imize either biological or economic productivity of the 

 resource. 



It is obvious that any tagging experiment should be 

 designed to answer specific questions which are clear- 

 ly stated at the time the experiment is being designed. 

 The suggestions offered here are general and 

 applicable to most tagging experiments regardless of 

 whether their primary objective is to determine 

 migratory routes, rates of travel, rates of exploitation 

 on various stock units, rates of escapement into par- 

 ticular watersheds, racial compositions of populations 

 in various fishing areas at different times, numerical 

 abundance of particular stock components, or any 

 combination of these. 



Establishment of an efficient data collection system 

 is essential. Accurate and complete records of tag 

 releases, recoveries, commercial effort (in terms of 

 standardized units and catches by time-area strata of 

 appropriate size), and timing and abundance of es- 

 capements into various watersheds are needed. 

 Release information should include hydrological and 

 meteorological data. Fish should be released in 

 groups, and conditions prevailing at the time of 

 release should be recorded. 



Exact locations and recapture times for a purse 

 seine fishery can probably be obtained only through a 

 logbook system. Logbooks also should be used to 

 collect data on the number and location of the sets 

 made each day by an individual boat. 



To determine the amount of incomplete recovery 

 and reporting of recaptured tags, trained fishery 

 biologists should sample the catch independently in a 

 manner that does not interfere with normal recovery 

 procedures used by fishermen. Sampling designs for 

 estimating the extent of incomplete reporting are 

 given by Paulik (1961). In some circumstances it is 

 advisable to include double tagging experiments to 

 determine tag loss (GuUand 1963). 



Data collection and record-keeping methods should 

 be standardized, and data forms that allow im- 

 mediate transfer of field data to punch cards should 

 be used. With proper mechanization and data-han- 

 dling procedures, it is feasible to carry out an analysis 

 of partial data from an experiment while the experi- 

 ment is still in progress. Feedback from this type of 

 simultaneous analysis permits an evaluation of ex- 

 perimental techniques that would not otherwise be 

 possible until it was too late to correct faulty 

 procedures. 



A special effort should be made to collect recovery 

 information on a daily basis in the immediate vicinity 

 of release sites. These data can be used to develop cor- 

 rection factors to adjust mortality rate estimates for 

 the bias caused by nonavailability of tagged fish 

 shortly after their release (see discussion of type C 

 error in Ricker 1958, p. 122-126). 



The method of capturing fish for tagging and the 

 types of tags applied should not interfere with the ob- 



jectives of the experiment. Installation of salmon 

 traps at key locations along primary migration routes 

 should be considered for major tagging experiments. 

 These traps would be used for the sole purpose of cap- 

 turing fish for tagging and would assure a fixed point 

 of release and permit closer control of tagging tech- 

 niques than would be possible with purse seines. 

 Traps should also provide holding facilities so that 

 fish could be released in batches of some minimum 

 critical size. Migratory delays should be held to a 

 minimum to reduce the possibility of abnormal 

 behavior. If tagging is carried out from both seines 

 and traps, simultaneous releases should be made to 

 allow a comparison between the two methods of tag- 

 ging. The numbers of tags released at any one time 

 should be as large as feasible, e.g., at least 500 fish, or 

 if a breakdown of recoveries into fine time-area strata 

 is desired, 1,000 fish. The use of re-releases of recap- 

 tured tagged fish and also the possibility of tagging a 

 selected subsample of fish with battery-powered 

 transmitting tags should be investigated for studies of 

 migratory behavior. 



The spatial and temporal distribution of the 

 releases should cover the entire geographical region- 

 time space occupied by the populations being studied. 

 Extrapolation of tagging results to populations found 

 at times and in places not included as part of the tag- 

 ging experiment is generally not advisable. In some 

 cases it may be necessary to make preseason releases, 

 closed-period releases, and postseason releases. 

 Preseason and closed-period releases provide data for 

 separating the fishing mortality rate from the escape- 

 ment rate and for testing various hypotheses about 

 temporal changes in the escapement rate. The 

 possibility of using simultaneous translocation ex- 

 periments to study migratory patterns should be in- 

 vestigated. 



It is advisable, if funds are limited, to make fewer 

 releases and to allow sufficient funds for an adequate 

 recovery program. Depending somewhat on the objec- 

 tives of the experiments, it is usually desirable to con- 

 duct spawning ground surveys in major watersheds 

 and to enumerate spawning populations wherever 

 possible. Even if great effort is expended on tagging, 

 the results of a tagging experiment will be in- 

 conclusive if the recovery effort is inadequate. 



When possible, new experiments should maintain 

 continuity with earlier experiments. Experimental 

 methods should be standardized and changes 

 thereafter should be made only with proper duplica- 

 tion using old and new techniques at the same time 

 and place to obtain correction factors that will allow 

 assimilation of the new findings with knowledge 

 gained from past experiments. 



It has now become technically feasible to integrate 

 a great deal of information concerning the dynamic 

 behavior of a complex system such as the pink salmon 

 fishery in southeastern Alaska into a large-scale com- 

 puter simulation program. Agencies concerned with 



37 



