Appendix A 



PREDICTION OF 1965 MALE KILL 



Douglas G. Chapman 

 16 November 1964 



Quantitative methods of predicting the an- 

 nual male fur seal harvest were first discussed 

 in 1959. 1 This discussion was concerned both 

 with preseason and inseason predictions. Sub- 

 sequently, only preseason predictions were 

 studied and actually made for the seasons 

 1961 to 1964. Several methods were used in 

 this prediction analysis. The predictions for 

 most years were only moderately successful, 

 though the forecast for 1964 was the most 

 accurate yet made. Nevertheless, the predic- 

 tion errors still are quite large. Some attempt 

 is made to assess these errors, but this is 

 difficult because each prediction may be a 

 composite of several partially independent 

 predictions. Moreover, various "adjustments" 

 have to be made in previous data for varying 

 termination dates of the kill and length limits, 

 and additional adjustments must be made in 

 the predictions to extrapolate St. Paul Island 

 estimates to both islands. 



As is obvious, the forecast of the 4-year 

 male kill is relatively easy, since the kill of 

 the 3-year-olds from this year class provides 

 a substantial amount of data on which to base 

 the forecast. The prediction of the larger 

 3-year-old component is based on much more 

 tenuous data; thus the errors of prediction for 

 a kill of this age class are much larger. 



Prediction of 4-year old male kill 



Predictions of 4-year-old male kills in 

 former years have been based on one or more 

 of (a) the estimate of 3-year male escapement, 

 (b) the population- return equation, (c) the 

 dead-pup count- return equation, and (d) the 

 temperature-return equation. 



Chapman (1964) showed that the estimates 

 of escapement based on estimating the return 

 of 3-year males after termination of killing and 

 the proportion of undersize or oversize ani- 

 mals were inconsistent with the estimates of 

 recruitment to the population of adult males. 

 Consequently, the escapement estimate was 

 replaced by a dir-ect regression of 4-year 

 male kill on 3-year male kill and the mean 

 date of the latter. 



1 Douglas G. Chapman [p. 49]. In Carl E. Abegglen, 

 Alton Y. Roppel, and Ford Wilke. Alaska fur seal investi- 

 gations, Prlbilof Islands, Alaska. Report of Field activ- 

 ities, June-September 1959. Bureau of Commercial Fish- 

 eries Marine Mammal Biological Laboratory, Seattle, 

 Wash. [Processed.] 



An analysis made in 1963 (appendix table 3; 

 Roppel, Johnson, and Chapman, 1965) showed 

 that the dead-pup count yielded essentially no 

 additional predictive information over the 

 temperature data, and thus this regression 

 was also eliminated. 



New data available in 1964 may be useful 

 for forecasting purposes, that is, the recovery 

 of seals selected and tagged as yearlings in 

 1962. This provides an estimate of the size 

 of the 1961 year class in 1962 when it is 

 hoped that the heaviest and most variable 

 component of mortality has already taken 

 place. If this is true, a small proportion can 

 be subtracted from this estimated number of 

 yearlings for mortality in the next 2 years, 

 and the balance will then either be harvested 

 or escape to join the breeding reserve. 



Method 1 . --Regression of 4-year-male kill 

 on 3-year-male kill and the mean date of the 

 kill of 3-year-olds. This regression is based 

 on the data in appendix table 1. 



Because there were two starting dates and 

 several termination dates for the kills during 

 the past several years, the adjustments indi- 

 cated in the footnotes to appendix table 1 were 

 made. The basis of these adjustments was 

 reported previously (Roppel, Johnson, and 

 Chapman, 1965). 



Appendix table 1. --The 3- and 4-year-old male kill and mean 

 date of the 3-year-old male kill, St. Paul Island, year 

 classes 1952-60 



1 Kill of 4-year-old males to 31 July, plus 80 percent of the 

 number of 3-year-old males taken after 31 July from the same 

 year class. 



2 Adjusted to the base of a common starting round on 27 June. 



26 



