ML070160217
| ML070160217 | |
| Person / Time | |
|---|---|
| Site: | FitzPatrick  |
| Issue date: | 12/31/1996 |
| From: | Dunning D, Kenna M, Menezes J, Ross Q, Tiller G New York Power Authority, Sonalyst |
| To: | Office of Nuclear Reactor Regulation |
| jmm7 | |
| References | |
| Download: ML070160217 (12) | |
Text
Nnrth American Journal of Fisheries Management 16:548-559. 1996 O Copyright by the American Fisheries Society 1996 Reducing Impingement of Alewives with High-Frequency Sound at a Power Plant Intake on Lake Ontario QUENTIN E. ROSS AND DENNIS J. DUNNING New York Power Authority 123 Main Street. White Plains, New York 10601. USA JOHN K. MENEZES, MARK J. KENNA, JR., AND GARY TILLER Sonalysts. Inc.. 215 Parkway North. Waterford. Connecticut 06385. USA Abstract.From April 22 through July 20, 1993, we conducted a follow-up study to confirm that high-frequency broadband sound (122-128 kHz) at a source level (in decibels fdB] in reference to 1 lxPa) of J90 dB reduced the impingement of alewives Alosa pseudoharengus at the James A.
FitzPatrick Nuclear Power Plant (JAF), located on Lake Ontario near Oswego, New York. During the first full-scale test in 1991, the sound field covered only the front of the JAF intake. In this second full-scale test, the sound field included the top, sides, and rear of the JAF intake to prevent fish from approaching the intake from those directions when the JAF reactor was shut down and the hot water discharge, located 57 m offshore from the intake, disappeared. Our study also provided the opportunity to evaluate the effectiveness of the deterrent system during a mass die-off of alewives that occurred in Lake Ontario during late spring and early summer in 1993. We used a before-after-control-impact pairs (BACIP) design to test and quantify the effectiveness of the deterrent system. The new sound field reduced the impingement of alewives by 81-84% during a year following an unusually cold winter and should reduce impingement by 87% during most years.
The stocking of salmonids that began in the ear-ly 1970s in Lake Ontario (Jones et al. 1993) gen-erated a major sport fishery with considerable eco-nomic benefits (Talhelm 1988). This valuable fish-ery is primarily dependent upon a single forage species, the alewife Alosa pseudoharengus (Brandt 1986; Elrod and O'Gorman 1991). However, ale-wife populations in Lake Ontario experience high mortality during or following unusually cold win-ters (O'Gorman and Schneider 1986; Bergstedt and O'Gorman 1989) and historically have exhib-ited periodic mass declines (Scott and Crossman 1973). This instability in alewife production has been compounded by high salmonid stocking rates that have pushed piscivore densities in Lake On-tario to record levels (Leach et al. 1987). In 1991, the predator demand in Lake Ontario was esti-mated to be equal to the total annual production of pelagic prey species (Schneider and Schaner 1994), and simulation modeling suggested that a 25% increase in winter-related mortality would cause the alewife population to crash (Jones et al.
1993). The fact that the alewife population in Lake Michigan experienced a massive decline in the ear-ly 1980s that was followed by a collapse of the fishery for chinook salmon Oncorhynchus tshaw-ytscha (Eck and Wells 1987) heightened concern about the sustainability of alewife production in Lake Ontario. Reductions in salmonid stocking rates were initiated in 1993 with the goal of sta-bilizing predator demand at 50% of the prey pro-duction by 1996 (Schneider and Schaner 1994).
However, the alewife population in Lake Ontario suffered what appeared to be the highest mortality observed in 10 years in the late spring of 1993 and is presently considered to be threatened (Schneider and Schaner 1994).
As the concern over alewife production in-creased, the New York State Department of En-vironmental Conservation (DEC) sought greater protection for alewives in Lake Ontario from an-thropogenic sources of mortality, including power plants. There are eight power plants along the New York shoreline of Lake Ontario with a combined cooling-water flow of about 290 m3/s at full power.
In response to DEC's concern, the New York Pow-er Authority reviewed the mitigation technologies available for reducing impingement of alewives on the intake screens at power plants and temporarily installed and tested an acoustic deterrent system at the James A. FitzPatrick Nuclear Power Plant (JAF), located on Lake Ontario (Ross et al. 1993).
This electronic system produced intense (190 deci-bels [dB] measured 1 m from the source and ref-erenced to 1 jxPa), high-frequency broadband (122-128 kHz) sound and was specifically de-signed to repel alewives from the JAF intake (Dunning et al. 1992). When ambient water tem-548
REDUCING IMPINGEMENT WITH SOUND 549 peratures were below 13°C, the deterrent system was very effective, and the results were consistent with the avoidance responses expected from a pe-lagic prey species. The system reduced the density of fish (number/100 m3) directly in front of the JAF intake by as much as 96%, and the effective-ness of the deterrent system increased as fish den-sities increased. However, the deterrent system did not cause a significant reduction in fish densities in front of the JAF intake when ambient water temperatures were 13°C or above. Because most of the alewife population moves offshore into deep water after spawning (Scott and Crossman 1973),
Ross et al. (1993) hypothesized that the alewives remaining in shallow water after temperatures reached 13°C were generally in poor condition, which made them less responsive to high-frequen-cy sound. However, if the response of alewives to high-frequency sound decreases when their con-dition declines, the deterrent system may not be effective when it is needed most, such as after an unusually cold winter when alewives are in poor condition (O'Gorman and Schneider 1986).
Ross et al. (1993) found that the deterrent sys-tem had little effect on impingement in 1991 when the JAF reactor was shut down and no hot water was being discharged through the 236-m-long dif-fuser located on the bottom, 57 m offshore from the intake. The acoustic field generated by the de-terrent system covered only the open, shoreward-facing side of the JAF intake. Ross et al. (1993) hypothesized that the hot-water discharge formed a thermal barrier that prevented alewives from ap-proaching the rear of the intake when the reactor was operating. Therefore, when the reactor shut down, the thermal barrier disappeared, enabling alewives to approach the rear of the intake and swim along its top and sides to the front, where the opening is located, without encountering the sound field.
We conducted a full-scale follow-up study in 1993. To prevent fish from approaching the rear of the JAF intake and to test the thermal barrier hypothesis, we increased the number of transduc-ers in the array used by Ross et al. (1993) so that high-frequency sound was produced on top, along the sides, and in back of the intake, as well as in front. We were able to assess the effectiveness of the deterrent system before and during a mass die-off of alewives during this follow-up study be-cause the winter of 1992-1993 was colder than average and produced the first late-spring die-off observed since 1984 (Schneider and Schaner 1994). Dead alewives were first observed in late May. The die-off intensified during June and con-tinued into July. We began testing on April 22 and operated the deterrent system continuously until July 20.
Methods Test Site The James A. FitzPatrick Nuclear Power Plant is located on the south shore of Lake Ontario at Nine Mile Point, near Oswego, New York. It with-draws water from the lake at up to 23.4 m-Vs through a single offshore intake located 274 m north-northeast of the plant in water 7.3 m deep.
Water is withdrawn only from the south (shore-ward) side of the intake to reduce recirculation of heated water from the discharge which is located 57 m farther offshore. The velocity through the intake openings is 0.4 m/s. Water flows from the intake through a tunnel into the forebay of the plant where traveling screens remove fish and debris be-fore the water circulates through the cooling sys-tems of the plant. Fish and debris, impinged on the traveling screens, are washed off and collected in a basket. The discharge tunnel extends into the lake and forks; one branch heads east, and the other heads west, nearly parallel to the shoreline. Heated water is discharged from each branch tunnel through three high-velocity diffuser heads, spaced 45.7 m apart and consisting of paired 0.76-m dis-charge nozzles that are directed away from shore.
The total length of the diffuser system is 236 m, and the depth of the diffuser heads ranges from 7.0 m for the most easterly head to 8.5 m for the most westerly head. The exit velocity of the water from the diffusers is 4.3 m/s and the discharge causes turbulence that reaches all the way to the surface when JAF is at full reactor power. The temperature of the discharge when JAF is at full power is 17.5°C higher than the temperature of the water drawn into the intake. When the JAF reactor was starting up, shutting down, or shut down dur-ing our study, the flow of water through the plant was reduced by about 33%, and the difference be-tween the intake and discharge temperatures ranged from 0 to 6°C.
Control Site The Nine Mile Point Unit 1 Nuclear Power Plant (NM1) is located 914 m due west of JAF and with-draws water from the lake at up to 16.9 m3/s through a single offshore intake located 259 m northwest of the plant in water 7.5 m deep. Water is withdrawn from all sides of the intake. The ve-locity through the intake openings is 0.5 m/s. Wa-
2.5m
- WR)EBEAM(1tt1(1ft3)
ADDITIONAL WIOEBEAM (1W)
NARROWBEAM (1W1,1993)
FIGURE I.Location of the wide-beam and narrow-beam transducers of the deterrent system at the intake structure of the James A. FitzPatrick Nuclear Power Plant (JAF) during 1991 and 1993.
ter flows from the intake through a tunnel into the forebay of the plant where traveling screens re-move fish and debris before the water circulates through the cooling systems of the plant. Fish and debris impinged on the traveling screens are washed off and collected in a basket. Heated water returns to Lake Ontario through a low-velocity (1.2 m/s) multiport (six) discharge located 102 m north-northwest of the plant in water 5.2 m deep, inshore and west of (he NM1 intake. At full reactor power the temperature of the discharge from NM1 is 17.3°C higher than the temperature of the water drawn into the intake.
Impingement collections from NM1 should pro-vide a good control for those at JAF because the NMJ intake is close (1.3 km) to the JAF intake but is beyond the effective range of the deterrent system (80 m). Furthermore, the large-scale cir-culation in Lake Ontario is counterclockwise, gen-erating a current that flows from west to east in a relatively narrow band along the south shore.
Thus, schools of alewives moving through the Nine Mile Point area with this current pass the control intake before encountering the test intake.
Deterrent System The acoustic deterrent system consisted of an array of electronic transducers, connecting ca-bling, impedance-matching devices, power ampli-fiers, a signal generator, and a personal computer for control and data logging. The transducer array contained 16 narrow-beam and 9 wide-beam trans-ducers (Figure 1). The 20-transducer array on the front of the intake was the same as that used by Ross et al. (1993) and produced a minimum sound pressure level (SPL) at 1 m from the transducers of 190 dB in a frequency band from 122 to 128 kHz. (As in Burdic [I984J, all SPLs are given as decibels referenced to 1 jiPathat is, 190 dB de-notes 190 dB//u,Pa). In addition, four wide-beam transducers were installed on the top and sides of the intake, and a fifth wide-beam transducer was mounted on a tripod on the lake bottom at the back of the intake structure to ensonify the back, sides, and top of the intake.
The sound fields produced by the five new trans-ducers did not overlap as much as those from the 20 transducers that were mounted on the front of the intake. However, given the sensitivity of ale-wives to high-frequency sound, the presence of the five new transducers was expected to generate a detectable reduction in impingement at JAF when the reactor was shut down and there was no ther-mal barrier behind the intake. Sound was produced for a 0.5-s duration every 1.5 s. Upon installation and before removal of the system, hydroacoustic measurements were taken to verify the design source levels and beam patterns transmitted.
Impingement Collections Paired, 24-h impingement samples were col-lected at JAF and NM1. The samples at NM1 were collected between 1200 and 1300 hours0.015 days <br />0.361 hours <br />0.00215 weeks <br />4.9465e-4 months <br />. Those at JAF were collected between 1300 and 1400 hours0.0162 days <br />0.389 hours <br />0.00231 weeks <br />5.327e-4 months <br />.
This 2-h period was selected because the abun-dance of alewives in the vicinity of the two power plants was generally low during the middle of the day (Ross et al. 1993) and the interruption in the
REDUCING IMPINGEMENT WITH SOUND 551 collection of impinged fish was not likely to con-found the daily totals.
Statistical Analyses Effectiveness of the deterrent system.A before-after-control-impact pairs (BACIP) design (Stew-art-Oaten et al. 1986) was used to test the differ-ences among selected sets of paired impingement samples. An observation in this design is the dif-ference between the impingement counts at JAF (the impact site) and NM1 (the control site) on the same day. The "after" samples consist of the paired daily impingement counts collected during the period from late April through late July in 1993 when the deterrent system was operating. The "be-fore" samples consist of paired daily impingement counts collected during the same period in years when the deterrent system was either not installed (1981, 1985-1987, and 1994) or was installed but not turned on (1991). The installation and removal of the high-frequency transducers did not affect the flow patterns or physical features at the JAF intake. Thus, the fact that some of the "before" samples were collected after the "after" samples should not affect their validity as controls. How-ever, it does make "before" an inappropriate label.
Therefore, we used "sound not produced" for "before" and "sound produced" for "after" in our tables.
We used two-sample /-tests for determining whether the average daily difference in the "after" samples was significantly different from that in the "before" samples. These tests assume normality, additivity, and independence. We transformed the daily counts (log^ c or log,, c + 1 when zeros were present) and used modified /-tests (Statistix 4.1; Analytical Software 1994) to protect against vio-lations of the normality assumption (Stewart-Oat-en et al. 1992).
Before testing the remaining assumptions, we separated the paired samples into two groups, those collected when the JAF reactor was shut down and those collected when it was at or near full power. We used the first group of samples to determine the effectiveness of the new transducer array. We used the second group to determine the effectiveness of the deterrent system after the un-usually cold winter of 1992-1993. Only those sam-ples collected on days when the JAF reactor was at or near full power were used in the second group because increases in cooling water flows associ-ated with rising power levels changed the mag-nitude and distribution of the water currents within TABLE 1.Number of alewives impinged on the intake screens of the James A. FitzPatrick Nuclear Power Plant (JAF) and the Nine Mile Point Nuclear Station Number 1 (NM1) on those days when high-frequency sound was pro-duced and when it was not and when only the NM 1 reactor was at or near full power, and the temperature of Lake Ontario was less than 13°C. Samples were partitioned into days when the number of alewives impinged at NM 1 ex-ceeded 1,000 (high-abundance block) and days when the number was between 100 and 999 (low-abundance block).
Sound produced Date JAF NM1 Sound Date not produced JAF NM1 High abundance block 1993 Apr 23 Apr 24 Apr 25 Apr 26 Apr 27 Apr 28 May 28 May 29 595 183 580 276 692 183 1,618 590 8,400 10,716 4.388 1,438 1.291 1.423 11.500 5.292 1985 Apr 26 Apr 28 Apr 29 Apr 30 May 1 May 2 May 5 May 6 May 7 May 8 May 9 May 10 May 11 May 13 May 14 May 15 May 16 May 30 May 31 1,692 1,652 1.850 3.312 5.236 6,608 8.884 7,688 5,032 4,412 9,488 6.732 6.864 11,788 7,316 10.804 9,124 4.168 3.868 10,424 10,224 8.416 15.464 10.136 15,952 2,688 8.704 9,000 12,4%
9.780 8.828 6.380 9.628 11,364 9,762 9.724 5,740 4.796 Low-abundance block 1993 Apr 21 Apr 22 May 18 May 19 May 20 May 21 May 22 May 23 May 24 May 25 May 26 May 27 78 172 56 31 48 55 34 40 26 34 75 88 138 579 491 376 608 670 626 844 341 207 487 547 1991 May 16 May 17 May 19 May 20 May 21 May 22 May 23 May 25 182 167 214 189 200 138 102 76 947 467 577 582 536 439 401 391 the JAF forebay and generated transient surges in impingement.
When the JAF reactor was shut down, the num-bers of alewives impinged at NM1 ranged from 138 to 15,952 (Table I). To reduce variance het-erogeneity and the effects of seasonal changes in behavior that might be associated with spawning, we partitioned the samples into a high-abundance block that included all days when the number of alewives impinged was at or above 1,000 at NM 1, and a low-abundance block that included all days when the number of alewives impinged was be-
tween 100 and 999 at NM1 (Table 1). The high-abundance block involved primarily prespawning and spawning alewives; the low-abundance block involved alewives impinged primarily after the pe-riod of peak impingement when spawning prob-ably occurred.
When the JAF reactor was at or near full power, ambient water temperatures ranged from 6 to 23°C.
During the 1991 study, the response to high-fre-quency sound disappeared when water tempera-tures were 13°C or above (Ross et al. 1993). There-fore, we divided the samples collected when the JAF reactor was at or near full power into two groups, one consisting of samples collected when water temperatures were below 13°C and the other consisting of samples collected when water tem-peratures were 13°C or above.
When water temperatures were below 13°C and the JAF reactor was at or near full power, the num-ber of alewives impinged at NM1 ranged from 95 to 12,960 (Table 2). To reduce variance hetero-geneity and the effects of seasonal changes in be-havior that might be associated with spawning, we again partitioned the samples into a high-abun-dance block that included all days when the num-ber of alewives impinged was at or above 1,000 at NM1 and a low-abundance block containing all days when the number of alewives impinged was between 95 and 999 at NM1 (Table 2). The high-abundance block involved prespawning and spawning alewives; the low abundance block in-volved alewives impinged after the period of peak impingement.
The samples collected when ambient water tem-peratures were 13°C or above and the JAF reactor was at or near full power included only alewives impinged after the period of peak impingement and impingement counts at NM1 ranged from 0 to 319 (Table 3). We did not partition this set of samples.
The assumption of additivity requires that the expected difference between the impact and con-trol sites be the same for all dates. We tested for additivity within each abundance block and within the high-temperature group by correlating the dif-ferences between, and the sums of, the transformed paired daily impingement counts from the two sites for the "sound not produced" treatment (a significant correlation indicated the presence of nonadditive effects). We did not test for additivity within the "sound produced" treatment because the deterrent system was expected to generate a significant correlation between the sums of the paired daily impingement counts from the two sites and the differences between the pairs by TABLE 2.Number of alewives impinged on the intake screens of the James A. FitzPatrick Nuclear Power Plant (JAF) and the Nine Mile Point Nuclear Station Number 1 (NM 1) on those days when high-frequency sound was pro-duced and when it was not and when both the JAF and NM 1 reactors were at or near full power, and the temper-ature of Lake Ontario was less than I3°C. Samples were partitioned into days when the number of alewives im-pinged at NM1 exceeded 1,000 (high-abundance block) and days when the number was between 95 and 999 (low-abundance block.
Sound produced Date JAF NMI Sound not produced Date JAF NMI High-abundance block 1993 Apr 30 May 2 May 3 May 4 May 5 May 7 May 8 May 9 May 10 May 11 May 12 May 13 May 14 495 749 465 254 266 477 164 100 318 218 241 286 166 2.731 3,730 1,523 1.653 1,635 2,135 4.906 1.597 6,841 1.594 1.894 3.091 1,075 1987 May 6 May 7 May 8 May 10 1994 May 7 May 13 May 14 May 15 May 21 May 22 May 27 1,336 414 1,336 2.838 1,920 3.024 6,450 4.670 2,936 5,387 9,450 1,170 1.340 1.0%
2.568 1.578 12,960 7,332 2,232 2.771 3,487 2,241 Low-abundance block 1993 May 1 May 6 May 15 May 16 May 17 Jun 2 Jun3 Jun 4 Jun 5 Jun 6 Jun 7 Jun 8 Jun 9 Jun 10 Jun 11 Jun 12 Jun 13 303 312 161 191 155 48 24 36 36 54 89 55 51 123 151 117 88 950 824 699 667 556 222 223 133 360 211 244 136 256 268 304 228 268 1987 May 12 May 13 May 14 May 17 May 21 May 22 May 27 1994 May 8 May 16 May 18 May 19 May 26 489 332 189 99 88 58 135 308 1,421 187 307 226 690 597 322 192 187 202 102 129 680 180 372 95 greatly decreasing the contribution of the test site to the sum of the impingement counts. Thus, the magnitudes of the difference between the counts from the two sites and the sum of the counts from the two sites would both be dependent upon the counts at NM 1.
Results from /-tests may be invalid when first-order autocorrelations are greater than 0.30 (Stew-art-Oaten et al. 1992). We tested for independence by estimating the first-order autocorrelations among the differences (generated from trans-formed data) for the treatments within each block.
REDUCING IMPINGEMENT WITH SOUND 553 TABLE 3.Number of alewives impinged on the intake screens of the James A. FitzPalrick Nuclear Power Plant (JAF) and the Nine Mile Point Nuclear station Number I (NM1) on those days when the high-frequency sound was produced and when it was not, when both the JAF and NM1 reactors were at or near full power, and when the temperature of Lake Ontario was >13°C.
Sound Date 1993 Jun 14 Jun 15 Jun 16 Jun 17 Jun 18 Jun 19 Jun 20 Jun 21 Jun 22 Jun 23 Jun 24 Jun 25 Jun 26 Jun 27 Jun 28 Jun 29 Jun 30 Jul 1 Jul 2 Jul 3 Jul 4 Jul 5 Jul 6 Jul 7 Jul 8 Jul 9 Jul 10 Jul 11 Jul 12 Jul 13 Jul 14 Jul 15 Jul 16 Jul 17 Jul 18 Jul 19 produced JAF 147 119 132 114 113 90 123 102 116 99 84 107 99 101 79 67 170 102 227 83 128 84 29 21 27 25 18 14 10 II 1
55 65 13 0
1 NMI 118 226 306 300 129 151 319 27 120 117 82 55 17 66 153 84 54 116 274 48 86 164 68 44 100 35 49 11 103 21 6
249 141 14 63 4
Sound Date 1981 Jul 31 1985 Jul 25 1986 Jul 2 Jul 25 1987 Jun 17 Jun 26 Jul 2 Jul 10 Jul 20 Jul 30 1994 Jun 14 Jun 23 Jun 28 Jul 5 Jul 13 not produced JAF NMI 373 232 472 269 140 178 371 476 68 15 144 63 83 49 10 92 17 22 10 10 75 7
4 2
1 29 2
2 1
0 The time series in all of the samples collected when sound was not produced were interrupted and could not be tested. There were time series within four of the five samples collected when sound was produced that were long enough to test. However, three of these time series were relatively short (in-volving 15 or fewer observations) and the power of these tests was probably low.
We obtained our longest test series, 36 d, during the period when water temperatures were 13°C or above and alewife abundance was low. Because most alewives move offshore into deeper and cool-er water after spawning, we suspected that the smaller numbers impinged during June and July came from alewives that were resident in shallow water. We hypothesized that these fish were in poor condition, which prevented them from moving off-shore into deeper water. If these fish were unable to leave shallow water, they should have been ex-posed to the high frequency sound field more than the prespawning or spawning alewives were. Thus, if acclimation to high-frequency sound were to occur, it should have been most apparent with these fish. We tested for acclimation in these fish by regressing the daily differences between the trans-formed impingement counts at JAF and NMI against time. A significant negative slope would indicate that alewives became less responsive to high-frequency sound over the 36-d period.
We used an alpha level of 0.05 in all tests of assumptions, the test for acclimation when water temperatures were 13°C or above, and the test of the effectiveness of the deterrent system when wa-ter temperatures were 13°C or above. To protect against inflation of alpha errors in our evaluation of the effectiveness of the new transducer array and to confirm the effectiveness of the deterrent system when the JAF reactor was at or near full power, we used an alpha level of 0.025 for the /-
tests in each block.
We estimated the effectiveness of the deterrent system using the equation percent change = (*on-off _ l) x 100, where "on" and "off" are the means from a BA-CIP comparison. Under the null hypothesis, the expected difference between the two means is zero, resulting in a 0% change. If the deterrent system reduced the impingement of alewives at JAF, the percent change was negative.
Analysis of the control data.If alewives avoid-ed high water temperatures near the JAF discharge, we expected that they would do the same at the NMI discharge. However, unlike the effect hy-pothesized for JAF (a reduction in impingement),
this response should result in an increase in im-pingement at NMI because the location of the in-take relative to the discharge was the opposite of that at JAF. The NMI discharge is located inshore and west of the NMI intake, and alewives at-tempting to avoid the NMI discharge by moving offshore would be carried into the vicinity of the NMI intake by the prevailing west to east current in the Nine Mile Point area. Intake flow also affects the numbers of fish impinged at a given site, and we expected the number of alewives impinged at JAF to be higher than that at NMI when both power plants are at or near full power. Therefore,
we expected to observe the greatest effect of the discharge at NM1 when the JAF reactor was shut down, at which time the intake flows at the two sites were comparable in this study.
We used the mean differences observed between the transformed impingement counts at the test and control sites (always subtracting the transformed counts at NM1 from the transformed counts at JAF) when no sound was produced to determine the effect of the NM1 discharge on the number of alewives impinged at NM1. We used one-sample r-tests and tested each mean difference against the null hypothesis that the average difference be-tween the transformed impingement counts from the two sites was equal to 0 when no sound was produced. We used an alpha level of 0.05 for each test.
Results Reliability of the Deterrent System The deterrent system operated continuously for 90 d. During this time, no systems or components failed. All validation measurements, taken with the receive hydrophone in the maximum response axis of each transducer, met or exceeded design values (a level of 170 dB or above at 10 m). None of the transducers, unlike the other underwater surfaces at or near the JAF intake, were fouled by zebra mussels Dreissena polymorpha. Cladophora spp.,
or other aquatic organisms.
Effectiveness of the New Transducer Array (at Water Temperatures below I3°C)
Tests for additivity.The test for additivity was not significant in either the high-abundance block or the low-abundance block (Table 1). The cor-relation between the difference in the transformed daily impingement counts and the sum of the trans-formed daily impingement counts from JAF and NMI when sound was not produced was 0.376 (P
- 0.113) in the high abundance block and 0.310 (P = 0.454) in the low-abundance block.
Tests for independence.The time series in the high-abundance block and in the "sound not pro-duced" treatment in the low-abundance block were too short to test for autocorrelations. The first-order autocorrelation for the treatment that could be tested was greater than 0.30 (0.58). An inspec-tion of the differences between the impingement counts from the test and control sites revealed a consistent 2-d pattern (Table 1). The second-order autocorrelation was very small (0.05). Therefore, we used the average counts for each 2-d interval as the independent observations in this data set to reduce the serial dependence. The sample size de-creased from 12 to 7.
BACIP tests.When alewife abundance was high, the deterrent system significantly reduced the number of alewives impinged at JAF (P = 0.001);
the estimated reduction was 81%. When alewife abundance was low, the deterrent system had no significant effect (P = 0.065); the estimated re-duction was 51%.
Effectiveness of the Deterrent System after an Unusually Cold Winter (at Water Temperatures below 13°C)
Tests for additivity.The test for additivity was not significant in either the high-abundance block or the low-abundance block (Table 2). The cor-relation between the difference in the transformed daily impingement counts and the sum of the trans-formed daily impingement counts from JAF and NMI when sound was not produced was 0.163 (P
= 0.631) in the high-abundance block and 0.238 (P = 0.456) in the low-abundance block.
Tests for independence.The assumption of in-dependence could not be tested in the "sound not produced" treatments because the time series were too short. In the "sound produced" treatments, we inserted 2 d from the low-abundance block (May 1 and May 6) into the time series from April 30 to May 14 in the high-abundance block to generate a longer time series for testing the autocorrela-tions. In the low-abundance block, the time series from June 2 through June 13 was long enough to lest. The first-order autocorrelation in the low-abundance block was less than 0,30 (0.27), but in the high-abundance block it was greater than 0.30 (0.42). An increase in the response to the deterrent system occurred after May 7 based on the differ-ences between JAF and NMI in the high-abun-dance block (Table 2). From April 30 through May 6, water temperatures were below 9°C; from May 7 through May 14, water temperatures were be-tween 9 and 11°C, except for May 8. Smith (1985) reported that alewives begin spawning at 11°C, and the heightened response to the deterrent system was probably caused by the increased activity as-sociated with the onset of spawning. We attempted to avoid the confounding effects of this change in behavior by testing the impingement samples col-lected before May 8 separately from those col-lected after May 7 (Table 2).
BACIP tests.The deterrent system significant-ly reduced the number of alewives impinged at JAF in both blocks. In the high-abundance block, the estimated reduction was 81 % (P < 0.001) prior
REDUCING IMPINGEMENT WITH SOUND 555 to May 8 and 92% (P < 0.001) after May 7. In the low-abundance block, the estimated reduction was 68% (P < 0.001).
Effectiveness of the Deterrent System When Water Temperatures Were I3°C or Above Tests for additivity.The test for additivity was not significant (Table 3). The correlation between the difference in the transformed daily impinge-ment counts and the sum of the transformed daily impingement counts from JAF and NMI under the "sound not produced" treatment was -0.223 (P
= 0.425).
Tests for independence.The assumption of in-dependence could not be tested in the "sound not produced" treatment because the time series was too short. In the "sound produced" treatment, the time series was long enough to test. The first-order autocorrelation was less than 0.30 (0.26).
BACJP test.The deterrent system significantly reduced the number of alewives impinged at JAF when water temperatures were 13°C or above (P
< 0.001): the estimated reduction was 96%.
Acclimation test.The regression of the daily differences between the transformed counts at JAF and NM 1 against time during this 36-d period was significant (R2 = 0.184; P = 0.009: slope =
-0.04).
Analysis of Control Data When water temperatures were below 13°C and both sites were at or near full power, the mean differences between the transformed impingement counts at JAF and NM 1 when no sound was pro-duced were not significantly different from 0 in both the high-abundance (P = 0.843) and low-abundance (P = 0.538) blocks. When water tem-peratures were below 13°C and the JAF reactor was shut down, the mean difference between the transformed impingement counts from the two sites observed when no sound was produced was negative and significantly different from 0 in both the high-abundance (P = 0.010) and low-abun-dance (P < 0.001) blocks. When water tempera-tures were 13°C or above and both sites were at or near full power, the mean difference between the transformed impingement counts from the two sites observed when no sound was produced was positive and significantly different from 0 (P <
0.001).
Discussion Analysis of Control Data When water temperatures were below 13°C.
many large schools of alewives moved from west to east through the Nine Mile Point area. If these schools avoided the hot water from the NM1 dis-charge as they did at JAF (Ross et al. 1993), some alewives would be deflected offshore toward the NM1 intake, which could increase the numbers of alewives impinged at NM1. This hypothesis is consistent with the differences observed between the transformed impingement counts at JAF and NM1 when water temperatures were below 13°C and no sound was produced. When both sites were at or near full power, the transformed impingement counts at JAF were not significantly greater than those at NM1, in spite of the greater intake flows at JAF. When the JAF reactor was shut down, the transformed impingement counts at NM 1 were sig-nificantly greater than those at JAF, in spite of the fact that the intake flows were comparable at the two sites.
When water temperatures were 13°C or above, the hypothesized effect of the NM1 discharge on impingement at NM1 disappeared. When both sites were at or near full power, the transformed impingement counts at JAF were significantly greater than those at NM 1, which is consistent with the difference between the intake flows at the two sites. We believe that the effect of the NM1 dis-charge on impingement at NM 1 was not detectable when water temperatures were 13°C or above be-cause large schools of alewives stopped moving through the Nine Mile Point area. Small schools of alewives were more likely to move away from the NM1 discharge without coming close to the NM 1 intake than large schools.
The disappearance of large schools of alewives from the Nine Mile Point area after spawning also provides an explanation for another result that only occurred when water temperatures were 13°C or above, an increase in the numbers of alewives im-pinged at NM1 when the deterrent system was op-erating. At temperatures below 13°C, alewives re-pelled by the deterrent system in the direction of the NMI intake would encounter many large schools of alewives that were moving through the Nine Mile Point area in the opposite direction, i.e.,
west to east. The alewife is a schooling species and the alewives swimming west away from the JAF intake were more likely to have joined a large school swimming east than they were to have con-tinued swimming west through it. As a result, all alewives repelled by the deterrent system would eventually join the west to east flow of alewives through the Nine Mile Point area. However, when small scattered schools of alewives were moving through the Nine Mile Point area, alewives swim-
ming west away from the JAF intake would have a lower probability of encountering a school of alewives swimming east and be more likely to reach the NM1 intake. Ross et al. (1993) found that large schools were more common when water temperatures were below 13°C; small schools were more common when water temperatures were 13°C or above. Thus, the probability of reaching the NM1 intake would be higher when water temper-atures were 13°C or above.
The low flow of alewives through the Nine Mile Point area when water temperatures were 13°C or above and the accumulation of responsive ale-wives in the area around the NM1 intake when the deterrent system was operating would also in-crease the relative abundance of unresponsive ale-wives in the area around the JAF intake, account-ing for the slight acclimation that occurred when water temperatures were 13°C or above and pro-viding an explanation for the absence of a statis-tically significant treatment effect in the 1991 study when water temperatures were 13°C or above (Ross et al. 1993).
Effect of the Unusually Cold Winter of 1992-1993 We evaluated the effect of the unusually cold winter of 1992-1993 by comparing the results from the 1991 and 1993 studies. In 1991, both JAF and NM1 were at full power during the first week in May (Ross et al. 1993). The most comparable data set from 1993 is the high-abundance block when both JAF and NM 1 were at or near full power and lake temperatures were generally between 9 and 11°C. We could not use impingement data from 1991 because the sample size was too small under these conditions. To provide a direct com-parison between the 1993 BACIP estimate derived from 24-h impingement samples and the 1991 study, we converted the diel estimates of the per-cent reduction in the density of fish in front of the JAF intake from Ross et al. (1993) into estimates of the percent reduction over a 24-h period by assigning equal weights to the number of hours in each diel period and to the average density of fish observed in front of the JAF intake during each diel period. The daytime period was twice as long as the nighttime period but the density of fish ob-served during the day was more than five times lower than that observed at night. The BACIP es-timate of the effectiveness of the deterrent system (92%) was almost identical to the 24-h estimate (91%) from the 1991 study, which suggests that the unusually cold winter of 1992-1993 did not affect the responsiveness of spawning alewives.
There are no data from 1991 that can be directly compared to the BACIP estimate generated from prespawning fish (81%). The remaining compari-son between 1991 and 1993 estimates involves samples collected when alewife impingement was declining and water temperatures were below 13°C. We did not calculate a BACIP estimate from the impingement data collected in 1991 during this period because the JAF reactor was shut down. We believe that the BACIP estimate would be biased under these conditions because the deterrent sys-tem covered only the front of the JAF intake in 1991. When the JAF reactor was shut down, more fish were sampled with hydroacoustics than by im-pingement collections. Therefore, the effect of small numbers of fish from the rear of the intake probably had less of an effect on the estimate of the effectiveness of the deterrent system generated from the hydroacoustic data. So, we converted the diel estimates of the percent reduction in the den-sity of fish in front of the JAF intake when the JAF reactor was shut down in 1991 into an esti-mate of the percent reduction over a 24-h period.
The resulting percent reduction over a 24-h period was 86%. This estimate included fish that ap-proached from the rear of the JAF intake into the area monitored in front of the intake, and thus, was probably biased low. The 24-h estimate when the JAF reactor was at full power, which was unbiased because the thermal discharge behind the JAF in-take blocked fish approaching from the rear of the intake, was 91%. Thus, the bias generated when the JAF reactor shut down could be as much as 5 percentage points. In 1993, when alewife abun-dance was low, water temperatures were below 13°C, and both JAF and NM1 were at or near full power, the BACIP estimate was 68%. The 18 point difference between these two estimates suggests that the unusually cold winter of 1992-1993 af-fected the responsiveness of alewives during the postspawning period when alewife impingement was declining and water temperatures were below 13°C.
Effectiveness of the New Transducer Array The new transducer array installed on the sides and back of the intake was at least as effective as the one in front of the intake, based on the 81%
BACIP estimate when alewife abundance was high, water temperatures were 8°C or less, and both power plants were at or near full power and when the JAF reactor was shut down under the
REDUCING IMPINGEMENT WITH SOUND 557 same conditions. The two samples collected on May 28 and 29, when water temperatures were below 13°C, the JAF reactor was shut down, and alewife abundance was high, confirmed that the sound field behind the JAF intake was at least as effective as the one in front of the intake. These samples were collected when there were high winds from the north, which should have moved postspawning fish from offshore waters directly into the rear sound field. The differences between the impingement counts at JAF and NMl on these two dates were close to the average value for the entire high-abundance block. The similarity be-tween the late April and late May samples also indicates that the offshore population of post-spawning alewives during late May was as re-sponsive as prespawning alewives were during late April and early May.
The new transducer array appeared to be less effective than the one in front of the intake when abundance of alewives was low and water tem-peratures were below 13°C. This result could be due to areas of low sound pressure within the acoustic field along the sides, rear, and top of the intake that permitted small schools of fish to ap-proach the intake. The acoustic coverage around the JAF intake was not uniform. In front of the intake, there was more overlap among the beam patterns of the transducers than among those along the sides, rear, and top of the intake. As the number of fish in a polarized school decreases (assuming similar nearest-neighbor distance), the attention field of the school, i.e., the volume of water within which the school reacts to stimuli, decreases (Nor-ris and Schilt 1988). Thus, smaller schools of ale-wives, with their smaller attention fields, may have been able to fit into the areas of low pressure sound that larger schools could not.
We believe that a better explanation for the ap-parent reduction in effectiveness of the new trans-ducer array at low abundance was a diminished ability of alewives near the intake to respond.
When abundance was low, alewives moved on-shore by the wind event during late May were more responsive to the deterrent system than those al-ready in shallow water. This explanation suggests that alewives that had been severely stressed by the unusually cold winter of 1992-1993 did not move offshore into deeper water as alewives usu-ally do after spawning (Scott and Crossman 1973).
As these fish died or recovered over time, the rel-ative abundance of severely stressed alewives within the shallow water zone would decrease, and thus the effectiveness of the deterrent system would increase. The differences among the dates when the three low-abundance samples were col-lected and the differences among the BACIP es-timates generated from these samples are consis-tent with this hypothesis. For example, most of the test samples collected when water temperatures were below 13°C, alewife abundance was low, and the JAF reactor was shut down were collected over the period from May 18 through May 27 (Table 1), right after the period when spawning probably occurred. The BACIP estimate from this test was 51%. Most of the test samples collected when wa-ter temperatures were below 13°C, alewife abun-dance was low, and both JAF and NM1 were at or near full power were collected either during early May before spawning or during June (Table 2),
after the beginning of the mass die-off of alewives in Lake Ontario in 1993 reported by Schneider and Schaner (1994). The BACIP estimate from this test was 68%. All of the test samples collected when water temperatures were 13°C or above, alewife abundance was low, and both JAF and NMl were at or near full power were collected after mid-June (Table 3). The BACIP estimate from this test was 96%.
If the underlying cause is the failure of severely stressed alewives to move offshore after spawning, the comparison of the 24-h estimate of the effec-tiveness of the deterrent system in 1991 when wa-ter temperatures were below 13°C, alewife abun-dance was low, and the JAF reactor was shut down (86%) and the BACIP estimate from the test con-ducted under the same conditions in 1993 (51%),
provides another measure of the effect of the un-usually cold winter of 1992-1993. This measure suggests that the full effect of the unusually cold winter was not expressed until after the alewives had spawned, which is consistent with the begin-ning of the mass die-off of alewives during late May.
The 24-h estimate of the effectiveness of the deterrent system from 1991 when water temper-atures were below 13°C, alewife abundance was low, and the JAF reactor was shut down also pro-vides a conservative estimate of the effectiveness of the deterrent system on prespawning alewives in 1991 because it was generated from samples collected immediately after the period when ale-wife abundance was high (Ross et al. 1993). The prespawning BACIP estimates in 1993 (81%) were lower than the 24-h estimate from 1991 (86%),
which suggests that the unusually cold winter of 1992-1993 slightly reduced the responsiveness of
prespawning alewives to broadband high-frequen-cy sound.
Overall Effectiveness of the Deterrent System The average daily water temperature recorded at the JAF intake from January through March over the 13-year period from 1981 through 1993 was 3.1°C. The average daily water temperature during this period in 1993 was 1.8°C, well below the 13-year average. In 1991, the average daily water temperature during the period from the beginning of January through the end of March was close (3.3°C) to the 13-year average.
We combined the prespawning, spawning, and postspawning estimates from 1993 to generate an estimate of the overall effectiveness of the deter-rent system following an unusually cold winter.
We used the ratio of the number of alewives im-pinged at NM1 during each test to the total number impinged at NM1 during all tests in 1993 as a weighting factor for the corresponding BACIP es-timate. We multiplied each BACIP estimate by its weighting factor and summed the weighted esti-mates to arrive at the estimate of the overall ef-fectiveness of the deterrent system. The overall effectiveness of the deterrent system during the 1993 study was 81% when the low-temperature, low-abundance, BACIP estimates were included.
When the low-temperature, low-abundance, BA-CIP estimates were replaced by the estimate of effectiveness for the offshore population of post-spawning alewives (81%), the overall effective-ness of the deterrent system was 84%.
To generate an estimate of the overall effec-tiveness of the deterrent system following an av-erage winter, we used the 24-h postspawning es-timate from 1991 (86%) in place of the 1993 pre-spawning estimates (81%), the 24-h spawning es-timate from 1991 (91%) in place of the 1993 spawning estimate (92%), and the 24-h post-spawning estimate from 1991 (86%) in place of the 1993 postspawning estimates when water tem-peratures were below 13°C (51% and 68%), the 1993 estimate (96%) when water temperatures were 13°C or above, and the weighting factors from the 1993 tests. The estimated overall effec-tiveness of the deterrent system following a milder winter is 87%.
Electronically produced, intense (190 dB//u,Pa) high-frequency broadband (122-128 kHz) sound consistently produced a strong and directional avoidance response from healthy alewives in rep-licated tests under controlled conditions with caged fish (Dunning et al. 1992) and later, in the field, with fish unaffected by capture and handling (Ross et al. 1993). The effectiveness of a full-scale deterrent system installed at JAF was confirmed by our field test which used a primary measure-ment variable and an analytical method different from those used by Ross et al. (1993). Collectively, these studies would constitute a successful dem-onstration of a new fish protection measure, except for the absence of comprehensive tests at a wide variety of sites (Tyus and Winter 1992; Cada and Sale 1993; OTA 1995). However, we believe that high-frequency broadband sound will be as effec-tive in decreasing the impingement of alewives at other sites as it was at JAF if the deterrent sound field has no holes, background noises do not mask the high-frequency signals generated by the de-terrent system, there are no strong reflections of the high-frequency signals that make it difficult for alewives to determine the location of the source, and there are no strong currents which pre-vent alewives from moving away from the deter-rent sound field. The similarity of our estimates of effectiveness with hydroacoustic and impingement data indicate that hydroacoustic methods can be used at facilities where a good control site is un-available, or the cost is high for collecting exten-sive time series of paired impingement samples at control and test sites both before and after the installation of a deterrent system.
Hastings et al. (1996) stated that high-frequency sounds, as used at JAF, could potentially damage the ears of alewives if these fish are exposed to an SPL of 180 dB, or even less, for an extended period of time. However, they also concluded that short-term stimulation with sound (e.g., minutes) or stimulation when fish are free to leave the sound field may have little effect on the ear and lateral line. Ross et al. (1993) demonstrated that schools of alewives in front of the JAF intake responded in less than 1 s to high-frequency sound at 156 dB. When fish were swimming toward the front of the intake, they reversed direction. When fish were swimming parallel to or away from the front of the intake, they continued in those directions.
When the sound was turned off, 2-7 min passed before the density of alewives (number/m3) in front of the JAF intake reached pre-sound levels.
The average reduction in the density of alewives in front of the JAF intake was 85% during a period of five weeks when high-frequency sound was pro-duced. These results indicate that alewives near the JAF intake strongly avoided sounds at inten-sities about one-sixteenth the SPL used by Has-
REDUCING IMPINGEMENT WITH SOUND 559 tings et al. (1996) to produce limited and incon-sistent damage to the ears of the oscar Astronotus ocellatus, that alewives did not remain in the sound field at JAF for an extended period of time after it was initially produced, and that relatively few alewives entered the sound field at JAF once it was established. Thus, we believe that a well-de-signed deterrent system is not likely to cause dam-age to the ear of alewives that are capable of swim-ming away from high-frequency sound.
Acknowledgments This project was conducted in cooperation with the Water Quality Subcommittee of the Empire State Electric Energy Research Corp. (ESEERCO).
It was funded by ESEERCO (EP 89-30) and the New York Power Authority (NYPA). We thank John Holsapple of ESEERCO; Mary Alice Ko-eneke of EA Engineering, Science, and Technol-ogy; Pete Dolan of Sonalysts, Inc.; Hugh Flanagan of Niagara Mohawk Power Corp.; Lucio Lombar-dozzi of NYPA; and the staff of the James A.
FitzPatrick Nuclear Power Plant for their assis-tance.
References Analytical Software. 1994. Statistix, version 4.1. An-alytical Software, Tallahassee, Florida.
Bergstedt, R. A., and R. O'Gorman. 1989. Distribution of alewives in southeastern Lake Ontario in autumn and winter: a clue to winter mortalities. Transactions of the American Fisheries Society 118:687-692.
Brandt, S. B. 1986. Food of trout and salmon in Lake Ontario. Journal of Great Lakes Research 12:200-205.
Burdic, W. S. 1984. Underwater acoustic system anal-ysis. Prentice-Hall, Englewood Cliffs, New Jersey.
Cada, G. F, and M. J. Sale. 1993. Status of fish passage facilities at non-federal hydropower projects. Fish-eries 18(7):4-13.
Dunning, D. J., and five coauthors. 1992. Alewives avoid high-frequency sound. North American Jour-nal of Fisheries Management 12:407-416.
Eck, G. W.. and L. Wells. 1987. Recent changes in Lake Michigan's fish community and their probable caus-es, with emphasis on the role of alewife (Alosapseu-doharengus). Canadian Journal of Fisheries and Aquatic Sciences 44(Supplement 2):53-60.
Elrod. J. H.. and R. O'Gorman. 1991. Diet of juvenile lake trout in southern Lake Ontario in relation to abundance and size of prey fishes, 1979-1987.
Transactions of the American Fisheries Society 120:
290-302.
Hastings, M. C, A. N. Popper, J. J. Finneran, and P. J.
Lanford. 1996. Effect of low-frequency underwater sound on hair cells of the inner ear and lateral line of the teleost fish Astronotus ocellatus. Journal of the Acoustic Society of America 99:1759-1776.
Jones, M. L.. J. F. Koonce, and R. O'Gorman. 1993.
Sustainability of hatchery-dependent salmonine fisheries in Lake Ontario: the conflict between pred-ator demand and prey supply. Transactions of the American Fisheries Society 122:1002-1018.
Leach, J. H., and five coauthors. 1987. A review of methods for prediction of potential fish production with application to the Great Lakes and Lake Win-nipeg. Canadian Journal of Fisheries and Aquatic Sciences 44(Supplemenl 2):471-485.
Norris. K. S., and C. R. Schilt. 1988. Cooperative so-cieties in three-dimensional space: on the origins of aggregations, flocks, and schools, with special ref-erence to dolphins and fish. Ethology and Socio-biology 9:149-179.
O'Gorman, R., and C. P. Schneider. 1986. Dynamics of alewives in Lake Ontario following a mass mortal-ity. Transactions of the American Fisheries Society 115:1-14.
OTA (Office of Technology Assessment). 1995. Fish passage technologies: protection at hydropower fa-cilities. U. S. Government Printing Office, OTA-ENV-641, Washington, D.C.
Ross, Q. E., and five coauthors. 1993. Response of ale-wives to high-frequency sound at a power plant on Lake Ontario. North American Journal of Fisheries Management 13:291-303.
Schneider, C. R, and T. Schaner. 1994. The status of pelagic prey stocks in Lake Ontario in 1993. New York State Department of Environmental Conser-vation 1994 Annual Report from Bureau of Fish-eries Lake Ontario Unit to the Lake Ontario Com-mittee and the Great Lakes Fishery Commission, Ann Arbor, Michigan.
Scott, W. B., and E. J. Crossman. 1973. Freshwater fishes of Canada. Fisheries Research Board of Can-ada Bulletin 184.
Smith, C. L. 1985. The inland fishes of New York State.
New York Stale Department of Environmental Con-servation. Albany.
Stewart-Oaten, A., J. R. Bence, and C. W. Osenberg.
1992. Assessing effects of unreplicated perturba-tions: no simple solutions. Ecology 73:1396-1404.
Stewart-Oaten, A., W. W. Murdoch, and K. R. Parker.
1986. Environmental impact assessment: pseudo-replication in time? Ecology 67:929-940.
Talhelm, D. R. 1988. Economics of Great Lakes fish-eries: a 1985 assessment. Great Lakes Fishery Com-mission Technical Report 54.
Tyus, H. M., and B. D. Winter. 1992. Hydropower de-velopment. Fisheries 17(l):30-32.