 

 Pittock, unpubl. data), and catch per unit effort in 

 the fishery as an index of shrimp density. Kruse 

 (1981) found inverse barometer-corrected sea level 

 at Newport and Neah Bay to be highly correlated 

 with sea-bottom shelf temperatures near Newport. 

 We tested each variable with no time lag (year t) 

 and a 1-year time lag (t - 1) to match growth in an 

 earlier life stage. Adjusted sea level at Newport was 

 tested both with and without the 1983 data point, a 

 year of abnormally high sea level caused by a strong 



El Nino event. Finally, we tested the average CPUE 

 for the years t and t - 1 combined, to represent the 

 average density encountered over the life of an age-1 

 shrimp. Second-order polynomial regression was also 

 used to test each variable for a significantly curvi- 

 linear relationship with the age-1 growth index (Ricker 

 1975). 



Correlation analysis with time-series data of short 

 duration is often of limited value, but does help to 

 generate initial hypotheses to be tested with the ac- 

 cumulation of future data (Ricker 1975). Short time- 

 series often exhibit unidirectional time trends causing 

 spurious correlations. For these reasons, we felt that 

 correlation analysis would be a relatively poor tool for 

 differentiating the relative importance of the various 

 factors to shrimp growth, but would help to identify 

 the factors which deserve future analysis. Consequent- 

 ly, correlation analysis was not pursued further in this 

 study. 



For analysis of trends in the sex composition data, 

 we once again relied upon four index-area months with 

 the most complete time-series. The four index-areas 

 employed were Areas 19 and 28 in October and Areas 

 21 and 22 in September. Pink shrimp are protandrous 

 hermaphrodites and mate primarily in September and 

 October each year (Pacific Fisheries Management 

 Council 1981). They usually mature first as males in 

 the fall at about IV2 years of age, and after spawning 



