9710 - Laserfiche WebLink

Home Browse Search

PERFORMANCE OF PIT TAG INTERROGATION SYSTEMS 409 TABLE 2.-Percen! of detections by river-right (RR) and river-left (RL) antenna~ in each array of a 3 X 2 ~'Ystem in Beaver and Rattlesnake creeks at high and low flow levels. Protocol I (see Table 1) was used to select passage events; n=the number offish detection events. Rattlesnake Creek Beaver Creek Direction Array Flow 11 RR RL Both 11 RR RL Both Downstream A High 55 60 33 7 51 51 40 8 Low 154 71 21 8 141 13 70 16 B Higl) 68 75 24 l 62 70 25 3 Low J69 92 6 2 140 47 32 19 C High 4l 15 83 2 54 22 73 4 Low 158 4 95 1 137 6 84 7 Upstream A High 31 55 39 6 16 44 54 0 Low 36 3\ 61 8 22 50 40 8 B High 19 37 58 5 13 39 53 6 Low 35 71 26 3 22 57 25 16 C High 17 35 59 6 16 56 20 22 Low 32 19 75 6 21 22 63 13 it was practical to run a separate analysis without cutthroat trout because of the difficulty in distinguish- ing hybrid individuals (with rainbow trout). For Beaver Creek, minimal difference (<0.01%) in system effi- ciency of the 3 X 2 system was noted when brook trout were removed from consideration (downstream: low flow, n = 3, high flow, n = 0; upstream: low flow, n = 2, high flow, II = I). Combining the more common salmonid species had the advantage of increasing the sample size of fish considered in the analysis. We used analysis of variance (ANOV A) to test for differences in detection efficiencies attributable to the direction in which fish were moving (downstream or upstream) and flow level (low or high). Stream sites (Rattlesnake and Beaver creeks) were considered replicates and, therefore, never included in interaction terms. When this stream factor did not significantly contribute to the variation of detection efficiencies (ANOV A, P > 0.05), it was dropped from the model. To test for differences in detection efficiency among the four designs (3 X 2, 3 X 1,2 X 2, and 2 X I), we used ANOV A, and when the design effect was significant (P < 0.05) we used Tukey's Studentized range test (Tukey's test) as a multiple comparison test to identify significant differences among the four designs. Because most values (29 of 32) for detection efficiencies of the systems for various combinations of direction of fish movement and flow level exceeded 80%, we trans- formed the detection efficiency variable by taking the arcsine of the square root of the estimated detection proportion to stabilize the variances (Olt 1977) before the statistical tests were run. Whenever the normality of the detection efficiency variable was testable (i.e., when n > 2), use of the Sbipiro-Wilk statistic (SAS Institute 1988) indicated that all groups were normal (P > 0.05) after the transformation procedure. Results Fish passage events were recorded at a maximum stage height of 1.94 m (flow, 6.31 m3/s) in Rattlesnake Creek and 2.03 m (4.23 m3/s) in Beaver Creek (Figure 5). During the overall period in which each system operated, a limited number of days qualified as high flow (Rattlesnake Creek, 21 % of 707 d; Beaver Creek, 7% of 596 d). These relatively rare high-flow days, however, accounted for relatively high portions of the downstream and upstream fish passage events (Rattle- snake Creek: 35% downstream and 50% upstream; Beaver Creek: 30% downstream and 39% upstream). Within the range of stage heights at which fish passage events were recorded, detection efficiency was high for the inten'ogation systems in Rattlesnake and Beaver creeks. The inten'Ogation system in Rattlesnake Creek had detection efficiencies that ranged from 96% to almost 100% for trout (Le., rainbow and cutthroat trout) moving downstream or upstream during low or high-tlow levels, whereas the system at Beaver Creek had detection efficiencies for salmonids (Le., rainbow trout, juvenile steelhead, and brook trout) that exceeded 99% for all combinations of direction and flow level (Figure 6). Although relatively minor overall differ- ences in detection efficiency were evident between the two systems, these systems were more efficient during low flow (mean = 99.9%, coefficient of variation [CV = 100 . SE/mean] = 0.2%) than during high flow (mean = 98.3%, CV = 1.5%) (ANOV A: df = 4,7; P = 0.024), averaged over the nonsignificant contribution of the direction of fish movement (P = 0.637). The performance of the pass-by and hybrid antenna types varied in a complex way depending on flow level and direction of fish movement (ANOV A, flow X direction X type interaction term: P < 0.001; Figure 7). The difference in mean efficiency of the hybrid arrays