258 



NORTH AMERICAN FAUNA 74 



that, whereas there may have been no conscious selection for females with 

 calves, these samples probably were drawn from herds in which such animals 

 prevailed. Because there is some mortality of fetuses, birth rate should tend to be 

 lower than pregnancy rate in representative samples. Data on rates of repro- 

 ductive success have indicated that only about 83 % of ovulations and 95 % of 

 conceptions result in live births (Table 35). 



Samples B and C in Table 47 are known to have been drawn mainly from 

 "nursery herds," which would account for their apparent overrepresentation of 

 cows bearing calves. I am confident that the fifth sample (E) was not biased in 

 the same way, because it was taken early in the spring, before the nursery herds 

 were formed. The probability of bias in the two samples from the Chukchi Sea 

 (A, D) is unknown to me, but I suspect that overrepresentation of cows bearing 

 calves is likely to be consistently high in harvests taken during spring and 

 summer, simply because exploitation of large rather than small herds is most 

 efficient in returns vs. time expended. 



These are the best empirical data available for estimation of gross produc- 

 tivity, but the birth rates suggested by them probably are higher than those of the 

 adult female population at large, because of overrepresentation from the nursery 

 herds. The pregnancy rates probably were not affected by the same bias, 

 however, for pregnant females tend to be less gregarious and are uncommon in 

 the nursery herds. Nevertheless, the pregnancy rates require some adjustment, 

 because they were not derived by the same methods. The age compositions of the 

 samples also need to be compared, since both fertility and fecundity vary with 

 age (Krylov 1962; Burns 1965). A sample made up mainly of young adults will 

 tend to show a higher pregnancy rate than one made up of older animals (Tables 

 34, 35). 



The age compositions of samples B, C, and E of Table 47 are known in part; 

 that of D is unknown; the composition of sample A can be derived from the given 

 body and tusk size classes relative to body and tusk growth curves (Figs. 19 and 

 81). Comparison of the age compositions of samples A to E with an ostensibly 

 random sample of females reported by Krylov (1968: Table 7), indicates 

 (Table 48) that they are sufficiently similar to be accepted as representative of 

 the females of breeding age. I assume that sample D also is representative. 



Table 48. Comparison of age compositions of the samples in Table 47 with an 

 ostensibly random sample of females of breeding age reported by Krylov 

 (1968). 



Sample and (in parentheses) sample size 



A B C E Random 



(110) (20) (153) (33) (32) 



ass (years) 



No. 



% 



No. 



% 



No. 



% 



No. 



% 



No. 



% 



6-9 



24 



22 



9 



45 



38 



25 



5 



15 



7 



22 



10-15 



39 



35 



9 



45 



70 



46 



18 



54 



13 



41 



16-20 



31 



28 



1 



2 



30 



20 



7 



21 



8 



25 



>20 



16 



14 



1 



2 



15 



10 



3 



9 



4 



12 



