estimate the number of tags applied to each 

 age class at the time of tagging. Table 25 

 sunnnnarizes the number of tags attached to 

 seals on St. Paul Island in 1966 and recovered 

 on the Pribilof Islands in 1967 and 1968. 



The number of males tagged at ages Z and 

 3 in 1966 can be used for making population 

 estimates. On the basis of recoveries of tags 

 in 1967 and 1968 (table 26), 25,000 3-year-old 

 males from the 1963 year class and 70,000 

 2-year-old males from the 1964 year class 

 were still living after the kill in 1966. The 

 estimates for both year classes from re- 



coveries in 1968 are much higher than esti- 

 mates from recoveries in 1967. Possible 

 causes for the discrepancy in the estimates 

 between years for a given year class are 

 that some of the animals that lose both tags 

 are not recognized and the number of aninials 

 with both tags lost would be greater 2 years 

 after tagging than 1 year later. 



The estimates of 25,000 males age 3 and 

 70,000 males age 2 remaining after the kill 

 in 1966 seem reasonable. A more critical 

 appraisal of the results will be possible when 

 additional data become available. 



FORECAST OF THE KILL OF MALE SEALS 



Forecasts of the number of male seals in 

 the kill have been made each year since 1959. 

 The accuracy of the various prediction methods 

 was reviewed in 1966 (Marine Mammal Biolog- 

 ical Laboratory, 1969), but no quantitative com- 

 parisons were made. Quantitative comparisons 

 for past data, forecasts of the kill of 3- and 4- 

 year-old males for St. Paul Island in 1969, and 



forecasts of the kill of naales 2 to 5 years 

 old for the Pribilof Islands in 1969 are dis- 

 cussed in this section. 



QUANTITATIVE COMPARISONS FOR 3- AND 

 4 -YEAR -OLD MALE SEALS 



Table 27 gives the results of five methods 

 of forecasting the kill of 4-year-old males. 



Table 27. —Comparisons of forecast!/ of kill of -i-year-old male seals, by 

 method, St. Paul Island, 1965-68 



Method 



Year 



1965 



1966 



1967 



1968 



-Thousands- 



Regression on kill of 3-year- 

 olds and mean round 14. 3(+) 



Regression on mean air 

 temperature 13. 5(+) 



Regression on pup weight 



10. 2(-) 14. 1(+) 10. 5(-) 



19.7(++) 20. 0(++) 4. 3(--) 



9. 2(-) 



Regression on the number of 

 pups born and count of dead 

 pups 



17.8(+) 



Forecast derived from 

 yearling estimate 



Combined estimate 



Actual kill- 



Not 



2/ 



1/ Erred positively by a moderate (+) or large (++) amount or negatively by 

 a moderate (-) or a large ( — ) amount. 



2/ The yearling estimate used was incorrect because no a]J.owance was made 

 for double tag-lost recoveries. If the tag-lost recoveries were considered, 

 the forecast would have been 12.0, which like the other forecasts derived from 

 yearling estimates is very accurate. 



3/ The season for killing males ended 9 August in 1965, 5 August in 1966 and 

 1967, and 2 August in 1968. 



22 



