876 



Fishery Bulletin 101(4) 



Figure 2 



Map of areas used in the study. SR=Shag Rocks. NG=North South Georgia, EG=East South Georgia, 

 SG=South South Georgia, and the 200-m, 1000-m and 3000-m bathymetric contours are shown. The 

 total number of hauls in each of these areas were as follow: SR=145, NG=50, EG=30, and SG=15. 

 Inset shows the position of the study area in relation to the South Atlantic. 



(Macpherson, 1988) were caught at intermediate depths. 

 Other influences on crab catch rates were investigated by 

 using two cruises in 2001, which covered the study area 

 more evenly (most hauls in cruise Gl were concentrated 

 in a small area around southeast Shag Rocks; see Fig. 2). 

 For each species two generalized linear models were con- 

 structed with Splus statistical software (version 2000, 

 Mathsoft Engineering & Education, Inc., Cambridge, UK): 

 a binomial link model for the probability of encountering a 

 crab of that species (pe) and a Gaussian link model for the 

 natural logarithm of CPUE (catch in numbers per pot) for 

 all nonzero catches. Both models were of the form A=pjX 

 depth+p.2xarea-(-p3Xvessel+p4Xsoak time, where for the 

 binomial models, A was set to 1 if crabs had been caught 

 and if they had not, and for the Gaussian models, A was 

 set to ln(numbers per pot) for all sets catching crabs. Area 

 and vessel parameters were factors. Depth and soak time 

 were modeled as linear continuous variables, except in the 

 case of the Gaussian model for P. forniosa, where a third- 

 order polynomial best described the relationship between 

 CPUE and depth. Predictions from the two models were 

 combined to predict crab catch rates per pot, 



(} = ;7(c)xexp[H±1.96x5£], 



where/'/ = the predicted In(CPUE) from the Gaussian 

 model; and 

 SE = the standard error from the Gaussian model. 



Biological data were collected from all crabs in randomly 

 selected pots. Carapace widths, carapace lengths, chela 

 height, and chela length were measured to the nearest 

 millimeter below by using calipers. Weights were measured 

 to the nearest 5 g below with spring-balanced scales. Sex, 

 maturity stage, condition of the carapace, and an index of 

 vitality (Table 1) were recorded for a subsample of crabs, 



selected as a random portion across species from the 

 contents of selected pots. Identification of the uncommon 

 species, Paralomis anamerae, Neolithodes diomedeae, and 

 Lithodes murrayi, was confirmed by using dried specimens 

 in London (at Imperial College and the British Museum of 

 Natural History). 



Male size at maturity was determined from the allometric 

 relationship between carapace length (L) and right (domi- 

 nant) chela size (height, CH, or length, CD. The slope of 

 the L-CH or L-CL relationship is assumed to change when 

 crabs reach maturity. Around South Georgia, Otto and 

 Macintosh (1996) used L and CH to determine the size at 

 maturity of male P. spinosissima, and around the Falkland 

 Islands Hoggarth (1991) used L and CL. For both P. spino- 

 sissima and P. fonnosa we found that the intersection of 

 the two lines corresponding to the onset of maturity was 

 not easy to identify from the relationship between CL and 

 L. We therefore used CH in our relationships. Following the 

 methods ofSomerton and Otto (1986), two linear regression 

 lines were fitted to natural logarithms of L and CH or CL. 

 These lines represent juvenile and adult phases in the L-CH 

 or CL relationship and intersected at a point taken as the L 

 at which males became mature. The regression lines of best 

 fit were determined by minimizing the combined residual 

 sums of squares, and standard errors were estimated by 

 using 500 bootstraps (sampling with replacement). 



Females were classified into two categories, "eggs absent" 

 and "eggs present." The vast majority of "eggs absent" fe- 

 males were small immature animals, but some large ani- 

 mals were also encountered in this category. Consequently, 

 for estimating size at maturity, females in the "eggs pres- 

 ent" category were defined as "mature" and those in the 

 "eggs absent" category were classified as "immature" up to 

 the size at which the proportion of females with eggs (i.e. 

 mature) reached 90%, after which they were classified as 

 "mature without eggs." Female size at maturity was deter- 



