in I960 and 1961. As previously pointed out, 

 however, the estimate for 1961 is based on 

 questionable data recovered at age 2. 



Estimates From Tagged Females 



As for males, the tag lost to tagged ratios 

 among females taken on the two islands were 

 compared. The ratios were similar for 4- and 

 5-year-old females, but were higher for 2- 

 year-old females taken on St. Paul Island 

 and higher for 3-year-olds killed onSt. George 

 Island. Only 14 2-year-old females with tags 

 or checkmarks were recovered on St. George 

 Island. This number is not adequate to provide 

 a valid comparison. There is no apparent 

 reason for the difference among 3-year-old 

 females. Because errors, if they exist, could 

 not be identified, changes in the total recoveries 

 were not made. 



Estimates of the pup populations given in 

 table 34 are based on data obtained from the 

 kill of female seals. Estimates of the 1958 and 

 1959 year classes are much higher than are 

 those based on data obtained from the kill of 

 males. The estimates for the I960 and 1961 

 year classes are only slightly higher than 

 estimates based on data from males. 



20,000 marked pups were proportioned to the 

 rookeries according to the distribution of the 

 harem bulls. On most rookeries, however, the 

 allotment was exceeded. 



As in previous years, the pup population was 

 sampled by obtaining a marked to unmarked ra- 

 tio from groups of 25 pups counted at regular in- 

 tervals along lines located on the ground per- 

 pendicular to the water's edge. The population 

 was sampled 19-21 August and again on 26 and 

 27 August to determine the effect of time on the 

 estimate and to determine if the estimates from 

 the two periods were consistent. 



During the second sampling period, counting 

 was more difficult, and the number of available 

 samples decreased because the pups were 

 more active in going to and into the water. 



The estimates from the two periods differ 

 by only 5,625 or 2.4 percent of the average 

 of 229,900 (table 35). The estimates for indi- 

 vidual rookeries differ by amounts varying 

 from 4 to 2 1 percent of the average estimate 

 for that rookery. 



The number of pups per harem bull was 

 determined from the average of the two esti- 

 mates (table 36). There were no exceptionally 

 large differences between rookeries; only 

 for two rookeries did the number of pups 



Table 34. — Estimates of the fall pup population from female tag 

 recoveries, Pribilof Islands, Alaska, year classes 1958-61 



Estimates From Pup Sampling 



Since 1961 the pup population has been 

 estimated from a marked to unmarked ratio 

 obtained by sampling the live pups on all St. 

 Paul Island rookeries. The marked pups used 

 for the ratio in 1961 and in 1962 were those 

 tagged with monel tags during the regular 

 tagging program. The tagged pups, however, 

 were difficult to identify during sampling and 

 were concentrated in certain locations on the 

 rookeries. These disadvantages were elimi- 

 nated in 1963 by shearing the fur from the 

 heads (fig. 6) of 21,919 pups and by distributing 

 the marking effort throughout the rookeries. 

 Also, the pups were marked in late July and 

 early August when they are not yet moving 

 about the rookery and before many have begun 

 to enter the water. Before shearing, a total of 



per harem bull differ from the mean by more 

 than five. 



Table 37 lists estimates of the St. Paul 

 Island pup population for 1961, 1962, and 1963 

 by rookery. Estimates for the 3 years are of 

 the same general magnitude even though the 

 method was changed. 



The live pups on Little Polovina were counted 

 11 August in 1963 as a check on the accuracy 

 of estimates from pup sampling. Beginning 

 at one end of the rookery, successive pods of 

 pups were rounded up and individuals of each 

 released slowly between two observers for 

 counting. Few pups willingly entered the water 

 on 11 August. The counts of the two observers 

 averaged 7,230. This count is believed to have 

 an error of less than + 10 percent. Experience 

 gained in counting live pups should result in 

 less error in the future. The estimate for 



26 



