in combination with plastic bars, nylon streamers, 

 or vinyl tubing (spaghetti). Tags used below the 

 dam were larger than the tags used above the 

 dam. It might be argued that a larger, different 

 shaped tag would aid recovery or observation of a 

 tagged fish released below the dam. In con- 

 sidering the conditions for tag recoveries and 

 observations, we do not agree with this argument. 

 At Tumwater Dam all tags were observed in clear 

 water from a distance of 3 or 4 feet. Most tags 

 were read for numbers; thus, differentiation of tags 

 from above and below the dam should have been 

 readily apparent. This also holds true for tag 

 observations at Zosel Dam. On the Okanogan 

 River spawning grounds, tags were recovered from 

 dead spawned-out fish or were observed during 

 surveys of the river. In the latter case, biologists 

 identified tags from 5 to 25 feet away and recogni- 

 tion was easy. In fact, when viewing tagged fish 

 from the Tumwater Dam suspension bridge and 

 from the railroad bridge just below Zosel Dam, 

 both suspended 20 to 30 feet above the water, we 

 could easily discern sizes and shapes of tags. 

 Biologists observing fish at Rock Island dis- 

 tinguished, from about 40 feet away, shapes, sizes, 

 and colors of tags on fish in the tailrace just below 

 the fish way. 



So far as tag colors are concerned, all tags used 

 below the dam had single-colored disks and single- 

 colored bars, streamers, or spaghetti. Tags used 

 above the dam had single or bicolored patterns on 

 the disks. It is difficult to say what the dif- 

 ferences in colors or color combinations mean in 

 terms of likelihood of recovery. Based on our 

 observations of hundreds of tags under many 

 circumstances, we believe that the tags, when 

 seen, can be differentiated by color and kind. 



ALTERNATIVE MODELS FOR ESTIMATING 

 MORTALITY 



Several models were considered for estimating 

 mortalities in addition to the one used (ratios of 

 recovery proportions of tagged fish from different 

 release areas). We were unable, however, to 

 accept the assumptions necessary for some. In 

 others, dubious and conflicting results were ob- 

 tained; hence, they are not given in detail in this 

 paper. Two models are discussed briefly. 



In one model we determined the racial composi- 

 tion of sockeye salmon in the tagged sample by 

 estimating the number of 4-year-old upriver fish 



from the relative abundance of 3-year-old Oka- 

 nogan fish in the tagged lots, and using the ratio 

 of fishway counts to numbers released, as an 

 adjustment factor in estimating the relative 

 abundance of Wenatchee tagged fish in different 

 release groups. Another model estimated racial 

 composition from the differences in the proportions 

 of 3-year-old sockeye in the populations at Rock 

 Island Dam and in the Okanogan River, caused 

 by a segment of the 4-year-old sockeye population 

 splitting off and entering the Wenatchee system. 

 In neither of these two models were definitive 

 results obtained regarding mortalities caused by 

 the dam. Mortalities were indicated for some 

 years and areas, not for others. Because of 

 these conflicting and nondefinitive results, we 

 feel the best approach to the problem is that 



given. 



MORTALITIES 



Populations of anadromous salmonids lose some 

 members passing large dams. For years, dead 

 salmon have been observed floating down the 

 Columbia River below Bonneville Dam, ap- 

 parently fish which failed to pass the dam. Using 

 the results of experiments involving the marking 

 and recovery of dead salmon carcasses, Merrell 

 and Collins 3 estimated a salmon loss at Bonneville 

 Dam. Thompson (1945) found that a serious 

 decline in Fraser River sockeye salmon runs was 

 related to an area of difficult passage and to an 

 obstruction in the river, even though salmon 

 were known to pass this area each season. In a 

 later paper on the Fraser River sockeye salmon, 

 Talbot (1950) found that fish delayed longer than 

 14 days at Hells Gate did not reach their spawning 

 grounds. 



The tagging experiments at Rock Island gave 

 conflicting results of mortalities caused by the 

 dam. We found no mortalities for 3-year-old 

 sockeye salmon when comparing recoveries of 

 tagged fish released above and below the dam. 

 We found mortalities in many experiments for 

 4-year-old sockeye when comparing tag returns 

 from above and below dam releases. In other 

 experiments involving these 4-year-old fish, no 

 mortalities were indicated. Most experiments 



' Merrell, Theodore R„ and Melvin D. Collins. An investigation of adult 

 chinook salmon mortality in the vicinity of Bonneville Dam, 1954 and 1955, 

 on the Columbia River. Fish Commission of Oregon, August, 1960. 150 p. 

 (Contract No. DA-35-026-eng-20892, U.S. Department of Interior, Fish and 

 Wildlife Service.) 



LOSS AND DELAY OF SALMON PASSING ROCK ISLAND DAM 



363 



