Myers, and Tampa, Fla. Estimates of the total 

 effort expended weekly on the Tortugas grounds 

 had to be projected from sample interviews (n) 

 with trawler captains landing their catches at 

 these ports. The procedure simply consisted of 

 estimating from the sample data the average 

 number of hours fished during each fishing trip 

 and multiplying this by the total number of 

 trips as determined by a canvass of processing 

 plant records. Thus, the total hours fished in any 

 week is given by 



A }\J n 



X=Nx=- x; x, 



n i 



i=l,2,3, . . .n 



where ./Vis the total number of trips by individual 

 trawlers to the fishing grounds, x ( is the number 

 of trip-hours fished as reported by the captain of 

 the i*" vessel entering a weekly sample, and x — 



n 



- is the sample estimate of the mean number 



of hours actually spent fishing during each trip. 

 The only difficulty arose when "uninterviewed" 

 trips (N—n) extended over periods of more 

 than 1 calendar week. In such cases, fractional 

 trips for each of any 2 adjoining weeks had to be 

 designated. These fractions were then summed 

 together with trips contained entirely within 

 respective weeks to obtain the weekly totals. 

 Variances of the sample estimates were also 

 calculated and, under the assumption of negligible 

 or constant sampling bias, provided the estimates 

 of total effort with an index of reliability. Hence 



r(x} = n(n-l) (-AT*) 



and 



a A . 



confidence interval (A r ): X±tN\r(r) 



where t is the value of the normal deviate corre- 

 sponding to the desired confidence probability. 

 The Tortugas experiment lasted about 14 

 weeks; table 2 gives effort statistics and esti- 

 tnates, together with total landings, for each 

 week plus t lie two that immediately preceded the 

 -tart of the experiment. To provide some indi- 

 cation of how the effort was distributed on the 

 fishing grounds, the weekly totals are also appor- 

 tioned (on the basis of interview data) among three 



well-defined depth zones. Landing figures are 

 not estimates but accumulations of actual catch 

 weights obtained through a continuing canvass 

 of all processing plants. 



One other matter that will warrant considera- 

 tion in a later section relates to the stage in age 

 group development at which the member shrimps' 

 commercial acceptance becomes complete, or 

 selectivity no longer constitutes a factor. As 

 employed here, the word "selectivity" does not 

 refer to the selective properties of fishing gear 

 but means, rather, the purposeful avoidance of 

 shrimp of undesirable size even though they may 

 be largely vulnerable to the gear in general use. 

 Such selection is ordinarily accomplished by 

 sorting catches or by test fishing for aggregations 

 in which most individuals are of marketable size. 



Unfortunately, the minimum marketable size 

 of shrimp varies within the industry (or according 

 to State law) from area to area around the Gulf, 

 and in some localities may be only vaguely 

 defined (or not stringently enforced). During 

 the Tortugas experiment the fishery did not 

 recognize a specific minimum size, with the result 

 that sorting catches for the purpose of discarding 

 small shrimp was not practiced. In fact, the 

 only form of selection evident was the weekly 

 expenditure of comparatively small amounts of 

 fishing effort prior to the period in which the 

 greater proportion of available shrimp reached a 

 size of about 70 headless-count. 



Examination of size (weight) frequency curves 

 obtained from landings during the study period 

 disclosed that selective fishing solely for medium 

 and large shrimp was, practically speaking, in- 

 operative (fig. 3). It suggested further that 

 shrimp somewhat below the size designated herein 

 as the minimum acceptable size (70 headless- 

 count) were, if not entirely, almost wholly vul- 

 nerable and actively sought out by a sizeable 

 though undetermined portion of the fishing fleet. 

 Figure 3 clearly shows that small (in many 

 instances, very small) shrimp predominated in 

 Tortugas catches during the last 15 weeks of 1961 . 

 This fact lends credibility to the assumption that 

 bias due to the selective action of the fishing gear 

 was negligible and that the retrieval of marked 

 shrimp of a size well below that tacitly set as a 

 commercial minimum was almost entirely governed 

 by the probability of a piece of gear encountering 

 and capturing them, even though such probability 



318 



U.S. FISH AND WILDLIFE SERVICE 



