PARRACK: FISHING EFFORT FROM AERIAL SEARCH DATA 



(Table 2) as theoretically should occur. Therefore, 

 when countries change fishing patterns from one 

 time period to the next so that PiF) differs, 

 P(F/N) also changes, thus introducing error in the 

 estimates made by estimator III, a condition also 

 true for P{G) and P{G/N) on which estimator II is 

 based. This results from changes in the mean 

 number of days fished (or days on grounds in the 

 case of estimator II) per vessel, a likely occurrence 

 if a particular fleet experiences difficulties in 

 finding fish, if weather conditions are unusually 

 unsuitable for fishing, or if equipment repair or 

 modifications demand excessive lost time in a 

 certain time period. 



Although these changes theoretically should not 

 produce changes in the ratio of reported to ob- 

 served fishing effort {R, estimator I), other factors 

 can conceivably produce such variation of that 

 ratio. Changes in visibility due to weather can 

 likely be an important factor. If fog or other 

 visibility-restricting weather conditions are more 

 prevalent in one time period than another, R may 

 be expected to be larger during that period. 

 Likewise, varying success of overffights in locat- 

 ing fleet concentrations is a factor. Unusually 

 successful searching may be expected to produce 

 ratios smaller than average while low success will 

 tend to increase R. 



In addition, changes in the accuracy of reported 

 effort (/andgr) will result in corresponding changes 

 in the accuracy of calculations of P{F/N), P{G/N), 

 R, and K for particular time periods. Since 

 reporting accuracy cannot be measured, such 

 deviations have been included in the error 

 coefficients as have the above listed sources of 

 error. 



Although a method of calculating probability 

 limits on estimates is presented, the methodology 

 utilizes the observed past performance of each 

 estimator to establish the probability of error. It 

 must be assumed, therefore, that the frequency 

 distribution of estimation error is correctly repre- 

 sented by these past data. Although this assump- 

 tion can reasonably be made if fisheries surveil- 

 lance flight patterns and fishing fleet movements 

 are generally constant, caution should be exercised 

 in this regard. If flight patterns or seasonal fleet 

 movements change drastically, the probabilities of 

 not sighting fishing effort (estimators II and III) 

 and the ratio of reported to sighted effort (es- 

 timator I) will likewise change so that they are not 



correctly represented by the range of past values. 

 Aberrant values will result if the fleets are exten- 

 sively concentrated in different areas than in the 

 past. Fleets will not be located by fisheries sur- 

 veillance flights as well as in the past and, there- 

 fore, effort will not be observed to the same extent 

 as in the past. As a result, values of PiF/N), and R 

 will be much greater than past values. Sizable 

 underestimates of days fished will occur with 

 probabilities greater than those represented by 

 past error frequencies. Conversely, if fleet loca- 

 tions are anticipated by surveillance flight per- 

 sonnel much more accurately than in the past, 

 these estimation constants will be much smaller 

 than represented by past data, so that probabili- 

 ties of overestimation will be much greater than 

 represented by past performance data. 



ACKNOWLEDGMENTS 



All overflight data utilized in this study was 

 collected by the National Marine Fisheries Ser- 

 vice, Law Enforcement and Marine Mammal 

 Protection Division, Northeast Region. I extend 

 my thanks to Charles Philbrook and William Beers 

 of that agency for their advice as to the contents of 

 those data. I am indebted to Judith Brennan and 

 Bradford E. Brown of the Northeast Fisheries 

 Center, National Marine Fisheries Service for 

 their advice in the development of these meth- 

 odologies and manuscript. I am especially indebt- 

 ed to Robert L. Edwards and Richard C. Hen- 

 nemuth, also of the Northeast Fisheries Center for 

 the continued support and advice which made this 

 study possible. 



LITERATURE CITED 



HOEL, P. G. 



1962. Introduction to mathematical statistics, 3rd ed. John 

 Wiley and Sons, Inc., N.Y., 427 p. 

 Pearson, E. S., and H. 0. Hartley (editors). 



1956. Biometrika tables for statisticians. Vol. I. Cambridge 

 Univ. Press, Lond., 238 p. 

 Scheffe, H. 



1959. The analysis of variance. John Wiley and Sons, Inc., 



N.Y.,477p. 



SlEGEL, S. 



1956. Nonparametric statistics for the behavioral sci- 

 ences. McGraw-Hill, N.Y., 312 p. 

 Steel, R. G. D., and J. H. Torrie. 



1960. Principles and procedures of statistics with special 

 reference to the biological sciences. McGraw-Hill, N.Y., 

 481 p. 



513 



