494 



Fishery Bulletin 102(3) 



O April 



A September 



(P>0.999) 

 (P=0.06) 



1.0 

 0.8 

 0.6 



CD CD 



Q- E 



• 0A \f 



i, 

 02OA 



O 





A. 



°^° 



O 



o 



A 



O 



A 



' C*°J 



0D 



0.2 



0.4 0.6 0.1 



1.0 



Percent legal-size 

 male crabs on transects 



Figure 4 



The percentage of male crabs of the 

 Dungeness crab (C. magister) popu- 

 lation that were legal size (>165 mm) 

 estimated from pots compared to the 

 percentage of male crabs estimated 

 from dive transects. Data from pots 

 and dive transects were compared for 

 each season with a paired sign test 

 and P-values are reported. 



0.8 



06 



g 

 o 



0.2 



Pots Transects 

 -o- •*■ 05 change/year 



003 change/year 



50 



100 150 



Sample size (n) 



250 



Figure 5 



Relationship between power and sample size (/!) in comparing 

 catch from pots and density on dive transects for male Dungeness 

 crabs (C. magister) at three levels of population change. 



In both April and September, pot sampling was signif- 

 icantly biased against females (Fig. 3A). When females 

 were categorized as ovigerous and nonovigerous, it was 

 clear that ovigerous females accounted for the major- 

 ity of the bias because pots were not biased against 

 nonovigerous females (Fig. 3B). Similar results have 

 been found for a closely related species, Cancer pagurus; 

 female C. pagurus readily enter pots when they are in 

 a nonovigerous reproductive state but are rarely cap- 

 tured when they are ovigerous (Bennett, 1995). Move- 

 ment studies of Dungeness crabs tagged with sonic 

 transmitters have demonstrated that ovigerous females 

 move less frequently and move slower than males or 

 nonovigerous females (O'Clair et al., 1990). Thus, one 

 explanation for the bias against ovigerous female crabs 

 is that their restricted movements make it less likely 

 they will be able to locate and become entrapped in 

 pots. In addition to being less mobile, ovigerous females 

 may be less attracted to bait than nonovigerous crabs. 

 In controlled feeding experiments, ovigerous females 

 had lower feeding rates than nonovigerous females, and 

 ovigerous females took longer to begin feeding (Schultz 

 et al., 1996; Schultz and Shirley, 1997 i. Therefore, ovig- 

 erous females may be less responsive to the bait plume 

 from a pot. 



The estimate of nonovigerous females from both pot 

 CPUE and dive transect density increased from April 

 to September (Fig. 2, C and D). As with males, the in- 

 crease in CPUE for nonovigerous females may be partly 

 due to an increase in catchability in September. How- 



ever, the fact that the density estimates from dives also 

 increased suggests that the number of nonovigerous 

 females actually increased between April and Septem- 

 ber. This explanation is supported by the decrease in 

 ovigerous crabs from April to September for both CPUE 

 (Fig. 2E) and density estimates (Fig. 2F i. 



The low catchability of ovigerous females makes it 

 problematic to monitor relative abundance of females 

 or changes in sex ratio through time. However, be- 

 cause pots were not biased against nonovigerous fe- 

 males (Fig. 3), the solution may be to estimate the rela- 

 tive abundance of females by sampling after females 

 hatch their eggs and before they extrude a new clutch 

 of eggs in the fall. In southeastern Alaska, most females 

 are nonovigerous in late July and early August (Stone 

 and O'Clair, 2001; Swiney et al., 2003); therefore this 

 would be the optimal time of year to sample females 

 or to measure sex ratio of Dungeness crab populations. 

 Unfortunately, this timing coincides with the summer 

 commercial fishing season, which could bias sampling 

 if there was "competition" between survey pots and 

 commercial pots. 



For both males and females, the power analyses of 

 the pot and dive data indicated that for most population 

 assessment applications it would be extremely difficult 

 to conduct enough dive transects to obtain sufficient 

 statistical power. Even if it were possible to conduct 

 as many dive transects as pot samples, the power of 

 a dive transect was still lower than that of a pot; the 

 higher power of the pots was due to lower variance 

 among pots. Pots work by attracting crabs with a bait 

 plume; thus the area and number of crabs sampled is 



