Acoustic deterrents do not reduce black drum predation on oysters.

ABSTRACT We studied the effectiveness of acoustic deterrents in limiting predation on Louisiana oyster leases by black drum Pogonias cromis. Three acoustic deterrents were used: (1) natural or synthesized alarm sounds; (2) synthesized, low frequency sound played over short time intervals and (3) mechanically produced sound. Fish locations were monitored in all approaches and feeding rates on oysters were recorded in approaches 1 and 3. Experiments in salt-water ponds indicated that alarm calls from males, or synthesized tapes (approach 1) did not depress fish feeding rates or cause fish to avoid transducers, in comparison with controls. Lower frequencies (<20 Hz, approach 2) displaced fish 8 m further from transducers (about 20% of the distance possible) relative to controls. Finally, a solar-powered hammer (approach 3) was designed as a more logistically feasible deterrent. Fish were weakly attracted (although control and experimental locations differed by only 5%). Because acoustic deterrents were either ineffective at displacing fish or lowering feeding rates on oysters, or required considerable electrical power to displace fish, we conclude that they are not practical to control losses of oysters on leases to black drum.

KEY WORDS: black drum, sound deterrents, oysters

INTRODUCTION

The black drum, Pogonias cromis (Linnaeus, 1766), is a significant predator of oysters in the northern Gulf of Mexico (Cave 1978, Sutter et al. 1986, Dugas 1986). Mortality rates are low in intertidal oyster beds because of aerial exposure and the aggregated growth form, but commercial, subtidal oyster leases often experience high mortality. Oyster mortality is especially high after small, individual "seed" oysters are bedded in the fall, or when black drum return to estuarine waters in the spring after breeding in coastal passes, in a survey of lease holders in Louisiana (LADWF 2000), 55% reported black drum caused losses to production. With powerful pharyngeal teeth, black drum crush and eat 30 oysters per night, and black drum >70 cm in length can feed on oysters 75 mm in length (Cave 1978). Our earlier work indicated mortality rates as high as 90% on experimental leases, and that scent deterrents (e.g., from dead con-specifics) were not effective (Brown et al., 2003).

Some evidence suggested that acoustic deterrents might be effective. Black drum are vocal (as are other Sciaenid fishes) and emit low frequency "drumming" sounds, which may attract females or act as alarm calls. These sounds can be heard by humans both in air and water and are produced by muscular contractions against an inflated swim bladder that acts as a resonator (Smith 1905). Black drum detect frequencies ranging from <100-800 Hz and are most sensitive to frequencies <500 Hz (Ramcharitar & Popper 2004).

Acoustic cues can also alter fish distributions and feeding rates in other fish. Haymes and Patrick (1986) excluded alewives from water intakes using low-frequency sounds. Ross and Dunning (1996) used 190 dB high frequency (122-128 kHz) broad-based sound, and reduced alewife abundance by 51% at power plant intakes in Lake Ontario. Knudsen et al. (1994) found that salmon (whose hearing range extends from 5-150 Hz), avoid low frequency infra-sound (10 Hz) but not high frequency sounds (150 Hz), and argued that low frequency sounds mimic the swimming of predators. Maes et al. (2004) used multiple sounds in a frequency range of 20-600 Hz to decrease fish abundance at an estuarine power plant by 60%.

Our research objective was to determine if acoustic deterrents either reduced the feeding rates of black drum on seeded oysters, or caused them to avoid oyster leases. Three general types of acoustic deterrents were used: (1) natural or synthesized alarm sounds; (2) synthesized low frequency sound played over short time intervals and (3) mechanically produced sound. Fish locations were monitored in all approaches and feeding rates on oysters were recorded in approaches 1 and 3.

MATERIALS AND METHODS

Experiments were conducted in four 0.1 ha saltwater ponds at the Lyle St. Amant Marine Laboratory on Grand Terre Island (29.3[degrees]N 89.99[degrees]W) operated by the Louisiana Department of Wildlife and Fisheries. Ponds contained pumped-in Barataria Bay sea water (over the experimental interval, mean temperature was 23.4 [+ or -] 0.6[degrees]C [+ or -]SE], salinity 20.6 [+ or -] 0.4 PSU, and dissolved oxygen 8.3 [+ or -] 0.4 mg/L) and had removable stand pipes to aid in draining ponds to recapture fish. However, pumps were not run during experiments to eliminate masking of experimental sounds. The ponds were large enough (25 m x 40 m x 1 m deep) so that sound would attenuate with distance (see later in discussion) and reflections would be minimized, allowing drum to find areas with low sound. Although low frequency sounds attenuate rapidly in shallow water over soft sediments (Rogers & Cox 1988), oyster leases are typically seeded under such conditions, and the ponds thus accurately model similar conditions on oyster leases.

In March 2001, using contiguous 30,000 L concrete raceways at the laboratory, we played back tapes with pure sinusoidal wave forms at various frequencies and observed individual black drum. Frequencies from 10 Hz to 100 Hz caused fish to move to the opposite end of the tank. However, there was considerable "crosstalk" (sound transfer among the concrete tanks) along with standing waves that complicated determining the strength of acoustic signals.

Similar studies however indicated a clear gradient in sound intensity along the length of the 0.1 ha ponds, similar levels on each side, and little cross talk among ponds (Fig. 1). An Argotech model 210 underwater transducer (positioned at the center of one end of the pond) produced a pure tone of 164 dB [micro]Pa/@ 1 m at a frequency of 100 or 300 Hz. Sound levels were measured with a "Dolphin Ear" DP100-6 hydrophone (frequency response 7 Hz to 22 KHz) and decibel meter at each corner near the transducer (~12.5 m distant), halfway down each side (~23.5 m), and at the center(~40 m) and each corner (~42 m) of the opposite end. At 100 Hz, transmission loss was approximately 90% at 12.5 m, 97% at 23.5 m, and 99% at 40-42 m. At 300 Hz, loss was 96.4% at 12.5 m, 95% at 23.5 m, and 99.2% at 40-42 m. Little variation in signal strength occurred across the pond at any distance, and signals were not detected in adjacent ponds, indicating no cross talk.

[FIGURE 1 OMITTED]

For approach 1, experiments were conducted from January to April 2002. Two Argotech model 210 underwater single-neoprene membrane transducers, at the center of one end of each experimental pond, were powered by a 200 W Peavy CS800X amplifier, connected to a cassette recorder. Two continuous-loop tapes were supplied by Argotech. The first was alarm "drumming" by a male in the raceways, edited to remove splashes, and with bandwidth limited to 4 KHz (frequency peaks were at 37, 70, 146, 210 and 300 Hz, Fig. 2). The second was a synthesized recording predicted to produce avoidance based on previous work by Argotech (Fig. 3, with peaks at 20, 40, 60 and 267 Hz).

[FIGURES 2-3 OMITTED]

Transducers were switched weekly, so treatment ponds were controls and the previous controls were treatments, to minimize pond-specific differences in fish feeding rates or habituation. Alter four weeks, the sound was changed to the synthesized tape (and all fish replaced) and another four, weeklong experiments were run in the same pattern.

Each pond had 5 black drum, with a mean total length of 82.4 [+ or -] 1.6 cm (SE) and weight of 9.9 [+ or -] 0.4 Kg. Fish were collected by trot line or rod and reel in nearby (<2 Km) Barataria Pass, and were held in raceways. Five fish were used to facilitate tracking (see below). Actual school size has yet to be studied in Black Drum, but we have observed them in both small (~5) and larger groups in Barataria Bay (K. Brown, pets. obs.).

To study feeding, 50 oysters of vulnerable size (50-150 g total wet weight, 5-10 cm shell length, Brown et al. 2003) were placed in each of six Nestler trays (0.67 [m.sup.2] bottom area mesh trays with 0.1 m high walls). Three trays were along the short edge of each pond, with the transducer at the center of the trays on one end. No other food was provided. Experiments lasted three days: oysters were replaced daily to determine feeding rates, and trays were then removed for three days to starve fish before the next experiment. Feeding rates were analyzed in a 1-way analysis of covariance, with time since the start of the experiments as a covariate to control for seasonal effects.

To follow movement, fish were tagged with numbered "spaghetti" tags inserted below the soft dorsal fin and attached to individually colored small (~2 x 7 cm) styrofoam floats by 200 pound test, monofilament line. Because the floats were small, and black drum are relatively large fish, they did not impair movement or grouping patterns. Locations (to the nearest 0.5 m) were recorded hourly from 10:00-16:00 on a grid laid out in the ponds, and distances from the transducer digitized. The hourly mean distance of the five fish from the transducer (or the same location in the control ponds) was analyzed in a 1-way repeated measures analysis of variance, contrasting the controls and two recorded sounds. The repeated measure was hour of day.

The second approach involved broadcasting pure tones at frequencies of 10, 15, 20, 25, 30, 40, 50 and 60 Hz in a single pond. The same amplifier powered a single, larger Argotech model 220 twin-neoprene membrane transducer at the center of one end. Although lower-frequency sounds attenuate rapidly in shallow, soft-bottomed environments (see earlier discussion), and we did not explicitly measure attenuation in this experiment, signals carried far enough to displace fish (see results), and were thus apparently audible. Following Maes et al. (2004), we contrasted fish positions during sound-on and sound-off periods. Each experiment had a five-minute interval with sound, and five minutes without. In this experiment, 4 tagged fish were used, and we used similar methods to record the positions of the tagged fish (mean length = 94.7 [+ or -] 2.3 cm, mean weight = 13.0 [+ or -] 0.9 KG) each minute. The amplifier was then turned off for five minutes before the next frequency was tested. The experiment was repeated the next day, so that 10 replicate observations were recorded for each frequency and its corresponding control period. The locations of the four fish relative to the transducer at each minute interval were analyzed in a 2-way analysis of variance (sound on or off versus frequency). Experiments were conducted from December 6 to 8, 2002.

For the final approach, we collaborated with biological engineers at Louisiana State University's Department of Biological Engineering to produce a mechanical, sound-producing device. The rationale was 2-fold. First, the underwater transducers required considerable power, and were prone to leaks. Second, striking a metal oil drum with a wooden stick near the pond edge caused fish to rapidly move away. The device (Fig. 4) consisted of two 55 gallon, sealed steel drums that provided buoyancy. One drum had an internal electric motor (Fig. 4, inset) activating a spring-loaded hammer. Both the time between impacts and the number of impacts per application were adjusted several times during the experiment by a dial on the top of the apparatus to limit habituation. The motor was powered by a 12-V automotive battery charged by a solar cell suspended above the drums.

[FIGURE 4 OMITTED]

The experiment was conducted in one experimental and one control pond. Four tagged fish (mean length = 86.4 [+ or -] 1.8 cm, mean weight = 10.3 [+ or -] 0.5 KG) were used in this final experiment in each pond. The apparatus could not be easily transferred among ponds, so experimental and control ponds were not reversed. Each experiment lasted two days. On the first evening, six trays filled with 50 oysters were added, and tagged fish locations determined for six hours the next day. In earlier experiments, hourly locations were a "snapshot", as fish moved constantly. We therefore noted average fish locations for the first 10 minutes of each hour in this experiment. A repeated measures analysis of variance indicated no repeated measures effect (P = 0.15), so all 60 observations (6 h x 10 observations) were used as replicates in a l-way analysis of covariance. The experiment was repeated five times during April to May 2003. Number of days since the start of the first experiment was the covariate.

RESULTS

For the initial approach, feeding rates were not depressed by alarm sounds ([F.sub.2,59] = 0.7, P = 0.52), in comparison with the synthetic sounds or the control (Fig. 5). On average, fish exposed to alarm sounds consumed 49% of the oysters provided daily, in comparison with 67% with synthetic tapes, and 61% in controls. The covariate was significant ([F.sub.1,59] = 22.6, P < 0.0001) because feeding increased with warmer temperatures or acclimation of fish.

[FIGURE 5 OMITTED]

Hour of observation did not affect fish location (Wilk's [lambda] = 0.83, [F.sub.6,46] = 1.5, P = 0.19), nor was there a treatment effect ([F.sub.2,51] = 0.4, P = 0.7). There was an interaction (Wilk's [lambda] = 0.63, [F.sub.12,92] = 2.0, P = 0.04), although inspection of the pattern of treatments across hours (Fig. 6) revealed no obvious trends.

[FIGURE 6 OMITTED]

In the second approach with lower-frequency sound, frequency was not significant ([F.sub.8,152] = 0.7, P = 0.69), but there was a difference in location between when the transducer was on and off ([F.sub.1,152] = 116.6, P < 0.0001), and an interaction between frequency and sound activation ([F.sub.8,152] = 5.7, P < 0.0001).

Fish averaged 17 m from the transducer when it was inactive, versus 25.1 m when it was broadcasting. For example, at 15 Hz (Fig. 7), fish moved from less than 10 m to over 33 m away from the transducer within 3 min, and returned to within 15 m after the transducer was switched off. The small standard errors in the figure reflect the close spatial clustering of the fish. The interaction occurred because experimental and control fish locations converged at frequencies >20 Hz (Fig. 8). Comparisons of locations with Tukey a posteriori tests indicated fish were further away from transducers at 20 Hz, and closer at 50 Hz, with the distances in all other treatments overlapping. However, eight of the nine highest displacements occurred when the transducer was on, again suggesting the fish were avoiding the lower frequency sounds.

[FIGURES 7-8 OMITTED]

For the final approach with the mechanical device, there was a difference in fish location between the experimental and control ponds ([F.sub.1,596] = 8.5, P < 0.01). However, the trend was opposite of that predicted, with fish on average 19.2 m away in the control pond, versus 17 m in the experimental pond (Fig. 9). There was also a difference in feeding rates ([F.sub.1,18] = 24.6, P < 0.0001) with rates in the control pond only 2% of those in the experimental pond (Fig. 9). The covariate did not have a significant effect ([F.sub.1,19] = 0.7, P = 0.42).

[FIGURE 9 OMITTED]

DISCUSSION

The first approach suggested that there was little effect of male alarm sounds either in depressing feeding, or in displacing black drum from the sound source. Although there were no significant main effects on fish location, variation among the treatments at different times of day caused a significant statistical interaction.

The second approach did indicate displacement of fish from the transducer over the short term at lower frequencies. However, the electrical power necessary to propagate these low frequency sounds underwater makes this approach logistically impractical as a deterrent in field applications (see discussion below). Underwater transducers were also prone to leaks and malfunctions, and had to be replaced on several occasions by Argotech during the experiments.

The results of the first two approaches must be interpreted carefully. Again, low frequency sound attenuates rapidly in shallow, soft-bottomed environments (Rogers & Cox 1988), possibly limiting fish responses. However, the initially measured sound field, although indicating considerable signal loss across the pond, also indicated a sound volume at the opposite end that was still above the threshold of black drum at 100 Hz, about 90 dB re l[micro]Pa (Ramcharitar & Popper 2004). Second, fish were evidently able to hear the low frequency sounds, because they were displaced. Finally, regardless of the mechanism (whether fish ignored the sounds or could not discern them), the first two approaches were either not effective (approach 1) or practical (approach 2), negating their use in management. We are confident similar results would occur on oyster leases, because the ponds again had similar depth and sediment type.

The last approach, a solar-powered hammer, was more cost effective but weakly attracted fish. However, the average distances differed by only 2 m, only 5% of the distance available. The higher feeding rates in the experimental pond must also be interpreted carefully, because the apparatus could not be moved between ponds. The experimental fish may have been more accustomed to feeding on oysters, similar to differences in feeding between individual groups of fish used in past experiments (Brown et al. 2003).

We conclude that the acoustic deterrents used in approach one and three were not effective for reducing predation by black drum, or could even attract fish. Using sound stimuli at low frequencies to displace fish (approach 2), although effective, requires more power, and we conclude costs to lease holders would outweigh benefits. Many Louisiana oyster leases are in remote areas, and supplying sufficient electrical power is again problematic. Conducting these lower frequency experiments over longer time intervals might also have revealed habituation in the responses.

ACKNOWLEDGMENTS

This research was funded by Sea Grant's Gulf Oyster Initiative Program (GOIP - 19). The Louisiana Department of Wildlife and Fisheries and the Lyle St. Amant Marine Laboratory provided experimental raceways and saltwater ponds, and also provided lodging and a logistical base. Mr. P. Vujnovich Jr. provided oysters used in feeding experiments in the ponds. Professor Randy Price, LSU Department of Biological Engineering, designed and constructed the solar-powered device. Argotech provided transducers, amplifiers and technician support during the initial experiments.

LITERATURE CITED

Brown, K. M., G. W. Peterson, P. D. Banks, B. Lezina, C. Ramcharan & M. McDonough. 2003. Olfactory deterrents to black drum predation on oyster leases. J. Shellfish Res. 22:589-595.

Cave, R. N. 1978. Predator-prey relationships involving the American oyster, Crassostrea virginica (Gmelin), and the black drum, Pogonias cromis (Linnaeus), in Mississippi Sound. Master's thesis. Southeastern Louisiana University, Hammond, LA.

Dugas, C. N. 1986. Food habits of black drum, Pogonias cromis, in southeast Louisiana with emphasis on their predation of the American oyster, Crassostrea virginica. Contri. Marine Res. Lab., 1980-1985. Louisiana Dept. Wildlife Fish. Tech. Bull. pp. 32-38.

Haymes, G. T. & P. H. Patrick. 1986. Exclusion of adult alewife, Alosa pseudoharengus, using low-frequency sound for application at water intakes. Can. J. Fish. Aq. Sci. 43:855-862.

Knudsen, F. R., P. S. Enger & O. Sand. 1994. Avoidance responses to low frequency sound in downstream migrating Atlantic salmon smolt, Salmo salar. J. Fish Bio. 45:227-233.

Louisiana Department of Wildlife and Fisheries. 2000. Black drum predation: results of a survey of Louisiana oyster ground lease holders. Final Report, Baton Rouge.

Maes, J., A. W. H. Turnpenny, D. R. Lambert, J. R. Nedwell, A. Parmantier & F. Ollevier. 2004. Field evaluation of a sound system to reduce estuarine fish intake rates at a power plant cooling water inlet. J. Fish Bio. 64:938-946.

Ramcharitar, J. & A. N. Popper. 2004. Masked auditory thresholds in sciaenid fishes: a comparative study. J. Acoust. Soc. Am. 116:1687-1691.

Rogers, P. H. & M. Cox. 1988. Underwater sound as a biological stimulus. In, J. Artema, R. R. Fay, A. N. Popper & W. N. Travolga, editors. Sensory biology of aquatic animals, New York, New York: Springer Verlag Press. pp. 131-149.

Ross, Q. E. & D. J. Dunning. 1996. Reducing impingement of alewives with high frequency sound at a power plant intake on Lake Ontario. N. Amer. J. Fish. Mgmt.

Smith, H. M. 1905. The drumming of the drum-fishes (Sciaenidae). Science 22:376-378.

Sutter, F. C., R. S. Waller, & T. D. McIlwain. 1986. Species profiles: life histories and environmental requirements of coastal fishes and invertebrates (Gulf of Mexico)--Black Drum. Report Number Biological-82 (11.51). Gulf Coast Research Laboratory, Ocean Springs, MS.

KENNETH M. BROWN, (1) * GARY W. PETERSON, (2) GERALD J. GEORGE (1) AND MICHAEL MCDONOUGH (3)

(1) Department of Biological Sciences; (2) Coastal Fisheries Institute; (3) Department of Oceanography and Coastal Studies Louisiana State University, Baton Rouge, Louisiana 70803

* Corresponding author. E-mail: [e-mail]