ANNALA MORTALITY OF ROCK LOBSTER 



On the basis of a 9-mo fishing season, the annual 

 estimate of F = 1.30 (Table 7). 



DISCUSSIOiN 



The weighting procedure used to derive the sea- 

 sonal size-frequency distribution was designed to 

 average out changes in the monthly distributions 

 due to fluctuations in catchability, recruitment, 

 and mortality, which affect the estimates of total 

 mortality rate. The estimates derived from the 

 seasonal size-frequency distribution using the 

 methods of Hancock ( 1965) and Van Sickle ( 1977) 

 were similar to the means of the respective 

 monthly estimates (Table 5). The estimates from 

 the monthly samples using the method of Bhat- 

 tacharya (1967) were highly variable, probably 

 due to the small sample sizes, and the mean of the 

 monthly estimates was considerably less than the 

 estimate from the seasonal distribution. 



The factor having the greatest potential effect 

 on the estimates of mortality derived from the 

 size-frequency distributions is probably the influx 

 of new recruits into the fishery by growth over the 

 minimum legal size. Male rock lobsters in the Gis- 

 borne local area exhibit marked periodicity in the 

 molt cycle, with most molting between October 

 and December. However, the monthly estimates 

 using the methods of Hancock (1965) and Van 

 Sickle (1977) do not indicate any changes in mor- 

 tality rate associated with this molting period. 

 Therefore, estimation of the total mortality rate 

 from the weighted seasonal size-frequency dis- 

 tribution is considered valid in this example. 



The estimates of total mortality rate from the 

 weighted seasonal size-frequency distribution 

 using the three methods gave similar results. The 

 method of Bhattacharya (1967) is considered 

 adequate when the sample size is large and an 

 estimate of growth rate is available to aid in fitting 

 the lines. However, when used in analyzing size- 

 frequency distributions whose sample sizes were 

 small, such as those from other areas (my unpubl. 

 data) and the monthly samples in this example 

 (Table 5), the results varied widely. Moreover, 

 when used with data without definite modes, the 

 abundance of the first component often appears to 

 be underestimated, perhaps due to the difficulty of 

 determining the 100'7f retention length, and 

 greater consistency is obtained if the first three 

 components are included for analysis. 



Method 2 of Van Sickle (1977) also requires 

 comprehensive size-frequency and growth data. 



One of Van Sickle's key assumptions is that the 

 method be applied to a stationary or steady state 

 population, which he defines (after Seber 1973) as 

 one having age and size structures that are cyclic, 

 with a period usually of 1 yr. Thus, size distribu- 

 tions observed at yearly intervals will appear 

 identical. However, he argues that the method can 

 be applied to annual "average" size distributions 

 rather than a distribution at one point in time 

 (Van Sickle 1977, quoting Van Sickle 1975). 

 Growth and mortality rates should not vary with 

 time, and seasonal and year-to-year changes in 

 recruitment and growth should be "averaged out" 

 of the data used. 



Estimates derived using the smaller of the mil- 

 limeter size groupings bracketing the annual 

 growth increment for this example (Tables 3, 5) 

 and for samples from other areas (my unpubl. data) 

 were usually lower than those derived using the 

 larger millimeter size grouping. These lower es- 

 timates may be due to the violation of one or more 

 of the above assumptions. Van Sickle's method is 

 very dependent on accurate estimates of growth 

 rates and densities at the boundaries of each size 

 class, and even minor fluctuations in recruitment 

 and for samples from other areas (my unpubl. 

 data) were usually lower than those derived using 

 the larger millimeter size grouping. These lower 

 estimates may be due to the violation of one or 

 more of the above assumptions. Van Sickle's 

 method is very dependent on accurate estimates of 

 growth rates and densities at the boundaries of 

 each size class, and even minor fluctuations in 

 recruitment could affect the estimates of the num- 

 bers in the boundaiy size groups. 



The success of the average annual growth in- 

 crement method of Hancock (1965) is also depen- 

 dent on the assumptions of constant recruitment 

 and growth rate over the size range considered. 

 However, this method is probably not as suscepti- 

 ble to fluctuations in recruitment as that of Van 

 Sickle ( 1977), because the use of broad size classes 

 based on average annual growth increments 

 should smooth out any small fluctuations. The ac- 

 curacy of both these methods may be improved by 

 combining size-frequency distributions obtained 

 over a number of years to reduce the effects of 

 changes in recruitment and growth rates. Con- 

 tinuous monitoring of size-frequency distributions 

 in the Gisborne fishery should result in improved 

 estimates in the future. 



In summary, the analyses of the size-frequency 

 distributions using the three methods gave gener- 



477 



