RALSTON and POLOVINA: COMMERCIAL DEEP-SEA IIANDLINE FISHERY 



is most conveniently formulated as instanta- 

 neous fishing mortality (F), measured over some 

 arbitrary interval of time, usually 1 yr (Ricker 

 1975). Frequently it is not possible to measure F 

 directly, however, and so a proportionate mea- 

 sure of F is selected, i.e., fishing effort or /. 

 The ideal choice of units for fishing effort results 

 in a linear correspondence between F and /, a 

 zero intercept, and minimal residual variance 

 (Rothschild 1977). Because F is frequently 

 unknown, it is often difficult to ascertain 

 whether these criteria are met and yet the 

 selection of an appropriate measure of fishing 

 effort is most critical to meeting the assumptions 

 of a stock-production analysis. Ample considera- 

 tion should be given to these factors before the 

 data are collected. 



No attempt has been made in the HDFG data 

 to record either the fishing effort or the fishing 

 power of individual fishermen. A suitable 

 measure of fishing mortality in this fishery 

 would be the cumulative number of hook-hours 

 or line-hours of fishing. While such figures are 

 currently unavailable, it has been possible to 

 determine the total number of fishing records 

 filed in a year which report the catch of a 

 particular species. This statistic, the number of 

 daily reports by fishermen who have caught any 

 one particular species, was frequently computed 

 and is termed catch-records. Figure 1 presents, 

 in addition to the catch, the total number of deep- 

 sea handline catch-records filed from 1959 to 

 1978 concerning all 13 species of bottom fish. 

 This measure of fishing effort does not always 

 correspond to the number of fisherman-days 

 because one operator may catch several species 

 during a single day of fishing. In this instance the 

 reporting of each particular species comprises 

 one catch-record. Thus, when aggregated spe- 

 cies groups are considered, the number of fish- 

 erman-days will always be fewer than the total 

 number of catch-records. When species are con- 

 sidered independently of one another the two fig- 

 ures are equal (catch-records = fisherman-days). 



Interpreting the meaning of a fisherman-day 

 as a unit of fishing effort in this fishery is 

 difficult. It was tabulated by following the daily 

 reports of individual fishermen, identified by 

 their commercial fishing license numbers. All 

 commercial fishermen in Hawaii, whether 

 captain or crew, must have a license. It is likely 

 that many catch reports are filed only by boat 

 captains who document the landings of an entire 

 fishing vessel, which may have a variable 



number of crew members. Thus a fisherman- 

 day, as defined here, may reasonably be thought 

 of as a vessel-day. However, because this unit of 

 effort is defined and specified on the basis of 

 commercial fishing licenses, in the interests of 

 exactitude, we have chosen to use the term 

 fisherman-day. 



RESULTS 



Clustering 



The usual method of aggregating catch data 

 would be to pool all 13 deep-sea handline species 

 into a single group and to analyze the total with 

 the TBSM. An alternative is to employ a 

 multivariate statistical procedure to assess the 

 degree of colinearity among species and to define 

 species groups based on the strength of inter- 

 species associations in the catch (Pope 1979). 

 Such an approach would identify those bottom 

 fish which tended to appear with one another in 

 the catch to the exclusion of others and would 

 measure the extent of correlation of fishing 

 mortality among species. Pope (1979) has termed 

 this "technological interaction" and has dis- 

 cussed its importance in multispecies fisheries. 

 Separate application of the TBSM to each species 

 group formed by clustering would constitute an 

 analysis performed at an intermediate level of 

 species aggregation. Conceptually this is desir- 

 able because in the Hawaiian offshore handline 

 fishery different species are known to exhibit 

 stratification by depth (Strasburg et al. 1968). 



Cluster analyses were performed with a com- 

 puter routine (Dixon 1977, program P1M) 

 where the 13 species of bottom fish comprised the 

 variables to be clustered and the catch from a 

 single day's fishing formed one case. Associa- 

 tions were computed on the basis of the landed 

 weight of each species. The average linkage 

 between groups defined the criterion for 

 amalgamating clusters and correlation coeffi- 

 cients were used as measures of similarity. 



Separate analyses were performed for each of 

 the four designated bank (Table 2) areas to assess 

 whether obvious differences exist among banks 

 with regard to interspecies associations. Simi- 

 larly, separate analyses were conducted for the 

 years 1959, 1965, 1971, and 1977 to see whether 

 temporal variation in species grouping is an im- 

 portant factor to consider. 



No striking differences or patterns emerged 

 from these various comparisons. The intrinsic 



439 



