the river, could be seen clearly. Because all flights 

 were made in the same manner, it is not important 

 whether we saw all floaters; the significant point 

 is that we determined the relative abundance of 

 floaters. 



Floaters could be seen easily at the low altitude 

 at which observations were made. Although 

 Chinook salmon could usually be distinguished 

 from other species by their relatively large size, 

 all floaters were combined in interpreting the ob- 

 servations. Adverse light conditions and waves 

 occasionally reduced the efficiency of the observa- 

 tions. To compensate for the glare on the water 

 surface, the plane was flown on the side of the 

 river toward the sun and the observers wore 

 Polaroid glasses. Waves and whitecaps were the 

 most serious problems because they reduced the 

 distance at which carcasses could be seen. For- 

 tunately, visibility was generally good on all 

 flights. We recorded all floaters except in the rare 

 instance when on the return trip we definitely 

 recognized a carcass that we had counted before. 

 We may have counted some twice but believe this 

 seldom occurred. 



We assumed that we saw a nearly constant 

 proportion of the floaters present on each flight. 

 Granting this assumption, the variations in num- 

 bers and distribution of floaters on different flights 

 reflect real differences. In 1954, between July 16 

 and September 17, floaters were counted on 10 

 flights over the Columbia River between Long- 

 view, Wash., and the McNary Dam (320 km.). 

 Figure 4 shows the numbers of floaters in 8-km. 

 sections of the river on each flight. In 1955, be- 

 tween May 6 and September 13, floaters were 

 counted on seven flights from the mouth of the 

 Columbia River at Astoria, Oreg., to its con- 

 fluence with the Snake River and up the Snake 

 River to Lewiston, Idaho — a total distance of 

 720 km. Figure 5 shows the numbers of floaters 

 in 8-km. sections of the area surveyed. The survey 

 between Astoria, Oreg., and Lewiston, Idaho, 

 extended over 2 days, July 14 and 15. 



Floaters in both years were not uniformly 

 distributed throughout the river but were con- 

 sistently concentrated at certain locations. The 

 greatest density was downstream from Bonne- 

 ville Dam when riverflows were high or when 

 relatively large numbers of migrating salmon were 

 present. Floaters were present, but at much lower 

 densities, throughout the river between Long- 



view and McNary Dam in July of both years — 

 probably as a result of rapid downstream dis- 

 persion from points of occurrence of high mortality 

 during this high-flow period. 



Most flights were at times when no experi- 

 mental dead chinook salmon (released by us) 

 were in the river. Only on the flights of Septem- 

 ber 9, 1954, and July 9, 1955, could experimental 

 fish have been present; six were seen below Bonne- 

 ville Dam on the first date and four on the second. 



The observations on the flights of July 14 and 

 15, 1955 (fig. 5), were of particular significance 

 because they included the greatest length of 

 river of any surveys and were made during the 

 period of our experiment to estimate chinook 

 salmon mortality. Visibility was good on both 

 days, and we believed our observations revealed 

 typical distribution of floaters at high river- 

 flows. The fact that floaters were most numerous 

 below Bonneville Dam suggests that dead fish 

 were originating near the dam. In the lowest 225 

 km. of the Snake River, where there were no 

 dams, only three floaters were seen, despite the 

 presence of large numbers of live sockeye and 

 chinook salmon that were migrating up the river. 



The aerial surveys were useful in showing that 

 floaters were usually more numerous below Bonne- 

 ville Dam than in other areas and in indicating 

 the times when the greatest numbers of floaters 

 were present. 



MARK-AND-RECOVERY EXPERIMENTS 

 TO ESTIMATE ABUNDANCE OF 

 DEAD CHINOOK SALMON 



In the mark-and-recovery technique we used 

 to measure mortality of chinook salmon near 

 Bonneville Dam during periods of high flow, 

 carcasses were tagged and introduced into the 

 river at the dam in 1954 and 1955. The recovery 

 sample consisted of the floaters found on boat 

 searches downstream from the dam at the search 

 stations shown in figure 3. The sample contained 

 the tagged carcasses as well as the carcasses of 

 chinook salmon that died naturally in the river. 

 Only the 1955 experiment had enough recoveries 

 for us to estimate the numbers of dead chinook 

 salmon. 



Our experiments differed from an experiment 

 that would lead to a standard Petersen-type of 

 population estimate in one important respect: 

 Instead of removing carcasses from the river, 



468 



U.S. FISH AND WILDLIFE SERVICE 



