Thomas C. Pratt, Davi R. Stanly, Sott Shlutr, Jak K.L. La Ros,Anrw Winstok, Paul T. Jaoson

a fisheries and Oceans Canada, Great Lakes Laboratory for fisheries and Aquatic Sciences, 1219 Queen Street East, Sault Ste. Marie, Ontario, P6A 2E5, Canada

bOntario Power Generation Inc., Environment Division, 14000 Niagara Parkway, Niagara on the Lake, Ontario, L0S 1J0, Canada

cUnited States fish and Wildlife Service, New York Field Office, 3817 Luker Road, Cortland, NY, 13045, USA

dOntario Ministry of Natural Resources and Forestry, Lake Ontario Management Unit, 41 Hatchery Lane, Picton, Ontario, K0K 2T0, Canada

eNew York Power Authority, Environment Division, 123 Main St, White Plains, NY, 10601, USA

fElectric Power Research Institute, 3420 Hillview Avenue, Palo Alto, CA, 94304, USA

ABSTRACT

Keywords:

American eel

Downstream passage

Behavioral guidance

Light diversion

1.Introduction

Anthropic activities have resulted in a network of barriers that limit the movement of aquatic organisms and have immense biodiversity implications (Grill et al., 2015; Liermann et al., 2012; Nilsson et al.,2005). Recent estimates indicate 48% of global river volume is severely impacted by barriers and/or flow regulation, and this will increase to 93% if planned projects are completed in the Amazon, Mekong and Congo basins (Grill et al., 2015; Winemiller et al., 2016). The increasing demand for power from hydroelectric generating stations is being driven by developing nations, and an estimated 3700 generating stations ＞1 MW are planned to meet this demand (Zarfl et al., 2015). These additional barriers will add further pressure to fish species where barriers to migration are particularly acute, including diadromous fishes that are obligate migrants such as eels from the family Anguillidae(Liermann et al., 2012).

Anguillid eels are globally distributed and have a complicated life history strategy. They are considered to be facultatively catadromous,meaning that they spawn in the ocean and spend their growth stage in freshwater lakes and rivers, brackish estuaries and nearshore oceanic coastal areas, often moving between these disparate habitats (Arai &Chino, 2012; Jacoby et al., 2015). Most species are panmictic and undergo long migrations to reach their continental rearing and oceanic spawning habitats (Tesch, 2003). Barriers, in particular hydroelectric generating stations, are an important threat for eel that reside in freshwater as they must safely pass both upstream (as juveniles) and downstream (as adults) of the stations to complete their migration.Despite their presumed capacity to adapt to anthropic pressures due to their wide distribution and use of a variety of habitats, north temperate anguillid eels are in decline (Drouineau et al., 2018). Four anguillid species, including the American eel (Anguilla rostrata), are considered Threatened by the International Union for Conservation of Nature Anguillid Specialist Sub-Group (Jacoby et al., 2015).

There is general agreement that the abundance of the American eel is lower than 30 years ago, though observed declines are not consistent across the species large geographic range (Atlantic States Marine fisheries Commission (ASMFC) (2017); Bowlby, 2018; Cairns et al., 2014;Haro et al., 2000). The most severe decline is recorded in one of the longest recruitment time series available, from eel ladders located on the St. Lawrence River on the U.S.-Canada border. One ladder, located at the Moses-Saunders hydroelectric generating station, was operated by the Ontario Ministry of Natural Resources and Forestry and Ontario Power Generation from 1972 to 2005; there is an observed decline of 95%-99% in the number of American eel recruiting to the upper St. Lawrence River/Lake Ontario (USLR/LO) since the ladder opened in 1972 (Cairns et al., 2014; Casselman, 2003). Starting in 2006 an additional ladder was added and operated by New York Power Authority. As a result of declines over the range of the species, the American eel was listed in the Province of Ontario as Endangered under the provincial Endangered Species Act, and has been assessed as Threatened in Canada by the Committee on the Status of Endangered Wildlife in Canada (Committee on the Status of Endangered Wildlife in Canada (COSEWIC), 2012).

The main concern for population protection is the downstream migration of adult (silver-stage) eel. Upstream passage for juvenile anguillids is considered relatively straightforward, with established design and operational parameters available (e.g., Haro, 2013; Knights& White, 1998). However, identifying safe downstream passage solutions for anguillids remains an immense challenge for fisheries managers worldwide (Drouineau et al., 2018). Most years, nearly all the water that leaves the Laurentian Great Lakes passes through the turbines of the Moses-Saunders hydroelectric generating station, with an estimated 26.4% mortality rate for out-migrating eel during turbine passage(Verreault & Dumont, 2003). Any American eel that pass up the eel ladder must out-migrate back down through the station when they reach maturity to successfully complete their life cycle.

There are solutions for protecting out-migrating eels from hydroelectric generating stations in small rivers, but these are more difficult in large rivers owing simply to large station footprint and greater volume of water passing through the station. Solutions include diversions from turbine passage (e.g., screening, behavioral diversion, bypasses) and temporary turbine shut-downs (e.g., Boub´ee & Williams, 2006; Calles et al., 2013; Eyler et al., 2016; Piper et al., 2015). The challenge of preventing eel from transiting downstream through turbines is magnified on larger rivers, as out-migrating eel follow the main flow of the river (Jansen et al., 2007). The St. Lawrence River is the out flow of the Laurentian Great Lakes, and has a considerable average annual discharge of 7410 m/s at the out flow of Lake Ontario (Benke & Cusing,2005). These extreme conditions make any of the downstream passage solutions developed for smaller hydroelectric generating stations infeasible on a river the size of the St. Lawrence.

The immense challenge of working towards a downstream passage solution on a large river has brought together hydroelectric producers and fisheries regulators to collaborate towards a common goal of costeffective downstream American eel passage on the St. Lawrence. The Eel Passage Research Center (EPRC) was developed and is managed by the Electric Power Research Institute (EPRI) to address this challenge.The Center recently completed an initial 5-year term to address the longterm goal of maximizing the survival rate of American eel that would otherwise pass through turbines along the St. Lawrence River without significantly reducing power production. In this manuscript, we provide detailed summaries of research developed by the EPRC and completed by consultants and contractors on their behalf. In particular, we assess the potential for manipulated flow fields, electricity, sound, and electromagnetic frequencies to be used as behavioral guidance techniques in large rivers. We also investigated the use of new sonar technologies to allow the assessment of out-migrating eel without the requirement of capturing and handling. Based on the outcomes of the research, we commissioned two white papers on the most promising guidance techniques, the use of sound and light, to guide or deter eel.

2.The St. Lawrence River and the development of the Eel Passage Research Center

The St. Lawrence River is ～1400 km in length from the outlet of Lake Ontario to the Gulf of St. Lawrence. There are two mainstem hydroelectric generating stations, the Robert Moses-Robert H. Saunders Power Dam (hereafter Moses-Saunders; 45024.01N, 744740.22W)jointly operated by Ontario Power Generation Inc. and the New York Power Authority, and the Beauharnois Generating Station (451849.72N, 735427.89W) operated by Hydro-Qu´ebec (Fig. 1). Nearly all the river’s flow passes through turbines at the Moses-Saunders dam,which has a maximum operational flow of 10,704 m/s and a total generating capacity of 1957 MW. Approximately 80% of the flow passes through turbines at the 1900 MW Beauharnois Generating Station,which has a maximum operational flow of 8200 m/s. Flows at Beauharnois are controlled by out flows from the Moses-Saunders at it operates as a run-of-river facility; Moses-Saunders has limited peaking and ponding ability, and also essentially operates as a run-of-river facility(EPRI, 2018a). Upstream of the Moses-Saunders is the Iroquois Water Control Structure (IWCS), a non-hydroelectric dam used to limit high water levels downstream in Lake St. Lawrence and for ice management in the winter (Fig. 1).

The EPRC was established in 2013 to address the challenge of providing safe downstream passage for out-migrating adult American eel on the St. Lawrence River (EPRI 2013). The EPRC is funded by Ontario Power Generation, Hydro-Qu´ebec, and the fish Enhancement,Mitigation, and Research Fund (FEMRF) administered by the U.S. fish and Wildlife Service. Duke Energy was an additional funder during the first term (2013-2018).

The bi-national, EPRI-led EPRC was established in 2013 to meet the need for coordinated, collaboratively-funded research to address the challenge of safe downstream passage of American eel at hydropower projects on the St. Lawrence River (EPRI 2013). The initiative is a direct outgrowth of long-standing collaboration among members and builds upon the substantial research effort directed toward American eel passage on the St. Lawrence River by the current EPRC funders over an extended period of time, both individually and collectively.

The large size of the St. Lawrence River and the two generating stations preclude screening and physical guidance technologies. Therefore, the EPRC has adopted an adaptive, collaborative process to plan and execute a research program on behavioral guidance of downstream migrating adult eel. The EPRC’s goal is to develop the technology to behaviorally guide eel to collection points for capture and transfer around the Moses-Saunders Power Dam and the Beauharnois Generating Station (Fig. 1).

Fig. 1.The St. Lawrence River, including the location of the two mainstem hydroelectric generating stations and the Iroquois Water Control Structure. Map courtesy of New York Power Authority.

Research is guided by a technical committee comprising representatives of EPRI, EPRC funding organizations, and resource management agencies with regulatory authority over hydropower activities on the St.Lawrence River. The technical committee collaboratively establishes the research priorities and plan, develops scopes of work, reviews proposals and drafts project reports. The EPRC has fully funded six major research and development projects and has supported several other projects relevant to the goals and objectives of the EPRC. The purpose, background, and major research findings of the EPRC are outlined below.

At the outset, the EPRC identified the following long-term goal for its R&D activities:

Maximize the survival rate of eels that would otherwise pass through turbines at Moses-Saunders Power Dam and Beauharnois Generating Station without significantly reducing power production.

Research and development to attain this goal support three management objectives, including 1) to concentrate adult eel for collection above Moses-Saunders and Beauharnois; 2) to collect and transfer adult eel downstream around turbines at Moses-Saunders and Beauharnois;and 3) to demonstrate effectiveness of the selected methods. Physical screening of intakes had been deemed infeasible given the size of the river and the facilities, thus research to address the first objective investigates and develops one or more technologies to guide eel to a collection or bypass location using the eels’ innate behavioral response to sensory stimulation (i.e., taxis). The eel literature indicated a number of stimuli to be investigated, including light, sound, electricity, electromagnetic field, water velocity, turbulence, or shear, and chemicals(either individually or in combination).

3.Assessment of behavioral guidance stimuli

3.1.Large flume electrical and flow guidance

The primary objective of the large flume study was to assess whether an electric guidance array, a flow velocity enhancement system, or the two stimuli working in concert could guide silver-stage American eel towards a downstream collection area.

3.1.1.Methods

This research was contracted to Alden Research Laboratory, Inc.,who worked with Blue Leaf Environmental, Scientific Solutions, Inc.,Smith Root, Inc., and Natural Solutions to deliver the science. A brief description of the methods follows, but full details are available (EPRI,2016).

The evaluation of the electrical and flow guidance systems with silver eel was conducted in a large re-circulating flume. Water depth in the flume was 2.4 m, and the testing area was 24.3 m in length and 5.2 m wide for the electric guidance and flow velocity enhancement system assessment. During testing, flows were kept at 0.9 m/s. The final design incorporated a fish acclimation and release pen, the electrical guidance array oriented at a 30angle, the flow enhancement system, and three collection “bins” at the downstream end of the channel (Fig. 2). The acclimation pen was moved from the left-hand wall to the top of the flume halfway through the experiment. The electrical guidance array contained three rows of electrodes spaced 1.07 m apart, suspended from the surface. The flow guidance system used an eductor nozzle that was 7.6 cm in diameter, 76 cm long, and produced a flow of 31 l/s. The goal for both guidance systems, whether used alone or in combination, was to move eel toward the bypass collection bin as they moved downstream in the flume.

Fig. 2.Plan view of large flume set-up for the electrical and flow guidance trials.

Silver-stage American eel for the electrical and flow guidance trials were captured in the St. Lawrence River estuary and transported to the Alden Research Laboratory in Holden, Massachusetts. Six hundred and sixty eel were tagged with model 795-LG acoustic tags which were programmed to transmit every 0.6-1.3 s, and individual eel movements were tracked via 15 hydrophones deployed at various depths in the flume using the HTI Acoustic Tracking System. The difference in arrival times of signals from each eel on the hydrophones were used to calculate a three-dimensional position in the flume.

The experimental design for the evaluation of the electrical and flow guidance trials included a goal of 150 replicate trials conducted with each of the three behavioral stimulus treatments (electrical guidance alone, flow guidance alone, and electrical and flow guidance in combination) and 150 control trials. Trials were initially divided into five blocks of replicate trials conducted with three behavioral cue treatments and a control (i.e., four test conditions) with 30 test animals per block.The controls were conducted in the same manner as the behavioral stimulus trials but with both stimulus devices turned off, and the electrodes removed from the water. An additional 60 control trials were conducted with the electrical guidance electrodes left in the flume to eliminate any possibility of eel behavior being affected by their presence. All trials were concluded after 2 h, completed at night, and in dark conditions. The response of eel to the electrical and flow stimuli were assessed by 1) determining which collection bin eel were detected in using a simple Pearson’s chi-square analysis, and 2) examining the three dimensional tracks of eel in the flume, which provided data on movement speed and direction. The spatial and temporal data obtained from the 3D telemetry tracks were consolidated into a small number of descriptive statistics capable of capturing important information about behavior before and after the initial interaction with one or both of the guidance stimuli; complete details on these tests are available from EPRI(2016).

3.1.2.Results

Movement and collection bin data were acquired from 652 American eel (mean length 902 mm, mean weight 1.54 kg), including 149 from electric guidance, 147 from flow guidance, 151 from electrical and flow guidance, and 145 from control trials. Data from some individual eels were not collected due to tag or hydrophone failures.

Assessment of the collection bin data showed no effect of electrical or flow guidance, either alone or in combination, on the expected location of American eel collection (Table 1). While there were significantdifferences in the number of eel captured in each bin, the bypass bin consistently had lower numbers of eel captured across all three treatments and in the two control treatments (Table 1). While the corresponding telemetry tracking data demonstrated that a number of eels were able to leave and re-enter the collection bins, making the data unreliable for assessing guidance efficiency, it was clear that most eel passed through the guidance stimuli during the trials. In addition, a surprising number of American eel remained upstream and did not enter the collection bins over the 2-h test, despite the relatively high flow (0.9 m/s) test conditions (Table 1).

Table 1Pearson’s chi-square analysis of the number eel of recovered at the end of each trial by collection location (bypass collection bin, middle collection bin, lefthand collection bin, or eel that remained upstream and were not caught in any bin) and guidance test type (EGS =electrical guidance, FVES = flow guidance, EGS/FVES =electrical and flow guidance in combination, control 1 =no guidance stimuli but with the electrical array in place in the flume, control 2=no guidance stimuli and the electrical array removed). P-values with an asterisk indicate statistically significant differences (p ＜0.05).

Both the density (Fig. 3) and individual tracking (Fig. 4) plots show a general tendency to prefer the side of the flume away from the electrical and flow stimuli, but these same preferences were also seen under control conditions. There were no statistical differences detected in the spatial distribution of individual eels between any of the treatments and the controls for either the entire 2 h trial, or during the 10 min time period evaluated prior to and after eel entered the area of a stimulus field. Ultimately, the analysis of eel movements following initial encounters with the stimulus fields did not detect any consistent differences between stimulus and control trials.

Fig. 3.Density plots developed from eel tracking data by test condition and trial block. Color contours indicate the proportion of eel that were detected in each cell of a horizontal grid over the 2 h trial duration. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4.An example of the detection tracks for each American eel tested in the large flume study. A single example is provided for each of the electrical guidance(EGS), flow velocity enhancement (FVES), electrical and flow guidance in combination (EGS/FVES), and the control trials. A black dot marks the first observation obtained for each eel, and the subsequent location for each recorded observation is plotted by a triangle symbol. The color of the triangle indicates time as indicated on the ten category scale on the left, with red symbols indicating early in the trial and blue symbols indicating near the end of the observations for that particular trial. Black symbols indicate eel that are in the guidance zone. The in fluence of the electrical guidance zone is shown in yellow (EGS plot) or pink (EGS/FVES plot),while the in fluence of the flow guidance zone is shown in yellow. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.2.Vibration (particle motion)

The primary objective of the vibration study was to determine if vibration could elicit behavioral responses from silver-stage American eel, including startle and/or directional avoidance.

3.2.1.Methods

The use of vibration-induced stimuli to guide eel away from turbines was also examined as part of the EPRC’s laboratory study of behavioral stimuli conducted by Alden Research Laboratory, using the same source of silver-stage American eel as the large flume guidance study. The vibration response evaluations were conducted in a small flume test facility using a closed-loop system measuring 24 m long by 1.8 m wide by 1.8 m deep. The test enclosures were divided into equally sized quadrants (i.e. 4.9 m apart) for the purpose of tracking fish locations and analyzing movements and responses based on spatial changes in eel distributions through time. A flow velocity measurement of 0.15 m/s was maintained throughout the duration of the study.

Two electrodynamic shakers were placed on the outside of the isolation screens at each end of the test enclosure in order achieve the targeted level of vibration. Vibration testing focused on frequencies within the known sensory range of eels. Preliminary testing showed that the three signals that elicited the strongest response included: (1) 100 ms, 10 Hz tone burst, (2) 100 ms, 50 Hz tone burst, and (3) 10 ms, 50 Hz tone half sine impulse.

Tracking eel movements within the test enclosures under dark test conditions required eel to be tagged with glow sticks attached to Floy tags inserted into the dorsal musculature. The study consisted of three blocks of trials conducted with each signal and a control for a total of 12 trials. Testing was randomized within each block and consisted of 15 eel with no previous exposure to the experimental vibration-induced stimuli, with each trial consisting of a 20-min pre-exposure, exposure and post-exposure period. Control trials utilized the same test parameters,but with no vibration-induced stimuli. fish were recorded in each quadrant at 1-min intervals during all three 20-min exposure periods.The response of eel to vibration-induced stimuli was determined by analyzing fish distributions through time during all three exposure periods. Complete details on the study design and analysis are available in EPRI (2016).

3.2.2.Results

Results from repeated measures ANOVAs revealed an avoidance reaction in silver-stage American eel responses to all three vibrationinduced stimuli, as the condition-by-period interaction statistics were statistically significant for the 100 ms-10 Hz tone (p =0.004), the 100 ms-50 Hz tone (p =0.024) and the 10 ms-50 Hz half sine impulse (p =0.038). While all three vibration-induced stimuli significantly in fluenced eel distribution, the 10 ms-50 Hz half sine impulse had the weakest avoidance reaction (Fig. 5).

Fig. 5.Comparison of displacement of the center-of-school measure for the experimental periods (pre-exposure, exposure, and post-exposure) of trials conducted with the 100 ms-10 Hz tone (left-hand figure), the 100 ms-50 Hz tone (middle figure), and the 10 ms-50 Hz half sine impulse (right-hand figure) vibration signals to the control trial.

3.3.Electromagnetic fields (EMF)

A laboratory study on electromagnetic fields (EMF) was conducted in parallel to the vibration study outlined above, with the objective of determining if EMF could elicit behavioral avoidance responses from silver-stage American eel.

3.3.1.Methods

The EMF study was conducted in Alden Research Laboratory’s small test flume with similar materials and methods used during the vibration testing. A smaller test enclosure (1.2 m ×2.4 m x 0.9 m) for the EMF study was constructed of a non-conductive PVC and supported by a fiberglass frame in order to not distort or affect the EMF field. Both ends of the enclosure had 1.9 cm mesh screens to allow flowing water to pass through while helping maintain a uniform flow velocity of 0.15 m/s during all trials. In order to produce a detectable EMF Field, a single EMF source (electromagnet) was installed at each end of the tank. The electromagnet in the half of the test enclosure where most eels were located at the end of a pre-exposure period was activated at the beginning of the exposure period.

All trials were conducted during the nighttime hours (at dusk) using the same tracking methods (i.e. glow stick attached to Floy tags) utilized in the vibration study. The study consisted of three blocks of trials using a full-strength EMF field and a control for a total of six trials. The target sample size was 10 eel per trial with each trial consisting of a 20-min pre-exposure, exposure and post-exposure period. Eel distribution was recorded at 2-min intervals and, similar to the vibration study, the response of eel to EMF was determined by analyzing fish distributions through time during all three exposure periods. Complete details on the study design and analysis are available in EPRI (2016).

3.3.2.Results

Silver eel did not demonstrate any discernible response during exposure to EMF during all study trials (Fig. 6). The results of a repeated measures ANOVA conducted with the mean number of eel and center-of school positions was not significant and no avoidance of the EMF field.In particular, the lack of statistical significance for the condition by period interaction statistic indicated that control and EMF exposed eel were distributed similarly over the three observation periods.

Fig. 6.Box and whisker plots comparing EMF exposure and control center-of-school data by observation period for the three trial blocks.

3.4.Small flume and tank electrical guidance

The EPRC felt that the surprising outcome (no consistent or significant guidance) of the large flume electrical guidance study described above necessitated a further evaluation of eel behavior in response to electrical fields in combination with several important environmental variables. In response, the EPRC commissioned a controlled, laboratorybased study of the behavioral responses of silver-stage American eel and European eel (A. anguilla) in response to pulsed direct current (PDC)fields. To help evaluate the potential of electrical fields as guidance stimuli for out-migrating American eel on the St Lawrence River, the research focused on the following objectives: 1) to determine the threshold field strengths for behavioral responses of large silver eel,including twitch, loss of orientation and paralysis; 2) to determine the effect of water conductivity on threshold response field strengths; and 3)to determine if the responses differ under static and flowing water conditions with water velocities similar to the St Lawrence River.

3.4.1.Static tank methods

Threshold response, conductivity and large tank electrode studies took place in static tanks located at the USGS S.O. Conte Anadromous fish Research Laboratory in Turners Falls, Massachusetts, USA. These studies focused on the responses of silver-stage American eel collected from the Connecticut River in October 2017.

Trials were carried out either at night or during the day with darkened tanks. Eel behavioral responses were recorded with video equipment and quantified, in some cases using software to aid in interpretation. The responses of eel to increases in electrical field voltage were classified into four distinct behaviors: no response, twitch, loss of orientation and paralysis. The threshold electrical field response of eel for each behavior was defined as the least voltage required to elicit a behavior in each fish. To characterize the field strength needed to generate specific behavioral responses, American eel were exposed to various PDC electrical fields as well as a control (no current) in small,static water tanks (EPRI, 2020a). Water conductivity was varied in the test tanks according to three treatments (50, 275 and 500 μS), to also evaluate the effect of environmental in fluences on the response of eel.Following a control period of no current, PDC stimuli was applied at 10 Hz (square wave, 10% duty cycle), 5v for 10 s followed by a 1-min rest period. This procedure was repeated with increasing field strength until paralysis response was observed. American eel were also exposed to a large-scale electric array in a larger test tank to measure their threshold electrical field response with more available space, a condition closer to afield environment or in-situ test (EPRI, 2020a). While the application of the electrical field stimuli was replicated from the conductivity tests, the water conductivity was held relatively constant at 93-99 μS.

3.4.1.1.Results.In the small tank static water conductivity tests, the lowest threshold strengths were observed for twitch (mean range =0.159-0.180 V cm-), followed by loss of orientation (mean range =0.207-0.263 V cm) and paralysis (mean range =0.274-0.386 V cm)responses resulting from the highest electrical field strengths (Fig. 7). A small decrease in mean threshold field strength was observed for nearly all responses with increasing water conductivity, though few differences showed statistical significance. Mean threshold field strengths and responses in the large static water tank experiment followed the same order as did those in the conductivity tests, though absolute strengths were significantly lower than for the most similar conductivity treatment observed in small tanks (Fig. 7).

Fig. 7.Threshold response field strengths for three potential behavioral responses in American eel (twitch, loss of orientation and paralysis) in three levels of water conductivity in a small tank test, and one level of conductivity in a large tank test. Asterisks and daggers indicate significant differences between pairs of mean threshold response field strengths within each group. Full details are available in EPRI (2020a).

3.4.2.Small flume methods

Building on the results from the static water tests above, a study investigating European eel responses to different electrical fields in flowing water were conducted by researchers at the University of Southampton, Southampton UK. Silver-stage European eel were supplied for this research from a commercial fishery source on the River Humber, UK. Experiments to measure the response of eel to electrical fields in flowing water were completed to explore how a river environment would in fluence a possible future guidance application. In these experiments, eel were subjected to two flow velocities within the range observed in the St Lawrence River and two electrical field strengths corresponding with the mean twitch and paralysis field strengths:

·low velocity, mean twitch field (0.5 m/s, 0.15 V/cm)

·low velocity, mean paralysis field (0.5 m/s, 0.30 V /cm)

·high velocity, mean twitch field (1.0 m/s, 0.15 V/cm)

·high velocity, mean paralysis field (1.0 m/s, 0.3 V/cm)

After an acclimatization period with flowing water in a flume tank,eel were exposed to electrical fields while behaviors were recorded for up to an hour or until the eel passed through the electrical array/field.The speed and direction of travel of eel were calculated from video footage of their responses to electrical fields. The responses were then classified according to several behaviors including: no response (no change in swimming speed or direction), acceleration (increase in speed), a switch in orientation (change in body directions of 90-360)and rejection (180change in direction and minor upstream movement).

3.4.2.1.Results.In the flowing water experiment, there were no differences in behavioral responses between the two field strengths but there was a statistically significant effect of flow velocity on behavioral response. In the low velocity condition, 74% of the eels exhibited a response, whereas in the high velocity condition only 31.2% of the eels responded to the electrical field. Specifically, the incidence of rejection declined significantly from 32.5% under the low velocity condition to 4.0% at the higher velocity (Fig. 8). Upon first encountering the electrical field under high flow conditions, 87.7% of tested eels either exhibited no response (67.5%) or acceleration (20.2%). Orientation switching behavior was observed in 8.3% of the eels.

Fig. 8.The effect of flow velocity, 0.5 m/s and 1.0 m/s on the percentage of four different behavioral responses observed.

A table summarizing the objectives, study design and outcomes of all of the guidance studies supported by the EPRC during its first 5 year term are presented in Table 2.

Table 2Summary of outcomes from behavioral guidance studies initiated by the Eel Passage Research Center.

Table 2(continued)

4.Acoustic detection of out-migrating eel

4.1.Assessment of acoustic technologies to study downstream migrating American eel approach and behavior at Iroquois Water Control Structure

An important consideration regardless of whatever technology is used for eel guidance is that the location of out-migrating eel in the water column must be determined to properly assess effectiveness. The location of guidance structures need to intercept eel migration paths in the river, and the eel need to be monitored to see if they are being attracted or repelled from the stimulus. Identifying a way to accomplish this without having to capture, handle and tag eels would be advantageous. The primary objectives of this study were to determine if existing acoustic technologies could be used to 1) estimate the relative abundance and distribution of out-migrating silver-phase American eel as they migrate down the St. Lawrence River and approach the IWCS; and 2) describe their behavior in the near field (＜15 m upstream) of IWCS(EPRI, 2017a). Behaviors of primary interest were diurnal variation in downstream movement, preference in horizontal and vertical distribution, reaction to structure, and whether eel out-migrate individually or in groups.

4.1.1.Methods

The study was conducted at the IWCS (44505.42N, 751810.66W), a flow-regulating dam, located 45 km upstream of the Moses-Saunders generating station on the St. Lawrence River. It stretches 600 m across the St. Lawrence River and consists of 32 sluiceways, each 15.2 m wide and equipped with a vertical lift gate (Fig. 9).

Five contiguous sluiceway openings with the highest flows were chosen for placement of the sonar gear, as sluiceways with high flows would increase the chances for eel encounters and subsequent visualization of eel by the sonars. Sluiceway openings for Gates 7 through 11 were selected as the sampling area for this study (Fig. 9).

Fig. 9.Overhead view of Iroquois Water Control Structure. The sampling area upstream of Gates 7 to 11 is illustrated by the red arrow. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Based on existing information and their wide use in fisheries applications three echosounding technologies were used; split-beam echosounders and multibeam sonars were two categories of sonar technologies that would most likely meet study objectives. The three sonar technologies selected for this evaluation were the Simrad 120 kHz EK60 echosounder with one 7circular split-beam transducer, a Sound Metrics ARIS Explorer 1800 sonar (1.1 and 1.8 MHz), and the Kongsberg Mesotech 500 kHz M3 multi-mode multibeam sonar.

Data for multiple sonar sampling configurations were collected from June through September 2015. Transducers were mounted and positioned to optimize acoustic coverage of sluiceways 7 through 11 and enhance acoustic backscattering of out-migrating adult eel. Complete details of the study are available in EPRI (2017a).

The intent of the planned sampling zone was to provide concurrent acoustic coverage by each of the technologies being tested. Maximum acoustic coverage of the water column for the EK60 was achieved by aiming the 7split-beam transducer horizontally, slightly upstream from the dam, and the ensonified area extended 88 m from the transducer,covering the entire water column (Fig. 10). To provide concurrent acoustic coverage with the multi-beam sonars, the split-beam was oriented to sample the same regions as the multi-beams. At 1800 kHz, the ARIS was able to sample approximately one sluiceway opening (15 m),while at 1100 kHz, the ARIS covered a distance of 33 m, or two sluiceway openings. The Mesotech M3 sonar was anticipated to provide intermediate coverage of up to 45 m of range, with a broader angle of coverage (Fig. 10).

Fig. 10.Aerial view of the initial location of the Simrad EK60 split-beam transducer (7°circular beam width), ARIS Explorer 1800 imaging sonar (28°horizontal x 28°vertical with spreader lens), and Kongsberg Mesotech M3 multibeam sonar (140°horizontal x 30°vertical in EIQ mode) installed on Iroquois Water Control Structure.

The first phase of the research involved determining software settings and transducer deployment options to find preferred methods to monitor eel. In phase 2, live yellow-stage American eel (size range 30-80 cm) were tethered to surface floats and released upstream of the sonar beams and allowed to swim through the study area at known depths.

4.1.3.Results

Phase 1 demonstrated that the ARIS sonar was effective at the low frequency (1100 MHz) mode (Fig. 11). The 28spreader lens provided sufficient intensity for detecting eel-sized targets out to a range of 15 m.The M3 sonar could ensonify the water over the entire water column to a distance of 50 m and the EK60 spilt beam had a range of 88 m and a beam width of approximately 10 m at that range.

Fig. 11.Images from an ARIS Explorer imaging multibeam sonar of (A) an eel (circled) moving over a debris-covered intake, and (B) the pronounced sawtoothshaped echo trace of an eel in an echogram.

The ARIS imagery had motion artifacts that made it difficult to distinguish eel from non-eel objects (e.g., a 1 m stick), especially when the image distortion of non-eel objects mimicked the anguilliform swimming motion of eels. The motion artifacts, in combination with the current speed, sampling area and the abundance of other confounding targets (i.e., the issue of false positives), made identifications of eel challenging at IWCS. The M3 and EK60 sonars were not capable of distinguishing eel from non-eel targets.

During Phase 2, targets were detected with both the ARIS and M3 sonars, but the wider M3field of view and lower frequency allowed targets to be detected over a larger area for this unit. However, identification of eel targets was not possible with the M3 sonar. Targets were detected by the three sonars at multiple ranges, but the range test demonstrated successful rate of identification of known tethered eel by the ARIS sonar was best at distances ＜10 m; identification rates then decreased at distances ＞10 m.

To test the potential of a relative index of eel abundance, a sonar monitoring period for out-migrating adult American eel at IWCS examined two periods from July 15-22 and September 17-19, 2015.During the July period when the ARIS sonar was mounted on Pier Nose 10, no targets detected by the ARIS sonar were identified as eel. However, in September, two targets at ranges 8.5 m and 11.2 m were identified as eel during nocturnal hours. For each of these targets, the echograms of maximum echo intensity clearly illustrated the diagnostic“sawtooth” echo pattern produced by the anguilliform swimming motion (Fig. 11).

4.2.Machine learning for eel detection in sonar data

While eel could be identified with relatively high classification accuracy by acoustic analysts in the ARIS sonar data, the analysis was time-consuming (and therefore expensive) and subjective. Adult eel in the upper St. Lawrence River migrate downstream over a period of time extending from the end of May to November (McGrath et al., 2003), and as a consequence monitoring of eel outmigration must occur over an extended period of time. The volume of data accumulated over a single outmigration season at a single location would be exorbitant to process.Thus, the sonar data was used in a machine learning exercise with the objective of automated identification of American eel targets.

4.2.1.Methods

EPRI and the Pacific Northwest National Laboratory (PNNL) secured funding from the U.S. Department of Energy’s Water Power Technology Office to demonstrate the feasibility of using deep learning (a form of machine learning/artificial intelligence) to automate the identification of eel in multibeam sonar data (EPRI, 2020b). The project exploited new data collected in the laboratory and the existing field data from the field study (EPRI, 2017a). The DOE-funded project developed and tested deep learning and other data analytic tools, including wavelet filtering, differencing for static object removal, and convolutional neural network analysis (Figs. 12 and 13) to provide proof of concept for automated identification of eel (EPRI, 2020b; Yin et al., 2020). Complete details of the automated learning study are available (EPRI, 2020; Yin et al., 2020,Zang et al., in review); the software tools developed by the project are posted at: https://github.com/xzang/Deep-learning-for-sonar-images[github.com].

Fig. 12.Work flow of sonar imaging processing and object extraction. (a) original image. (b) differenced image obtained by differencing four adjacent frames from the focal frame. (c) differenced image denoised by wavelet transform. (d) binary image obtained by applying a binary filter with threshold pixel intensity. A 61 ×61 pixel window slides across the image frame to screen potential objects of interest. Grayscale bars denote pixel intensity. Source: Zang et al. (in review).

Fig. 13.Architecture of the convolutional neural network for classification of extracted sonar images of eel and non-eel objects. Source: Zang et al. (in review).

4.2.2.Results

The analysis of the laboratory data demonstrated feasibility of the approach, revealed object characteristics observed with the sonar that distinguish eel from similarly sized and shaped acoustic targets (i.e.,aspect ratio variability and orientation angle variability), and provided additional data for algorithm selection and training. Deep learning algorithms trained and tested on the laboratory data alone achieved accuracy rates of greater than 98% when classifying acoustic images of eel and similar-sized neutrally buoyant sticks. The algorithm trained and tested on the pre-existing field data alone identified 9.3% false positives and 13.3% false negatives when distinguishing between eel and sticks/PVC pipes based on video clips and a 70% threshold for multiple,consecutive images. The trade-off between false positive and false negative classification rates can be adjusted according to the image-level threshold chosen (Fig. 14).

Fig. 14.Classification results of the sonar video clips from the field experiments (algorithm trained on field data only). The percentage threshold is the ratio of the number of images classified as eel by CNN to the total number of extracted images in the testing sonar video clip. Source: Zang et al. (in review).

The deep learning algorithm trained on a combination of video clips obtained in the laboratory and the field and tested on video clips from the field was able to distinguish eel from sticks and PVC pipes (a river debris analog) of similar size with 100% accuracy.

5.Recent research on the effect of light on out-migrating eel and recent advancements in lighting technology

Having already had a large-scale proof of concept study of light guidance at the IWCS (Versar, 2009), the EPRC decided to 1) provide an update on research related to the guiding of out-migrating eel with light,and 2) document advancements in lighting technology (EPRI, 2017b).The report provides an update to a previous report (Versar, 2009), and focuses on developments between 2007 and 2015. Primary and gray literature worldwide was reviewed, as well as the targeted canvassing of eel researchers to identify unpublished research. In addition, information gaps were identified, and St. Lawrence River site-specific guidance structure designs and recommendations were provided.

The findings suggest there is continued support for light as an effective behavioral guidance stimulus. Three relevant studies were published since 2007, two of which demonstrate effective silver eel guidance infield situations with LED strobe lights. Broad-spectrum white and narrow-spectrum blue light were recommended for future studies due to eel-specific sensitivity. LED lights have numerous advantages over other light sources and are recommended for further testing. Compared to other light sources, LED lights are less expensive,have increased longevity, increased electrical efficiency, and are fully programmable. Their light characteristics are highly flexible and can be operated in continuous or various flashing modes to improve visibility,reduce habituation, and reduce effects on non-target species. A reduction in biofouling on underwater housings can be achieved with the addition of ultraviolet (UV) diodes to reduce light array cleaning and maintenance costs. The report (EPRI, 2017b) was completed by Turnpenny Horsfield Associates Ltd, AECOM-Trois-Rivieres, School of Aquatic and fishery Sciences - University of Washington, and Milieu,Inc.

6.White paper investigation of the potential use of sound to guide out-migrating American eel near Iroquois Water Control Structure and the Beauharnois power Canal on the St. Lawrence River

Given the generally consistent deterrence response observed in the vibration study (section 3.3), a state of the knowledge white paper was commissioned with the objectives to provide an update on research related to the use of sound for guiding out-migrating eel, including the current understanding of sound detection in fish, the identification of data gaps, and both general and site-specific research recommendations(EPRI, 2018b). The report updates the findings in a previous report on the topic (Versar, 2009). A thorough literature review was conducted including published and unpublished research, in addition to questionnaires and interviews of technical experts in the field. The findings of the report suggest there are relatively few new papers on the topic since 2008, none of which provide useful insight into how sound might be used to guide out-migrating eel. A new understanding of sound detection by fish, including eels, is presented. In general, fish are more sensitive to the particle motion (directional) than pressure (omnidirectional) components of sound, enabling them to detect and respond in a directional way to a sound source. Particle motion data are critically important to interpret fish responses to a sound stimulus; however, confounding the issue is that much of the published research on fish hearing focuses on sound pressure. The report recommends focused research efforts on data gaps for developing and implementing an acoustic guidance system on the Upper St. Lawrence River. The major data gaps identified include the basic hearing ability of the American eel, behavioral response to sound in their environment, site-specific ambient acoustic environment, and sound propagation in the Upper St. Lawrence River. This report (EPRI,2018b) was completed by AKRF, Inc.

7.Discussion

The concept of using behavior for deterring eel from passing downstream through turbines while attracting them to bypasses is not novel(reviews by Richkus & Dixon, 2003; Versar, 2009). That said, the summaries of the studies funded by the EPRC and presented here demonstrate the challenge of protecting out-migrating eel in large river catchments where the majority or all of the flow exits through hydroelectric generating stations (Table 2). The sheer size of the generating stations in combination with high flows means that solutions such as screening that can work at smaller stations are infeasible, and non-contact behavioral guidance solutions are required to guide eels away from turbines and toward bypasses or collection devices. Some of the potential guidance techniques examined in the research sponsored by the EPRC showed no effect (electromagnetic fields, enhanced flow guidance), or no effect at the flows encountered during outmigration(electrical guidance) (Table 2). The only guidance technique examined in our studies to show promise for deterring silver-stage eel was sound,and those tests were only completed in low flow situations (0.15 m/s).We recognize that the findings in these behavioral studies were potentially complicated by experimental chamber effects (e.g., artifacts of flume confinement) than the response to guidance stimuli. Further, a study examining whether sonar technology could be used to assess eel outmigration demonstrated that, among the sonars tested, only the Sound Metrics Adaptive Resolution Imaging Sonar (ARIS) Explorer 1800 sonar system was capable of detecting and identifying out-migrating eel,and only within an ～15 m detection range. It is anticipated that these studies will be of use for those involved in the global protection of out-migrating anguillid eels in large rivers.

7.1.Assessment of behavioral guidance technologies

Out-migrating anguillid eels generally follow the main flow of the river, until encountering constricted flows (Jansen et al., 2007; Piper et al., 2015). Observations of a change in behavior above numerous hydroelectric generating stations provides an expectation that behavioral guidance can be used as a tool for reducing entrainment and turbine mortality. The use of electricity to exclude fishes from specific areas is generally well established (e.g., Parker et al., 2016; Smith & Tibbles,1980), but there few examples of successful downstream electrical guidance. Pugh et al. (1970) assessed the effectiveness of electrical guidance on three species of out-migrating salmon smolts and determined that guidance was only feasible at low (＜0.3 m/s) water velocities. Similarly, electrical guidance was successful only at low flows(＜0.25 m/s) for out-migrating juvenile sea lamprey (Petromyzon marinus) (Johnson & Miehls, 2014; Miehls et al., 2017). We saw similar outcomes in our electrical guidance work, where eel clearly exhibited behavioral responses (twitch, loss of orientation and eventually paralysis) in static water tests, but eel readily went through electrical fields that resulted in behavioral responses in static water when moving down the experimental flumes in flows similar to what they would experience on the St. Lawrence River. Concerns about human and wildlife safety,questions about acclimation over a long array, along with the difficulty in finding the balance between the strength of electrical field required for guidance versus paralysis (because a fish that is paralysed by the electrical field will drift down through the guidance array) have meant that electrical guidance is not regularly considered when attempting to mitigate turbine mortality at hydroelectric generating stations (Schilt,2007; Versar, 2009). While there were concerns about possible effects of the experimental flume on the outcome of the electrical and flow guidance studies, with study animals preferring one side of the flume and evidence for individuals moving among collection bins in our large flume study, both the large and small flume electrical guidance studies demonstrated that under high flow (0.9-1.0 m/s) eel would pass through the electrical array, and not be guided. This indicates that despite the successful use of electrical barriers to exclude fishes from certain areas, and the observations that eel are sensitive to electrical fields, electrical guidance is likely not feasible for out-migrating silver-stage eel on large rivers.

Anguillid eels are sensitive to flow fields, and Piper et al. (2015)suggested that this offers the potential to use flow for fish passage solutions when traditional physical screening is infeasible. Anthropic disturbance of natural flow fields have been shown to impede outmigration of both European and American eel, where searching behavior above hydroelectric generating stations is commonly observed(e.g., Brown & Castro-Santos, 2009; Bruijs et al., 2009; Piper et al.,2015). The use of enhanced flows has been proposed to guide out-migrating salmonid smolts to safe bypass routes (Coutant, 2001),and a framework for understanding the relationship between hydrodynamic flow and out-migrant migration patterns in salmonids was developed (Nestler et al., 2008). However, being able to practically match the theory and observations of natural out-migrants with the use of induced flows to guide out-migrating fish when they encountered anthropic flow disturbances has proven difficult. There was no evidence that flow guided eel in our study, either alone or in combination with the electrical guidance system, with a caveat that we were unable to measure the exact zone of flow in fluence in the flume. Although water velocity profiles were generated for several transects along the length and the width of the test flume, the data were not sufficient for mapping the FVES plume at the test velocity so a model designed to estimate the penetration and spreading of a jet into flow was used to provide an estimate of the FVES plume area of coverage during eel trials (Blevins,1984). Given the absence of obvious response in the 5.2 m wide flume, it is difficult to see how flow could effectively be used for guidance in the St. Lawrence River, which is ～1 km wide with an average flow of 7410 m/s. Further study of flow to assist in guidance over smaller spatial scales may still be important if a technique for broader-scale guidance is identified first.

The use of sound propagation to guide fishes away from hydroelectric generating stations has generated much interest because most fishes use sound to assess and respond to their surroundings, and because of the inherent properties of sound underwater (e.g., slow attenuation,highly directional, and unaffected by light or water levels; Putland &Mensinger, 2019). The best example of successful behavioral guidance in fish is the use of high frequency sound (＞20,000 Hz) to deter clupeids from unsafe areas, and provide opportunities for discovery of safe bypass channels (review by Putland & Mensinger, 2019). Studies on anguillid eels have demonstrated mixed results; a laboratory study found eel attracted to a 1000 Hz sound (Patrick et al., 2001), but the vast majority of research has identified a deterrent effect of infrasound (＜20 Hz) and the audible sound range (20-20,000 Hz; reviews by Popper et al., 2020;Putland & Mensinger, 2019). Our small flume study demonstrated a similar deterrent effect, with American eel moving away from the three signals tested. While the identification of a behavioral guidance technique that has demonstrated success in anguillid eels is positive, studies to date indicate that sound propagation itself is unlikely to be successful in guiding eel on its own (Deleau et al., 2020). In general, sound deterrents in the field have shown a relatively modest response by eels,with deterrence rates reported between 28 and 57% (Maes et al., 2004;Piper et al., 2019; Sand et al., 2000). Popper et al. (2020) state that it is premature expect to use sound propagation in a large-scale acoustic guidance system for anguillid eels; more basic research on sound detection in eel, their responses to sound stimuli during outmigration,and the nature of the site-specific acoustic environment are required first. Our results demonstrate that sound may have the ability to repel eel and should be examined further for its potential application in the field. It is assumed that sound alone is unlikely to be a suitable stimulus for guiding out-migrating eel, due to the technological difficulties of producing sufficiently high sound levels detectable to out-migrating eel in such a large river as the St. Lawrence River. More realistically, sound(both sound pressure and particle motion) might be coupled with another stimulus (e.g., light) in a multimodal approach to in fluence behavior. This information will help inform our consideration of including sound in a multimodal approach to guide out-migrating eel in the St. Lawrence River.

Anguillid eels appear sensitive to electromagnetic fields (EMF) as they possess magnetite in their lateral line (Moore & Riley, 2009) and seem to use a magnetic compass when migrating (Durif et al., 2013).However, there is very little evidence to suggest that EMF might be useful for behavioral guidance in eel. Studies have documented potential behavioral changes in European eel when in the vicinity of underwater cables that would expose individuals to EMF, but these changes were considered minor (¨Ohman et al., 2007). EMF levels generated in our test enclosure were stronger than in previous studies that noted similar responses of fishes to EMF (＞30 Gauss units), yet we found no response to the stimulus. These results likely suggest that the use of EMF would not be an effective method for guiding eel away from intakes during outmigration; the absence of directional movement away from the EMF generators during the test study suggests that EMF’s are unlikely to be an effective stimulus for guiding eel downstream of dams.

7.2.Acoustic detection and machine learning of acoustic targets

The phased approach of this study demonstrated lessons and challenges of three sonar technologies for detecting eel in situ, their deployment strategies, and environmental considerations to guide future research on remotely monitoring out-migrating American eel. A 120-kHz EK60 split-beam echosounder, 500-kHz M3 multimode multibeam sonar, and 1100/1800 kHz ARIS Explorer imaging sonar were successfully deployed from the upstream side of IWCS to sample overlapping volumes of the water column for eel detection and identification. Split-beam echosounder technology can estimate the target range,acoustic size (target strength [TS]), location, and speed and direction over multiple detections.

Split-beam echosounder surveys have become a widely accepted technique for estimating fish abundance in coastal and ocean waters as well as to monitor fish passage in riverine waters (Simmonds &Maclennan, 2005). However, the use of fixed-location split beam sonar for monitoring migrating anguillid eels has not been documented. The Simrad 120 kHz EK60 echosounder used in this study had one 7circular split-beam transducer and is commonly used to monitor fish biomass in other applications (e.g., Godlewska et al., 2009). The other advantages of this sonar were the capability of sampling longer ranges, locating targets in three-dimensional space within the beam, and providing quantitative acoustic backscatter (e.g., target strength or volume backscattering strength). In our study, however, the EK60 sonar was only able to detect known eel targets, and did not provide the resolution necessary to distinguish eel from other targets.

Two multibeam sonars were used in this study, the Sound Metrics Adaptive Resolution Imaging Sonar (ARIS) Explorer 1800 sonar and the Kongsberg Mesotech M3 multi-mode multibeam sonar. The application of multibeam sonar in fisheries research has been well established (e.g.,Gerlotto et al., 1999; Mayer et al., 1998; Moursand et al., 2003),including fixed-location monitoring of migratory and spawning movements of fishes in rivers and inlets, including anguillid eels (Bilotta et al.,2011; McCarthy et al., 2008; Mueller et al., 2010). The Kongsberg Mesotech M3 multibeam sonar is a 500-kHz multibeam sonar that provided a wider sampling coverage than the ARIS sonar. This sonar was designed for maritime inspection and has been shown anecdotally to be capable of detecting aggregations of fish at distances of about 50 m range and large individuals within 10-15 m over a 120swath (Melvin &Cochrane, 2014). However, similar to the EK60 sonar, the Mesotech M3 was not able to distinguish eel from non-eel targets.

The ARIS sonar produces a series of high-resolution images up to 15 frames per second yielding a camera-quality video with sufficient detail to allow recognition and classification of the anguilliform swimming behavior of eel species. The ARIS sonar has shown promise in detecting and identifying out-migrating eel with relatively high classification accuracy (McCarthy et al., 2008; Mueller et al., 2010) but may be range-limited with ranges between 15 and 30 m. Our study demonstrated similar findings; we could feasibly identify and monitor American eel abundance with the ARIS sonar during the adult outmigration at IWCS. There were some drawbacks for the ARIS sonar though, including a limited range (＜15 m in our study) and the data processing requirements to identify eel.

Data from the ARIS sonar was used in the machine learning project to achieve proof of concept for automated identification of eel in multibeam sonar data. Future work should focus on acquisition of additional data for more robust algorithm training and testing; modification of the software tools to accommodate multiple acoustic targets in the acoustic field at a given time; identification of additional object classes; incorporation of motion in the object identification and classification algorithms; operationalizing the software tools, including integration with other existing sonar data analysis tools; and partnering with hardware and software providers for distribution of the software tools with their commercial products.

7.3.Summary and next steps

Concern for the population status of American eel in the St. Lawrence basin remains high, as recruitment remains an order of magnitude lower than 4 decades ago (Cairns et al., 2014). Eel outmigration from Lake Ontario is expected to decline sharply in the next several years due to cumulative escapement of the finite cohorts of stocked eel (Pratt &Threader, 2011) and the low numbers of naturally recruiting eel returning to Lake Ontario. This compounds the urgency (from both a conservation and scientific perspective) to develop means to mitigate turbine mortality. A sharp drop off in eel abundance amplifies the biological significance of turbine mortality and it reduces the number of fish available to serve as research subjects for testing and optimizing mitigation technologies. Consequently, it will be important to rapidly transition from adaptive, exploratory research and development to adaptive mitigation and management utilizing early stage guidance,collection, and monitoring technologies. This can be accomplished by rapid, iterative design, deployment, and testing of guidance and collection technologies.

This approach will be initiated with design and deployment of a subscale, prototype guidance and collection structure utilizing light as a guidance tool. An earlier, large-scale proof-of-concept study on the St.Lawrence River in 2002 demonstrated the effectiveness of light as a potential guidance tool (Versar, 2009). An 80 m platform equipped with 84, 1000 W halogen lights was oriented at a 30angle to the current, and an estimated 77-85% of the eel were successfully guided by the array(Versar, 2009). Recent advances in LED light technology mean that further efforts to guide eel with light might be more effective, as LEDs can incorporate UV anti-biofouling diodes to reduce cleaning requirements, and can be programmed to operate in continuous or flashing mode to improve visibility and potentially reduce habituation(EPRI, 2018a). Recent studies have used LED lights to deter anguillid eels (Elvidge et al., 2018; Kruitwagen, 2013), so the technology exists to deploy at a larger scale. This adaptive mitigation approach will enhance mitigation effectiveness over the near term and develop more effective technology over the mid-to long term.

Acknowledgements

The research conducted by the EPRC has been funded by the fish Enhancement, Mitigation, and Research Fund, administered by the U.S. fish and Wildlife Service; Hydro-Qu´ebec and Ontario Power Generation.Additional funding was received from Duke Energy and the Electric Power Research Institute. These funding organizations as well as New York Power Authority, fisheries and Oceans Canada, Ontario Ministry of Natural Resources and Forestry, Qu´ebec Ministry of Forests, Wildlife and Parks, and New York State Department of Environmental Conservation have made technical and in-kind contributions to the research effort.The following organizations have conducted research described in this paper under contract to EPRI: AECOM, AKRF Inc., Alden Research Laboratory, Aquacoustics, Arthur Popper (consultant), Blue Leaf Environmental, Coutant Aquatics, Elgin Perry (consultant), Johnson fisheries Science, Kleinschmidt Associates, Loughfine Ltd., Natural Solutions, Normandeau Associates, Scientific Solutions Incorporated,Smith-Root, Turnpenny Horsfield Associates, University of Southampton, U.S. Geological Survey, and WSP Canada. This document is Publication # 2020-02 of the Eel Passage Research Center.