The regression determined from the data 

 is: 



Y= -15.10 + 0.60 xi + 5.29 x 2 



where xi = kill of 3-year-old males to 31 July 



X£ =mean date of the 3-year-old male 

 kill in days past 15 July 



Y = kill of 4-year-old males to 31 July, 

 plus 80 percent of the 3-year-old 

 male kill after 31 July 



This regression, which is based on 1 year 

 of additional data, is very similar to that 

 given in 1963 (Roppel, Johnson, and Chapman, 

 1965), and to two decimals, R2 is 0.82 as it 

 was in 1963. 



For the 1961 year class, which will be 4 

 years old in 1965, x, = 18, x-> = 4, so that 

 Y = 16.9. 



Deleting 3,900 (80 percent of the 3-year 

 male kill of 1964 after 31 July) and adding 

 10 percent for the 4-year male kill expected 

 in August 1965, yields an estimate of 14,300 

 for the 1965 4-year-old male kill. 



Method 2 . --Temperature-return regression. 

 The basic data for this regression are given 

 in appendix table 2. 



The regression is: 



Y = 16.0 + 0.99 T 



where Y = adjusted kill 



T = mean temperature in tenths of a 

 degree above 32° 



Appendix table 2. --The 3- and 4-year-old male kill and mean 

 temperature, St. Paul Island, year classes 1950-60 



Year 

 class 



Adjusted 

 kill 1 



Temperature 



1950 

 1951 

 1952 

 1953 

 1954 

 1955 

 1956 

 1957 

 1958 

 1959 

 1960 



Kill prior to 31 July, plus 80 percent of the number of 3- 

 year-olds taken in August. 



2 Mean temperature for 12-month period ending 30 June of 

 the indicated year, measured in tenths of a degree above 32°. 



For 1961, the value of T is 18, so that Y = 

 33.8. The kill of 3-year-old males in 1964 

 prior to 31 July was 17,600, and 80 percent 

 of the kill in August (4,861) is 3,900, so that 

 there remains for 1965 (to 31 July), 12,300. 

 An additional 10 percent expected in August 

 1965 brings the total kill of 4-year-old males 

 to 13,500. 



Method 



_3.- -Estimates from the yearling 



population estimate. Of 621 male yearlings 

 tagged or retagged in 1962, 259 were recov- 

 ered in 1963 as 2-year-olds or in 1964 as 

 3-year-olds. Additionally, in 1963 and 1964, 

 nine were recovered as 3- and 4-year-olds, 

 respectively, indicating a small degree of 

 error in the estimating of ages during tagging. 

 These nine seals were the survivors of 2 

 years' natural mortality at a rate of approxi- 

 mately 10 percent per year. Also, the 3-year- 

 old kill is approximately equal to the 4-year- 

 old kill plus escapement, so that it may be 

 estimated that these 9 suggest that a total of 

 20 of the original 621 were older than 1 at the 

 time of tagging. The total 2-year-old kill on 

 both islands in 1963 was 2,019, and the 3-year- 

 old kill in 1964 totaled 29,416. Combined, 

 these kills total 31,435. Hence, the estimate 

 of the number (N) of yearlings from both 

 islands alive in 1962 is: 



N 



1962 



(31,436)(602) 

 260 



72,786 or 72,800 



Reducing 72,800 by 10 percent for the annual 

 mortality from age 1 to age 2 (a very rough 

 estimate) and the 2-year-old kill, and further 

 reducing it by 10 percent for the next year's 

 natural mortality and a kill of 29,400, leaves 

 a balance of 27,800. A further overwintering 

 mortality in 1964-65 of 10 percent and an 

 escapement of about 10,000 imply a balance of 

 15,000 for the 4-year-old male kill in 1965. 

 Eighty percent of this is 12,000, which would 

 be the St. Paul Island estimate by this method. 

 Despite the obvious inadequacies of the data 

 used in this method, it agrees very closely 

 with the previous two estimates. 



Prediction of 3-year-old Male Kill 



In the past several years, the prediction of 

 the 3-year-old male kill has been based on 

 one or more of (a) the population- return equa- 

 tion, (b) the dead-pup count-return equation, 

 and (c) the temperature-return equation. In 

 addition, attempts have been made to utilize 

 the 2-year-old male kill data on the cumula- 

 tive estimates of the female population for 

 forecast purposes. These have beenunsuccess- 

 ful, though some additional discussion on the 

 cumulative estimate and its relationship to 

 the 3-year-old kill is given below. Also, a 

 forecast based on the return from yearling 

 tagging is obtained. 



27 



