pups by tagging small animals in late Sep- 

 tember and early October (see section on mark- 

 ing) . Population estimates based on recoveries 

 of tagged animals subsequently killed on the 

 Pribilof Islands are discussed in this section. 



Our data are not complete for all males re- 

 covered that were or had been tagged. We 

 could not, for example, determine the age of 

 a seal after the head (and canine teeth, on 

 which age is based) had become separated from 

 the carcass. The ages of seals that had lost 

 both tags (as recognized by a tag scar on each 

 front flipper) were determined, but there was 

 no way to ascertain the tag series applied to 

 these animals. Therefore, we adjusted for 

 these incomplete data before making popula- 

 tion estimates (1) by distributing animals of 

 unknown ages within each tag series according 

 to the age distribution for the series and (2) 

 for double-tag loss by adjusting the number 

 of tags of each series applied. The latter ad- 

 justment differed from that used in 1968 (Ma- 

 rine Mammal Biological Laboratory, 1970b) 

 when males that had lost both tags were dis- 

 tributed in proportion to the number with com- 

 plete data by age and tag series. The pro- 

 portion of the animals that were actually age 

 1 when tagged was determined from the age 

 distribution of the tagged animals recovered. 

 For example, of 809 males believed to be year- 

 lings and marked with 1-R series tags in 1965, 

 301 were recovered through 1969. Of the lat- 

 ter, 280 were actually age 1 when tagged. Thus, 

 the estimated number of yearling males tagged 

 in 1965 is (280/301) (809) = 753. If we add 

 64 animals that had been tagged as pups in 

 1964 and were given another tag in 1965, the 

 total number of males that were actually year- 

 lings was 817. The rate of tag loss among 

 lR-series recoveries was 0.33. To compensate 

 for loss of both tags, we adjusted the number 

 of yearling males tagged by multiplying by 

 the factor l-(0.33) 2 . Therefore, 728 yearling 

 males were effectively tagged with lR-series 

 tags in 1965. The number of effective tags is 

 used for population estimates. 



Population estimates based on recoveries of 

 tags applied to males older than pups are given 

 in tables 15 and 16. Except for 2T-series data, 

 the estimates for a given tag series increase 

 with an increase in age at recovery. Increased 



tag loss may be responsible for the apparent 

 rise in the estimates, but the data in table 7 

 do not indicate that it has been increasing suf- 

 ficiently to account for large decreases in re- 

 covery rates. Behavior, tag-caused mortality, 

 and selection for tags during killing are other 

 possible factors. Until we know what is caus- 

 ing a decrease in the recovery rate we can have 

 little confidence in population estimates based 

 on tag recoveries. Data to be collected in 1970 

 and 1971 from the 1965 and 1966 year classes 

 may help us to understand why the recovery 

 rates have changed. 



Ancel M. Johnson 



FORECAST OF THE KILL OF 

 MALE SEALS IN 1970 



Our forecasts of the male kill for 1966 and 

 1967 were extremely accurate, but subsequent 

 forecasts have been less satisfactory. For ex- 

 ample, our forecasts of 49,000 for 1968 and 

 57,000 for 1969 exceeded the actual kills by 

 5,000 and 18,000, respectively. We should, 

 therefore, ask whether these errors represent 

 chance variations in the forecast and merely 

 reflect the fact that our methods of forecasting 

 are still rather imprecise, or whether there has 

 been a change in the relation of the kill to basic 

 factors used in forecasting. The latter, if true, 

 might have been caused by undetected climatic 

 changes, an unanticipated result of our manip- 

 ulation of the population, or a reduction in the 

 amount of food available to the seals as a result 

 of the vast trawl fishery that moved into the 

 Bering Sea during the past decade. 



A necessary condition for accurate fore- 

 casting is that we have observations of a factor 

 or group of factors to be used as independent 

 variables that accurately reflect survival of fur 

 seals to age 3. Only the estimates from tags 

 applied to yearling males seem to fall in this 

 category. The variability in these estimates, 

 however, and the small number of recoveries 

 at age 2 decrease considerably the value of 

 yearling tagging as a way to provide a sub- 

 stantial increase in forecasting accuracy. 

 Therefore, unless the results from yearling tag- 

 ging improve or we find additional ways to 

 predict, it is unlikely that we will be able to 

 forecast with confidence. 



18 



