Le

Da

b

c

a

ARRAA

KBBCFI

1

a2ddeowsswAwpt

BU

h0

Ecological Engineering 67 (2014) 95–103

Contents lists available at ScienceDirect

Ecological Engineering

jou rn al hom ep age: www.elsev ier .com/ locate /eco leng

aboratory experiments demonstrate that bubble curtains canffectively inhibit movement of common carp

.P. Zielinskia,∗, V.R. Vollera, J.C. Svendsenb, M. Hondzoa, A.F. Mensingerc, P. Sorensenb

Department of Civil Engineering and St. Anthony Falls Laboratory, University of Minnesota, Minneapolis, USADepartment of Fisheries, Wildlife, and Conservation Biology, University of Minnesota, St. Paul, MN, USADepartment of Biology, University of Minnesota – Duluth, Duluth, MN, USA

r t i c l e i n f o

rticle history:eceived 22 April 2013eceived in revised form 7 January 2014ccepted 24 March 2014vailable online 21 April 2014

eywords:ehavioral barrier

a b s t r a c t

Although bubble curtains have been proposed many times as practical and inexpensive solutions to hinderthe movement of invasive fish, few studies have examined why or how they might work. By understandinghow bubble curtains influence fish behavior, management tools could be developed to control movementof invasive fish. In this study, the common carp (Cyprinus carpio L.) was used to examine the performanceof three different bubble curtains (fine-, graded-, and coarse-bubble) and acoustically enhanced systemsin an indoor channel. Trials revealed that the graded- and coarse-bubble systems reduced common carppassage across the curtain by 75–85% in both up- and down-stream directions. Concurrent acoustic field

ubble curtainommon carpish guidancenvasive fish management

measurements revealed that these bubble curtains generated sound near 200 Hz at approximately 130 dB(ref 1 �Pa), well above the common carp hearing threshold. Further testing with speaker arrays andlighting indicated that carp avoidance of the bubble curtain involved responses to sound and fluid motionrather than visual cues. Although field tests are warranted, our results suggest that bubble curtains maybe a viable and inexpensive deterrence system to limit common carp movement.

© 2014 Elsevier B.V. All rights reserved.

ho

baelbpeK12i

. Introduction

Fish guidance technologies have long been part of fisheries man-gement efforts to control invasive fishes (Taft, 2000; Lavis et al.,003; Noatch and Suski, 2012). Physical or mechanical barriers (i.e.ams, screens, or traps) can be effective at stopping both up- andown-stream movement of invasive fish; however, these barri-rs can be extremely difficult and expensive to maintain becausef clogging (Bainbridge, 1964). Consequently, behavioral barriers,hich utilize stimuli such as sound and light to target fish sen-

ory systems and guide fish in taxon-specific manners, have beenuggested for sites where mechanical or physical barriers are notell suited (Popper and Carlson, 1998; Noatch and Suski, 2012).

behavioral barrier of particular interest is the bubble curtain,

hich produces a wall of bubbles (e.g. by forcing air througherforated pipes). Bubble curtains are inexpensive, require rela-ively little maintenance, and generate complex sound, visual, and

∗ Corresponding author at: Department of Fisheries, Wildlife, and Conservationiology, Minnesota Aquatic Invasive Species Research Center, 135E Skok Hall, 2003pper Buford Circle, St. Paul, MN 55108, USA. Tel.: +1 612 624 7785.

E-mail address: ziel0064@umn.edu (D.P. Zielinski).

Wggptsmnfi

ttp://dx.doi.org/10.1016/j.ecoleng.2014.03.003925-8574/© 2014 Elsevier B.V. All rights reserved.

ydrodynamic fields which may be optimized to deter fish withoutbstructing water flow.

Initial development of bubble curtain technologies was driveny both commercial fishing (Kuznetsov, 1971), and the need to findlternative solutions to reduce fish impingement at power gen-ration facilities (Taft, 2000; Michaud and Taft, 2000). Althoughaboratory and field studies have reported fish to be deterred byubble curtains, these studies did not quantify sound fields or otherhysical characteristics needed to assess the factors driving theffectiveness of the barrier systems (Brett and MacKinnon, 1953;uznetsov, 1971; Zweiacker et al., 1997; Leiberman and Muessig,978; Stewart, 1982; Sager et al., 1987; EPRI, 1998, 2004; Sprott,001; Welton et al., 2002; Dawson et al., 2006). Further, stud-

es have reached contradictory conclusions (Patrick et al., 1985;elton et al., 1997). For example, while Patrick et al. (1985) sug-

ested bubble curtains act as a visual deterrent due to a ∼20%reater avoidance by gizzard shad (Osmerus mordax), alewife (Alosaseudoharengus), and smelt (Dorosoma cepedianum) under low lighthan in darkness, Welton et al. (1997) described the opposite, and

uggested that Atlantic salmon smolt (Salmo salar) were deterredore during night than daytime trials (42% compared to 0%). Alter-

atively, Kuznetsov (1971) suggested fish respond to the acousticelds generated by bubble curtains based on nighttime commercial

9 al Engineering 67 (2014) 95–103

fiibta

mia(d1olMlttcw(ltKlnv(

odBamtf2hcwihsfiotbco(

aua(bsfi

2

2

(

Fig. 1. Schematic of behavioral trial tank with approximate location of bubble cur-toi

etflU

wa5tIprwC

p3ccrarcttctafccb

2

tss

6 D.P. Zielinski et al. / Ecologic

shing. Overall, studies appear to suggest that bubble curtainsnhibit fish movement, but the abiotic parameters that affect fishehavior remains unclear. Importantly, no studies have attemptedo individually and collectively assess the influence of sound, visual,nd hydrodynamic fields on fish behavior.

By understanding how bubble curtains influence fish behavior,anagement tools could be developed to control movement of

nvasive fish. Bubble curtains may influence fish visual, auditory,nd lateral line systems by generating visual, sound, and tactilee.g. fluid flow) stimuli. Sound is generated by bubbles as theyetach from the diffuser (Leighton and Walton, 1987; Leighton,994; Lin et al., 1994), which at the continuum limit (the curtainf bubbles works as a collection of coupled oscillators) results inow frequency (<1000 Hz) sound emissions (Nicholas et al., 1994;

anasseh et al., 2004). The radiating sound field is comprised ofongitudinal particle motion and local pressure oscillations. For alleleost fish, the inner ear detects the particle motion component ofhe sound wave; however, ostariophysian fish (including commonarp) have an anatomical link between swim bladder and inner earhich provides indirect audition of the pressure component as well

Popper and Fay, 2011). Rising bubble plumes also generate turbu-ence with distinct recirculation currents that are dependent onhe upward velocity and density of the bubble plume (Brevik andristiansen, 2002; Soga and Rehmann, 2004). The mechanosensory

ateral line is the main sensory system for these hydrodynamic sig-als (Webb et al., 2008). Finally, bubble curtains may serve as aisual barrier by obscuring a fish’s line of sight past the barrierPatrick et al., 1985; Sager et al., 1987).

The present study investigated the impact of a bubble curtainn common carp (Cyprinus carpio L.), a cyprinid responsible foregrading water quality in shallow water ecosystems (Weber andrown, 2009). In Midwestern North America, common carp, here-fter termed carp, often inhabit stable, deep, normoxic lakes foruch of the year, but enter interconnected unstable (susceptible

o hypoxic conditions), shallow lakes to spawn, so the latter areasrequently serve as recruitment ‘hotspots’ (Bajer and Sorensen,010). Reducing or stopping the migration of adult fish to spawningabitat or young carp back to the stable lakes could dramati-ally decrease recruitment. Existing barrier technologies are notell suited for the conditions characteristic of streams connect-

ng stable and unstable lakes which typically have low hydraulicead. Bubble curtains could provide a targeted, safe, and inexpen-ive alternative for sites involving downstream movement of smallshes, especially in waters where reduction, not total eliminationf movement, is the management goal. Additionally, bubble cur-ains could also be readily removed or re-positioned, if needed. Aehavioral barrier employing acoustic stimuli – such as the bubbleurtain – may also be potentially useful for targeting carp becausef their relatively broadband hearing (50–3000 Hz) and sensitivity>65 dB re: 1 �Pa) (Popper, 1972).

The main objectives of the present study were to: (1) developnd test the efficacy of a bubble curtain to inhibit carp movementnder controlled laboratory conditions; (2) identify the acousticnd hydrodynamic flow fields generated by the bubble curtain; and3) determine the effect of visual and auditory components of theubble curtain to inhibit movement. This study appears to repre-ent the first attempt to quantify the biologically relevant stimulields generated by a bubble curtain that inhibits fish movement.

. Materials and methods

.1. Experimental setup

Common carp [mass: 204 ± 77 g; total length: 259 ± 29 mmmean ± S.D.)] were caught in Lake St. Catherine, MN, USA by

tawT

ain and PIT tag interrogation system (PIT antennas are labeled Ant #1-4). Theutside diameter of the tank is 3 m and the inside diameter is 1 m, and water depths 25 cm.

lectrofishing in July 2010 and transported to the laboratory, wherehe carp were maintained in large tanks supplied with continuouslyow-through 20 ◦C well water. Carp were fed pellets (Silver Cup,tah) once a day between 10.00 h and 16.00 h.

Passive integrated transponder (PIT) tags (OregonRFID, OR, USA)ere implanted into a third of the fish. Carp were anesthetized in

0.05% solution of buffered tricaine methanesulfonate (MS222), a mm incision was made between their pelvic and pectoral fins andhe 23 mm-long half duplex PIT tag placed inside their body cavity.ncisions were allowed to heal for three weeks (Skov et al., 2005)rior to the experiments and tagging resulted in no mortality. Theemaining carp were left untreated. All experimental proceduresere approved by the University of Minnesota Institutional Animalare and Use Committee.

Experiments were performed in a round tank (3 m diameter)rovided with an insert to create a circular channel (I.D. 1 m × O.D.

m) and water depth of 25 cm (Fig. 1). Water was supplied to thehannel through a submerged pipe, producing an average 5 cm s−1

urrent. Carp were tested in groups of three to facilitate natu-al shoaling behavior. To track carp movement, a PIT antennarray was constructed using the Oregon RFID Multi-Antenna HDXeader, powered by a 12-V deep cycle marine battery. Each antennaonsisted of 5 turns of 16 gauge solid wire (1 m × 0.3 m hoop), tunedo an inductance of ∼60–80 �H. All antennas were connected touning modules, which were connected to the PIT reader by twinoaxial cable. Each time a tagged carp passed through an antenna,he time of passage, PIT identification number, antenna number,nd time between detections were logged onto a memory cardor analysis. The antennas were equally spaced (∼1.6 m) along theircular channel at the quarter points, centered about the bubbleurtain (Fig. 1). Manual testing indicated that the detection proba-ility of each antenna was >99%.

.1.1. Tests of simple bubble curtainsResponses of carp to three different bubble curtain systems were

ested. Bubble curtain systems varied in size, configuration, and airupply to help identify features influencing fish behavior. The firstystem was a fine-bubble system comprised of two 2.5 cm diame-

er porous polyethylene pipes (Genpore, PA, USA). These pipes hadn average pore size of 25 �m over their entire surface. The pipesere placed 30 cm apart on the bottom of the test channel (Fig. 2a).

wo S41 regenerative air-blowers (Aquatic Ecosystems, FL, USA), in

D.P. Zielinski et al. / Ecological Engineering 67 (2014) 95–103 97

F (a) fia fish pf

sfi(o

slo1dstota

aa

cflibw

wtrulfpt

ig. 2. Plan view schematic of bubble curtain and speaker arrangements tested:rray/fine-bubble systems. The darkened area is filled by plastic mesh, preventingurther details of curtain positioning in the circular tank.

eries, were used to supply 5.6 L s−1 m−1 of air at 28 kPa. This con-guration was similar to the systems investigated by Welton et al.1997) and Patrick et al. (1985), with the difference being the usef two pipes instead of just one.

The second system was a graded-bubble system comprised ofix 2.5 cm diameter pipes. The system produced incrementallyarger bubbles across the length of the system (Fig. 2b), employingne porous polyethylene pipe (∼25 �m pores), four PVC pipes with

mm diameter holes spaced at 1 cm, and one PVC pipe with 3 mmiameter holes spaced at 5 cm. Pore sizes increased in the down-tream direction. The total air flow was 30.1 L s−1 m−1 at 5 kPa inhe drilled PVC pipes and 28 kPa in the porous pipe. The total widthf the system was 50 cm. The increasing bubble size produced byhe graded-bubble system was chosen to provide an increasingly

dverse stimuli field across the length of the barrier.

The third system was a coarse-bubble (CB) system comprised of grid layout of 2.5 cm diameter PVC pipes with 3 mm holes spacedt 5 cm over the entire surface (Fig. 2c). This layout provided bubble

t1it

ne-bubble; (b) graded-bubble; (c) coarse-bubble; (d) speaker array; (e) speakerassage along tank edge. Water flow is from left to right at 5 cm s−1. See Fig. 1 for

urtains that were both perpendicular and parallel to the waterow, and reduced the potential for fish to maneuver through gaps

n the curtain. The air flow was increased to 108.0 L s−1 m−1 at 5 kPa,y using four S41 regenerative air blowers in parallel. The totalidth of the system was 50 cm.

All behavioral trials were conducted between 20.00 and 6.00 hith all lights off in the testing facility and a black tarp covering

he experimental tank so that no light was visible, minimizing theole of any visual stimulus. For each trial, one PIT tagged and twontagged carp were selected at random from the holding popu-

ation, placed into the circular channel, and allowed to acclimateor 10 min before the trial began. One PIT-tagged carp was useder trial along with two untagged fish because the PIT tag sys-em can only detect one tag at a time. Preliminary tests using

wo tagged carp and video recording showed that carp maintained–2 cm spacing throughout the entire testing period and remained

n a close group when challenging a bubble curtain. Based onhese preliminary observations, the movement data was assumed

9 al Eng

tgcaictstfi(dtwttbp

2e5trr

t&01afAwaadad

2a

tltsw

2pnwilap

2iaat(o

tfgtuft

2twcflaew(astctaetpa(((

2

otwuoscsAedbtlasbsDt

ppuwt

8 D.P. Zielinski et al. / Ecologic

o describe movement of the entire group. Carp were tested inroups of three to promote natural behavior in the test tank (e.g.arp are social) (Sloan et al., 2013; Huntingford et al., 2010; Sislernd Sorensen, 2008). Movement of the PIT tagged carp was mon-tored for 7 h for all trials. Carp were then removed from the testhannel and placed into separate holding tanks to eliminate any re-esting during a given treatment. All trials were carried out over thepan of 10 days. A 3 week rest period was then provided betweenreatments, after which the next set of trials started by re-selectingsh randomly from the holding population. Sisler and Sorensen2008) previously demonstrated that carp held in the laboratoryo not alter their behavior if tested at 3 week intervals. Controlrials, consisting of bubble curtain pipes in place but no bubbles,ere performed before (N = 8) and after (N = 7) all bubble curtain

reatments to confirm that carp behavior remained constant overime. Trials with the fine- (N = 7), graded- (N = 8), and coarse- (N = 7)ubble curtains were performed consecutively, with a 3 week resteriod between each.

.1.1.1. Properties of the bubble curtain. The sound pressure fieldncountered by the carp during trials was mapped at depths of, 12.5, and 20 cm below the water surface at 10 cm intervals inhe quadrant of the bubble curtain and at 25 cm intervals in theemaining space. In this study, descriptions of the acoustic field areestricted to sound pressure measurements.

Acoustic pressure measurements were obtained using a minia-ure (50 mm length, 9.5 mm diameter) BK 8103 hydrophone (Bruel

Kjær, Denmark). This hydrophone has a frequency range of.1 Hz–180 kHz and a sensitivity of approximately −211 dB ref

V/�Pa. The output signal from the hydrophone was amplified by 144 L low noise voltage amplifier (DL instruments, NY, USA) anded into a National Instruments SC-2345 signal conditioning board.

custom LabVIEW program (National Instruments, TX, USA) wasritten to control and record from the conditioning board, and

custom Matlab (Mathworks, MA, USA) program was written tonalyze and transform the pressure waveform into the frequencyomain. At each measurement location, ten signal ensembles wereveraged to improve the signal-to-noise ratio. The hydrophoneata record was 10.0 s in duration and was sampled at 100 kHz.

.1.2. Testing the role of sound and light stimuli on the efficacy of bubble curtain

Once the coarse-bubble system was identified as the most effec-ive bubble curtain, we investigated the influence of sound andight on the efficacy of the coarse-bubble system. To accomplishhis, responses of carp to an illuminated coarse-bubble system,peaker array system, and bubble curtain system supplementedith a speaker array were tested as outlined below.

.1.2.1. Visual stimuli. Trials using the coarse-bubble system wereerformed between 20.00 and 6.00 h with three 40 W 120 V (Sylva-ia, MA, USA) light bulbs installed approximately 1.0 m above theater surface to illuminate the circular channel (average surface

llumination: 370 lx). These data (N = 7) were compared to data col-ected in darkness. Following testing, the fish were provided with

3 week rest period after which a set of control trials (N = 3) waserformed to ensure carp behavior was unchanged.

.1.2.2. Sound stimuli. Bubble curtains create visual and flow fieldsn conjunction with sound. To test the role of sound alone, a speakerrray was used in place of the coarse-bubble system. The speaker

rray system comprised of two UW30 underwater speakers (Elec-rovoice, MN, USA) on either side of an AQ339 underwater speakerClark Synthesis, CO, USA) (Fig. 2d). The signal was a 10 s recordingf the sound produced by the coarse-bubble system which was set

d

E

ineering 67 (2014) 95–103

o loop continuously. Acoustic pressure measurements were per-ormed to ensure that the maximum sound pressure level (SPL)enerated by the speaker array closely matched those created byhe coarse-bubble system, thereby creating a bubble curtain stim-lus without bubbles or turbulence. Speaker array trials (N = 7)ollowed the same testing procedure as previous bubble curtainrials.

.1.2.3. Sound and bubble stimuli. Another trial used a bubble cur-ain that was supplemented with a speaker array to investigatehether adding a sound field to a weak bubble curtain impacted

arp movement. Addition of the bubble curtains provided minimaluid flow and influenced the acoustic field by acting as a resonantmplifier (Manasseh et al., 2004) re-transmitting the sound gen-rated by the speaker array. The speaker array/fine-bubble systemas a combination of the fine-bubble and speaker array system

Fig. 2e). The signal of the speaker array/fine-bubble system was 10 s recording of the coarse-bubble system set to maintain theame the maximum SPL. Acoustic mapping was used to quan-ify any differences in the acoustic field dynamics between theoarse-bubble, speaker array, and speaker-array/fine-bubble sys-ems. Data from these trials (N = 7) were compared with the speakerrray and coarse-bubble data. A final set of control trials (speak-rs and bubble pipes in place but off) (N = 6) was performed afteresting of the speaker array/fine bubble system and 3 week resteriod, to confirm constant carp behavior. Overall, the order of tri-ls was as follows: (1) control; (2) fine-bubble; (3) graded-bubble;4) coarse-bubble; (5) control; (6) coarse-bubble with illumination;7) control; (8) speaker array; (9) fine-bubble/speaker array; and10) control.

.2. Statistical analysis

A Shapiro-Wilk test determined that the numbers of crossingsver each bubble curtain system during control and bubble cur-ain trials were not normally distributed, so nonparametric testsere required (Conover, 1980). First, a Kruskal–Wallis test wassed to confirm that no change in the number of passages wasbserved over time during all control trials. Because there was noignificant difference (P > 0.5), in the number of crossing betweenontrol trials over time, control data were combined into a univer-al control group for comparison with bubble curtain treatments.

Kruskal–Wallis test (Conover, 1980) was used to compare differ-nces between the number of passages in the up- and down-streamirection between the bubble curtain and control trials (e.g. fine-ubble, graded-bubble, and coarse-bubble, and universal control),he visual stimuli trials (e.g. coarse-bubble group with and withoutight), and speaker array trials (e.g. coarse-bubble, speaker array,nd speaker array/fine-bubble groups). The average number of pas-ages in each direction was used to compare the efficacy of eachubble curtain design. When the Kruskal–Wallis test suggestedignificance (P < 0.05), Mann–Whitney pair-wise comparisons withunn-Sidak correction for multiple comparisons were performed

o determine which pair differed.Additionally, the total activity of the carp over the entire test

eriod was quantified by summing the number of times a carpassed through any consecutive antenna. A Kruskal–Wallis test wassed to compare the total activity levels between all trials to assesshether the bubble curtains impacted all carp movement in the

ank or just passage over the bubble curtain.The efficacy (E) of each bubble curtain was calculated for each

ataset in the following manner:

= 100% ·(

1 − Ncurtain

Ncontrol

)

D.P. Zielinski et al. / Ecological Engineering 67 (2014) 95–103 99

Fig. 3. Total number of barrier crossings in the up- and down-stream directions made by common carp for each trial (e.g. control and bubble-curtain type). Dashed linesseparate trial groups compared with Kruskal–Wallis and Mann–Whitney tests. For each comparison, all controls were combined into a universal control group. Note, thecoarse-bubble trial is shown repeatedly for clarity of trials being statistically analyzed. Statistical significance was determined by Mann–Whitney pair-wise comparisonsw ents.

m

wcc

2

a(dutcin(s

s(

fipd(dim(ott

bsct

ith Dunn-Sidak correction for multiple comparisons between controls and treatminimum and maximum values. *P < 0.05, **P < 0.01, ***P < 0.001.

here Ncurtain is the mean number of carp passages through theurtain while it is on (test trials) and Ncontrol is the mean number ofarp passages over the curtain while it is off (control trials).Results

.3. Tests of simple bubble curtains

Carp tended to swim in the downstream direction nearly twices often as in the upstream direction during the control trialsMann–Whitney U-test: P < 0.01) (Fig. 3). There was no significantifference observed between the number of passages in either thep- or down-stream directions during the different sets of controlrials across time (Kruskal–Wallis: P > 0.5), so the controls wereombined into a universal control for the purpose of perform-ng statistical comparisons with bubble curtain trials. The meanumber of crossings during each control test was 60[28.5, 80]mean[1st, 3rd quartiles]) and 90[64,126] in the up- and down-

tream direction (Fig. 3).

A difference in the number of passages in the up- and down-tream direction was found among the bubble curtains trialsKruskal–Wallis: P < 0.001, both directions) (Fig. 3). While the

Mabp

Box plots illustrate data quartiles, mean, median (squares) and whiskers represent

ne-bubble curtain was not found to reduce the number ofassages in either direction (38[23,61] upstream and 31[8,51]ownstream), the graded-bubble curtain reduced passage by 78%Mann–Whitney U-test: P < 0.05, 20[2,30]) in the downstreamirection and 72% (Mann–Whitney U-test: P < 0.05, 16[5.5, 36])

n the upstream directions. The coarse-bubble curtain was theost effective system, resulting in a downstream efficacy of 87%

Mann–Whitney U-test: P < 0.001, 11[1,24]), and upstream efficacyf 82% (Mann–Whitney U-test: P < 0.005, 10[0,21]). Additionally,here were two graded-bubble and three coarse-bubble curtainests that had zero crossings in the downstream direction.

The swimming activity of the carps was unaffected by theubble curtains. The total number of crossings between con-ecutive antennae (total activity) between bubble curtain andontrol trials did change (Kruskal–Wallis: P < 0.05), but a reduc-ion was only observed during graded-bubble trials (46% reduction,

ann–Whitney U-test: P < 0.01). No additional change in totalctivity was observed over all other trials, resulting in a mean num-er of times carp passed between any two antennae at ∼1000/testeriod (or nearly 3 m traveled every 30 s).

100 D.P. Zielinski et al. / Ecological Eng

Fig. 4. Sound pressure level power spectrum in the behavioral trial tank duringcoarse-bubble (solid black line), graded-bubble (solid blue line), and fine-bubble((t

2

sapf(dpsafi3aftsb

FaiaT

Siabi

2

isUffass7e

ss1ffiee2dabp

3

solid red line) trials and the common carp hearing threshold documented by Popper1972) (dashed black line). Bubble curtain sound data was obtained at the center ofhe bubble curtain, 5 cm above the channel bottom.

.3.1. Identifying the characteristic sound field of a bubble curtainThe bubble curtains investigated in this study generated a broad

pectrum acoustic field with a peak frequency between 100 Hznd 300 Hz for all bubble curtain treatments (Fig. 4). The contourlot of the sound pressure field in the behavioral tank at 200 Hzor the coarse-bubble system was typical for all bubble curtainsFig. 5). The graded-bubble curtain produced a sound pressure gra-ient across the barrier length of 100–130 dB (ref 1 �Pa). The soundressure contours at 200 Hz for the fine-bubble and graded-bubbleystems displayed similar characteristics (i.e. no extraneous soundway from the bubble curtain and rapid attenuation in the neareld). The maximum sound pressure level (SPL) between 100 and00 Hz was 120 dB (ref 1 �Pa) for the fine-bubble system. The up-nd down-stream maximum SPL between 100 and 300 Hz wasound to be 100 and 130 dB (ref 1 �Pa) for the graded-bubble sys-

em, resulting from increased size of bubbles along the length of theystem. The maximum SPL between 100 and 300 Hz of the coarse-ubble system was 134 dB (re 1 �Pa). Power spectrum plots of the

ig. 5. Plan view of sound pressure field in dB (ref 1 �Pa) in the behavioral trial tankt a depth of 12.5 cm at 200 Hz with the coarse-bubble system. The bubble curtains located to the right of the center of the tank, perpendicular to the y-axis centeredt a radial distance of 0 cm. The background SPL at 200 Hz was ∼80 dB (ref 1 �Pa).he shaded area denotes the approximate location of the coarse-bubble system.

bgcbsfif

ala(sitirtcpaama

ict

ineering 67 (2014) 95–103

PL generated by the coarse-bubble system at additional locationsn the tank are provided in Appendix A and B, along with additionalcoustic and hydrodynamic field measurements of single pipe bub-le curtains under varying depth, air-flow, and water flow obtained

n an earlier pilot study.

.4. Influence of sensory fields on efficacy of bubble curtain

The number of passages over the coarse-bubble system whenlluminated was not found to be different in the up- or down-tream direction compared to the dark trial (Mann–Whitney-test: P > 0.05, 22[3,30] and 22[2,63]) (Fig. 3). Likewise, no dif-

erences in the number of up- or down-stream passages wereound between the coarse-bubble, speaker array, and speakerrray/fine-bubble trials (Kruskal–Wallis: P > 0.05) (Fig. 3). Thepeaker array/fine-bubble system resulted in a reduction of passageimilar to the coarse-bubble system, with an upstream efficacy of3% (Mann–Whitney U-test: P > 0.05, 16[6,25]) and down-streamfficacy of 76% (Mann–Whitney U-test: P < 0.05, 19[8,29]).

An area of high sound pressure was concentrated near thepeaker array with a maximum SPL between 100 and 300 Hz for thepeaker array and speaker array/fine-bubble system of 134 dB (ref

�Pa) (Fig. 6). The sound pressure field attenuated radially awayrom the speaker array system (Fig. 6b), while the sound pressureeld generated by the speaker array/fine-bubble system remainslevated between the fine-bubble diffusers located x/D = 0.3 onither side of the speakers at approximately 125 dB (ref 1 �Pa) at00 Hz (Fig. 6c), before attenuating rapidly in the channel wiseirection similar to that of the coarse-bubble system (Fig. 6a). Inddition to introducing a tactile stimulus, the presence of the fine-ubble curtain serves to re-orient the sound pressure gradientroduced by the speaker array alone.

. Discussion

This study demonstrated that common carp exhibit strongehavioral responses to bubble curtains in shallow water. Theraded-bubble and coarse-bubble systems produced similar effica-ies and reduced 75–85% of the number of barrier crossings madey common carp in the behavioral tank. Behavioral trials includingpeaker arrays and lighting indicated that sound and hydrodynamicelds influenced common carp movement more than visual cues

rom the bubble curtains.Results of the speaker array and speaker array/fine-bubble tri-

ls suggest that avoidance behavior near a bubble curtain is mostikely a response to some combination of stimuli (hydrodynamicnd acoustic), and not just an individual stimulus or characteristice.g. sound amplitude). Although the speaker array exhibited theame maximum sound pressure level of the coarse-bubble system,t did not match the level of passage reduction observed for bothhe coarse-bubble and speaker array/fine-bubble systems. Interest-ngly, the inclusion of the fine-bubble curtain to the speaker arrayesulted in an efficacy similar to that of the bubble curtain sys-ems (upstream 67%, downstream 82%). As expected, the bubbleurtains acted as resonant amplifiers which re-oriented the soundressure gradient in a stream-wise direction, rather than attenu-ting radially. A more abrupt change in the sound pressure fieldlong with improved efficacy appears to suggest that the trans-ission of sound through a bubble curtain, providing tactile and

uditory stimuli, may play a role in carp avoidance.

Sound, a variable not addressed in earlier bubble curtain stud-

es, appears in part to influence carp deterrence. Graded-bubble andoarse-bubble systems generated sound pressure levels well abovehe carp hearing threshold, and were sufficiently loud (roughly

D.P. Zielinski et al. / Ecological Engineering 67 (2014) 95–103 101

F coarsec to rigx config

32asbfowMwpbsa(r(ubnslatHas

mqtfast

(Hetsaibbbat

cpiitwcdte

scrSee

ig. 6. Sound Pressure Level (SPL) contours in dB (re 1 �Pa) at 200 Hz for the (a)

hannel along a straight line through the center of the bubble curtain (flow is left

/D = 0. Approximate locations of speakers and bubble diffusers are shown for each

0–40 dB (ref 1 �Pa) above background of 80 dB (ref 1 �Pa) at00 Hz) to prevent sound masking due to ‘cocktail party’ effects,

phenomena where the target signal is masked by backgroundounds (Popper and Carlson, 1998). Although bubble sizes differedetween curtain designs, the primary frequency of sound emittedrom the bubble curtains remained <300 Hz as result of collectivescillations of bubbles, reduced speed of sound in the bubble-ater mixture, and restrictive channel dimensions (Lamarre andelville, 1994; Nicholas et al., 1994). The sound pressure fieldas also observed to attenuate rapidly (i.e. exhibit a high soundressure gradient), an important feature for acoustic deterrentsecause it reduces the risk of fish from acclimating to the soundource. Rapid attenuation observed during the bubble curtain tri-ls was likely a result of the shallow water ‘cutoff phenomena’Urick, 1975; Akamatsu et al., 2002). The measured attenuationate of sound at 200 Hz for each bubble curtain was ∼107 dB/mref 1 �Pa), which is in close agreement with the theoretical atten-ation rate of 109 dB/m (ref 1 �Pa) derived from empirical relationsy Akamatsu et al. (2002). At such an attenuation rate, carp wouldot be expected to detect the presence of the bubble curtain viaound pressure (in 25 cm deep water) at distances beyond two bodyengths away (∼50 cm). Although sound is likely not responsible forll deterrence behaviors, the sound produced by these bubble cur-ains was biologically relevant and likely to influence carp passage.owever, at this time it remains unclear whether acoustic pressurend/or acoustic particle motion dominates the carp’s response to aound field (Popper and Fay, 2011).

Our finding of sound and fluid flow stimuli influencing carpovement through a bubble curtain is purely correlative. To

uantify the role of sound and fluid motion, further paired trialshat isolate each stimuli field is needed. The most logical option

or reducing/eliminating the influence of tactile stimuli could beccomplished by ablating the lateral line; however, results maytill be confounded since particle motion is detected not just byhe lateral line but by the inner ear as well.

9fiia

-bubble, (b) speaker array, and (c) speaker array/fine-bubble systems in the testht). Axes are normalized by the depth D = 25 cm and the barrier is centered abouturation.

A minimal reduction in efficacy to limit downstream passage87–75%) was observed during the illuminated coarse-bubble trial.ad the bubble curtain acted as a visual deterrent, an increase infficacy would have been expected for the illuminated case similaro what Mussen and Cech (2012) observed for lit and unlit fishcreen trials. This result was consistent with Welton et al. (2002)nd Patrick et al. (1985), who posited that when the bubble curtains well lit, fish are able to navigate through gaps in and around theubble curtain. In contrast, the lack of light in the remainder of theehavioral trials left no visible gaps and forced carp to avoid theubble curtain mainly using non-visual cues. This suggests that,t least for common carp, the role of visual cues to deter passagehrough a bubble curtain is limited.

Statistical comparisons between graded- and coarse-bubbleurtains suggest that both are equally effective at inhibiting carpassage. Despite an increased sound pressure level, fewer gaps

n the bubble curtain (result of grid configuration), and likelyncreased turbulence, the coarse-bubble curtain was only quali-atively better than the graded-bubble curtain. Although no testsere conducted to differentiate between effects of bubble curtain

onfiguration independent of bubble diameter, lack of statisticalifference between these bubble curtains suggests that changeso both bubble diameter and curtain layout minimally improvedfficacy.

Although responses to bubble curtains are known to be speciespecific, the avoidance of carp to the graded- and coarse-bubbleurtains is consistent with behaviors of other species found to beepulsed by bubble curtains (Stewart, 1982; Patrick et al., 1985;ager et al., 1987; McIninch and Hocutt, 1987). For example, Patrickt al. (1985) reported that gizzard shad, alewife, and smelt wereffectively excluded by a bubble curtain at a rate between 51% and

8%. Since other studies did not quantitatively measure the stimulield, direct comparisons is difficult. The 75–85% efficacy observed

n this study appears to be near the upper limit of bubble curtainslone since the graded- and coarse-bubble curtains used ∼30 and

1 al Eng

∼1ateehswswsi

v5iflpcc

4

tuctstoituba

A

NCas

A

i2

R

A

B

B

B

B

C

D

E

E

H

K

L

L

L

LL

L

M

M

M

M

N

N

P

P

P

P

P

R

S

S

S

S

S

S

S

T

T

02 D.P. Zielinski et al. / Ecologic

100 time more air than the next largest reported flow rate of Ls−1 m−1 (Welton et al., 1997) which only deterred salmon smoltt a rate of 40%. While others have developed bioacoustics systemshat combine bubble curtains with high power speaker arrays tonhance efficacy (Perry et al., 2012; Ruebush et al., 2012; Taylort al., 2005; Welton et al., 2002), these systems could be cost pro-ibitive for the small applications envisaged for the bubble curtainstudied here nor is it clear that they would work well in shallowater. The consistent activity level and steady control trial pas-

age rate observed throughout the entire testing period, coupledith and the avoidance of the graded- and coarse-bubble curtains

upports the hypthotheses that the bubble curtains can partiallympede fish passage.

Physical limitations of the experimental tank inflow system pre-ented behavioral testing at a background velocity greater than

cm/s. As the bubble curtain system was envisaged to be placedn low head streams that interconnect carp infested waters, whereow velocities are typically low, a velocity of 5 cm/s was deemedractical for this initial study. To better predict the effect of bubbleurtains under variable flow conditions, future evaluations shouldonsider in situ testing during natural flow regimes.

. Conclusion

This research represents the first effort to relate bubble cur-ain characteristics to fish deterrence. Our findings indicate thatnder laboratory conditions (25 cm deep and low flow), bubbleurtains provide a simple and inexpensive, yet effective meanso reduce common carp passage up to 80% in the up- and down-tream direction. Although the results of this study may not be fullyranslatable to field conditions, they do provide important insightn the responses of common carp to bubble curtains. In the future,

n-situ experiments of these bubble curtains should be performedo determine the efficacy for inhibiting common carp movementnder natural conditions. We recommend that improvement ofubble curtains should be based on enhancing hydrodynamic andcoustic cues rather than visual cues.

cknowledgements

This study was funded by the Minnesota Environmental andatural Resources Trust Fund as recommended by the Legislative-itizen Commission on Minnesota Resources (LCCMR). Aaron Clausnd Brian Moe helped with fish care and Zachary Sudman (Univer-ity of Minnesota) helped with data collection.

ppendix A. Supplementary data

Supplementary material related to this article can be found,n the online version, at http://dx.doi.org/10.1016/j.ecoleng.014.03.003.

eferences

kamatsu, T., Okumura, T., Novarini, N., Yan, H.Y., 2002. Empirical refinements appli-cable to the recording of fish sounds in small tanks. J. Acoust. Soc. Am. 112,3073–3082.

ainbridge, R., 1964. The problem of excluding fish from water intakes. Ann. Appl.Biol. 53, 508–515.

ajer, P., Sorensen, P., 2010. Recruitment and abundance of an invasive fish, thecommon carp, is driven by its propensity to invade and reproduce in basinsthat experience winter-time hypoxia in interconnected lakes. Biol. Inv. 12,

1101–1112.

rett, J.R., MacKinnon, D., 1953. Preliminary experiments using lights and bubblesto deflect migrating young spring salmon. J. Fish. Res. Board Can. 10, 548–559.

revik, I., Kristiansen, Ø., 2002. The flow in and around air-bubble plumes. Int. J.Multiph. Flow 28, 617–634.

U

W

ineering 67 (2014) 95–103

onover, W.J., 1980. Practical Nonparametric Statistics. John Wiley & Sons, NewYork, NY, pp. 493.

awson, H.A., Reinhardt, U.G., Savino, J.F., 2006. Use of electric or bubble barrier tolimit the movement of Eurasion Ruffe (Gymnocephalus cernuus). J. Great LakesRes. 32, 40–49.

lectric Power Research Institute (EPRI), 1998. Evaluation of Fish Behavioral Barriers.Technical Report 109483. Palo Alto, CA.

lectric Power Research Institute (EPRI), 2004. Chapter 22: Behavioral Methods, FishPassage Manual Chapter Updates. Technical Report 1011448. Palo Alto, CA.

untingford, F.A., Andrew, G., Mackenzie, S., Morera, D., Coyle, S.M., Pilarczyk Kadri,S., 2010. Coping strategies in a strongly schooling fish, the common carp Cyprinuscarpio. J. Fish Biol. 76, 1576–1591.

uznetsov, Y.A., 1971. The behavior of fish in the zone affected by a curtain of airbubbles. In: Alekeseeu, A.P. (Ed.), Fish Behavior and Fishing Techniques. NationalMarine Fish Service, NOAA, National Technical Information Service Transactions,71-50010. , pp. 103–110.

amarre, E., Melville, W.K., 1994. Void-fraction measurements and sound-speedfields in bubble plumes generated by breaking waves. J. Acoust. Soc. Am. 95(3), 1317–1328.

avis, D.S., Hallett, A., Koon, E.M., McAuley, T., 2003. History of and advances inbarriers as an alternative method to suppress sea lampreys in the Great Lakes.J. Great Lakes Res. 29 (Suppl. 1), 362–372.

eiberman, J.T., Muessig, P.H., 1978. Evaluation of an air bubbler to mitigate fishimpingement at an electric generating plant. Estuar. Coasts 1 (2), 129–132.

eighton, T.G., 1994. The Acoustic Bubble. Academic Press, San Diego, CA, pp. 913.eighton, T.G., Walton, A.J., 1987. An experimental study of the sound emitted from

gas bubbles in a liquid. Eur. J. Phys. 8, 98–104.in, J.N., Banerji, S.K., Yasuda, H., 1994. Role of interfacial tension in the formation

and the detachment of air bubbles. Langmuir 10, 936–942.anasseh, R., Nikolovska, A., Ooi, A., Yoshida, S., 2004. Anisotropy in the sound field

generated by a bubble chain. J. Sound Vib. 278, 807–823.cIninch, S.P., Hocutt, C.H., 1987. Effects of turbidity on estuarine fish response to

strobe lights. J. Appl. Icthyol. 3, 97–105.ichaud, D.T., Taft, E.P., 2000. Recent evaluations of physical and behavioral barriers

for reducing entrainment at hydroelectric plants in the upper Midwest. Environ.Sci. Policy 3 (1), 499–512.

ussen, T.D., Cech Jr., J.J., 2012. The roles of vision and the lateral-linesystem in Sacramento splittail’s fish-screen avoidance behaviors: evalu-ating vibrating screens as potential fish deterrents. Environ. Biol. Fish.,http://dx.doi.org/10.1007/s10641-012-0094-2.

icholas, M., Roy, R.A., Crum, L.A., Oguz, H., Prosperetti, A., 1994. Sound emissionsby a laboratory bubble cloud. J. Acoust. Soc. Am. 95 (6), 3171–3182.

oatch, M.R., Suski, C.D., 2012. Non-physical barriers to deter fish movements. Envi-ron. Rev. 20 (1), 71–82.

atrick, P.H., Christie, A.E., Sager, D., Hocutt, C., Stauffer Jr., J., 1985. Responses of fishto a strobe light/air-bubble barrier. Fish. Res. 3, 157–172.

erry, R.W., Romine, J.G., Adams, N.S., Blake, A.R., Burau, J.R., Johnston, S.V., Liedtke,T.L., 2012. Using a non-physical behavioral barrier to alter migration routing ofjuvenile Chinook salmon in the Sacramento-San Joaquin River Delta. River Res.Appl., http://dx.doi.org/10.1002/rra.2628.

opper, A.N., 1972. Pure-tone auditory thresholds for carp, Cyprinis carpio. J. Acoust.Soc. Am. 52 (6), 1714–1717.

opper, A.N., Carlson, T.J., 1998. Application of sound and other stimuli to controlfish behavior. Trans. Am. Fish. Soc. 127, 673–707.

opper, A.N., Fay, R.R., 2011. Rethinking sound detection by fishes. Hearing Res. 273(1/2), 25–36.

uebush, B.C., Sass, G.G., Chick, J.H., Stafford, J.D., 2012. In-situ tests of sound-bubble-strobe light barrier technologies to prevent range expansions of Asian carp.Aquat. Inv. 7, 37–48.

ager, D.R., Hocutt, C.H., Stauffer, J.R., 1987. Estuarine fish responses to strobe light,bubble curtains and strobe light/bubble-curtain combinations as influenced bywater flow rate and flash frequencies. Fish. Res. 5, 383–399.

isler, S.P., Sorensen, P.W., 2008. Common carp and goldfish discern conspecificidentity using chemical cues. Behaviour 145 (10), 1409–1425.

kov, C., Brodersen, J., Brönmark, C., Hansson, L.-A., Hertonsson, P., Nilsson, P.A.,2005. Evaluation of PIT-tagging in cyprinids. J. Fish Biol. 67, 1195–1201.

loan, J.L., Cordo, E.B., Mensinger, A.F., 2013. Acoustical conditioning and retentionin the common carp (Cyprinus carpio). J. Great Lakes Res. 39 (3), 507–512.

oga, C.L.M., Rehmann, C.R., 2004. Dissipation of turbulent kinetic energy near abubble plume. J. Hydraul. Eng. 130 (5), 441–448.

prott, T., 2001. Preliminary Report on the Field Testing of an Air-Curtain Screento Minimize Fish Passage onto Submerged Floating Drydocks. In: The NationalShipbuilding Research Program. NSRP 0589.

tewart, P., 1982. An investigation into the reactions of fish to electrified barriersand bubble curtains. Fish. Res. 1, 3–22.

aft, E.P., 2000. Fish protection technologies: a status report. Environ. Sci. Policy 3,349–359.

aylor, R.M., Pegg, M.A., Chick, J.H., 2005. Response of bighead carp to a bioacousticbehavioral fish guidance system. Fish. Manage. Ecol. 12, 283–286.

rick, R.J., 1975. Principles of Underwater Sound. McGraw-Hill, New York, NY, pp.

423.

ebb, J.F., Montgomery, J.C., Mogdans, J., 2008. Bioacoustics and the lateral linesystem of fishes. In: Webb, J.F., Fay, R.R., Popper, A.N. (Eds.), Fish Bioacoustics.Springer Science+Business Media, New York, pp. 17–48.

al Eng

W

W

W

D.P. Zielinski et al. / Ecologic

eber, M.J., Brown, M.L., 2009. Effects of common carp on aquatic ecosystems 80years after carp as a dominant: ecological insights for fisheries management.

Rev. Fish. Sci. 17 (4), 524–537.

elton, J.S., Beaumont, W.R.C., Ladle, M., Masters, J.E.G., 1997. Smolt Trapping UsingAcoustic Techniques. Phase 1, Literature Review and Initial Investigation ofAcoustic Bubble Screens. Environmental Agency, R&D Technical Report W66,Bristol.

Z

ineering 67 (2014) 95–103 103

elton, J.S., Beaumont, W.R.C., Clarke, R.T., 2002. The efficacy of air, sound andacoustic bubble screens in deflecting Atlantic salmon Salmo salar L., smolts in

the River Frome, UK. Fish. Manage. Eco. 19, 11–18.

weiacker, P.J., Gaw, J.R., Green, E., Adams, C., 1997. Evaluation of an air-bubblecurtain to reduce impingment at an electric generating station. In: Proceedingsof the Annual Conference of SE Association Fish Wildlife Committee 31, pp.343–356.