FISHERY BULLETIN: VOL. 86, NO. 4 



On the other hand, the way that the tag return 

 data are handled causes it to appear that there were 

 more sets with more than one tagged fish than was 

 actually the case. Ideally, all tagged fish would be 

 recovered by fishermen as soon as they are caught, 

 and then set aside for later examination by an 

 lATTC employee, and the tag numbers, locations, 

 and dates of recapture would be recorded so that 

 each fish could be assigned to the proper set. In 

 reality, however, less than half the tagged fish 

 recaptured ax^e recovered by the fishermen, and 

 since the fish from different sets are mixed in the 

 wells of the vessels, the chance of assigning tagged 

 fish to specific sets is lost, except when enough fish 

 are caught in one set to fill an entire well. Virtually 

 all of the tagged fish which are not recovered by 

 fishermen are recovered later by unloaders and can- 

 nery workers. The unloaders and cannery workers 

 usually inform the lATTC employee to whom they 

 return the tagged fish (or the tag without the fish 

 attached to it) that the fish was found in or had come 

 from a particular well or pair of wells of a particular 

 boat. The lATTC employee who receives the tagged 

 fish or tag records this information, and later 

 another lATTC employee compares this information 

 with an abstract of the vessel's logbook and assigns 

 the fish to the set which contributed the greatest 

 weight of fish of the species in question to that well 

 or pair of wells. For example, if a particular well 

 contained fish from sets with 12, 15, 20, and 13 tons 

 made on 1, 2, 3, and 4 June, respectively, and each 

 included one tagged fish, all recovered by unloaders 

 and cannery workers, all four would be assigned to 

 the 3 June set. This would make it appear that the 

 tagged fish tend to remain together more than is 

 actually the case. Another way to handle a situation 

 such as this would be to allocate the four fish among 

 the four sets in proportion to the weights of fish in 

 them, in this case one to each set. This is not done, 

 however, because tagged fish from the same trip of 

 the same vessel are often returned to the lATTC 

 over a considerable period of time, and it is not feas- 

 ible to keep the tags for long periods waiting for all 

 of them to be returned before processing them. Fur- 

 thermore, allocation of tagged fish in the way just 

 described would tend to make it appear that the 

 tagged fish remain together less than they actually 

 do if they are not randomly mixed with the untagged 

 ones. 



In addition to the problems created by failure to 

 recover all the tagged fish as soon as they are 

 caught, there are probably problems created by false 

 data. Sometimes the persons recovering tagged fish 



keep the tags they have recovered over a period of 

 several days, weeks, or even longer, and then return 

 them to an lATTC employee, telling him that they 

 were all recovered that day in the well or pair of 

 wells that was unloaded that day. The likelihood that 

 the data are false can often, but not always, be 

 detected by an alert lATTC employee. When the 

 likelihood that the data are false is not detected, it 

 will appear that the tagged fish remain together 

 more than is actually the case. It is believed that, 

 in spite of attempts to detect data which are likely 

 to be false, some false data are included in the 

 analyses and make it appear that the tagged fish 

 remain together more than is actually the case. 



The fact that the numbers of tag returns were 

 small, coupled with the requirement for the Chi- 

 square contingency tests that the categories for the 

 expected numbers of tagged fish be equal to or 

 greater than 5, tends to make it appear that the 

 tagged fish mix more rapidly with the untagged ones 

 than is actually the case. The biases resulting from 

 the mixing of the fish caught in different sets in the 

 same wells and from false data tend to make it ap- 

 pear that the tagged and untagged fish mix less 

 rapidly than is actually the case. Thus the two fac- 

 tors tend to cancel each other out, at least partial- 

 ly, although the first bias may be stronger than the 

 second. If so, the tentative conclusion made above 

 that the tagged and untagged fish mix thoroughly 

 within about 3 to 5 months may be incorrect; that 

 time could be somewhat longer. For the binomial 

 homogeneity tests only the second bias exists, so the 

 rate of mixing of the tagged and untagged fish in- 

 dicated by these tests is probably somewhat slower 

 than is actually the case. 



Sharp (1978) stated that it is likely that skipjack 

 tuna in the same school are "related," but it seems 

 unlikely that tunas could school together for their 

 entire lives, an implication attributed to Sharp (1978) 

 by Lester et al. (1985). During the egg and larval 

 stages, the fish are at the mercy of their environ- 

 ment, and individuals which were together at one 

 time would often be separated by the currents. Fur- 

 thermore, large tunas of the genus Thunnus occur 

 mostly in subsurface waters at depths to nearly 300 

 m (Suzuki et al. 1977). Although there are areas of 

 greater and lesser concentrations of large, subsur- 

 face-dwelling fish, there is no evidence that they 

 form concentrated schools such as those which oc- 

 cur at the surface. The present study indicates that 

 skipjack tuna in the size range of about 3.4 to 7.0 

 pounds (about 43 to 53 cm in length) of the same 

 school mix randomly with those of other schools 



642 



