58 



FISHERY BULLETIN OF THE FISH AND "WILDLIFE SERVICE 



crabs that molted during the current year from 

 those that did not molt. The crabs that molted 

 in the winter and early spring have had their 

 shells no more than 6 months at the time of sum- 

 mer sampling, whereas those not molting have had 

 their shells not less than 1 year. Although shell 

 condition is a subjective classification, the differ- 

 ence in discoloration and marking of the exoskel- 

 eton is distinct. 



Confidence in the ability to distinguish between 

 the current year molters and those that molted in 

 the previous year may be shown by an examuia- 

 tion of shell-condition classifications of tagged 

 crabs, recorded at release and again at recovery. 

 The bulk of the recoveries and the classifications, 

 were made aboard the Japanese mothership by a 

 biologist following, for the most part, our written 

 description of the various shell conditions. Ex- 

 cluding all tagged crab recoveries showing changes 

 in length measurements, and therefore indicative 

 of having molted, there were 595 tag returns with 

 shell-condition data available for study. Table 1 

 shows the shell conditions recorded at release and 

 recovery of the crabs and their periods of free- 

 dom. 



Of the 417 recoveries of new-shell releases, one 

 recovered after a year of freedom was classified as 

 new-shell, and by our criteria of shell conditions 

 is considered in error. An additional six were 

 classified as new-old, indicating some doubt. The 

 six doubtful cases were recorded in 1956, and after 

 the 1957 season the definitions of the shell condi- 

 tions were made more explicit. Of the old-shell 

 releases, two recoveries within the year of release 

 were classified as new shells on recoveiy and are 

 considered misclassified. The amount of error in 

 classification appears to be no more than 1.5 per- 

 cent and may be as low as 0.5 percent if the six 

 doubtful cases are not included. 



Table 1. — Shell condition classification at recovery of non- 

 molting tagged crais 



1 Six additional crabs were recovered but classified as new-old and are not 

 included. 



The amount of growth per molt is determined by 

 an examination of the tagged crab measurement 

 data that were taken at release and again at re- 

 covery. Preliminary analysis of the relation of 

 width and length of tagged crabs indicated some 

 measurement error. Therefore, width on length 

 regression and a 99 percent confidence interval 

 around this regression were calculated from a ran- 

 dom sample of 744 crabs. All tag recoveries where 

 measurements fell beyond the intei-val were not 

 considered in the analysis. A few recoveries were 

 also discarded due to illogical length to shell con- 

 dition relations, for example, an increase in cara- 

 pace length inconsistent with a logical change in 

 shell condition. 



In order to determine the range of measure- 

 ment error, we examined 128 within-season tag 

 recovery measurements (appendix table 2) rea- 

 soning that variations in measui'ements for this 

 group must result from error or bias. Plotting the 

 deviations of recovery from release measurements 

 sliows that 99 percent of the deviations lie be- 

 tween plus and minus 4.4 mm. This is shown 

 graphically by the shaded histogram in figure 3. 



All tagged crabs, that measured 5 nmi. or more 

 larger when recovered, and which had a cori-e- 

 sponding increase in width, are considered to rep- 

 resent crabs that gi-ew during their i>eriods of 

 freedom. The deviations of the lengths at recoveiy 

 from the lengtlis at release for 325 male crabs de- 

 picting growth are shown by the unshaded liisto- 

 gram in figure 3. Considering the shell condition 

 and the length of time at liberty, 15 crabs with 

 length increments greater than 23 mm. were con- 

 sidered to have molted at least twice, and therefore 

 are not used in the analysis. 



GROWTH BY SIZE FREQUENCIES 



Length measui-ements of all male king crabs 

 taken during station pattern sampling each year 

 since 1955 (Appendix table 1) w^ere smoothed by 

 a moving average of three ; the i-esulting numbers 

 at each millimeter of length were expressed as i^er- 

 centages of each year's total. Percentages were 

 used to compensate for varying numbers between 

 years. To emphasize the dominant size groups 

 and their progressions, the percentage deviation 

 of each years size frequency distribution from the 

 1955 through 1959 mean distribution was calcu- 

 lated. The resulting yearly positive and negative 

 deviations are plotted on figure 4. Examination 



