year was almost entirely age 5, as contrasted 

 with the normal pattern of about 50 percent 

 4-year fish (Ossiander, 1966--see footnote 1). 

 Improved forecasting of age at return is out- 

 side the scope of this report but is treated at 

 length in Ossiander (footnote 2). 



Coefficients of determination (r^ ) for cor- 

 relations of "hindcasts" and published fore- 

 casts with actual returns measure the amounts 

 of variation in actual returns (1961-67) that 

 are accounted for by each type of estimate. 

 The values of r^ for the birectilinear hind- 

 casts and published forecasts were simi- 

 lar (0.46 and 0.39, respectively). The ad- 

 vantage, if any, lies with the birectilinear 

 estimates. 



It is only fair to note, in considering the 

 above comparison, that the computational tech- 

 niques underlying the published forecasts were 

 materially improved during 1961-67. In par- 

 ticular the "probability tree" of Ossiander 

 (footnote 2) achieved much more accurate 

 allocation of estinnated returns among return 

 years than the average age distributions used 

 earlier (and also used here). Preliminary data 

 indicate that published forecasts for 1968 and 

 1969 were closer to the actual runs that would 

 have been predicted by the birectilinear ap- 

 proach. Runs for 1967-69 were all snnall, how- 

 ever, and it remains to be seen which method 

 will be superior for large runs such as those 

 of i960 and 1965. Use of the estimated re- 

 turns from the birectilinear curves and the 

 allocation from the probability tree might 

 represent a valuable combination of meth- 

 ods. 



200.0 



100.0 

 80.0 

 60.0 



40.0 

 S 20.0 



10.0 

 8.0 

 6.0 



4.0 



2.0 



OKvichok 



Wood / Noknek 

 O /O OEgeqik 



OUqoshik 



/ 

 OBronch 



ONuyokuk 



— r^ lO (T> — fO CO 



•" r- 0(\| <» to — 



— — — {\f ift 



LAKE ARE A (SQUARE KILOMETERS) 



Figure 12. — Regression of capacity on lake area. Right- 

 hand limbs are from figs. 2 to 11; lake areas from 

 Burgner et al. (1969). Lines are least-squares fits. 

 Lake areas were from a prepublication version of Burg- 

 ner et al.; minor corrections were made in the pub- 

 lished version, but they would have no significant effect 

 on the fitted curve. 



TABLE 2. — "Hindcasts" (estimates and actual nujnbers) of Bristol Bay sookeye salmon in thousands of 

 fish, by use of birectilinear curves, 1958-67 



Stream system 



1958 



1959 



1960 



1961 



1962 



1963 



1967 



Togiak 287 234 332 283 138 256 352 346 271 211 



Igushik 230 327 747 695 337 375 753 778 539 135 



Snake 19 67 46 17 32 76 86 46 16 64 



Nuyakuk 225 77 50 99 192 241 170 256 180 144 



tood 1,634 2,389 2,545 1,517 2,046 2,836 2,900 2,030 2,127 2,329 



Nushagak-Mulchatna^ . 49 49 49 49 49 49 49 49 49 49 



Kvlchak 1,774 524 18,924 21,616 7,462 2,489 29,285 30,906 13,813 5,517 



Naknek 1,172 2,294 1,786 2,459 2,373 1,648 2,533 2,767 1,878 2,402 



Branch/ Alagnak ^ 605 ^605 ^605 664 382 888 1,169 638 351 534 



Ifeegik 2,319 1,958 1,560 2,289 1,893 1,365 2,408 2,937 2,876 2,917 



Ugashik 1,770 1,688 675 1,350 998 1,108 956 1,590 1,525 1,198 



North side Alaska 



Peninsula^ 1,408 1,408 1,408 1,408 1,408 1,408 1,408 1,408 1,408 1,408 



Total 11,492 11,620 28,727 32,446 17,310 12,739 42,069 43,751 25,033 16,908 



Actual run'^ 7,974 15,278 42,645 26,274 12,266 8,617 13,386 63,055 20,656 12,526 



Published forecasts' -- — — 22,000 19,900 15,600 17,800 28,970 32,654 15,726 



^ For Nushagak-Mulchatna used mean 1956-65 from Ossiander, 1966, table 15 ( see text footnote 1) . 



^ For Branch or Alagnak, 1958-60, used mean return from 1955-62, escapement Ibid ., 

 tables 19, 20. 



^ For north side Alaska Peninsula, used mean 1956-65, Ibid., table 26. 



'• Actual run, 1958-65 from Ibid., table 1; 1966 from Ossiander, 1967, App. table 22 (see text 

 footnote 2); 1967 from Ossiander, 1968, App. table 2 (see text footnote 3). 



' From Royce, 1961, 1962 (see text footnotes 4, 5); Tait, 1963 (text footnote 6); and 

 Ossiander, 1964, 1965, 1966, 1967 (see text footnotes 1, 2, 7, 8). 



