float, most carcasses were found within a few 

 miles downstream from the dam. In the summer of 

 1955, many fresh dead salmon were found on the 

 bottom in shallow water on gravel bars immedi- 

 ately below the dam. Finally, very few floating 

 carcasses were observed during aerial surveys of 

 the forebay above the dam. (See also the data on 

 p. 15, last paragraph.) 



Another assumption is that no tags were lost 

 from recoverable carcasses. This assumption is 

 supported by the fact that there was no evidence 

 of missing tags on untagged chinook salmon 

 floaters. The tags were fastened to the jaw in 1954 

 and to the caudal peduncle in 1955; both locations 

 are exceptionally secure anchoring points for 

 tags on dead salmon. 



1954 EXPERIMENT 



To secure carcasses for releases in 1954, 1,095 

 chinook salmon were placed in frozen storage in 

 the summer and fall of 1953. All were ice glazed 

 to retard dehydration and oxidation. About 245 

 of these fish were spring chinook salmon of both 

 sexes that had been killed during construction of 

 Lookout Point Dam on the Middle Fork of the 

 Willamette River; the rest were fall male chinook 

 salmon from Bonneville and Oxbow Hatcheries on 

 the Columbia River. 



Frozen chinook salmon float when placed in 

 water; to ensure that they would sink like a natu- 

 rally killed salmon, the frozen carcasses were 

 thawed in air for 24 hours before being tagged and 

 released. The tags were sequentially numbered, 

 bright-colored nylon or plastic ribbons fastened 

 to the jaw. The tagged carcasses were released 

 during two periods: The first group (280 carcasses) 

 was released August 4 to 6 when spillway flows 

 averaged 3,300 c.m.s. and powerhouse flows 

 4,100 c.m.s.; the second group (815 carcasses) was 

 released September 2 to 3 when spillway flows 

 averaged 1,300 c.m.s. and powerhouse flows 3,700 

 c.m.s. 



Earlier evidence suggested that the spillway 

 channel was the source of most dead salmon near 

 the dam. Therefore, most of the tagged carcasses 

 (240 in August and 635 in September) were re- 

 leased into the spillway channel either by dropping 

 them from the dam immediately above the gates 

 or into the rollback below the gates (fig. 2). 

 Another 150 carcasses (20 in August and 130 in 

 September) were dropped into the draft tube dis- 



charge at the downstream face of the powerhouse. 

 Finally, 50 were dropped into the river in Sep- 

 tember from the Bridge of The Gods, 8 km. above 

 the dam, and 20 were released in August from a 

 boat at Oneonta, 11 km. below the dam. 



Less than 1 percent of the 1,095 carcasses were 

 recovered. Only one of the 280 carcasses released 

 in August was recovered; it was found at Oneonta 

 on August 10 at the same location where it had 

 been released on August 5. Eight carcasses were 

 recovered from the 815 released in September — 

 six from the powerhouse and two from the spillway 

 releases. No carcasses were recovered from the 

 group dropped from the Bridge of The Gods. 



The significance of the apparent difference in 

 the rates of recovery of the carcasses released at 

 the two sites in September can be evaluated by a 

 chi-square test for homogeneity. The null hy- 

 pothesis is that the probability of a carcass being 

 recovered does not depend on release location. 

 To test this hypothesis, we constructed the follow- 

 ing fourfold contingency table: 



For these data, x" = 15.35 with 1 d.f., and the 

 hypothesis of homogeneity is strongly rejected. 

 (The observed frequency for recoveries from spill- 

 way releases is smaller than is generally consid- 

 ered desirable for this statistical treatment, but 

 in this experiment the conclusion is so clear-cut 

 that we need not be greatly concerned over this 

 fact.) The conclusion is that carcasses released in 

 the spillway are less likely to be recovered than 

 carcasses released at the powerhouse. 



The small number of carcasses recovered from 

 the spillway release indicated that many of them 

 had probably disintegrated in the extremely 

 turbulent flow of the spillway rollback (fig. 2). 

 Such carcasses would have a reduced chance of 

 floating and being recovered. The less turbulent 

 flow of the powerhouse discharge probably con- 

 tributed to the higher recovery rate of carcasses 

 released there. 



Although the 1954 experiment did not produce 

 adequate data to estimate precisely the number 



472 



U.S. FISH AND WILDLIFE SERVICE 



