Table 3. — Optimum allocation requh-ed for 0.15Y, 90% 

 confidence limits in 1967 compared with true optimum 

 allocation of sample size for 1968. 



Strata 



1967 

 Allocation 



1968 

 Allocation 



n h 1967 

 n 



n h 1968 



n 



I 



20.9 



3.9 



.073 



.021 



II 



13.2 



2.9 



.016 



.016 



III 



6.9 



3.7 



.021 



.021 



IV 



1.3 



6.2 



.015 



.035 



V 



32.9 



25.1 



.115 



.110 



VI 



39-6 



12. H 



.138 



.069 



VII 



12.2 



9.9 



.012 



.055 



VIII 



26.0 



32.8 



.091 



.183 



IX 



32.6 



11.2 



.111 



.230 



X 



29.1 



HO. 8 



.102 



.228 



XI 



13-5 



— 



-151 



— 



XII 



25-7 



— 



.089 



— 



n 



287.2 



178.9 







of the days by year from 1966 (partial year), 

 through 1970 (Table 4). 



These estimates include many catch, effort, 

 and catch-per-unit-of-effort categories that are 

 not reported in "Maine Landings." The com- 

 parable estimates of catch in pounds, numbers, 

 and number of traps that are reported in 

 "Maine Landings" must exceed the estimates 

 from the survey because of the necessary con- 

 straints of the sampling period and the fact 

 that we cannot efficiently sample individual 

 fishermen who retail their catches. 



Aside from the absolute need of detailed 

 catch, effort, and catch-per-unit-of-effort data 

 in order to make management recommenda- 

 tions, the expanded estimates might have the 

 following additional useful purposes: 



(1) Gulland (1965) and others have advocated 

 the use of catch-per-unit-of-effort sub- 

 samples in relation to the actual total 

 catch (as reported in "Maine Landings") 

 in order to estimate the total effort in 

 more pertinent categories than just the 

 number of traps. 



(2) The survey totals by month or year could 

 serve as indices by category of what 

 actually occurs in the entire fishery. 



(3) These indices after a series of years 

 might make it possible to again compile 

 a figure of total catch with effort by year 



with the juxtaposed regulations and then 

 make some meaningful determinations 

 about the fishery, particularly since this 

 effort could be in several categories 

 rather than the only previously available 

 category of number of traps. 



Cluster Samples 



The cluster samples of 10 lobsters per boat 

 are vitally important to this study not only for 

 the lengths, but also for the weights, and per- 

 centages of females, culls, and shedders. All 

 of these categories have varying degrees of 

 importance on the assessment of the popula- 

 tion. The following sections demonstrate how 

 each category is used. 



Length frequency analysis. — In this 

 paper, lobster lengths are the basic building 

 blocks for estimating most population param- 

 eters. With this degree of importance, we 

 included the compilation of the number of 

 lobsters by size, sex, month and year (Table 

 5). These data will also make it possible for 

 the reader to make any other determinations 

 that he wishes. 



We used actual numbers or percent frequen- 

 cies to analyze the data in two ways: (1) 14% 

 groupings of length and (2) 1-mm increments 

 of length with the probability method. 



Analysis by 14% increments. — I chose 14% 

 increments because they closely approximate 

 the calculated percent increase in carapace 

 length with ecdysis for legal-sized lobsters 

 from the premolt and postmolt section and from 

 the study by Wilder (1953). 



On this basis, we separated the carapace 

 lengths in millimeters into groupings of 81 

 through 92, 93 through 106, 107 through 122, 

 and 123 through 127 (the legal maximum size 

 in this State). It is not logical to assume that 

 an age or molt group starts at 81 mm rather 

 than extending below this size. I will discuss 

 this in the section comparing 14% increments 

 with the probability modes. 



Silliman (1943), Beverton and Holt (1957), 

 and Ricker (1958) have discussed the assump- 

 tions that must be met when using length fre- 

 quencies in place of the age composition. We 

 cannot determine the age of any lobster that 



13 



