- Email: [email protected]
Traffic noise masks acoustic signals of freshwater stream fish
Biological Conservation 187 (2015) 27–33
Contents lists available at ScienceDirect
Biological Conservation journal homepage: www.elsevier.com/locate/biocon
Traffic noise masks acoustic signals of freshwater stream fish Daniel E. Holt, Carol E. Johnston ⇑ Fish Biodiversity Lab, School of Fisheries, Aquaculture and Aquatic Sciences, Auburn University, Auburn, AL 36849, United States
a r t i c l e
i n f o
Article history: Received 20 January 2015 Received in revised form 30 March 2015 Accepted 5 April 2015 Available online 22 April 2015 Keywords: Blacktail Shiner Noise pollution Bioacoustics Acoustic propagation
a b s t r a c t In order for an acoustic signal to be an effective source of communication, the signal must be successfully detected and interpreted by the intended receiver. One potential barrier to acoustic communication is background noise. Lotic systems contain a wide variety of habitats including riffles, shoals and waterfalls that can become quite noisy. The increasing prevalence of road and train crossings over small streams, and increased boat traffic in navigable rivers and lentic systems also presents potential anthropogenic noise sources with which vocal fishes did not evolve. The present study investigates the relationship between vocalizations and the natural soundscape of a common fish of the Southeastern United States, the Blacktail Shiner (Cyprinella venusta), and the potential effects anthropogenic noise from bridge crossings may have on the soundscape and acoustic communication in this species. Results revealed a particularly close association of a quiet window in the natural soundscape of C. venusta and dominant frequencies of the courtship vocalization of C. venusta. Results also indicated that C. venusta’s acoustic signals propagate short distances, following predictions based on the calculated cutoff frequency of the streams they inhabit, and were masked by noise generated from bridge crossings. Our calculations suggest that road traffic noise propagates to an extent that virtually entire watersheds are impacted by this noise pollution, especially in urban areas. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Because sound has the capacity to carry information, can be used intermittently, and does not require a line of sight, it is used extensively by animals as a mode of information transfer. Sound production has been documented in over 800 species of fishes representing 109 families within the infraclass Teleostei (Kasumyan, 2008). Despite the fact that a large number of fishes likely utilize sound for communication, numerous factors concerning the properties of the projected sound, hearing abilities of the receiver, constraints imposed by the physical environment, and ambient noise levels of the environment must fit together properly in order for acoustic communication to be effective. Ambient noise from both biotic and abiotic (wind, rainfall, turbulence) sources can decrease the signal-to-noise ratio of signals, or make temporal information more difficult to extract (Wysocki and Ladich, 2005). Studies on terrestrial species have shown that these environmental noise sources can act as strong selective pressures in the evolution of signal structure (Waser and Waser, 1977; Wiley and Richards, 1982; Jouventin et al., 1999; Narins et al., 2004). Unlike natural biotic and abiotic noise sources, the relatively recent development and rapid expansion of human activities such ⇑ Corresponding author. Tel.: +1 334 844 1781; fax: +1 334 844 9207. E-mail address: [email protected] (C.E. Johnston). http://dx.doi.org/10.1016/j.biocon.2015.04.004 0006-3207/Ó 2015 Elsevier Ltd. All rights reserved.
as urbanization, shipping, motorized recreational activities, drilling, and seismic explorations (Myrberg, 1990; Popper, 2003) do not provide the time necessary for the evolution of acoustic signals. Efforts have been made to determine the effect of anthropogenic noise on marine mammals (Southall et al., 2007; Hastings, 2008), and primarily marine fishes (Codarin et al., 2009; Ladich, 2013; Radford et al., 2014; Voellmy et al., 2014a,b). Elevated noise levels have been shown to reduce egg survival, reproduction and growth rates in fishes (Banner and Hyatt, 1973) and shrimp (Lagardère, 1982). Studies have also shown that anthropogenic noises can affect fish hearing or behaviors (Fernandes et al., 2000; Vabø et al., 2002; Handegard et al., 2003), which can potentially have detrimental effects on fitness. Amoser et al. (2004) found, for example, that noise from powerboats racing on an alpine lake was loud enough to be detected by otophysine fishes (fishes possessing a hearing specialization; see Popper and Fay, 2010) at up to 400 m away. Vasconcelos et al. (2007) found that the noise from ferry boats in the Tagus River estuary (Portugal) caused significant hearing threshold shifts in the Lusitanian toadfish (Halobatrachus didactylus), and that females ability to detect male signals would be significantly diminished under ship noise. Despite efforts that have been made in other habitats, we are currently unaware of any study that has looked at potential anthropogenic noise sources in small freshwater streams, and how these noise sources may impact the ability of small, vocal
28
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
fishes to communicate acoustically. While several studies have examined signal propagation with regard to the ambient environment in shallow water systems, these studies have not included the effects of anthropogenic noise (Fine and Lenhardt, 1983; Ghahramani et al., 2014; Locascio and Mann, 2011; Lugli and Fine, 2007). Because small freshwater systems are home to a disproportionately large percentage of the imperiled fishes of the southeastern United States, an initiative must be taken to better characterize anthropogenic noise transmitted into these systems. The current study was aimed at describing the interaction between the natural soundscape and acoustic repertoire of the Blacktail Shiner (Cyprinella venusta) in east Alabama. We also examined the effect that bridge traffic, one particular anthropogenic noise source common in low order streams, may have on the natural soundscape and the ability of C. venusta to communicate acoustically. The results provide a better understanding of how a common source of anthropogenic noise may affect the acoustic soundscape of small freshwater streams and rivers. 2. Materials and methods 2.1. Natural ambient noise measurements All natural ambient noise measurements were made on 12 May 2011 between 1300 and 1600 h. A large shoal on Little Uchee Creek (Lee Co., AL, U.S.A., 32° N, 85° W), which is a tributary of the Chattahoochee River, was chosen to describe the natural soundscape of C. venusta. Water temperature was 27.2 °C. This location was chosen because it offered a wide variety of suitable spawning habitats for C. venusta and during the reproductive summer months, the water is periodically shallow and clear enough to allow a researcher to locate the exact locations of nests by watching the fish behave from the bank. Nine active nest sites were identified by observation from exposed bedrock using polarized sunglasses. Nest sites were typically found at the confluence of an area of high flow and a pool. However, spawning aggregations were also observed directly within rapidly flowing chutes. A hydrophone (Hi-Tech HTI-96-MIN, sensitivity 164.4 re 1 V/lPa, frequency response: 0.002–30 kHz) and digital recorder (Marantz PMD 661, sampling rate 44.1 kHz) were used to record 1 min of ambient noise in each of the 9 sites. In sites with substantial flow, an effort was made to place the hydrophone in a low flow area adjacent to the flow to reduce hydrodynamic noises. Sounds were imported into Raven 1.4 (Cornell University, Ithaca, NY), where three, 1 s segments were randomly selected from the recording made at each site. Two power spectra of each segment were then calculated using the power spectrum function of Raven (Hamming window, 50% time overlap, FFT length: 2048 samples, analysis bandwidth: 21.5 Hz; FFT length: 512 samples, analysis bandwidth: 86.1 Hz). Two separate power spectra, each with a unique analysis bandwidth had to be produced for natural ambient noise and anthropogenic noise sources so that SNR’s could be calculated between C. venusta acoustic signals and noise. When calculating SNR’s, it is necessary to analyze all sounds to be included in the analysis at the same frequency resolution. The analysis bandwidths of 21.5 and 86.1 Hz result from the typical duration of growls and knocks, respectively. For each analysis bandwidth, the three power spectra from each nest site were exported into Microsoft Excel where they were averaged to produce a single power spectrum for each of the 10 nest sites. Kendall’s concordance test was used to determine whether spectrum shape (using the spectrum curve with analysis bandwidth of 21.5 Hz) of natural ambient noise between 21.5 and 1999.5 Hz was significantly different across active nesting sites (Lugli and Fine, 2003). No difference was observed in the shape of curves from different nesting sites, and so power spectra from all sites were
averaged to generate a single, composite power spectrum for natural ambient noise. Spectrum levels were calibrated to represent absolute levels using the sensitivity of the hydrophone and a GW GOS-6xxG dual trace Oscilloscope, and by taking into consideration the gain applied to the signal by the Marantz and when importing sounds into Raven. The frequency range of major spectral components (such as bandwidth of the quiet window found in the natural field recordings) of natural and anthropogenic sounds were defined as the range of frequencies within 3 dB of the peak frequency of the average power spectrum. This is standard for determining general tuning properties of sounds (Bennet-Clark, 1999), and has been used in a similar context by Lugli (2010). 2.2. Anthropogenic noise source measurements and propagation Source levels and propagation of semi-trailer trucks crossing streams were measured at 6 road crossings, all located within Lee County and Macon County, Alabama (Table 1). Recordings were made between 3 and 14 March 2010. Water temperatures were not recorded. All crossings were beam bridges supported by a piling at the junction of each bridge segment. For all recordings, the hydrophone was placed approximately 8 cm off the substrate. This depth was chosen because the substrate was mostly sand and gravel, and potential nests in this type of habitat are usually close to the substrate. The hydrophone was mounted to the end of a 17.7 cm PVC pipe, which was secured between two submerged sandbags and positioned in such a way that the sandbags were downstream of the hydrophone. Water depth and flow velocity were not recorded, however water depth never exceeded approximately 84 cm (the maximum depth at which the hydrophone apparatus could be set up with without the researcher having to submerge their head). At each road crossing, one hydrophone was fixed 1 m upstream or downstream of the bridge’s edge (direction was determined by accessibility), while a second hydrophone was moved different distances away from the first hydrophone in the direction opposite the bridge. Both hydrophones recorded simultaneously onto separate channels of the Marantz digital recorder. Prior to going into the field, gain on each channel of the Marantz was made equal by recording a tone of known amplitude with the same hydrophone successively on each channel, and adjusting gains on each channel until the level was the same on both channels. Distances separating the hydrophones varied between 2 and 16 m for each site (Table 1), and depended on the locations of suitable (not flowing rapidly) and accessible (shallow enough to set hydrophone up on substrate) habitat. Because of hydrophone cable length constraints, recordings with both hydrophones simultaneously were not possible beyond 16 m. However, at three of the six sites, recordings were made with a single hydrophone at several distances up to 82 m. Recording was performed for several minutes at each distance, and the time at which semi-trailer trucks passed was noted. Care was taken to ensure that at least 3 trucks passed over, and that at least one 10 s period with no traffic occurred during the recording period at each distance. Table 1 Recording sites and hydrophone distances from the source of anthropogenic noise. Location
Total bridge length (m)
County
Hydrophone separation (m)
I-85 at Choctafaula Cr. I-85 at Hodnett Cr. I-85 at tributary to Choctafaula Cr. I-85 at Uphapee Cr.
62.5 92.5 72.5
Lee Macon Macon
2, 4, 12, 16 2, 4, 13, 40, 80 2, 4, 6, 12, 16, 36
200.1
Macon
I-85 at Cubahatchee Cr. C.R. 40 at Calebee Cr.
129.2 152.8
Macon Macon
2, 4, 6, 9, 12, 14, 23, 49, 82 2, 4, 6, 12, 16 4, 7.5, 15
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
To calculate source levels of semi-trailer truck noise, a total of three truck passes were selected from the recordings made by the hydrophone 1 m away from the bridge at each site. As before when measuring natural environmental noise, two power spectra with different analysis bandwidths were generated at each distance. Power spectra from the three truck passes at each site were exported to Excel, and were averaged to generate a final power spectra at each analysis bandwidth representing semi-trailer truck noise 1 m downstream of bridges. To calculate attenuation of semi-trailer truck noise, three truck passes at each distance from the bridge, at each site, were selected from the audio files of both hydrophones. Power spectra were calculated for each of the three passes using Raven (FFT length: 32,768 samples, analysis bandwidth = 1.35 Hz), and exported to Excel. The average power in 1/3 octave bands was then calculated for bands with center frequencies at 100 and 315 Hz (representing the lower and upper frequency bands of C. venusta growls) and 160 and 630 Hz (representing the lower and upper frequency bands of C. venusta knocks). Attenuation in each of the 1/3 octave bands was calculated by subtracting the sound pressure level between the two hydrophones, and averaging this attenuation across the three semi-trailer truck passes recorded at each distance. Kolmogorov– Smirnov tests were used to determine whether attenuation rates differed between the 1/3 octave bands, and bands that did not differ were combined. Attenuation rates of combined frequency bands were plotted as an X–Y scatter, and the best fitting curve for the attenuation rates was calculated. A logarithmic curve estimation procedure was used to generate the best fit lines, and ANOVA was used to test the acceptability of each model. 2.3. Measurement of C. venusta signal spectrum level C. venusta knock and growl spectrum levels were measured in the laboratory. A detailed description of C. venusta acoustic signals and behaviors, along with locality information for where fish were captured and how they were stored is provided by Holt and Johnston (2014a), and a description of how spectrum levels were determined under quiet conditions is provided by Holt and Johnston (2014b). An artificial nest was set up within a 76 L aquarium, and a hydrophone was placed in front of the nest. Two males and one to two females were placed in the tank at a time, and allowed to acclimate for between 20 min and 9 h, depending on the time taken for normal behaviors and nest guarding behavior to resume. Vocalizations produced during approach and lateral display behaviors were recorded onto Raven using a Brüel and Kjaer 8103 hydrophone, and a Brüel and Kjaer 2635 charge amplifier. Growls were collected from a total of 19 different males, and knocks were collected from a total of 14 different males. Only signals recorded within 15 cm of the hydrophone (to minimize distortion caused by interaction with the 3726 Hz resonant frequency of the aquarium), and that were less than 11.6 ms (for knocks) and 46.4 ms (for growls) in duration were used for analysis. Analyzed signals were limited to this duration so that power spectra with a uniform analysis bandwidth could be generated using Raven. Power spectra were exported to Excel where an average knock power spectrum, and growl power spectrum were generated. 2.4. Propagation of C. venusta acoustic signals Attenuation of fish signals was measured at three different spots adjacent to one of the natural nesting sites by playing the sounds back through an underwater speaker (University Sound UW-30, Oklahoma City) and recording with the Marantz digital recorder at different distances. The playbacks were conducted on 7 June 2011 between 1345 and 1416 h, and on 20 June 2011 between 1036 and 1220 h. Water temperature was 25 °C for the
29
recordings made on 7 June 2011, and 28.3 °C for the recordings made on 20 June 2011. The speaker was positioned with the center of its cone 20 cm above the substrate at all three locations, and 16.4 cm below the water’s surface at the first location, 11.2 cm below the water’s surface at the second location, and 18.1 cm below the water’s surface at the third location. All three locations were within a 10 m radius of the nesting site, and the speaker was always arranged in such a way that it projected sound parallel to the water’s surface with no structures between the front of the speaker and the hydrophones. A recording of a knock vocalization was played from a Blackberry curve 8520, and amplified by a Pioneer GM-X372 240W amplifier. The playback was calibrated so that the spectrum level of the sound’s dominant frequency was close to that measured in the laboratory. One hydrophone was placed 5 cm directly in front of the speaker, and a second hydrophone was moved away from the first in increments of 5 cm up to 40 cm. Playbacks were recorded by both hydrophones at each distance and imported into Raven. Spectrum levels of the sounds were calculated at each hydrophone using power spectra (analysis bandwidth = 86.1 Hz). Attenuation of the 86.1 Hz bands centered at 100, 172, 344 and 603 Hz were calculated by subtracting the spectrum level at the distant hydrophone from the source hydrophone. Repeated measures ANOVA was used to compare attenuation rates between the four frequency bands, and bands that did not differ in attenuation were combined. A single attenuation rate of fish signals was then derived from the linear best fit line of the combined data.
2.5. Active area of truck noise and C. venusta signals under natural noise The distance at which C. venusta signals may emerge from the semi-trailer truck noise with a SNR of 10 at particular frequencies was estimated using the following equation:
ðs 10Þ n ¼ mðlog10 xÞ
ð1Þ
where x is the distance being solved for, s is the signal spectrum level, n is the noise spectrum level, and m is the slope of the attenuation of truck noise at that frequency. A conservative estimate assuming conditions that promote the largest active area was made using the mean plus one standard deviation of C. venusta signal spectrum level, the mean minus one standard deviation of truck noise spectrum level, and the mean plus the standard error of the attenuation rate. A liberal estimate assuming conditions that promote the smallest active area was calculated using similar logic. The active area of knocks and growls under natural ambient noise conditions at particular frequencies was estimated using the following equation:
ðs 10Þ n ¼ mðxÞ
ð2Þ
where x is the distance being solved for, s is the signal spectrum level, n is the natural ambient noise spectrum level, and m is the attenuation rate of C. venusta signals in dB/m. This was performed using the mean of each value, with conservative and liberal estimates made using similar logic to that used above. Fishes are generally unable to detect pure tones with SNRs below 15–20 dB if the noise and signal are coming from the same source, and 10–15 dB if the signal and noise are coming from the different directions (Buerkle, 1969; Chapman, 1973; Chapman and Sand, 1974; Fay, 1974; Hawkins and Sand, 1977; Mann and Lobel, 1997). We chose to assume a SNR of 10 dB, as reflected in Eq. (2).
30
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
3. Results Average source levels of semi-trailer truck noise varied little between sites (Fig. 1). Most of the truck noise occurred at low frequencies, with an average spectrum level of 116.2 ± 4.6 dB re 1 lPa (analysis bandwidth = 21.5 Hz) at 65 Hz. Truck noise spectrum levels rapidly decreased as frequency increased, reaching the upper standard deviation of natural ambient noise levels by approximately 475 Hz. Attenuation of semi-trailer truck noise showed no significant difference between the two lowest 1/3 octave bands (100 and 160 Hz, Kolmogorov–Smirnov Z64 = 0.750, P = 0.627), or the two highest 1/3 octave bands (315 and 630 Hz, Kolmogorov–Smirnov Z64 = 0.875, P = 0.428). Both low frequency bands differed significantly in attenuation rate from both high frequency bands (100 and 315 Hz, Z64 = 1.750, P = 0.004; 100 and 630 Hz, Z64 = 2.0, P = 0.001; 160 and 315 Hz, Z64 = 1.250, P = 0.088; 160 and 630 Hz, Z64 = 1.500, P = 0.022), with the low frequency bands showing a higher rate of attenuation than the high frequency bands. The best fit curve for the combination of 100 and 160 Hz 1/3 octave bands was statistically significant (ANOVA, F1 = 98.012, P < 0.001, R2 = 0.613; Fig. 2a), as was the curve for the combination of 315 and 630 Hz (ANOVA, F1 = 40.962, P < 0.001, R2 = 0.398; Fig. 2b). Equations for the best fit lines of the low frequency and high frequency composites were (y = 5.524 ln(x) 1.271) and (y = 4.215 ln(x) + 2.018), respectively. Natural ambient noise showed a peak amplitude of 79.7 ± 3.5 dB re 1 lPa (in 21.5 Hz bins) at 43 Hz (Fig. 1a). Above 43 Hz, spectrum level dropped rapidly into a quiet window occurring between 172 and 366 Hz, where the average spectrum level
Fig. 1. In plates (a) and (b), the thick green lines represent mean power spectrum of the natural soundscape. The thin green lines represent one standard deviation above and below the mean natural soundscape. The thick red lines represent the mean truck noise 1 m from the bridge. The thin red lines represent one standard deviation above and below the mean truck noise 1 m from the bridge. An analysis bandwidth of 21.5 Hz and 86.1 Hz was used in plate (a) and (b), respectively. The broken black line in plate (a) shows average power spectrum of bursts, and the broken black line in plate (b) shows average power spectrum of knocks under quiet conditions in the laboratory. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 2. Attenuation of semi-trailer truck noise between two hydrophones. The onethird octave bands centered at 100 and 160 Hz are shown in plate (a), and the onethird octave bands centered at 315 and 630 Hz are shown in plate (b). Because there was no significant difference in the attenuation rate of the 100 and 160 Hz bands, the linear best fit line in plate (a) represents a combination of the two bands. Likewise, the linear best fit line in plate (b) represents a combination of the 315 and 630 Hz bands.
was 56.9 ± 1.2 dB re 1 lPa (in 21.5 Hz bands). Above this quiet window, power levels increased to a second, broad peak from 581 to 1140 Hz, where the average power was 66.5 ± 0.9 dB (in 21.5 Hz bands). Significant concordance in spectrum shape of naturally occurring ambient noise between 21.5 and 1999.5 Hz was found among the active nesting sites at Moffits Mill (W = 0.428, chi-square = 232.64, P < 0.001, df = 8). Repeated measures ANOVA with a Greenhouse-Geisser correction showed that the attenuation rates of knocks played back through the speaker did not differ between frequencies (F1.5, 34.5 = 53.938, P = 0.212), although a greater degree of variation was apparent at the lowest frequency band. A linear best fit line plotted through the average attenuation of all frequencies for knocks was statistically significant (ANOVA F1 = 241.356, P < 0.001, R2 = 0.720), and showed an attenuation rate of 64.2 dB/m, with an equation of y = 0.642x 3.998 (Fig. 3). In comparison, losses from cylindrical spreading are typically 3 dB/distance doubled. Cylindrical losses over the same distance would result in a transmission loss of 12 dB, which equates to 16 dB/m. Sounds produced by C. venusta are broadband, with the greatest power at low frequencies. The average spectrum level of the lowband frequency of growls (107.5 Hz) was 84.1 ± 7.9 dB re 1 lPa (analysis bandwidth = 21.5 Hz), and the high-band (301 Hz) was 80.7 ± 8.1 dB re 1 lPa (analysis bandwidth = 21.5 Hz; Fig. 1a). The low-band frequency of knocks (172.2 Hz) had an average spectrum level of 116.2 ± 5.1 dB re 1 lPa (analysis bandwidth = 86.1 Hz) and the high-band (602.7 Hz) had an average spectrum level of 116.5 ± 8.2 dB re 1 lPa (analysis bandwidth = 86.1 Hz; Fig. 1b). The average SNR of growl vocalizations to the natural ambient stream noise was 20.4 dB at 107.5 Hz, and 23.9 dB at 301 Hz, while
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
Fig. 3. Attenuation of speaker playbacks of C. venusta signals. Although attenuations from speaker playback were only calculated from knock sound types, the frequency bands containing the fundamental frequencies of knocks (172 Hz: open squares, and 603 Hz: symbols) and growls (86 Hz: solid diamonds, and 344 Hz: open circles) are provided along with a linear best fit line for all data combined.
the average SNR of knock vocalizations to the natural ambient noise was 43 dB at 172.2 Hz, and 51.1 dB at 602.7 Hz. At 172.2 and 602.7 Hz, the spectrum level of natural noise was 73.2 ± 3.4 and 65.4 ± 10.5 dB re 1 lPa (analysis bandwi dth = 86.1 Hz), respectively. C. venusta knocks should therefore propagate between 38.2 and 64.6 cm (mean 51.4 cm) at 172.2 Hz and between 34.9 and 93.1 cm (mean 64.0 cm) at 602.7 Hz before dropping below a SNR of 10.0 in natural conditions. At 107.5 and 301 Hz, the natural noise was 63.7 ± 4.7 and 56.8 ± 5.6 dB re 1 lPa (analysis bandwidth = 21.5 Hz), respectively. Growls should propagate 0–35.8 cm (mean 16.2 cm) at 107.5 Hz and 0.3–43.0 cm (mean = 21.7 cm) at 301 Hz before dropping below a SNR of 10 dB in natural conditions. The spectrum level of semi-trailer truck noise was 112.3 ± 4.2 dB re 1 lPa (analysis bandwidth = 86.1 Hz) at 172.2 Hz, and 76.2 ± 4.3 dB re 1 lPa (analysis bandwidth = 8 6.1 Hz) at 602.7 Hz. Knocks should therefore emerge from truck noise with a SNR above 10 at distances between 0.6 and 22.3 m (mean = 3.0 m) from the bridge at 172.2 Hz and should already have a SNR above 10 at 1 m from the bridge at 602.7 Hz. Semi-trailer truck noise was 109.8 ± 3.1 dB re 1 lPa (analysis bandwidth = 21.5 Hz) at 107.5 Hz, and 86.2 ± 2.6 dB re 1 lPa (analysis bandwidth = 21.5 Hz) at 301 Hz. Growls should therefore emerge from semi-trailer truck noise with a SNR above 10 dB at distances between 58.0 m and 12,113 m (mean = 640.0 m) from the bridge at 107.5 Hz, and between 2.7 m and 1584.2 m (mean = 39.5 m) at 301 Hz. 4. Discussion Our results suggest that road traffic noise masks the acoustic signal of a representative stream fish, C. venusta, and recovery to the natural soundscape may be as far as 12,113 m, depending on the frequency. Semi-trailer truck noise was shown to diminish the normal active area of both growls and knocks to some degree, depending on the frequency of concern and the distance from the noise source. Growls showed the potential to be more severely affected than knocks, due to their lower amplitude and greater spectral overlap with anthropogenic noise. It is difficult, however, to assess the degree to which anthropogenic noise sources affect communication and overall reproductive success of C. venusta, although this species was shown to exhibit the Lombard Effect (producing higher amplitude signals) in response to the traffic noise recorded in this study (Holt and Johnston, 2014b). It is not known whether there is a difference in the information content of the low-band and high-band frequencies, which are affected
31
differently by truck noise. Several studies have shown correlations between signal frequency and male size, which can be considered an indicator of male fitness. In the case of C. venusta, the 301 Hz band of growls is negatively correlated with fish standard length (Holt and Johnston, 2014a). Disturbance of the quiet window in the natural soundscape caused by anthropogenic noise may interfere with the ability of females to attain important information about males from acoustic signals. Other signal parameters such as call duration have also been found to be correlated with male condition (Amorim et al., 2010). If call duration carries such information in other freshwater fishes, a smaller active area induced by anthropogenic noise events could cause growls to be heard for a shorter period of time by a female, which could diminish perceived male fitness and have adverse effects on reproductive success. Interfering with call detection could especially be true of many benthic species of fishes that produce lengthy tonal calls. Call duration has been shown to be related to male quality, and preferred by females in the Lusitanian toadfish (H. didactylus). Numerous darters in the subgenus Catonotus (Johnston and Johnson, 2000; Speares et al., 2011), freshwater gobies such as Padogobius martensii (Lugli and Fine, 2007), and the banded sculpin (Cottus carolinae; personal observation) are examples of benthic, freshwater species who produce tonal sounds during reproductive or aggressive encounters. Masking of even part of these sounds could disrupt the communicative function of these calls. Despite the consistent presence of natural noise at C. venusta nesting sites, the presence of a quiet window in the natural noise spectrum helps increase SNR of vocalizations. Knocks show spectrum levels that are much higher than the natural ambient noise, are always produced at a close distance to the receiver, and are therefore not likely dependent on the integrity of the natural quiet window. Growl are much quieter with SNR’s less than half that of knocks, and therefore rely much more on the quiet window for maintenance of their active area. Although we did not investigate it, noise from bridge crossings should definitely show temporal variation at both short term and long term scales. We have observed peak spawning times for Cyprinella venusta in the morning before water temperatures begin to rise. An overlap between peak morning traffic and spawning times could exacerbate any communication problems caused by anthropogenic noise. The close association between C. venusta vocalizations and the quiet window within the natural ambient noise is not unique to C. venusta. Reports of quiet windows in the background noise of freshwater streams have been made by several authors (Wysocki et al., 2007; Speares et al., 2011; Lugli and Fine, 2003; Amoser, 2006; Crawford et al., 1997), and Lugli (2010) reports the close association of vocalization frequencies with quiet windows in the habitat for several goby species inhabiting freshwater, brackish, and marine environments. Because hydrology largely dictates natural environmental noise, and fish are often bound to certain habitat types due to specific life history strategies, the utilization of a naturally occurring quiet window in the ambient noise is probably an adaptation for intraspecific communication. However, because physical conditions largely dictate both the natural underwater soundscapes and preferred nesting habitats, the nature of the quiet window and the effect of anthropogenic noise on it is more likely to be conserved across a larger geographic areas. Detection distances of the different call types do not appear to be arbitrary and seem to fit appropriately with the behaviors they accompany. This correlates well with the types of behaviors that are typically associated with knocks. Knocks are usually produced by males during aggressive encounters with other males around the nesting site, with males typically less than 15 cm apart when vocalizing during a conflict. We have also observed knock production during chases between nest guarding males and sunfish that
32
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
would attempt to eat eggs from the nest. The ability of C. venusta to theoretically detect knocks at over half of a meter away suggests that sounds produced by a male may be detectable by males that are nearby, but not actively engaged in an aggressive interaction with the vocalizing male under normal ambient noise conditions. If knocks serve as an honest signal of male fitness to other males, a larger active area of the sound should benefit the sound producer. Because there are often many other reproductive males at a given nesting site, a territorial male should benefit from sending honest signals about their intentions and motivational level to both the intruding male and peripheral males simultaneously. Although results from the speaker playback showed no difference in attenuation rates of different frequency bands of fish calls, the greater variation in attenuation seen at the lowest frequency band may have been due it being at the lower end of the speaker’s frequency response. Attenuation rates of truck noise at frequencies below the cutoff frequency (where the energy of the sound begins to be reduced) of the streams were much lower than fish sounds because truck noise can be transmitted into the stream through the air, as well as through the substrate. Propagation of sounds that originate within the water column are subjected to the limitations of the cutoff frequency, a phenomenon by which acoustic signals below a certain frequency dictated in large part by the water depth attenuate extremely rapidly (Rogers and Cox, 1988). We are unsure of the contribution of noise transmitted from the air versus the substrate, but it is feasible for sounds to propagate down the bridge pilings into the substrate where it could re-emerge into the water column. Because the composition of the bottom below the immediate substrate (sand and gravel) is unknown, we cannot address the mode of propagation at this point. The findings of the current study offer a new perspective with which to view reproductive behaviors in fishes inhabiting small, freshwater systems, providing a description of a previously undocumented and potentially widespread source of anthropogenic noise in these habitats. The noisy environment in which C. venusta spawn has a convenient window in the noise spectrum, which is exploited by C. venusta for the purpose of communication with females during reproductive behaviors. Our results show that this quiet window is disrupted by road traffic noise. This discovery should be followed up by studies investigating the behavioral, and stress responses of C. venusta, as well as other more imperiled fishes, to noise from bridge crossings. Future efforts should also be made to characterize noise from different bridge styles, daily temporal patterns of noise, and attenuation at longer distances.
Acknowledgements We would like to thank D. Higgs, G. Hill, D. Mann, M. Mendonça, and P. Noel for valuable input and comments concerning the study.
Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biocon.2015.04. 004. These data include Google maps of the most important areas described in this article.
References Amorim, M.C.P., Simões, J.M., Mendonça, N., Bandarra, N.M., Almada, V.C., Fonseca, P.J., 2010. Lusitanian toadfish song reflects male quality. J. Acoust. Soc. Am. 213, 2997–3004. Amoser, S., 2006. Year-round changes of ambient noise in freshwater habitats and temperature effects on hearing in fishes. J. Acoust. Soc. Am. 120, 3056.
Amoser, S., Wysocki, L.E., Ladich, F., 2004. Noise emission during the first powerboat race in an alpine lake and potential impact on fish communities. J. Acoust. Soc. Am. 116, 3789–3797. Banner, A., Hyatt, M., 1973. Effects of noise on eggs and larvae of two estuarine Fishes. Trans. Am. Fish Soc. 1, 134–136. Bennet-Clark, H.C., 1999. Which Qs to choose: questions of quality in bioacoustics. Bioacoustics 9, 351–359. Buerkle, U., 1969. Auditory masking and the critical band in Atlantic Cod (Gadus morhua). J. Fish. Res. Board Can. 26, 1113–1119. Chapman, C.J., 1973. Field studies of hearing in teleost fish. Helgoländer Meeresun. 24, 371–390. Chapman, C.J., Sand, O., 1974. Field studies of hearing ink two species of flatfish Pleuronectes platessa (L) and Limanda limanda (L) (Family Pleuronectidae). Comp. Biochem. Physiol. – A 47, 371–385. Codarin, A., Wysocki, L.E., Ladich, F., Picciulin, M., 2009. Effects of ambient and boat noise on hearing and communication in three fish species living in a marine protected area (Miramare, Italy). Mar. Pollut. Bull. 58, 1880–1887. Crawford, J.D., Jacob, P., Bénech, V., 1997. Sound production and reproductive ecology of strongly acoustic fish in Africa: Pollimyrus isidori, Mormyridae. Behaviour 134, 677–725. Fay, R.R., 1974. Masking of tones by noise for the goldfish (Carassius auratus). J. Comp. Physiol. Psychol. 87, 708–716. Fernandes, P.G., Brierley, A.S., Simmonds, E.J., Millard, N.W., McPhail, S.D., Stevenson, P., Squires, M., 2000. Fish do not avoid survey vessels. Nature 604, 35–36. Fine, M.L., Lenhardt, M.L., 1983. Shallow-water propagation of the toadfish mating call. Comp. Biochem. Physiol. 76A, 225–231. Ghahramani, Z.N., Mohajer, Y.J., Fine, M.L., 2014. Developmental variation in sound production in water and air in the blue catfish Ictalurus furcatus. J. Exp. Biol. 217, 4244–4251. Handegard, N.O., Michalsen, K., Tjostheim, D., 2003. Avoidance behavior in cod (Gadus morhua) to a bottom-trawling vessel. Aquat. Living Resour. 16, 265–270. Hastings, M.C., 2008. Coming to terms with the effects of ocean noise on marine animals. Acoust. Today 4, 22–34. Hawkins, S.D., Sand, O., 1977. Directional hearing in the median vertical plane by the cod. J. Comp. Physiol. A 122, 1–8. Holt, D.E., Johnston, C.E., 2014a. Sound production and associated behaviors in Blacktail Shiner (Cyprinella venusta): a comparison between field and lab. Environ. Biol. Fish. 97, 1207–1219. Holt, D.E., Johnston, C.E., 2014b. Evidence of the Lombard effect in fishes. Behav. Ecol. http://dx.doi.org/10.1093/beheco/aru028. Johnston, C.E., Johnson, D.L., 2000. Sound production during the spawning season in cavity-nesting darters of the subgenus Catonotus (Percidae: Etheostoma). Copeia 2000, 475–481. Jouventin, P., Aubin, T., Lengagne, T., 1999. Finding a parent in a king penguin colony: the acoustic system of individual recognition. Anim. Behav. 57, 1175– 1183. Kasumyan, A.O., 2008. Sounds and sound production in fishes. J. Ichthyol. 48, 981– 1030. Ladich, F., 2013. Effects of noise on sound detection and acoustic communication in fishes. In: Brumm, H. (Ed.), Animal Communication and Noise. Springer-Verlag, Berlin, Heidelberg, pp. 65–90. Lagardère, J.P., 1982. Effects of noise on growth and reproduction of Crangon crangon in rearing tanks. Mar. Biol. 71, 177–185. Locascio, J.V., Mann, D.A., 2011. Localization and source level estimates of black drum (Pogonias chromis) calls. J. Acoust. Soc. Am. 130, 1868–1879. Lugli, M., 2010. Sounds of shallow water fishes pitch within the quiet window of the habitat ambient noise. J. Comp. Physiol. A 196, 439–451. Lugli, M., Fine, M.L., 2003. Acoustic communication in two freshwater gobies: ambient noise and short-range propagation in shallow streams. J. Acoust. Soc. Am. 114, 512–521. Lugli, M., Fine, M.L., 2007. Stream ambient noise, spectrum and propagation of sounds in the goby Padogobius martensii: sound pressure and particle velocity. J. Acoust. Soc. Am. 122, 2881–2892. Mann, D.A., Lobel, P.S., 1997. Propagation of damselfish (Pomacentridae) courtship Sounds. J. Acoust. Soc. Am. 101, 3783–3791. Myrberg, A.A., 1990. The effects of man-made noise on the behavior of marine mammals. Environ. Int. 16, 575–586. Narins, P.M., Feng, A.S., Lin, W., Schnitzler, H.U., Denzinger, A., Suthers, R.A., Xu, C., 2004. Old world frog and bird vocalizations contain prominent ultrasonic harmonics. J. Acoust. Soc. Am. 115, 910–913. Popper, A.N., 2003. Effects of anthropogenic sounds on fishes. Fish. Res. 28, 24–31. Popper, A.N., Fay, R.R., 2010. Rethinking sound detection by fishes. Hear. Res. 273, 25–36. Radford, A.N., Kerridge, E., Simpson, S.D., 2014. Acoustic communication in a noisy world: can fish compete with anthropogenic noise? Behav. Ecol. 25, 1022–1030. Rogers, P.H., Cox, M., 1988. Underwater sound as a biological stimulus. In: Atema, J., Fay, R.R., Popper, A.N., Tavolga, W.N. (Eds.), Sensory Biology of Aquatic Animals. Springer, New York, pp. 131–149. Southall, B.L., Bowlesn, A.E., Ellison, W.T., Finneran, J.J., Gentry, R.L., Green Jr., C.R., Kastak, D., Ketten, D.R., Miller, J.H., Nachtigall, P.E., Richardson, W.J., Thomas, J.A., Tyack, P.L., 2007. Marine mammal noise exposure criteria: Initial scientific recommendations. Aquat. Mammals 33, 411–521. Speares, P.D., Holt, D.E., Johnston, C.E., 2011. The relationship between ambient noise and fundamental frequency of vocalizations in two species of darters (Percidae: Etheostoma). Environ. Biol. Fish. 90, 103–110.
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33 Vabø, R., Olsen, K., Huse, I., 2002. The effect of vessel avoidance of wintering Norwegian spring spawning herring. Fish. Res. 58, 59–77. Vasconcelos, R.O., Amorim, C.P., Ladich, F., 2007. Effects of ship noise on the detectability of communication signals in the Lusitanian toadfish. J. Exp. Biol. 210, 2104–2112. Voellmy, I.K., Purser, J., Flynn, D., Kennedy, P., Simpson, S.D., Radford, A.N., 2014a. Acoustic noise reduces foraging success in two sympatric fish species via different mechanisms. Anim. Behav. 89, 191–198. Voellmy, I.K., Purser, J., Simpson, S.D., Radford, A.N., 2014b. Increased noise levels have different impacts on the anti-predator behaviour of two sympatric fish species. PLoS ONE 9, e102946. Waser, P.M., Waser, M.S., 1977. Experimental studies of primate vocalization: specializations for long distance propagation. Z. Tierpsychol. (J. Comp. Ethol.) 43, 239–263.
33
Wiley, R.H., Richards, D.G., 1982. Adaptations for acoustic communication in birds: sound transmission and signal detection. In: Kroodsma, D.E., Miller, E.H., Quellet, H. (Eds.), Acoustic Communication in Birds, vol. I. Academic Press, New York, pp. 131–278. Wysocki, L.E., Ladich, F., 2005. Effects of noise exposure on click detection and the temporal resolution ability of the goldfish auditory system. Hear. Res. 201, 27– 36. Wysocki, L.E., Amoser, S., Ladich, F., 2007. Diversity in ambient noise in European freshwater habitats: noise levels, spectral profiles, and impact on fishes. J. Acoust. Soc. Am. 121, 2559–2566.
Contents lists available at ScienceDirect
Biological Conservation journal homepage: www.elsevier.com/locate/biocon
Traffic noise masks acoustic signals of freshwater stream fish Daniel E. Holt, Carol E. Johnston ⇑ Fish Biodiversity Lab, School of Fisheries, Aquaculture and Aquatic Sciences, Auburn University, Auburn, AL 36849, United States
a r t i c l e
i n f o
Article history: Received 20 January 2015 Received in revised form 30 March 2015 Accepted 5 April 2015 Available online 22 April 2015 Keywords: Blacktail Shiner Noise pollution Bioacoustics Acoustic propagation
a b s t r a c t In order for an acoustic signal to be an effective source of communication, the signal must be successfully detected and interpreted by the intended receiver. One potential barrier to acoustic communication is background noise. Lotic systems contain a wide variety of habitats including riffles, shoals and waterfalls that can become quite noisy. The increasing prevalence of road and train crossings over small streams, and increased boat traffic in navigable rivers and lentic systems also presents potential anthropogenic noise sources with which vocal fishes did not evolve. The present study investigates the relationship between vocalizations and the natural soundscape of a common fish of the Southeastern United States, the Blacktail Shiner (Cyprinella venusta), and the potential effects anthropogenic noise from bridge crossings may have on the soundscape and acoustic communication in this species. Results revealed a particularly close association of a quiet window in the natural soundscape of C. venusta and dominant frequencies of the courtship vocalization of C. venusta. Results also indicated that C. venusta’s acoustic signals propagate short distances, following predictions based on the calculated cutoff frequency of the streams they inhabit, and were masked by noise generated from bridge crossings. Our calculations suggest that road traffic noise propagates to an extent that virtually entire watersheds are impacted by this noise pollution, especially in urban areas. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Because sound has the capacity to carry information, can be used intermittently, and does not require a line of sight, it is used extensively by animals as a mode of information transfer. Sound production has been documented in over 800 species of fishes representing 109 families within the infraclass Teleostei (Kasumyan, 2008). Despite the fact that a large number of fishes likely utilize sound for communication, numerous factors concerning the properties of the projected sound, hearing abilities of the receiver, constraints imposed by the physical environment, and ambient noise levels of the environment must fit together properly in order for acoustic communication to be effective. Ambient noise from both biotic and abiotic (wind, rainfall, turbulence) sources can decrease the signal-to-noise ratio of signals, or make temporal information more difficult to extract (Wysocki and Ladich, 2005). Studies on terrestrial species have shown that these environmental noise sources can act as strong selective pressures in the evolution of signal structure (Waser and Waser, 1977; Wiley and Richards, 1982; Jouventin et al., 1999; Narins et al., 2004). Unlike natural biotic and abiotic noise sources, the relatively recent development and rapid expansion of human activities such ⇑ Corresponding author. Tel.: +1 334 844 1781; fax: +1 334 844 9207. E-mail address: [email protected] (C.E. Johnston). http://dx.doi.org/10.1016/j.biocon.2015.04.004 0006-3207/Ó 2015 Elsevier Ltd. All rights reserved.
as urbanization, shipping, motorized recreational activities, drilling, and seismic explorations (Myrberg, 1990; Popper, 2003) do not provide the time necessary for the evolution of acoustic signals. Efforts have been made to determine the effect of anthropogenic noise on marine mammals (Southall et al., 2007; Hastings, 2008), and primarily marine fishes (Codarin et al., 2009; Ladich, 2013; Radford et al., 2014; Voellmy et al., 2014a,b). Elevated noise levels have been shown to reduce egg survival, reproduction and growth rates in fishes (Banner and Hyatt, 1973) and shrimp (Lagardère, 1982). Studies have also shown that anthropogenic noises can affect fish hearing or behaviors (Fernandes et al., 2000; Vabø et al., 2002; Handegard et al., 2003), which can potentially have detrimental effects on fitness. Amoser et al. (2004) found, for example, that noise from powerboats racing on an alpine lake was loud enough to be detected by otophysine fishes (fishes possessing a hearing specialization; see Popper and Fay, 2010) at up to 400 m away. Vasconcelos et al. (2007) found that the noise from ferry boats in the Tagus River estuary (Portugal) caused significant hearing threshold shifts in the Lusitanian toadfish (Halobatrachus didactylus), and that females ability to detect male signals would be significantly diminished under ship noise. Despite efforts that have been made in other habitats, we are currently unaware of any study that has looked at potential anthropogenic noise sources in small freshwater streams, and how these noise sources may impact the ability of small, vocal
28
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
fishes to communicate acoustically. While several studies have examined signal propagation with regard to the ambient environment in shallow water systems, these studies have not included the effects of anthropogenic noise (Fine and Lenhardt, 1983; Ghahramani et al., 2014; Locascio and Mann, 2011; Lugli and Fine, 2007). Because small freshwater systems are home to a disproportionately large percentage of the imperiled fishes of the southeastern United States, an initiative must be taken to better characterize anthropogenic noise transmitted into these systems. The current study was aimed at describing the interaction between the natural soundscape and acoustic repertoire of the Blacktail Shiner (Cyprinella venusta) in east Alabama. We also examined the effect that bridge traffic, one particular anthropogenic noise source common in low order streams, may have on the natural soundscape and the ability of C. venusta to communicate acoustically. The results provide a better understanding of how a common source of anthropogenic noise may affect the acoustic soundscape of small freshwater streams and rivers. 2. Materials and methods 2.1. Natural ambient noise measurements All natural ambient noise measurements were made on 12 May 2011 between 1300 and 1600 h. A large shoal on Little Uchee Creek (Lee Co., AL, U.S.A., 32° N, 85° W), which is a tributary of the Chattahoochee River, was chosen to describe the natural soundscape of C. venusta. Water temperature was 27.2 °C. This location was chosen because it offered a wide variety of suitable spawning habitats for C. venusta and during the reproductive summer months, the water is periodically shallow and clear enough to allow a researcher to locate the exact locations of nests by watching the fish behave from the bank. Nine active nest sites were identified by observation from exposed bedrock using polarized sunglasses. Nest sites were typically found at the confluence of an area of high flow and a pool. However, spawning aggregations were also observed directly within rapidly flowing chutes. A hydrophone (Hi-Tech HTI-96-MIN, sensitivity 164.4 re 1 V/lPa, frequency response: 0.002–30 kHz) and digital recorder (Marantz PMD 661, sampling rate 44.1 kHz) were used to record 1 min of ambient noise in each of the 9 sites. In sites with substantial flow, an effort was made to place the hydrophone in a low flow area adjacent to the flow to reduce hydrodynamic noises. Sounds were imported into Raven 1.4 (Cornell University, Ithaca, NY), where three, 1 s segments were randomly selected from the recording made at each site. Two power spectra of each segment were then calculated using the power spectrum function of Raven (Hamming window, 50% time overlap, FFT length: 2048 samples, analysis bandwidth: 21.5 Hz; FFT length: 512 samples, analysis bandwidth: 86.1 Hz). Two separate power spectra, each with a unique analysis bandwidth had to be produced for natural ambient noise and anthropogenic noise sources so that SNR’s could be calculated between C. venusta acoustic signals and noise. When calculating SNR’s, it is necessary to analyze all sounds to be included in the analysis at the same frequency resolution. The analysis bandwidths of 21.5 and 86.1 Hz result from the typical duration of growls and knocks, respectively. For each analysis bandwidth, the three power spectra from each nest site were exported into Microsoft Excel where they were averaged to produce a single power spectrum for each of the 10 nest sites. Kendall’s concordance test was used to determine whether spectrum shape (using the spectrum curve with analysis bandwidth of 21.5 Hz) of natural ambient noise between 21.5 and 1999.5 Hz was significantly different across active nesting sites (Lugli and Fine, 2003). No difference was observed in the shape of curves from different nesting sites, and so power spectra from all sites were
averaged to generate a single, composite power spectrum for natural ambient noise. Spectrum levels were calibrated to represent absolute levels using the sensitivity of the hydrophone and a GW GOS-6xxG dual trace Oscilloscope, and by taking into consideration the gain applied to the signal by the Marantz and when importing sounds into Raven. The frequency range of major spectral components (such as bandwidth of the quiet window found in the natural field recordings) of natural and anthropogenic sounds were defined as the range of frequencies within 3 dB of the peak frequency of the average power spectrum. This is standard for determining general tuning properties of sounds (Bennet-Clark, 1999), and has been used in a similar context by Lugli (2010). 2.2. Anthropogenic noise source measurements and propagation Source levels and propagation of semi-trailer trucks crossing streams were measured at 6 road crossings, all located within Lee County and Macon County, Alabama (Table 1). Recordings were made between 3 and 14 March 2010. Water temperatures were not recorded. All crossings were beam bridges supported by a piling at the junction of each bridge segment. For all recordings, the hydrophone was placed approximately 8 cm off the substrate. This depth was chosen because the substrate was mostly sand and gravel, and potential nests in this type of habitat are usually close to the substrate. The hydrophone was mounted to the end of a 17.7 cm PVC pipe, which was secured between two submerged sandbags and positioned in such a way that the sandbags were downstream of the hydrophone. Water depth and flow velocity were not recorded, however water depth never exceeded approximately 84 cm (the maximum depth at which the hydrophone apparatus could be set up with without the researcher having to submerge their head). At each road crossing, one hydrophone was fixed 1 m upstream or downstream of the bridge’s edge (direction was determined by accessibility), while a second hydrophone was moved different distances away from the first hydrophone in the direction opposite the bridge. Both hydrophones recorded simultaneously onto separate channels of the Marantz digital recorder. Prior to going into the field, gain on each channel of the Marantz was made equal by recording a tone of known amplitude with the same hydrophone successively on each channel, and adjusting gains on each channel until the level was the same on both channels. Distances separating the hydrophones varied between 2 and 16 m for each site (Table 1), and depended on the locations of suitable (not flowing rapidly) and accessible (shallow enough to set hydrophone up on substrate) habitat. Because of hydrophone cable length constraints, recordings with both hydrophones simultaneously were not possible beyond 16 m. However, at three of the six sites, recordings were made with a single hydrophone at several distances up to 82 m. Recording was performed for several minutes at each distance, and the time at which semi-trailer trucks passed was noted. Care was taken to ensure that at least 3 trucks passed over, and that at least one 10 s period with no traffic occurred during the recording period at each distance. Table 1 Recording sites and hydrophone distances from the source of anthropogenic noise. Location
Total bridge length (m)
County
Hydrophone separation (m)
I-85 at Choctafaula Cr. I-85 at Hodnett Cr. I-85 at tributary to Choctafaula Cr. I-85 at Uphapee Cr.
62.5 92.5 72.5
Lee Macon Macon
2, 4, 12, 16 2, 4, 13, 40, 80 2, 4, 6, 12, 16, 36
200.1
Macon
I-85 at Cubahatchee Cr. C.R. 40 at Calebee Cr.
129.2 152.8
Macon Macon
2, 4, 6, 9, 12, 14, 23, 49, 82 2, 4, 6, 12, 16 4, 7.5, 15
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
To calculate source levels of semi-trailer truck noise, a total of three truck passes were selected from the recordings made by the hydrophone 1 m away from the bridge at each site. As before when measuring natural environmental noise, two power spectra with different analysis bandwidths were generated at each distance. Power spectra from the three truck passes at each site were exported to Excel, and were averaged to generate a final power spectra at each analysis bandwidth representing semi-trailer truck noise 1 m downstream of bridges. To calculate attenuation of semi-trailer truck noise, three truck passes at each distance from the bridge, at each site, were selected from the audio files of both hydrophones. Power spectra were calculated for each of the three passes using Raven (FFT length: 32,768 samples, analysis bandwidth = 1.35 Hz), and exported to Excel. The average power in 1/3 octave bands was then calculated for bands with center frequencies at 100 and 315 Hz (representing the lower and upper frequency bands of C. venusta growls) and 160 and 630 Hz (representing the lower and upper frequency bands of C. venusta knocks). Attenuation in each of the 1/3 octave bands was calculated by subtracting the sound pressure level between the two hydrophones, and averaging this attenuation across the three semi-trailer truck passes recorded at each distance. Kolmogorov– Smirnov tests were used to determine whether attenuation rates differed between the 1/3 octave bands, and bands that did not differ were combined. Attenuation rates of combined frequency bands were plotted as an X–Y scatter, and the best fitting curve for the attenuation rates was calculated. A logarithmic curve estimation procedure was used to generate the best fit lines, and ANOVA was used to test the acceptability of each model. 2.3. Measurement of C. venusta signal spectrum level C. venusta knock and growl spectrum levels were measured in the laboratory. A detailed description of C. venusta acoustic signals and behaviors, along with locality information for where fish were captured and how they were stored is provided by Holt and Johnston (2014a), and a description of how spectrum levels were determined under quiet conditions is provided by Holt and Johnston (2014b). An artificial nest was set up within a 76 L aquarium, and a hydrophone was placed in front of the nest. Two males and one to two females were placed in the tank at a time, and allowed to acclimate for between 20 min and 9 h, depending on the time taken for normal behaviors and nest guarding behavior to resume. Vocalizations produced during approach and lateral display behaviors were recorded onto Raven using a Brüel and Kjaer 8103 hydrophone, and a Brüel and Kjaer 2635 charge amplifier. Growls were collected from a total of 19 different males, and knocks were collected from a total of 14 different males. Only signals recorded within 15 cm of the hydrophone (to minimize distortion caused by interaction with the 3726 Hz resonant frequency of the aquarium), and that were less than 11.6 ms (for knocks) and 46.4 ms (for growls) in duration were used for analysis. Analyzed signals were limited to this duration so that power spectra with a uniform analysis bandwidth could be generated using Raven. Power spectra were exported to Excel where an average knock power spectrum, and growl power spectrum were generated. 2.4. Propagation of C. venusta acoustic signals Attenuation of fish signals was measured at three different spots adjacent to one of the natural nesting sites by playing the sounds back through an underwater speaker (University Sound UW-30, Oklahoma City) and recording with the Marantz digital recorder at different distances. The playbacks were conducted on 7 June 2011 between 1345 and 1416 h, and on 20 June 2011 between 1036 and 1220 h. Water temperature was 25 °C for the
29
recordings made on 7 June 2011, and 28.3 °C for the recordings made on 20 June 2011. The speaker was positioned with the center of its cone 20 cm above the substrate at all three locations, and 16.4 cm below the water’s surface at the first location, 11.2 cm below the water’s surface at the second location, and 18.1 cm below the water’s surface at the third location. All three locations were within a 10 m radius of the nesting site, and the speaker was always arranged in such a way that it projected sound parallel to the water’s surface with no structures between the front of the speaker and the hydrophones. A recording of a knock vocalization was played from a Blackberry curve 8520, and amplified by a Pioneer GM-X372 240W amplifier. The playback was calibrated so that the spectrum level of the sound’s dominant frequency was close to that measured in the laboratory. One hydrophone was placed 5 cm directly in front of the speaker, and a second hydrophone was moved away from the first in increments of 5 cm up to 40 cm. Playbacks were recorded by both hydrophones at each distance and imported into Raven. Spectrum levels of the sounds were calculated at each hydrophone using power spectra (analysis bandwidth = 86.1 Hz). Attenuation of the 86.1 Hz bands centered at 100, 172, 344 and 603 Hz were calculated by subtracting the spectrum level at the distant hydrophone from the source hydrophone. Repeated measures ANOVA was used to compare attenuation rates between the four frequency bands, and bands that did not differ in attenuation were combined. A single attenuation rate of fish signals was then derived from the linear best fit line of the combined data.
2.5. Active area of truck noise and C. venusta signals under natural noise The distance at which C. venusta signals may emerge from the semi-trailer truck noise with a SNR of 10 at particular frequencies was estimated using the following equation:
ðs 10Þ n ¼ mðlog10 xÞ
ð1Þ
where x is the distance being solved for, s is the signal spectrum level, n is the noise spectrum level, and m is the slope of the attenuation of truck noise at that frequency. A conservative estimate assuming conditions that promote the largest active area was made using the mean plus one standard deviation of C. venusta signal spectrum level, the mean minus one standard deviation of truck noise spectrum level, and the mean plus the standard error of the attenuation rate. A liberal estimate assuming conditions that promote the smallest active area was calculated using similar logic. The active area of knocks and growls under natural ambient noise conditions at particular frequencies was estimated using the following equation:
ðs 10Þ n ¼ mðxÞ
ð2Þ
where x is the distance being solved for, s is the signal spectrum level, n is the natural ambient noise spectrum level, and m is the attenuation rate of C. venusta signals in dB/m. This was performed using the mean of each value, with conservative and liberal estimates made using similar logic to that used above. Fishes are generally unable to detect pure tones with SNRs below 15–20 dB if the noise and signal are coming from the same source, and 10–15 dB if the signal and noise are coming from the different directions (Buerkle, 1969; Chapman, 1973; Chapman and Sand, 1974; Fay, 1974; Hawkins and Sand, 1977; Mann and Lobel, 1997). We chose to assume a SNR of 10 dB, as reflected in Eq. (2).
30
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
3. Results Average source levels of semi-trailer truck noise varied little between sites (Fig. 1). Most of the truck noise occurred at low frequencies, with an average spectrum level of 116.2 ± 4.6 dB re 1 lPa (analysis bandwidth = 21.5 Hz) at 65 Hz. Truck noise spectrum levels rapidly decreased as frequency increased, reaching the upper standard deviation of natural ambient noise levels by approximately 475 Hz. Attenuation of semi-trailer truck noise showed no significant difference between the two lowest 1/3 octave bands (100 and 160 Hz, Kolmogorov–Smirnov Z64 = 0.750, P = 0.627), or the two highest 1/3 octave bands (315 and 630 Hz, Kolmogorov–Smirnov Z64 = 0.875, P = 0.428). Both low frequency bands differed significantly in attenuation rate from both high frequency bands (100 and 315 Hz, Z64 = 1.750, P = 0.004; 100 and 630 Hz, Z64 = 2.0, P = 0.001; 160 and 315 Hz, Z64 = 1.250, P = 0.088; 160 and 630 Hz, Z64 = 1.500, P = 0.022), with the low frequency bands showing a higher rate of attenuation than the high frequency bands. The best fit curve for the combination of 100 and 160 Hz 1/3 octave bands was statistically significant (ANOVA, F1 = 98.012, P < 0.001, R2 = 0.613; Fig. 2a), as was the curve for the combination of 315 and 630 Hz (ANOVA, F1 = 40.962, P < 0.001, R2 = 0.398; Fig. 2b). Equations for the best fit lines of the low frequency and high frequency composites were (y = 5.524 ln(x) 1.271) and (y = 4.215 ln(x) + 2.018), respectively. Natural ambient noise showed a peak amplitude of 79.7 ± 3.5 dB re 1 lPa (in 21.5 Hz bins) at 43 Hz (Fig. 1a). Above 43 Hz, spectrum level dropped rapidly into a quiet window occurring between 172 and 366 Hz, where the average spectrum level
Fig. 1. In plates (a) and (b), the thick green lines represent mean power spectrum of the natural soundscape. The thin green lines represent one standard deviation above and below the mean natural soundscape. The thick red lines represent the mean truck noise 1 m from the bridge. The thin red lines represent one standard deviation above and below the mean truck noise 1 m from the bridge. An analysis bandwidth of 21.5 Hz and 86.1 Hz was used in plate (a) and (b), respectively. The broken black line in plate (a) shows average power spectrum of bursts, and the broken black line in plate (b) shows average power spectrum of knocks under quiet conditions in the laboratory. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 2. Attenuation of semi-trailer truck noise between two hydrophones. The onethird octave bands centered at 100 and 160 Hz are shown in plate (a), and the onethird octave bands centered at 315 and 630 Hz are shown in plate (b). Because there was no significant difference in the attenuation rate of the 100 and 160 Hz bands, the linear best fit line in plate (a) represents a combination of the two bands. Likewise, the linear best fit line in plate (b) represents a combination of the 315 and 630 Hz bands.
was 56.9 ± 1.2 dB re 1 lPa (in 21.5 Hz bands). Above this quiet window, power levels increased to a second, broad peak from 581 to 1140 Hz, where the average power was 66.5 ± 0.9 dB (in 21.5 Hz bands). Significant concordance in spectrum shape of naturally occurring ambient noise between 21.5 and 1999.5 Hz was found among the active nesting sites at Moffits Mill (W = 0.428, chi-square = 232.64, P < 0.001, df = 8). Repeated measures ANOVA with a Greenhouse-Geisser correction showed that the attenuation rates of knocks played back through the speaker did not differ between frequencies (F1.5, 34.5 = 53.938, P = 0.212), although a greater degree of variation was apparent at the lowest frequency band. A linear best fit line plotted through the average attenuation of all frequencies for knocks was statistically significant (ANOVA F1 = 241.356, P < 0.001, R2 = 0.720), and showed an attenuation rate of 64.2 dB/m, with an equation of y = 0.642x 3.998 (Fig. 3). In comparison, losses from cylindrical spreading are typically 3 dB/distance doubled. Cylindrical losses over the same distance would result in a transmission loss of 12 dB, which equates to 16 dB/m. Sounds produced by C. venusta are broadband, with the greatest power at low frequencies. The average spectrum level of the lowband frequency of growls (107.5 Hz) was 84.1 ± 7.9 dB re 1 lPa (analysis bandwidth = 21.5 Hz), and the high-band (301 Hz) was 80.7 ± 8.1 dB re 1 lPa (analysis bandwidth = 21.5 Hz; Fig. 1a). The low-band frequency of knocks (172.2 Hz) had an average spectrum level of 116.2 ± 5.1 dB re 1 lPa (analysis bandwidth = 86.1 Hz) and the high-band (602.7 Hz) had an average spectrum level of 116.5 ± 8.2 dB re 1 lPa (analysis bandwidth = 86.1 Hz; Fig. 1b). The average SNR of growl vocalizations to the natural ambient stream noise was 20.4 dB at 107.5 Hz, and 23.9 dB at 301 Hz, while
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
Fig. 3. Attenuation of speaker playbacks of C. venusta signals. Although attenuations from speaker playback were only calculated from knock sound types, the frequency bands containing the fundamental frequencies of knocks (172 Hz: open squares, and 603 Hz: symbols) and growls (86 Hz: solid diamonds, and 344 Hz: open circles) are provided along with a linear best fit line for all data combined.
the average SNR of knock vocalizations to the natural ambient noise was 43 dB at 172.2 Hz, and 51.1 dB at 602.7 Hz. At 172.2 and 602.7 Hz, the spectrum level of natural noise was 73.2 ± 3.4 and 65.4 ± 10.5 dB re 1 lPa (analysis bandwi dth = 86.1 Hz), respectively. C. venusta knocks should therefore propagate between 38.2 and 64.6 cm (mean 51.4 cm) at 172.2 Hz and between 34.9 and 93.1 cm (mean 64.0 cm) at 602.7 Hz before dropping below a SNR of 10.0 in natural conditions. At 107.5 and 301 Hz, the natural noise was 63.7 ± 4.7 and 56.8 ± 5.6 dB re 1 lPa (analysis bandwidth = 21.5 Hz), respectively. Growls should propagate 0–35.8 cm (mean 16.2 cm) at 107.5 Hz and 0.3–43.0 cm (mean = 21.7 cm) at 301 Hz before dropping below a SNR of 10 dB in natural conditions. The spectrum level of semi-trailer truck noise was 112.3 ± 4.2 dB re 1 lPa (analysis bandwidth = 86.1 Hz) at 172.2 Hz, and 76.2 ± 4.3 dB re 1 lPa (analysis bandwidth = 8 6.1 Hz) at 602.7 Hz. Knocks should therefore emerge from truck noise with a SNR above 10 at distances between 0.6 and 22.3 m (mean = 3.0 m) from the bridge at 172.2 Hz and should already have a SNR above 10 at 1 m from the bridge at 602.7 Hz. Semi-trailer truck noise was 109.8 ± 3.1 dB re 1 lPa (analysis bandwidth = 21.5 Hz) at 107.5 Hz, and 86.2 ± 2.6 dB re 1 lPa (analysis bandwidth = 21.5 Hz) at 301 Hz. Growls should therefore emerge from semi-trailer truck noise with a SNR above 10 dB at distances between 58.0 m and 12,113 m (mean = 640.0 m) from the bridge at 107.5 Hz, and between 2.7 m and 1584.2 m (mean = 39.5 m) at 301 Hz. 4. Discussion Our results suggest that road traffic noise masks the acoustic signal of a representative stream fish, C. venusta, and recovery to the natural soundscape may be as far as 12,113 m, depending on the frequency. Semi-trailer truck noise was shown to diminish the normal active area of both growls and knocks to some degree, depending on the frequency of concern and the distance from the noise source. Growls showed the potential to be more severely affected than knocks, due to their lower amplitude and greater spectral overlap with anthropogenic noise. It is difficult, however, to assess the degree to which anthropogenic noise sources affect communication and overall reproductive success of C. venusta, although this species was shown to exhibit the Lombard Effect (producing higher amplitude signals) in response to the traffic noise recorded in this study (Holt and Johnston, 2014b). It is not known whether there is a difference in the information content of the low-band and high-band frequencies, which are affected
31
differently by truck noise. Several studies have shown correlations between signal frequency and male size, which can be considered an indicator of male fitness. In the case of C. venusta, the 301 Hz band of growls is negatively correlated with fish standard length (Holt and Johnston, 2014a). Disturbance of the quiet window in the natural soundscape caused by anthropogenic noise may interfere with the ability of females to attain important information about males from acoustic signals. Other signal parameters such as call duration have also been found to be correlated with male condition (Amorim et al., 2010). If call duration carries such information in other freshwater fishes, a smaller active area induced by anthropogenic noise events could cause growls to be heard for a shorter period of time by a female, which could diminish perceived male fitness and have adverse effects on reproductive success. Interfering with call detection could especially be true of many benthic species of fishes that produce lengthy tonal calls. Call duration has been shown to be related to male quality, and preferred by females in the Lusitanian toadfish (H. didactylus). Numerous darters in the subgenus Catonotus (Johnston and Johnson, 2000; Speares et al., 2011), freshwater gobies such as Padogobius martensii (Lugli and Fine, 2007), and the banded sculpin (Cottus carolinae; personal observation) are examples of benthic, freshwater species who produce tonal sounds during reproductive or aggressive encounters. Masking of even part of these sounds could disrupt the communicative function of these calls. Despite the consistent presence of natural noise at C. venusta nesting sites, the presence of a quiet window in the natural noise spectrum helps increase SNR of vocalizations. Knocks show spectrum levels that are much higher than the natural ambient noise, are always produced at a close distance to the receiver, and are therefore not likely dependent on the integrity of the natural quiet window. Growl are much quieter with SNR’s less than half that of knocks, and therefore rely much more on the quiet window for maintenance of their active area. Although we did not investigate it, noise from bridge crossings should definitely show temporal variation at both short term and long term scales. We have observed peak spawning times for Cyprinella venusta in the morning before water temperatures begin to rise. An overlap between peak morning traffic and spawning times could exacerbate any communication problems caused by anthropogenic noise. The close association between C. venusta vocalizations and the quiet window within the natural ambient noise is not unique to C. venusta. Reports of quiet windows in the background noise of freshwater streams have been made by several authors (Wysocki et al., 2007; Speares et al., 2011; Lugli and Fine, 2003; Amoser, 2006; Crawford et al., 1997), and Lugli (2010) reports the close association of vocalization frequencies with quiet windows in the habitat for several goby species inhabiting freshwater, brackish, and marine environments. Because hydrology largely dictates natural environmental noise, and fish are often bound to certain habitat types due to specific life history strategies, the utilization of a naturally occurring quiet window in the ambient noise is probably an adaptation for intraspecific communication. However, because physical conditions largely dictate both the natural underwater soundscapes and preferred nesting habitats, the nature of the quiet window and the effect of anthropogenic noise on it is more likely to be conserved across a larger geographic areas. Detection distances of the different call types do not appear to be arbitrary and seem to fit appropriately with the behaviors they accompany. This correlates well with the types of behaviors that are typically associated with knocks. Knocks are usually produced by males during aggressive encounters with other males around the nesting site, with males typically less than 15 cm apart when vocalizing during a conflict. We have also observed knock production during chases between nest guarding males and sunfish that
32
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33
would attempt to eat eggs from the nest. The ability of C. venusta to theoretically detect knocks at over half of a meter away suggests that sounds produced by a male may be detectable by males that are nearby, but not actively engaged in an aggressive interaction with the vocalizing male under normal ambient noise conditions. If knocks serve as an honest signal of male fitness to other males, a larger active area of the sound should benefit the sound producer. Because there are often many other reproductive males at a given nesting site, a territorial male should benefit from sending honest signals about their intentions and motivational level to both the intruding male and peripheral males simultaneously. Although results from the speaker playback showed no difference in attenuation rates of different frequency bands of fish calls, the greater variation in attenuation seen at the lowest frequency band may have been due it being at the lower end of the speaker’s frequency response. Attenuation rates of truck noise at frequencies below the cutoff frequency (where the energy of the sound begins to be reduced) of the streams were much lower than fish sounds because truck noise can be transmitted into the stream through the air, as well as through the substrate. Propagation of sounds that originate within the water column are subjected to the limitations of the cutoff frequency, a phenomenon by which acoustic signals below a certain frequency dictated in large part by the water depth attenuate extremely rapidly (Rogers and Cox, 1988). We are unsure of the contribution of noise transmitted from the air versus the substrate, but it is feasible for sounds to propagate down the bridge pilings into the substrate where it could re-emerge into the water column. Because the composition of the bottom below the immediate substrate (sand and gravel) is unknown, we cannot address the mode of propagation at this point. The findings of the current study offer a new perspective with which to view reproductive behaviors in fishes inhabiting small, freshwater systems, providing a description of a previously undocumented and potentially widespread source of anthropogenic noise in these habitats. The noisy environment in which C. venusta spawn has a convenient window in the noise spectrum, which is exploited by C. venusta for the purpose of communication with females during reproductive behaviors. Our results show that this quiet window is disrupted by road traffic noise. This discovery should be followed up by studies investigating the behavioral, and stress responses of C. venusta, as well as other more imperiled fishes, to noise from bridge crossings. Future efforts should also be made to characterize noise from different bridge styles, daily temporal patterns of noise, and attenuation at longer distances.
Acknowledgements We would like to thank D. Higgs, G. Hill, D. Mann, M. Mendonça, and P. Noel for valuable input and comments concerning the study.
Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biocon.2015.04. 004. These data include Google maps of the most important areas described in this article.
References Amorim, M.C.P., Simões, J.M., Mendonça, N., Bandarra, N.M., Almada, V.C., Fonseca, P.J., 2010. Lusitanian toadfish song reflects male quality. J. Acoust. Soc. Am. 213, 2997–3004. Amoser, S., 2006. Year-round changes of ambient noise in freshwater habitats and temperature effects on hearing in fishes. J. Acoust. Soc. Am. 120, 3056.
Amoser, S., Wysocki, L.E., Ladich, F., 2004. Noise emission during the first powerboat race in an alpine lake and potential impact on fish communities. J. Acoust. Soc. Am. 116, 3789–3797. Banner, A., Hyatt, M., 1973. Effects of noise on eggs and larvae of two estuarine Fishes. Trans. Am. Fish Soc. 1, 134–136. Bennet-Clark, H.C., 1999. Which Qs to choose: questions of quality in bioacoustics. Bioacoustics 9, 351–359. Buerkle, U., 1969. Auditory masking and the critical band in Atlantic Cod (Gadus morhua). J. Fish. Res. Board Can. 26, 1113–1119. Chapman, C.J., 1973. Field studies of hearing in teleost fish. Helgoländer Meeresun. 24, 371–390. Chapman, C.J., Sand, O., 1974. Field studies of hearing ink two species of flatfish Pleuronectes platessa (L) and Limanda limanda (L) (Family Pleuronectidae). Comp. Biochem. Physiol. – A 47, 371–385. Codarin, A., Wysocki, L.E., Ladich, F., Picciulin, M., 2009. Effects of ambient and boat noise on hearing and communication in three fish species living in a marine protected area (Miramare, Italy). Mar. Pollut. Bull. 58, 1880–1887. Crawford, J.D., Jacob, P., Bénech, V., 1997. Sound production and reproductive ecology of strongly acoustic fish in Africa: Pollimyrus isidori, Mormyridae. Behaviour 134, 677–725. Fay, R.R., 1974. Masking of tones by noise for the goldfish (Carassius auratus). J. Comp. Physiol. Psychol. 87, 708–716. Fernandes, P.G., Brierley, A.S., Simmonds, E.J., Millard, N.W., McPhail, S.D., Stevenson, P., Squires, M., 2000. Fish do not avoid survey vessels. Nature 604, 35–36. Fine, M.L., Lenhardt, M.L., 1983. Shallow-water propagation of the toadfish mating call. Comp. Biochem. Physiol. 76A, 225–231. Ghahramani, Z.N., Mohajer, Y.J., Fine, M.L., 2014. Developmental variation in sound production in water and air in the blue catfish Ictalurus furcatus. J. Exp. Biol. 217, 4244–4251. Handegard, N.O., Michalsen, K., Tjostheim, D., 2003. Avoidance behavior in cod (Gadus morhua) to a bottom-trawling vessel. Aquat. Living Resour. 16, 265–270. Hastings, M.C., 2008. Coming to terms with the effects of ocean noise on marine animals. Acoust. Today 4, 22–34. Hawkins, S.D., Sand, O., 1977. Directional hearing in the median vertical plane by the cod. J. Comp. Physiol. A 122, 1–8. Holt, D.E., Johnston, C.E., 2014a. Sound production and associated behaviors in Blacktail Shiner (Cyprinella venusta): a comparison between field and lab. Environ. Biol. Fish. 97, 1207–1219. Holt, D.E., Johnston, C.E., 2014b. Evidence of the Lombard effect in fishes. Behav. Ecol. http://dx.doi.org/10.1093/beheco/aru028. Johnston, C.E., Johnson, D.L., 2000. Sound production during the spawning season in cavity-nesting darters of the subgenus Catonotus (Percidae: Etheostoma). Copeia 2000, 475–481. Jouventin, P., Aubin, T., Lengagne, T., 1999. Finding a parent in a king penguin colony: the acoustic system of individual recognition. Anim. Behav. 57, 1175– 1183. Kasumyan, A.O., 2008. Sounds and sound production in fishes. J. Ichthyol. 48, 981– 1030. Ladich, F., 2013. Effects of noise on sound detection and acoustic communication in fishes. In: Brumm, H. (Ed.), Animal Communication and Noise. Springer-Verlag, Berlin, Heidelberg, pp. 65–90. Lagardère, J.P., 1982. Effects of noise on growth and reproduction of Crangon crangon in rearing tanks. Mar. Biol. 71, 177–185. Locascio, J.V., Mann, D.A., 2011. Localization and source level estimates of black drum (Pogonias chromis) calls. J. Acoust. Soc. Am. 130, 1868–1879. Lugli, M., 2010. Sounds of shallow water fishes pitch within the quiet window of the habitat ambient noise. J. Comp. Physiol. A 196, 439–451. Lugli, M., Fine, M.L., 2003. Acoustic communication in two freshwater gobies: ambient noise and short-range propagation in shallow streams. J. Acoust. Soc. Am. 114, 512–521. Lugli, M., Fine, M.L., 2007. Stream ambient noise, spectrum and propagation of sounds in the goby Padogobius martensii: sound pressure and particle velocity. J. Acoust. Soc. Am. 122, 2881–2892. Mann, D.A., Lobel, P.S., 1997. Propagation of damselfish (Pomacentridae) courtship Sounds. J. Acoust. Soc. Am. 101, 3783–3791. Myrberg, A.A., 1990. The effects of man-made noise on the behavior of marine mammals. Environ. Int. 16, 575–586. Narins, P.M., Feng, A.S., Lin, W., Schnitzler, H.U., Denzinger, A., Suthers, R.A., Xu, C., 2004. Old world frog and bird vocalizations contain prominent ultrasonic harmonics. J. Acoust. Soc. Am. 115, 910–913. Popper, A.N., 2003. Effects of anthropogenic sounds on fishes. Fish. Res. 28, 24–31. Popper, A.N., Fay, R.R., 2010. Rethinking sound detection by fishes. Hear. Res. 273, 25–36. Radford, A.N., Kerridge, E., Simpson, S.D., 2014. Acoustic communication in a noisy world: can fish compete with anthropogenic noise? Behav. Ecol. 25, 1022–1030. Rogers, P.H., Cox, M., 1988. Underwater sound as a biological stimulus. In: Atema, J., Fay, R.R., Popper, A.N., Tavolga, W.N. (Eds.), Sensory Biology of Aquatic Animals. Springer, New York, pp. 131–149. Southall, B.L., Bowlesn, A.E., Ellison, W.T., Finneran, J.J., Gentry, R.L., Green Jr., C.R., Kastak, D., Ketten, D.R., Miller, J.H., Nachtigall, P.E., Richardson, W.J., Thomas, J.A., Tyack, P.L., 2007. Marine mammal noise exposure criteria: Initial scientific recommendations. Aquat. Mammals 33, 411–521. Speares, P.D., Holt, D.E., Johnston, C.E., 2011. The relationship between ambient noise and fundamental frequency of vocalizations in two species of darters (Percidae: Etheostoma). Environ. Biol. Fish. 90, 103–110.
D.E. Holt, C.E. Johnston / Biological Conservation 187 (2015) 27–33 Vabø, R., Olsen, K., Huse, I., 2002. The effect of vessel avoidance of wintering Norwegian spring spawning herring. Fish. Res. 58, 59–77. Vasconcelos, R.O., Amorim, C.P., Ladich, F., 2007. Effects of ship noise on the detectability of communication signals in the Lusitanian toadfish. J. Exp. Biol. 210, 2104–2112. Voellmy, I.K., Purser, J., Flynn, D., Kennedy, P., Simpson, S.D., Radford, A.N., 2014a. Acoustic noise reduces foraging success in two sympatric fish species via different mechanisms. Anim. Behav. 89, 191–198. Voellmy, I.K., Purser, J., Simpson, S.D., Radford, A.N., 2014b. Increased noise levels have different impacts on the anti-predator behaviour of two sympatric fish species. PLoS ONE 9, e102946. Waser, P.M., Waser, M.S., 1977. Experimental studies of primate vocalization: specializations for long distance propagation. Z. Tierpsychol. (J. Comp. Ethol.) 43, 239–263.
33
Wiley, R.H., Richards, D.G., 1982. Adaptations for acoustic communication in birds: sound transmission and signal detection. In: Kroodsma, D.E., Miller, E.H., Quellet, H. (Eds.), Acoustic Communication in Birds, vol. I. Academic Press, New York, pp. 131–278. Wysocki, L.E., Ladich, F., 2005. Effects of noise exposure on click detection and the temporal resolution ability of the goldfish auditory system. Hear. Res. 201, 27– 36. Wysocki, L.E., Amoser, S., Ladich, F., 2007. Diversity in ambient noise in European freshwater habitats: noise levels, spectral profiles, and impact on fishes. J. Acoust. Soc. Am. 121, 2559–2566.