286 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



Each year is divided into the more productive 

 and less productive parts; for convenience these 

 are called summer and winter, respectively (rows 

 1 and 2 for each boat and year). The number of 

 trips per biweekly period (col. 7) and number of 

 day's fishing per biweekly period (col. 9) are 

 greater during summer, but the ratio of trips to 

 day's fishing (col. 10) is not markedly different in 

 the two seasons, although trips are somewhat 

 longer in winter. 



Nonproductive trips (col. 5) occur with greater 

 frequency (col. 6) in winter than in summer, and 

 there is a tendency for the number of productive 

 trips per day's fishing to be greater in summer. 

 Differences in the number of productive trips per 

 day's fishing are also apparent between years, as 

 in 1953 there were generally more productive trips 

 per day's fishing than in 1952. 



To summarize the performances of the two 

 boats, it appears that the ratio of productive trips 

 to the nvmiber of day's fishing is greater during 

 times of good fishing and smaller during times of 

 poor fishing. Since trips are shorter when the fish- 

 ing is good and there are also fewer nonproductive 

 trips, the actual effort in terms of day's fishing will 

 usually be underestimated dui"ing the periods of 

 poor fishing as compared witli periods of good 

 fishing. 



Variations from the official number of crew will 

 also affect the accuracy of the estimate of fishing 

 effort. The official number is a maximum and 

 variations will usually mean that fewer than the 

 official number are aboard. In the Hawaiian skip- 

 jack fleet, boats ordinarily carry the maximum 

 number of crew during the summer season, after 

 which some men leave to find other employment. 

 In this study, since the official (maximum) num- 

 ber of crew has been used throughout the year as 

 a weight for the individual trip, the "fisherman" 

 factor in the productive fisherman-trip is probably 

 overestimated during the times of poor fishing. 



Thus, the biases in the productive fisherman-trip 

 between times of good and poor fishing tend to 

 cancel each other because during the winter season 

 and years of generally poor fishing, longer trips 

 and the increased frequency of nonproductive 

 trips cause an underestimation of the actual time 

 spent fishing while, at the same time, the actual 

 number of fishermen on the boat is likely to be 

 fewer than the official number. The converse will 



hold true during the summer season and in years 

 when good fishing attracts the maximum number 

 of fishermen to the fleet. 



Information is not available to permit the ex- 

 amination of the actual variations in the number 

 of fishermen and the extent to which they offset 

 the bias introduced by nonproductive trips, but if 

 boat ^4 is assumed to represent the average situa- 

 tion, the number of productive trips per day's fish- 

 ing (col. 11) appears to be about 10 percent greater 

 in summer than in winter. Since the crew of the 

 average skipjack boat is about 10 men, the absence 

 of one of these men on the average during the 

 winter season represents a 10-percent overestima- 

 tion of the number of fishermen. Thus, the ab- 

 sence of one fisherman per boat during the winter 

 season would be sufficient to equalize the bias in 

 the productive trip factor introduced during the 

 winter season. 



SOURCES OF ERROR 



Unreported catches or forms containing incom- 

 plete or inaccurate information are an obvious 

 source of error. Yamashita (1958, p. 258) esti- 

 mates that the reported portion of the 1952 catch 

 included di percent of the pounds, but only 88 per- 

 cent of the trips, indicating a bias in favor of the 

 reporting of large catches. Since small catclies are 

 most likely to occur in the slack part of the year, 

 there may be a tendency for an estimate of the fish- 

 ing effort, which is a function of the number of 

 trips, to be correspondingly reduced. 



Inaccurate information is difficult to detect 

 without data from other sources with wliicli to 

 compare the catch records. On the basis of inter- 

 view records, Yamashita (1958, p. 258) estimates 

 that only 45 percent of the statistical areas indi- 

 cated in the 1952 catch reports were reasonably 

 accurate. By means of bi-oad geographical divi- 

 sions to summarize the data (fig. 4), it is assumed 

 that the effects of such erroneous information will 

 be minimized. 



ERROR IN DETERMINATION OF FISH SIZE 



Dividing the total weight caught by the esti- 

 mated number as indicated in the catch report, 

 yields the average size of fish caught, but provides 

 no indication of the range or variability of sizes. 

 Since the entire catch is assigned to eitlier the 

 small or large category on the basis of the average 



