Evaluating and Assessing the Relative Effectiveness of Acoustic Deterrent Devices and other Non-Lethal Measures on Marine Mammals

Marine Scotland commissioned a research project aimed at gathering literature and data into the effectiveness of non-lethal measures of deterring marine mammals from a range of activities (e.g. fish farms, renewable developments etc.). This review attempt

7 Concerns over the Use of ADDs

The primary function of acoustic deterrents is to reduce the impact of human activities on a particular group of marine mammals. In order to deliver an ecologically coherent assessment of ADDs, the benefits created for the target group, or species, must be compared against any potential negative effects upon that group, or any other affected group. Only once a reliable assessment of the likely environmental impacts has been made, can an informed appraisal be made comparing and evaluating the predicted merits with the likely costs.

It is also important to note that what may be regarded as a negative impact in one application ( e.g. the displacement of certain species) could be considered as a desired impact in another application. The discussion below relating to disturbance, exclusion and behavioural effects considers these impacts in regards to non-target species, where any impact is conceptually negative. In another instance, such as pile-driving mitigation, where all species might be described as 'target species', less species discrimination will be necessary and these impacts should not be considered as negative.

7.1 Reduction in Responsiveness over Time

Many of the studies discussed in this document have reported a reduction in responsiveness over time (Arnold, 1992; Rueggeberg and Booth, 1989; Sepulveda and Oliva, 2005). This is often loosely referred to as "habituation" but could in fact result from a combination of effects. Habituation can be defined as a decrease in a behavioural response to a recurring stimulus. In the case of humans, we know that habituation in this sense results when we no longer pay attention to the recurrent stimulus. Reduced effectiveness of ADDs at fish farms, for example, might also result from animals learning strategies to avoid responding to these signals, or to reduce their effects. For example, some have reported seals swimming with their heads above water, presumably minimising the impact of underwater sound ( e.g. Mate et al., 1986a). Animals might also learn to approach powerful ADDs between transmissions or find "holes" or "shadows" in the sound field. Northridge et al. (2010) report an instance where seal depredation at a site started after the failure of a single transducer in a multi-transducer system. This may have provided a gap in the acoustic field allowing the seals to reach the net. Reduced effects of ADDs on seals could also result from permanent threshold shift, a reduced sensitivity through hearing damage. Götz (2008) highlighted the fact that the early stage of hearing damage often affects the outer hair cells which act to amplify signals within the cochlea.

In practical terms it may be important to distinguish between these different mechanisms for reduced effectiveness in order to find strategies to counteract them. For example, habituation is known to be stimulus specific, and the behavioural response (deterrence) may return when presented with a new stimulus or if the sound source is active only intermittently. A model of ADD which has been designed to generate a diversity of signals, presumably to minimise habituation is the Terecos DSMS-4 ( section 2.2.1).

The distinction between habituation and hearing damage is also very important in this context. When the fatiguing stimulus is withdrawn from a habituated animal for a period of time, the response is known to recover at least partially (Rankin et al., 2010). This is obviously not the case for permanent hearing damage.

While reduction in efficacy has been reported in the majority of studies, exceptions include the work of Graham et al. (2009) who tested ADDs in Scottish salmon rivers and stated that they found no evidence of reduced effectiveness over a five month trial period, and that of Kastelein et al. (2006) who found that while captive animals exhibited slight habituation within a 45 minute sound exposure, this did not transfer between days. Neither of these studies was conducted at a fish farm and for both of them the sound source was active for relatively short periods of time. Similarly, Dawson et al. (2013) found that there was no evidence of habituation, or any diminution of the response of cetaceans (as measured by bycatch rates) to long-term exposure to pingers on gillnets, and Morton and Symonds (2002) found no reduction in the displacement of killer whales by ADDs over several years. In part, this variation can probably be explained by motivation, or lack thereof. In the context of pingers preventing bycatch in gillnets, there is no strong motivation for cetaceans to approach nets, whereas for seals at aquaculture sites, this may not be the case. Gotz and Janik (2010) for example showed that captive seals rapidly stopped showing an aversive response to a received level of up to 146 dB re 1 μPa ( RMS) when food was provided as a motivation to stay close to the loudspeaker.

The work of Götz and Janik (2011) measuring startle responses is unique in that in their trials the animals seemed to show sensitisation and increasing responsiveness with repeated exposure. As part of work to develop a more effective ADD based on the startle reflex they exposed seals in captivity to loud, fast onset sounds, designed to elicit a startle response. Startle sound were preceded by a quieter alerting signal. They report increasing responsiveness to the signal. Eventually animals would react to just the alerting signal by leaving the feeding station and hauling out. Trials are now underway to explore whether this will translate into real-world application in ADDs that are really effective in context of a commercial fish farm.

7.2 Hearing Damage

7.2.1 Thresholds and Criteria for Hearing Damage from Sound Exposure

To effectively manage risks resulting from the exposure of marine mammals to loud sounds, regulators need to work with agreed criteria for acceptable exposure, and to establish thresholds below which exposure might be considered to be of minimal concern. This is a difficult task, in part because, until recently, very little information on the effects of noise on marine mammal hearing existed and, although studies have been carried out over the last decade or so, information is still sparse. Regulators in Europe and the UK have not proposed any science-based thresholds themselves; however, more progress has been in made in North America. Here, the US National Oceanographic and Atmospheric Administration ( NOAA) funded a series of workshops for a panel of experts charged with developing criteria for noise exposure for marine mammals. This group was able to find little basis for proposing generally applicable thresholds for behavioural responses to sound. Effects of sound on marine mammal hearing however, which largely result from a combination of mechanical and physiological processes, have proven more amenable to prediction. A review of much of the research that supported their deliberations, a detailed explanation of how these were then used to determine criteria and the proposed thresholds themselves, were published in a peer reviewed paper (Southall et al., 2007). These criteria are often termed the "Southall Criteria".

Noise can result in hearing damage via two mechanisms. Exposure to extremely loud acoustic pressures or impulses can cause instantaneous damage mechanically. Sound exposures at lower levels over longer periods of time can also result in permanently impaired hearing which is more likely to be related to metabolic exhaustion of sensory cells from over-stimulation. In this case, as a first approximation , the total amount of sound energy received over a time period, the sound exposure level ( SEL), is a more useful metric than sound pressure level ( SPL).

Reflecting these two mechanisms, Southall et al. (2007) proposed a dual set of criteria: sound pressure level thresholds determining the maximum allowable peak pressure exposure, however brief; and sound exposure level thresholds defining the maximum allowable dose of acoustic energy received over an extended period (up to 24 hours).

7.2.2 Measuring Hearing Damage (Temporary and Permanent Threshold Shifts)

It is considered unethical to directly damage the hearing of marine mammals experimentally, so instead, as is the case with most human research, the phenomenon of temporary threshold shift ( TTS) has been studied. As its name suggests, TTS is an impermanent reduction in sensitivity (increase in threshold) resulting from exposure to sound. Hearing returns to pre-exposure levels after a recovery period of minutes to hours. Temporary threshold shift is not in itself considered harmful and i is a phenomenon that we all experience and adapt to in our daily lives. It is considered unlikely that occasional TTS is of biological significance for wild animals. Its importance in this context is as an indicator of the exposure levels at which hearing damage might occur. Generally, the greater the sound exposure, the greater will be the reduction in sensitivity. Southall et al. (2007) reviewed available marine mammal TTS studies which included data for two species of odontocete, bottlenose dolphins and beluga, and three pinnipeds, the harbour seal, the elephant seal and the California sea lion. In reviewing this literature and the more detailed research with humans and other terrestrial mammals they also found general support for the contention that the total acoustic dose, the sound exposure level ( SEL), correlated well with TTS onset over a range of different exposure periods. This relationship is the basis for the so called "equal energy" hypothesis, which states that equal amounts of acoustic energy (measured as SEL) will cause equal amounts of hearing impairment, regardless of how this energy is distributed over time. The studies available provide data on sound exposures leading to TTS for a limited number of marine mammals. It was necessary to extrapolate from these to exposures likely to result in permanent threshold shift ( PTS). Based largely on studies of terrestrial mammals and humans, Southall et al. (2007) proposed levels of additional exposure required to induce PTS for several different sound types and species groups. They proposed that for continuous sound exposures, levels for PTS should be the levels causing TTS plus 20 dB for all marine mammals. For single or multiple pulses PTS threshold should be that for TTS plus 15 dB.

7.2.3 Frequency Weighting

Different species show both a difference in absolute sensitivity ( i.e. in the quietest sounds they can hear) and also some variation in their relative sensitivity at different frequencies. Typically, this frequency dependent variability in auditory sensitivity reflects a species' life style and the spectral range of its vocalisations. Thus, within marine mammals, high frequency specialists such as the harbour porpoise, have extremely good sensitivity in the high ultrasonic, specifically around 120 kHz - the dominant frequency in their echolocation clicks. Seals have best sensitivity at lower frequencies (in the mid-10s of kHz), as well as having poorer overall sensitivity than porpoises and dolphins. Baleen whales, which predominantly produce low frequency vocalisations, have auditory systems that have adapted to be sensitive mainly to low frequency sound (Ketten, 1997; Ketten, 1998).

It is likely that their differential hearing sensitivities make species more or less vulnerable to the damaging effects of noise at different frequencies ( e.g. high frequency specialist might be more likely to have their hearing affected by high frequency fatiguing noise than would low frequency specialists). The most common way of measuring frequency-dependent differential sensitivity is to measure the quietest pure tones that can be just detected at a series of frequencies across the animal's hearing range. A plot of these minimum thresholds against frequency is called an audiogram. For marine mammals, audiograms can be obtained either behaviourally, where a captive animal is trained to respond in a particular way when a sound is detected; or electro-physiologically, by measuring electrical signals from the auditory brain-stem response ( ABR) using surface electrodes. Behavioural audiograms are considered superior, but are very difficult and time consuming to obtain. Currently, audiograms have only been measured from a limited subset of marine mammal species and much of the available audiogram data are summarised in Nedwell et al. (2004). An audiogram can be a useful basis for determining parameters directly related to the detection of low level sounds, such as the maximum range at which a sound can be detected in a low noise environment. However, it may not be a reliable or appropriate metric for predicting hearing damage caused by exposure to intense sounds.

KG No. Knowledge Gap
50 Reliable audiogram data (or equal loudness contours) are not available for several of the species found in Scottish waters ( e.g. minke whale, white-beaked [ Lagenorhynchus albirostris] and Atlantic white-sided [ Lagenorhynchus acutus] dolphins).

Patterns of differing sensitivity to different frequencies reflect in how loud the sound is. "Loudness" is a psychological term (not a direct physical one) which describes how a subject perceives sounds of different intensities and is measured in phons. Measuring loudness for a human is quite straight forward. For example, a subject might be asked to compare their perception of the loudness of tones at different frequencies and adjust the levels of two tones until they are perceived as being the same. In this way plots of how loudness is perceived at different frequencies can be derived (Fletcher and Munson, 1933). By convention, loudness is referenced to the perception of a tone at 1 kHz. Plots of equal loudness (frequency versus loudness in phons) for very quiet sounds generally follow the u-shaped curve of an audiogram. However, as the intensity of signals being tested increases, plots of equal loudness tend to become "flatter". In other words, the differences in perception of loudness with frequency become less pronounced as a sound's intensity increases ( Figure 13). The risk of inducing hearing damage from sounds of different frequency is thought to reflect these "flattened" phon plots for more intense sounds.

Figure 13 Fletcher-Munson curves: plots of equal loudness for sounds of different intensities and frequencies (Fletcher and Munson, 1933)

Figure 13

Appropriate Fletcher-Munson curves (or their revised modern equivalents) are the basis for the acoustic filters used to provide frequency weighting when assessing the effects of different types of noise on humans. Thus, when considering annoyance effects from relatively low level noise, the so-called "A-weighting", based on the 40 phon curve, is applied, while for the assessment of the effects of intense sounds a "C-weighting" filter based on the equal loudness curve at 100 phons is more appropriate. (40 phons is approximately the noise level in a quiet home, 100 phons would be experienced close to noisy machinery such as a petrol-driven chainsaw.)

Equal loudness contours had not been measured for any marine mammals when Southall et al., (2007) were reviewing the available literature for their report (some data, however, are now available for bottlenose dolphins and harbour porpoises). Given this lack of data, Southall et al., (2007) derived frequency-selective weighting functions for four groups of marine mammals based on the shape of the human C-weighting function and knowledge of the functional hearing range of the species groups concerned. Their intention was that, given the considerable uncertainty in this area, the application of these filters should lead to precautionary assessment of hearing risks. The functional hearing groups for which they proposed frequency weighting filters were:

  • Low frequency cetaceans (baleen whales)
  • Mid-frequency cetaceans (57 species of odontocetes ranging from sperm whales to oceanic dolphins)
  • High-frequency cetaceans (20 species producing narrow band very high frequency clicks including porpoises, Kogia spp. and Cephalorhynchus spp. dolphins)
  • Pinnipeds in water
  • Pinnipeds in air

7.2.4 Thresholds for Permanent Threshold Shifts ( PTS)

Southall and colleagues were able to calculate sound exposure thresholds for two different types of criteria (sound pressure level and sound exposure level) over the four different marine mammal auditory groups (low frequency cetaceans, mid frequency cetacean, high frequency cetaceans, pinnipeds in water and pinniped in air) for three different sound types (single pulses, multiple pulses and non-pulsed). They achieved this by combining results from available studies of TTS in marine mammals, adding the suggested additional exposure required to induce PTS and applying appropriate frequency-dependent filters. The calculated thresholds are outlined in Table 3.

The sound pressure level thresholds are only likely to be reached as a result of explosions or be found close to powerful impulsive activities such as pile driving. These levels are therefore not relevant for exposures to ADDs used at Scottish salmon farms, but they may be at marine construction sites. In the context of this review, it is the sound exposure level ( SEL) thresholds that are more likely to be exceeded through prolonged exposure to ADD sound fields. These thresholds relate to cumulative exposure over an extended period. To assess this, in addition to the sound source level, propagation loss and sound field, one needs to consider the duty cycle of the signal and, most importantly, how focal animals behave and move in its vicinity. In fact, in many cases it is this simple behavioural information which remains as the most critical data gap, limiting the calculation of more realistic thresholds. Behavioural responses to sound will be modified by many factors including experience, learning and motivational state and may be fundamentally unpredictable. Behavioural responses can, however, be directly measured, and in most cases they must be.

Table 3 Proposed injury criteria for individual marine mammals exposed to "discrete" noise events (either single or multiple exposures within a 24-h period)

Sound type
Marine Mammal Group Single Pulses Multiple Pulses Non-Pulsed
Low-frequency cetaceans
Sound pressure level 230 dB re: 1 μPa (peak) (flat) 230 dB re: 1 μPa (peak) (flat) 230 dB re: 1 μPa (peak) (flat)
Sound exposure level 198 dB re: 1 μPa 2 s -1 (Mlf) 198 dB re: 1 μPa 2 s -1 (Mlf) 215 dB re: 1 μPa 2 s -1 (Mlf)
Mid-frequency cetaceans
Sound pressure level 230 dB re: 1 μPa (peak) (flat) 230 dB re: 1 μPa (peak) (flat) 230 dB re: 1 μPa (peak) (flat)
Sound exposure level 198 dB re: 1 μPa 2 s -1 (Mmf) 198 dB re: 1 μPa 2 s -1 (Mmf) 215 dB re: 1 μPa 2 s -1 (Mmf)
High-frequency cetaceans
Sound pressure level 230 dB re: 1 μPa (peak) (flat) 230 dB re: 1 μPa (peak) (flat) 230 dB re: 1 μPa (peak) (flat)
Sound exposure level 198 dB re: 1 μPa 2 s -1 (Mhf) 198 dB re: 1 μPa 2 s -1 (Mhf) 215 dB re: 1 μPa 2 s -1 (Mhf)
Sound pressure level 199.7 dB re: 1 μPa (peak) (flat) 199.7 dB re: 1 μPa (peak) (flat) 199.7 dB re: 1 μPa (peak) (flat)
Sound exposure level 179.3 dB re: 1 μPa 2 s -1 (Mhf) 179.3 dB re: 1 μPa 2 s -1 (Mhf) 184.3 dB re: 1 μPa 2 s -1 (Mhf)
Pinnipeds (in water)
Sound pressure level 218 dB re: 1 μPa (peak) (flat) 218 dB re: 1 μPa (peak) (flat) 218 dB re: 1 μPa (peak) (flat)
Sound exposure level 186 dB re: 1 μPa 2 s -1 (Mpw) 186 dB re: 1 μPa 2 s -1 (Mpw) 203 dB re: 1 μPa 2 s -1 (Mpw)
Pinnipeds (in air)
Sound pressure level 149 dB re: 20 μPa (peak) (flat) 149 dB re: 20 μPa (peak) (flat) 149 dB re: 20 μPa (peak) (flat)
Sound exposure level 144 dB re: 20 μPa 2 s -1 (Mpa) 144 dB re: 20 μPa 2 s -1 (Mpa) 144.5 dB re: 20 μPa 2 s -1 (Mpa)

7.2.5 Relevant findings since Southall et al. (2007)

One of the strengths of Southall et al. (2007) is that it lays out a logical framework for determining thresholds and meticulously describes how this was applied using the information available at the time. This makes it possible to apply the same method to new research findings as they become available. Indeed, facilitating this process of revision was Southall et al.'s stated intention. Here we review some relevant work in this area which has been completed since the publication of their report. Some of this work provides new information on required data, such as TTS thresholds, while other findings address certain aspects of their approach. Temporary Threshold Shifts ( TTS) in Porpoises

Harbour porpoises are the most common marine mammal in Scottish coastal waters and the species of cetacean most likely to come into contact with, and be affected by, ADDs at Scottish aquaculture sites or marine energy development sites. Southall et al. (2007) did not include any data on hearing effects on porpoises or other high frequency specialists. However, the bioacoustics of this species are quite different from that of the better studied mid-frequency odontocetes and some earlier papers ( e.g. Verboom, 2000) had suggested that harbour porpoise would be more vulnerable to auditory damage than mid-frequency odontocetes.

Concerns about the effects that pile driving might have on the hearing of harbour porpoises led to a series of experiments in which captive porpoises were exposed to impulses from a small (20 cubic inches) airgun. The airgun produced powerful low frequency sound pulses with peak frequency below 500Hz, although significant energy also extended to frequencies up to 20 kHz. With these acoustic characteristics the airgun served as a convenient surrogate sound source for pile driving noise (Lucke et al., 2008; Lucke et al., 2009). Hearing thresholds were measured at frequencies of 4, 32 and 100 kHz before and after exposure. Exposure levels were increased during trials until a clear TTS was evident. TTS was induced at 4 kHz (but not at the two higher test frequencies) after relatively low exposure of 199.7 dB re 1 µPa peak to peak (193.7 dB re 1 µPa peak) and a sound exposure level of 164.3 dB re 1µPa 2 s.

These results were noteworthy for several reasons. They were the first data on TTS for any phocoenid. It was also notable, and perhaps surprising, that TTS could be induced by noise so far below the frequency range of best hearing (which in porpoises is at around 100 kHz). In fact the peak frequency and the bulk of the sound energy from the airgun pulse would fall outside the frequency-weighting filters for high frequency cetaceans proposed by Southall et al. (2007). In other words, if the Southall process was applied to these new data, the effective SEL of an airgun exposure would be rather low (see Table 4).

Some more recent studies provide further evidence that phocoenid auditory systems might be particularly vulnerable to being damaged by noise. Popov et al. (2011) report on an extensive set of trials with Yangtze finless porpoise ( Neophocaena phocaenoides asiaeorientalis). They exposed two study animals (one male, one female) to half-octave band noise and measured thresholds at frequencies of 32, 45, 64 and 128 kHz. Greatest levels of TTS were measured when the noise band centre frequency was 0.5 octaves below that of the test frequency. Lower frequency noise seemed to have a stronger effect in inducing TTS than did high frequencies. The study mainly focused on patterns of TTS development and recovery and threshold exposures for TTS are not explicitly stated. However, their results indicated that large TTS, up to 30 dB, was induced by an exposure of 150 dB re 1 µPa for 1 minute, equivalent to an SEL of 168 dB re 1 μPa 2 s -1. This was very much in line with Lucke et al.'s threshold value for harbour porpoise of 164.3 dB. Their results did not fully support the "equal energy hypothesis" in that, for signals of equivalent acoustic energy, higher amplitude sounds appeared to cause greater threshold shifts than longer duration sounds. A possible explanation for this would be that some recovery had taken place during the exposure period. The time taken for TTS to recover to pre-exposure levels however, seemed to be more effected by exposure duration than signal amplitude, particularly for low frequency sounds. While this paper does not provide a threshold value for TTS, the substantial TTS they induced at relatively low SEL is in line with Lucke et al.'s (2009) suggestions of low thresholds for TTS in harbour porpoises.

Table 4 Revised thresholds for PTS for porpoise and harbour seals calculated by applying the "Southall et al." method to new TTS threshold data

Marine Mammal Group Single pulses Multiple pulses Non-pulsed
Lucke et al., 2009
Sound exposure level 179.3 dB re: 1 μPa 2 s -1 179.3 dB re: 1 μPa 2 s -1 184.3 dB re: 1 μPa 2 s -1
Kastelein et al., 2012b
Sound exposure level 166 dB re: 1 μPa 2 s -1 166 dB re: 1 μPa 2 s -1 171 dB re: 1 μPa 2 s -1
Harbour Seal (in water)
Kastelein et al. 2012a
Sound exposure level
(short exposures- 15 mins)
193 dB re: 20 μPa 2 s -1 (Mpa) 193 dB re: 20 μPa 2 s -1 (Mpa) 198 dB re: 20 μPa 2 s -1 (Mpa)
Sound exposure level
(long exposures- 60 mins)
185 dB re: 20 μPa 2 s -1 (Mpa) 185 dB re: 20 μPa 2 s -1 (Mpa) 195.5 dB re: 20 μPa 2 s -1 (Mpa)

The most recent and most complete set of measurements of TTS in porpoises is reported in Kastelein et al. (2012b). They exposed a young male harbour porpoise to octave band noise centred at 4 kHz at sound exposure levels ranging from 151 to 190 dB re 1 μPa 2 s -1. They achieved this using 18 different combinations of three sound pressure levels (124, 135 and 148 dB re 1 μPa) ) and 6 exposure durations ranging from 7.5 to 240 minutes. The lowest SEL that induced a significant TTS was 151 dB re 1 μPa 2 s -1 (124 dB re 1 μPa for 7.5 mins) while the greatest exposure, an SEL of 190 dB re 1 μPa 2 s -1 (148 dB re 1 μPa for 240 minutes) caused a TTS of 15 dB. This study indicates a threshold for TTS of 151 dB re 1 μPa 2 s -1, even lower than that indicated by Lucke et al. (2009). Comparison of TTS induced by equivalent SEL produced by exposures of different durations indicated that longer exposures at lower sound pressure levels were more effective in inducing TTS. Temporary Threshold Shifts ( TTS) in Harbour Seals

An extensive exploration of TTS in harbour seal have recently been reported by Kastelein et al. (2012a). They exposed two harbour seals to octave band white noise centred at 4 kHz and measured changes in threshold at 4 kHz. Exposures were made up from a combination of three sound pressure levels 124, 136 and 148 dB re 1 μPa and six different durations from 7.5 to 240 minutes. The thresholds for significant TTS were at 170 dB re 1 μPa 2 s -1 for longer exposures (60 minutes at 136 dB) and 178 dB re 1 μPa 2 s -1 for shorter exposures (148 dB over 15 minutes). These are somewhat lower than the value of 183 dB re 1 μPa 2 s -1 for onset of TTS in harbour seals used in Southall et al. (2007) which were derived from results reported by Kastak et al. (2005). They too found that longer exposures were more effective in eliciting TTS than were shorter exposures of more intense sounds with the same SEL. Observations of the Relative Effects of Intensity and Duration on TTS Usually, a longer exposure to sound at a particular intensity will result in a greater TTS. The method of Southall et al. (2007) assumed that sound intensity and duration contribute equally to TTS and that a dose of a particular amount of acoustic energy would have the same effect, however it is administered. As we have seen above, several marine mammal TTS studies do not support this assumption of a simple exchange between level and duration.

A study designed specifically to explore this phenomenon was reported in Mooney et al. (2009a). They exposed a male bottlenose dolphin to octave band noise (4-8 kHz) at a range of sound pressure levels and over durations from 2 to 30 minutes. When sound exposure level was held constant they found that the size of TTS induced increased with the exposure duration suggesting that duration had a stronger effect than sound pressure level on TTS. Mooney et al. (2009a) fitted a model to their results which suggested a logarithmic relationship between duration, SPL and TTS development. At this stage however, it is not clear whether this is a general relationship that could be applied to other species and other types of fatiguing noise or one that is specific to this particular case.

Its seems then that the equal energy assumption incorporated in the Southall et al. method may be an over-simplification, but it is as yet not clear what should replace it. What is evident though is that TTS thresholds based on observations of short-term exposure are likely to underestimate the levels of TTS induced by long-term exposures. Much of the early work on marine mammal TTS in the USA involved short exposures to high intensity fatiguing noise, for example 1 second pure tones (Schlundt et al., 2000) and short intense water gun pulses (Finneran et al., 2000). Thus, the existing thresholds proposed by Southall et al. (2007) may be far less precautionary than these authors intended.

In the context of ADD exposures these findings might indicate that we should perhaps be more concerned about the effects of noise on marine mammals that remain within a few hundred metres of an ADD for extended periods than the effects of occasional short, high level exposures.

Mooney et al. (2009a) also point out that other properties of noise, such as its acoustic characteristics and duty cycle may also influence how effectively noise induces TTS. Given these complexities and resulting uncertainties, it will always be safest to make an assessment of hearing damage risk using data from the species and noise of concern with levels and patterns of exposures that closely match those likely to be encountered in real life situations.

A complex relationship between noise and hearing damage in marine mammals should surely be expected. The auditory system of marine mammals is the product of millions of years of evolution, which have seen it adapt to function as animals moved, over evolutionary time, between the hugely different acoustic media of air and water at least five times. It is hardly unexpected then, that this exquisitely sensitive but idiosyncratic organ, should be vulnerable to being damaged by intense sound in a variety of ways, and that relationships between acoustic dose and hearing impacts are likely to be complex and non-linear.

KG No. Knowledge Gap
51 Equal energy hypothesis for TTS does not seem to hold in all circumstances. A universal relationship between signal duration, intensity and hearing impact has not yet been described. Species Specific Frequency Weighting

It is likely that species will be more vulnerable to TTS from noise at frequencies to which they are most sensitive. However, simple audiograms do not provide a good indication of what the frequency weighting function should be. In humans, plots of perceived loudness for higher intensity sounds have been used to derive frequency dependent weighting filters for use in assessments of noise impacts. Measurements of a dolphin's perception of "loudness" for sounds of different frequencies and intensities have recently been made from a bottlenose dolphin and reported by Finneran (2012). For these experiments, a dolphin was trained to indicate to its trainers which of two sounds it perceived as being louder. Data collected over a long series of trials were combined to generate a series of equal loudness plots for sounds of the same intensity at different frequencies (similar to the human derived Fletcher Munson plots shown in Figure 13). These plots indicated that, much as has been shown in humans, equal loudness contours were flatter for more intense sounds. If more data of this type were to be obtained from a wider range of species they could provide the basis for more reliable frequency weighting functions for species groups, though it might be noted that even in humans the extent to which equal loudness contours improve predictions of hearing loss is still an area of research. Ron Kastelein's group, working in Holland, are exploring the potential of a different approach, which had previously been used to measure loudness contours in monkeys by Stebbins (1966). This makes use of the fact that response time to a signal correlates well with its perceived loudness. If animals are trained to respond to a signal in a specific way then their response latency can be measured using video analysis. This method does not require the same degree of training and should therefore yield results much more quickly, allowing a broader range of species to be tested. An initial study tested the feasibility of the techniques with a harbour seal using only low level signals close to the animal's hearing threshold (Kastelein et al., 2011). Results from an extensive series of trials with a harbour porpoise to derive equal loudness plots for sounds of differing frequencies and intensities are expected soon. Summary

Southall et al. (2007) remains the most comprehensive attempt to provide a consensus for science-based thresholds for hearing damage. It has been a helpful document providing criteria and guidance that have been widely adopted, including by regulators in the UK and Scotland. However, it is clear that some revision is needed to incorporate both new measurements of sound levels causing TTS and new research findings that question some fundamental assumptions within the Southall et al. (2007) model, for example the equal energy rule. Most of the new studies have indicated that marine mammals are more vulnerable to hearing damage than was assumed by Southall et al., thus, the assessments of risk provided by applying the Southall et al. procedure should not be considered precautionary.

7.2.6 Biological Significance of Hearing Damage

Despite its apparent complexity, reduction of hearing sensitivity is a straight-forward sensory phenomenon; understanding the biological significance of a threshold shift, however, is less clear-cut. It is widely accepted that acoustics is the primary sensory modality for long-range underwater sensing in marine mammals. They make use of the vocalisations of conspecifics to maintain contact and to communicate; odontocetes detect the faint echoes of their echolocation signals to navigate and hunt prey and they attend to the myriad of passive acoustic cues in the environment (the acoustic scene) to provide information on prey location, to detect predators and for orientation.

A loss in sensitivity means an animal is able to hear fewer of the quieter sounds and this equates to a reduced range over which they can detect acoustic cues. The scaling between changing threshold and number of cues within range will vary depending on propagation conditions. If we assume spherical spreading, however, then an increase in threshold of 6 dB would equate to a halving of detection range (Mohl, 1981). If acoustic sound sources ( e.g. prey items) are distributed randomly in 3-dimensional space then this halving in detection range is equivalent to an 8 times (2 3) reduction in the number of prey items that are within detection range. Thus, if the detection of quiet signals is biologically important, the effect of even a small shift in threshold could be very substantial. A small degree of hearing damage can also degrade frequency discrimination, and thereby reduce the ability to classify sounds (Götz and Janik, 2013).

When the fatiguing noise has a restricted frequency band, TTS appears to be most pronounced over a frequency range centred at about half an octave above that of the fatiguing noise. Thus, the effect on detection range will be greatest for signals at these frequencies. For a species such as the harbour porpoise, which produces signals in a narrow frequency band, the effects of changes in detection of these signals, on the efficiency of echolocation or communication for example, may be limited to only those threshold shifts affecting hearing sensitivity in that narrow band.

7.2.7 Commercial Significance of Hearing Damage

Hearing damage is not solely a concern from an animal welfare perspective, but also because it is likely to reduce the effectiveness of ADDs themselves as the depredating animals become decreasingly sensitive within the targeted hearing range (Götz and Janik, 2013). Seals rely to some extent on passive acoustic detection of prey items and loss or reduction of the ability to discriminate frequencies and classify sounds could lead to increased reliance on predictable and 'low cost' prey, including farmed fish (Götz and Janik, 2013).

7.2.8 Likelihood of ADDs Causing Hearing Damage

Gordon and Northridge (2002) attempted to assess risks of hearing damage to marine mammals from ADDs by extrapolating from human damage risk criteria. However, we suggest that the process outlined in Southall et al. (2007) and new data on threshold shifts in marine mammals that have been published since then (see section 7.2.5), should supersede those efforts.

Lepper et al. (In Review) provides an exhaustive analysis of the source levels of ADDs used at Scottish salmon farms and the propagation losses (especially within 500m) predicted by appropriate propagation models for a range of typical Scottish salmon farm sites. They compared the "sound fields" that would be expected from these with the thresholds for auditory damage sound exposure from Southall et al. (2007) and from the more recent findings of Lucke et al. (2009).

As discussed above, Southall et al. (2007) proposed two sets of thresholds beyond which they predicted the onset of permanent threshold shift: one for the maximum instantaneous exposure to un-weighted peak pressure levels and a second based on cumulated sound exposure levels ( SEL) to sound after appropriate species specific frequency weighting filters had been applied.

Assessing the likelihood of exceeding the first of these, the sound pressure level threshold, is straight forward because it is an instantaneous measure. The threshold provided by Southall et al., for seals, is 218 dB re 1µPa while Lepper et al.'s interpretation of Lucke et al. (2009)'s findings in this context suggested a threshold for porpoise of 206 dB re 1 µPa. Published source levels for ADDs are usually provided as root mean square ( RMS) levels and peak levels may be somewhat higher than this. Even so it is unlikely that any ADD peak levels would ever reach these thresholds. Thus, even at 1m range, accepted instantaneous injury ( PTS) exposure thresholds are unlikely to be reached.

Assessing the likelihood of exceeding the SEL based thresholds, however, is more complex. This is because SEL is a measure of cumulative exposure over a period of many hours and is therefore a function of the sound field around the device, its duty cycle (which are fairly easy to predict for ADDs that are activated continuously) and also the animals' movements within this field over this time period. Such movements could of course be highly variable and, unfortunately, this key information has never been measured. There are observations, however, supporting instances of the two extremes. For example, seals have been repeatedly sighted within ca. 50m of fish farms with operating devices over periods of days (Northridge et al., 2013), while porpoises have been observed moving quickly away from ADDs as soon as they are activated (Brandt et al., 2012b; Johnston, 2002).

Lepper et al. (In Review) considered the simplest scenario, that of an animal remaining stationary at a particular range and calculated the time to reach threshold for injury to such an animal at different ranges out to 500 m. They repeated these scenarios for both seals and porpoises for the Airmar, Ace Aquatec and Terecos devices. Their results are summarised in Table 5.

For the Airmar device, a seal at 100 m was predicted exceed threshold after 3.3 hours for a single transducer but within 1.6 or 1.1 hours if two or three devices were deployed (as is often the case at Scottish aquaculture sites). For a porpoise exposed to a single Airmar, threshold for injury would be exceeded at 200 m in 2.8 hours while for a site with three Airmar devices the time to exceed threshold at 300 m would be ca. 1 hour.

For a Terecos device, exposure to a seal would exceed the threshold if it remained at 100 m for around 9 hours or spent 24 hours at 200 m. For porpoise the exposure threshold at 100 m was exceeded after 2.5 hours while the safe range for 24 hour exposure was beyond 500 m.

With an Ace Aquatec ADD, a seal at 100 m would receive a dose exceeding the threshold after 3 hours and the threshold range for a 24 hour exposure would be 350 m.

As expected, harbour porpoises are substantially more vulnerable to damage than seals and farms which utilise several Airmar units at the same time seem to pose the greatest theoretical risk. However, seals are known to spend extended periods close to fish farms and, presumably, seals that are attempting to get close to nets to attack salmon must be exposed to much higher levels than these simulations with static animals allow.

Clearly, the "movement model" employed here is an unrealistic one. Data on the movement of animals in the vicinity of fish farms or other sites with operating ADDs remain the largest source of uncertainty. Better data on this would certainly allow more complex and realistic modelling but, if movement patterns are highly variable between individuals, new data on movements from a few individuals may do little to clarify the real risks. What is evident, however, is that there does seems to be a real danger that the hearing of marine mammals can be permanently damaged by ADDs and that seals, which appear to be motivated to spend extended periods close to fish farms, even when ADDs are active, may be particularly vulnerable. Whether or not there may be similar motivations for seals or cetaceans to remain near to operational ADDs in other circumstances, for example around construction or turbine sites, remains to be determined.

KG No. Knowledge Gap
52 Realistic movement models for animals (particularly porpoises and seals) in the vicinity of ADDs.
53 Hearing damage caused by ADDs on wild populations of seals in particular seems possible, but has not yet been proven.

Table 5 Plots of time to exceed thresholds for injury based on Southall et al., (2007) for seals and porpoise with three different ADD devices. Propagation conditions assume a sandy bottom with a range of water depths from 20 to 120m. The red line indicates 24 hours.

Table 5: Plots of time to exceed thresholds for injury based on Southall

7.3 Disturbance, Exclusion and Behavioural Effects on Non-Target Species

Sound provides the principal modality for long range detection in marine mammals and these animals are known to respond to some acoustic signals at very low received levels. For example, gray whales responded negatively to playbacks of killer whale calls at just perceptible levels (Cummings and Thompson, 1971) and an aversive response might be expected where a signal is perceived by the receiver as having similar characteristics. Similar responses have been observed to other signals that might be associated with threats. For example, beluga whales were shown to respond negatively to ice breakers at very substantial ranges (25 - 50 km) and it is likely that the vessels were only just audible to these animals at these ranges (Cosens and Dueck, 1993; Finley, 1990; Richardson et al., 1995). Negative effects of disturbance of marine mammals include both disruption of biologically important behaviour and exclusion from habitat.

Acoustic deterrent devices used in the Scottish aquaculture industry are generally intended to have a strong behavioural effect on seals. However, they can also have unintended impacts on the behaviour of cetaceans, many of which have more sensitive hearing than pinnipeds. Here we review studies of the effects of the types of ADDs on the behaviour of non-target cetacean species.

7.3.1 Harbour Porpoises

One of the earliest, and still one of the most comprehensive investigations on the effects of ADDs on harbour porpoises was carried out in 1994 by the Department of Fisheries and Oceans, Canada, in British Colombia. Results were presented in both a research report (Olesiuk et al., 1995) and a peer reviewed paper (Olesiuk et al., 2002). The field site for this study was in the Broughton Archipelago, an area of sheltered and enclosed deep water, not unlike many fish farm sites on the west coast of Scotland. Olesiuk and colleagues used a floating platform to establish an observation position with a 6.4 m eye-height close to an existing salmon farm site. The study took place over an 18 week period (29 th June to 31 st October 1994) during which observers made systematic scans with the naked eye and binoculars and measured ranges to sighted porpoises using a combination of reticule binoculars and known land marks. An Airmar ADD array was established about 80m offshore from the observation station and could be turned on or off under the control of the research team. The study period was divided into three six week sampling periods. In each of these the first three week period was a control, with no ADD, while for the second three weeks the Airmar ADD was active. This design, with its repeated alternating trials, helped to control for seasonal changes in porpoise density and sighting conditions.

The results were clear and striking. As soon as the ADD was activated a substantial and significant decline in porpoise sighting rates was evident. The mean sightings per scan fell to between 1.7% and 3.7% of control values for scans with the naked eye and binoculars respectively. Porpoises were also visible for shorter periods with the number of sightings during the tracking of a porpoise pod falling from around 13 per track to around 1.5, suggesting that animals that were in the area were spending less time there. No porpoises were seen within 200 m of the device when it was active and the proportion seen at ranges of 200 to 399, 500 to 599 m, 600 to 2499 m and 2500, 3500 m were 0.2%, 1.4%, 2.5%, 3.3% and 8.1% respectively of those seen in the same zones during control periods. The local topography meant that 3500 m was the maximum range at which observations could be made and it is clear that this is unlikely to represent the full extent of these effects. There was no sign of habituation or a reduction in the size of effects over the three week duration of any of the trials. However, sighting rates recovered within a few days of the ADD being switched off.

Another study, conducted on the Canadian East Coast, from the island of Grand Manan in the Bay of Fundy, used a different approach that aimed to measure responses of individual animals. Johnston (2002) established a tracking station on an elevated location (eye-height ca. 30m) and used a theodolite to fix the position of porpoises and track their movements. An Airmar dB Plus II ADD was deployed from a boat about 450 m offshore. On each observation day, the ADD was either turned on or was left inactive. ADD state was determined randomly and the observation team ashore were not informed of the treatment. Experiments lasted for 2 hours and only one was conducted a day. Observations were restricted to days with good visibility and a sea state of one or less. Data were collected on 16 observation days: 9 days with the ADD active and 7 controls. The observation team searched with binoculars and recorded the locations of all sightings within 1500 m of the ADD using a theodolite and group movements were tracked as far as possible. There were substantial differences in detection rates between ADD active and inactive days. When ADDs were active the mean detection rate was reduced to 0.22 ( SD 0.44) porpoise sightings per scan from a mean of 2.91 ( SD 1.29) on control days. There was also evidence in the data of animals leaving the site soon after the ADD was activated. Porpoise sightings within 1500 m were lower in the 5 minutes after ADD activation but not significantly so. Low numbers were seen during the first 30 minutes of scanning and no sightings at all were made in the remaining 1.5 hours of experimental exposure.

Porpoise movements were tracked wherever possible, with a total of 69 tracks being recorded: 60 during control periods and 9 when the device was active. It was clear from these data that porpoises maintained a greater range from the ADD when it was active. The mean closest approach of tracked animals was 364 m ( SD 261 m) on control days but 991 m ( SD 302 m) on days when the ADD was active. No porpoises were observed within 645 m of the ADD when it was active. By applying an appropriate propagation model to measured source levels for their ADD, Johnston calculated that the received level at 645 m would have been 128 dB re 1 µPa.

Research in Scottish waters to explore the effects of ADDs on porpoise densities in fish farming areas was presented by Northridge et al. (2010). This work differed from the earlier Canadian studies in two respects. In the first place the study sites were close to operating fish farms with no or limited experimental control over when ADDs were active or inactive. Secondly, the data on porpoise presence and relative densities were collected acoustically, using both static passive acoustic devices ( T-PODs and C-PODs, Chelonia Research Ltd) moored at different ranges from fish farms with ADD devices and with simple towed hydrophone arrays.

PODs were used to collect data at two different salmon farm sites, both using Airmar ADDs on the west coast of Scotland; one was at Fiunary in the Sound of Mull and the other at Laga Bay in Loch Sunart. At Fiunary, PODs were deployed at monitoring stations with similar water depths and distance from the shore at ranges of 200, 500, 1000, 1500 and 3000 m from the fish farm site. PODs were deployed nearly continuously for over five months. For the last two months of monitoring the fish farm had been harvested and no ADD was present. At the Laga Bay site PODs were deployed at monitoring stations at distances of 240, 1100, 1700, 3000 and 8000 m from the cages with ADDs. Here the PODS were deployed for 23 days while ADDs were active. After this the ADD was turned off and after three weeks of ADD inactivity the PODs were redeployed for seven weeks. The trial then had to be abandoned because seal activity at the site resulted in the farm manager wishing to resume use of the ADDs.

The number of porpoise click train detection positive minutes ( DPM) per day was used as an index of porpoise density. Changes in detection rate with distance from active farm sites were less clear cut than the changes in sighting rates reported by Olesiuk et al. (2002) and Johnston (2002). Complete exclusion was not evident even at the closest monitoring sites and substantial inter-site differences in detection rates, which were likely due to habitat factors, tended to obscure effects of range to ADDs. Indeed at one farm location the POD which was closest to the ADD had the highest detection rate overall, probably because it was adjacent to deeper water. However, significant increases in detection rates were evident in the data after the ADDs had been turned off. At the Laga Bay site DPMs per day increased by factors of 7, 4 and 9 times at monitoring stations at distances of 200, 1100 and 4000 m respectively. The PODs at the other stations at this site were either lost or malfunctioned. Sound levels in the ADDs main frequency band ( ca. 10 kHz) measured at ranges of 240, 1100 and 8000 m were 146, 128 and 105 dB re 1 µPa.

The use of static acoustic loggers for this work allowed monitoring to extend over several months and to continue 24 hours a day where previous studies had been limited to daylight hours. While the data show an effect of ADDs with no sign of it being reduced at a range of 4000 m, these results seem less dramatic than those reported by Olesiuk et al. (2002) and Johnston (2002). The more opportunistic approach adopted, which involved collecting data around real operating farm sites, provided a messier, less controlled dataset, but had the advantage of being more representative of real-world situations. Porpoises in this area are exposed to ADDs from a range of fish farm sites throughout their home range and it is likely that animals will not have been naïve to these signals. It is possible that this resulted in a degree of reduced responsiveness. It is notable, however, that even considering this level of long-term exposure, full habituation had not occurred.

Northridge et al. (2010), like the earlier Canadian studies, investigated responses to just one type of ADD, the Airmar dB II. However, at least four different makes of ADDs are used by the Scottish salmon farming industry and, according to Northridge et al. (2010), nearly half (42%) of sites were using Terecos devices. Northridge et al. (2013) used PODs in a similar manner to the studies described above, to investigate porpoise responses to a Terecos device deployed at a fish farm site in Loch Hourn. Nine PODs were deployed at matched monitoring sites at ranges between 300 and 4500 m. PODs collected data for 65 days during which time the Terecos ADD was alternately either active or inactive following an approximate seven day cycle. Overall, there was no significant difference in detection rate when the ADD was active. Detection rates were reduced, though not significantly, at the four closest sites, which were all within 1000 m. Terecos ADDs are known to have a lower acoustic power output than the Airmar dB Plus II (Lepper et al., 2004), see Table 2. The Airmar system at Laga Bay was measured with a calibrated hydrophone to be 185 dB 1 µPa, higher than the previously reported Terecos SPL of 179 dB re 1 µPa ( RMS). These source level differences may go some way to explaining the lack of a pronounced response to the Terecos. As far as we are aware, this is the only trial of the effects of Terecos devices on harbour porpoises. The work should be repeated with a more complete set of trials, but if this finding proves to be robust, there would be a strong case for preferring the Terecos to some other ADD types on the grounds that its effects on harbour porpoise densities seems to be minor or non-existent.

KG No. Knowledge Gap
54 Is the Terecos device consistently less aversive to harbour porpoises than other ADDs?

Work to develop an ADD which would be effective in deterring seals while having a minimal effect on harbour porpoises and other cetaceans is described by Götz (2008) and also summarised in section 1.3 of this report. To make the device aversive to seals, psycho-physiological research was reviewed and these findings were used to design a signal which would induce a startle reflex in seals (Gotz and Janik, 2010). In order to minimise the effects of these sounds on odontocetes, signals were used of relatively low frequency, to which seals were more sensitive than porpoises and dolphins. Götz (2008) reported field trials to test porpoise responses to a prototype system. Observations were made, and animals were tracked, from the shore using a similar approach to that of Johnston (2002). Porpoise sighting rates were not significantly lower when the candidate ADD signal was being broadcast and neither were ranges of closest approach any greater. In fact, porpoises were observed as close as 8m from the speaker broadcasting the signal at full level. Commercial ADDs using the same signal type are now under development. If porpoises and other small cetaceans show the same minimal level of reaction to signals from these devices as they did to the experimental playbacks described above, it would provide a strong basis for the use of these new ADDs at Scottish salmon farm sites, assuming they are seen to be at least as effective as more established devices in reducing depredation (something which itself is unquantified).

A project recently conducted in the Baltic and North Sea, designed to investigate how ADDs could be used as an aversive signal for mitigating potentially dangerous activities such as pile driving and reported by Brandt et al. (2012b), is the most recent sizeable piece of research measuring the effects of ADD signals on harbour porpoise. As the intention of this work was to investigate ADDs for deterring porpoises from pile-driving operations, the methodology is described more fully in section of this report. In this case, the ADD being tested was a Lofitech seal scarer and fieldwork was conducted in two contrasting locations - an offshore site in the North Sea and an inshore site in the Baltic close to an elevated onshore observation location. Acoustic detection rates, collected by an array of 16 PODs at the offshore site, were compared before, during and after active ADD deployments. There was a clear and dramatic reduction in detections when the ADD was active. Porpoise detections were almost completely absent at the zero range POD and even at 7500 m detection rates were around 96% lower during broadcast trials. Received sound levels from the ADD at 7500 m were estimated to be 115 dB re 1 µPa. A visual aerial survey of a 30x30 km survey block, centred on the playback location, was conducted before and during an ADD transmission trial. Results from this survey were compelling and also provide strong evidence that changes in POD detection rates were really indicative of porpoises leaving the area, rather than merely a change in acoustic behaviour. Sighting rates fell by 86% during transmission periods. In the pre-transmission survey, nine porpoises were sighted within 7500 m of the device location while during transmission there was only one sighting in this area, at a range of 6300 m. This was the closest observed approach during transmission.

The experiments conducted in the Baltic (reported in Brandt et al., 2013) complimented the offshore trials and were useful in revealing the behaviour of individual animals in an inshore region. They showed that porpoises responded as soon as transmissions commenced and that, even though the experimental protocols meant that porpoise groups were usually being tracked at the time transmissions started, the animals typically "disappeared" during transmissions which was taken as an indication of a very pronounced disturbance effect. During a series of six trials designed to measure responses of animals at ranges of greater than 1500 m there was no obvious avoidance response at ranges of 2100 to 3300 m. Propagation loss at this site was found to be much greater at this inshore location than at the North Sea site. Brandt et al. (2012b) suggested that the Lofitech device has a significant disturbance effect at sound levels above 119 dB re 1µPa ( RMS), and that near complete deterrence occurred at received levels of greater than 132 dB re 1µPa ( RMS).

Brandt et al.'s studies provide the first data on the effects of Lofitech devices on harbour porpoises of which we are aware. The range, in the offshore area, over which such dramatic responses are evident, is striking. There was a 96% reduction in detection rate at a range of 7.5 km and no indication that this was the maximum range at which effects would be evident in offshore conditions. The Lofitech produces a narrow band high frequency signal rather similar to that of the Airmar (see Figure 3). It is perhaps to be expected, therefore, that it would also cause a high level of disturbance, seemingly over even greater ranges than has been reported for the Airmar (Olesiuk et al., 2002).

Another relevant piece of work that might be mentioned is a study of harbour porpoise presence at aquaculture sites reported by Haarr et al. (2009). They monitored for the presence of harbour porpoises at salmon farm sites in the Canadian Bay of Fundy over two summer seasons using both shore based visual observation and static acoustic monitoring devices ( T-PODs). ADDs were not in use at these sites at this time. However, the occurrence of other forms of disturbance such as large and small boat traffic and net cleaning were noted. It was clear that for most of the time, porpoises were not avoiding the farm sites. Mothers and calves in particular seemed to prefer to be among the cages at these sites and it was suggested the cages might provide some shelter and protection. POD data indicated that there were more detections at night than during the day. Vessel traffic and activities such as cage cleaning caused short-term disturbance leading to lower densities within the immediate area. These results suggest that, in the absence of ADDs, porpoises will make full use of fish farm sites; there were even indications that some components of the population might prefer them because they offered some protection and possibly feeding opportunities.

7.3.2 Killer Whales

Salmon farms sites in British Colombia are located within the home range of one of the world's best known populations of resident killer whales. Nearly every whale encountered in this area can be individually recognised and the life histories of most individual whales have been followed since the 1970s. Several research groups are continuously engaged in studying and monitoring the population here. Morton and Symonds (2002) reported on changes in killer whale detection rates and residence patterns at two locations: the Broughton Archipelago, a salmon farming area, and the mouth of Johnstone Strait, some 25km away, over a period of 15 years between 1985 and 2000. Airmar ADDs were introduced at four farm sites in the Broughton Archipelago in 1993 and remained active there until 1999. At Johnstone Strait, by contrast, no ADDs were active. Both sites were very intensively monitored using a combination of cabled hydrophone systems, which allowed constant real time shore-side monitoring, a system of VHF communication with experienced observers and water users, and a combination of both land-based and boat-based searches. Killer whale pods in this area can be identified reliably by their call types so the acoustic monitoring provided a particularly complete information set. The authors believed that it would be unlikely that whales would pass through either site without being detected. Morton and Symonds (2002) reported that while whale presence (the proportion of days in which killer whales were detected in the area) remained stable in the Johnstone Strait location, their presence in the Broughton Archipelago fell substantially (by a factor of over 3) and significantly during the five years when ADDs were active there. Whale occurrence, however, returned to levels that were not significantly different to pre-exposure values once the ADD had been removed. This pattern was particularly clear when the occurrence of just the resident (salmon eating) killer whale pods was analysed independently of the presence of mammal eating "transient" whales. Analysis of photo-identification data showed that the same pods were using the area throughout the study, so these changes in density were unlikely to be due to any larger scale population changes. There were also no indications of changes in food (salmon) availability when killer whale occurrence was low. Indeed the high level of escapement from some farms led to a rise in salmon availability in the Broughton Archipelago area at that time.

Several aspects of this study are worth commenting on. In the first place it is intriguing to find that killer whales should be so strongly affected by ADDs. Research summarised above ( section 7.3.1), has shown the harbour porpoise to be particularly vulnerable to disturbance. For many researchers this is not unexpected. Harbour porpoises are known to be shy animals and easily scared, and are described as neophobic (for example, they rarely interact with boats). They are predated upon by killer whales and also attacked by bottlenose dolphins (Ross and Wilson, 1996) and crypsis and avoidance of novel sounds may well be their anti-predator strategies. Killer whales, by contrast, are large robust animals with no known predators, which are not afraid of vessels and often approach them. Another interesting finding from this research is the extended period, some six years, over which avoidance and partial exclusion was demonstrated. This indicates an absence of significant accommodation or habituation over that period by individuals, in a population of known resident animals. Morton and Symonds (2002) collected photo- ID data that showed the same animals were repeatedly observed in the area, throughout the period of ADD activity. They would therefore have been repeatedly exposed to ADD signals and some degree of habituation might thus have been expected, but was not observed.

7.3.3 Pacific White-Sided Dolphins

Morton (2000) reported on the occurrence of Pacific white-sided dolphins ( Lagenorhyncus obliquidens) in the waters around the Broughton Archipelago between October 1984 and December 1989. This was the same study area as for the study on killer whales (Morton and Symonds, 2002) reported above, and the two studies also overlapped in time. During the first years of this study measures of occurrence of dolphins increased dramatically, increasing from being present on only 0.4% of days in 1984 to being recorded on 19% of days in 1994. Morton (2000) suggested that rising water temperature and increases in prey abundance may have led to these changes. In 1994, ADDs were introduced at the local fish farm sites and dolphin occurrence subsequently fell, with dolphins being present on only 2% of days in 1998. Data after 1998, when ADDs were removed from the fish farms in the area, were not included by Morton (2000). However, Alexandra Morton reported, as a pers. comm. in Gordon and Northridge (2002), that rates of occurrence increased in the first two years after ADDs were removed. In 2001, however, sightings of white sided dolphins and several other species were reduced, which might have been attributable to unusual oceanographic conditions in that year.

This study, like that of Morton and Symonds (2002), was an opportunistic one. There are indications that ADDs may have had an effect on the frequency of white-sided dolphin sightings. The fact that the population distribution appeared to be quite dynamic (for other reasons), as well as the limited time series available for analysis and the lack of a control area, all combine to make the case less clear than that made for killer whales and porpoises.

7.3.4 Baleen Whales

There have been no dedicated studies investigating the effects of aquaculture ADDs on baleen whales. However, a reduction in the number of sightings of humpback, gray and minke whales in the Broughton Archipelago over the period when salmon farms in the area were using Airmar ADD devices, followed by a substantial recovery in sighting rates after ADD use was halted in 1999, was reported by Morton (1997). Sightings rates were very also low in 1999 but this seems to be attributable to oceanographic conditions that provided exceptionally good feeding conditions in another nearby location (Morton, pers. comm.).

Other opportunistic observations suggesting that humpback whales vacated areas where high intensity ADDs were in operation were provided by Lien et al. (1995).

The minke whale is the only baleen whale routinely encountered in Scottish inshore waters and is likely to come within range of aquaculture ADDs. In the winter of 1993 a minke whale "took up residence" for at least ten weeks in Loch Grimshader on the Island of Lewis. This small sea loch had a shallow entrance and it was not clear whether this animal had become embayed or was choosing to stay there to feed on dense fish schools which were present in the loch at the time. A team from the Mull-based Sea Life Surveys research group collected behavioural data from this animal to assess its well-being. A fish farm within the Loch was equipped with an Airmar dB plus II ADD. This had not been active when the whale entered the loch. However, because the farm felt they might need to use the device in the near future it was decided to turn it on for a 24 hour trial during which time the animal's behaviour would be monitored allowing any responses to be assessed. When the ADD was turned on the animal changed from a 'resting' to a 'feeding' mode of behaviour but continued to make full use of the loch. In fact, at one stage it seemed to be actually investigating the device (Fairbairns et al., 1994; Gordon and Northridge, 2002). The indications from this very short trial are that the whale could detect the device but that its behaviour did not change in any manner that was an immediate cause for concern.

Götz and Janik (2013) recommend the use of lower frequency signals than are currently employed by ADDs in order to minimise impacts on odontocetes (1 - 2 kHz). They stress, however, that impacts on baleen whales (and hearing specialist fish species) should be investigated before such devices be used commercially.

KG No. Knowledge Gap
55 Responses of baleen whales (and several other less studied species) to ADDs in Scotland are not clear at present.

7.3.5 Non-Target Species Discussion

Much of the evidence for displacement of non-target marine mammals by ADDs is drawn from one area (the Broughton Archipelago in Canada) and involved a single type of ADD. We do not know whether animals in different areas with different motivations would necessarily be affected in the same way, nor do we fully understand the likely different responses to different devices. Recent experiments with Lofitech ADDs, for example, suggest an even greater degree of displacement of porpoises with these devices (Brandt et al., 2012c), whereas another experiment using a Terecos device seemed to induce little if any response beyond in Loch Hourn (Northridge et al., 2013). Our lack of understanding or ability to predict the behavioural responses of a range of species to different devices in a range of contexts is another significant knowledge deficiency.

7.4 Summary

Most published reports have shown significant and long lasting behavioural responses from cetaceans to ADDs. Harbour porpoises seem to be particularly vulnerable, with good evidence that densities can be reduced substantially at ranges of many kilometres for at least two devices types in multiple locations. The majority of studies have investigated responses of animals to one particular type of ADD, the Airmar dB Plus II. Responses to other devices may be quite different. There are indications that harbour porpoises may respond even more strongly to the Lofitech seal scarer while the Terecos ADD may have much smaller impacts. From a Scottish perspective, there is an obvious requirement to measure responses to the range of devices available to Scottish salmon farms including the newly developed "cetacean friendly" ADD (Götz, 2008).

It is clear that some, if not all, of the ADDs currently being used on Scottish salmon farms have an effect on local densities of porpoises (and possibly some other species). This raises two questions: is this likely to be of any biological significance for local cetacean populations, and how should these devices be managed and permitted under existing regulations?

The biological significance of acoustic disturbance for marine mammals is a question that has attracted the attention of both scientists and regulators over the last decade. The US National Research Council held a series of workshops to explore this question which are summarised in two publications (National Research Council, 2003; National Research Council, 2005). One outcome from the latter of these was a model or conceptual framework for the population consequences of acoustic disturbance ( PCADs), outlining the steps by which a sound might cause disturbance that could eventually result in biologically significant population consequences. This framework is outlined in Figure 14.

Figure 14 The National Academy of Science Population Consequences of Acoustic Disturbance Model ( PCADS). The model considers the stages from a sound being produced, being detected and affecting an animal's behaviour and the potential consequences of this for individuals and populations. In each box the number of "+" signs indicates how easily the parameter can be measured while the number of "+" signs next to the arrows linking the boxes indicates how reliably one set of parameters can be inferred from the earlier ones.

Figure 14

It shows a fairly straight-forward cascade of events from a sound being produced, being affected by propagation loss, being received and perceived by an animal, resulting in a change in behaviour that could cause a change to a life process that might in turn result in a change in vital rates finally translating to changes at the population level. At each stage, the number of "+" signs within the box indicates how readily the required data type can be measured while the number of "+" signs next to the arrows linking the process show how reliably one type of data can be inferred from that proceeding it. It can be seen that, according to these authors at least, all of the parameter types can be measured directly, though with varying levels of ease. The "transfer functions" used to infer one set of parameters from those preceding them in the chain are particularly poorly known however, and, in the view of these authors at least one of these steps (inferring vital rates from changes in life functions) is not possible.

A pragmatic conclusion from this framework would be that the PCAD model provides a useful conceptual framework for the process but that it is very unlikely that one can use this as a practical tool for inferring the biological significance of a particular sound type. All of the steps in the process are amenable to being measured to some extent but it will also be important to establish cause and effect. While factors such as population size or viability might be the parameters of most interest, it will be very difficult to infer that observed changes at the population level were caused by any particular type of disturbance, especially in wide-ranging and long-lived species which will be affected by many factors in a complex environment. In addition, by the time any effects could be measured at this level it would be very late for taking useful management action. Measuring processes earlier in the chain in addition to population monitoring is thus essential for both establishing cause and effect and for allowing timely management to be put in place.

Having said this, some groups are actively exploring the extent to which population consequences can be predicted from short-term disturbance. It seems that, in the case of some of the best studied marine mammal populations in areas of high disturbance, it may be possible to infer consequences at the population level. In our opinion, no populations of Scottish cetaceans are sufficiently well studied to allow this.

A simpler perspective might be to regard displacement as exclusion and consider the areas from which animals have been displaced as representing a form of habitat loss. Plots such as figures 5 and 6 could be used to explore the extent of such 'habitat loss'. This could, however, severely underestimate or overestimate the true biological significance of displacement. For example, if animals were displaced from an area, but there was a lot of 'unused' habitat outside that area in which they could feed and function just as efficiently as before, then the effects might be very minor and/or short-lived. This might occur, for example, if the population was being kept below carrying capacity by something other than overall food limitation. Alternatively, if animals are displaced into a habitat that is already 'full' then they would be competing for food and other resources with animals that were already established there. The resulting competition and disruption could lead to reduced foraging success for many more animals than just those (Gill et al., 2001; Gill and Sutherland, 2000), at least initially. Animals could, in this way, be indirectly affected by the noise causing disturbance, without having heard it themselves.

KG No. Knowledge Gap
56 Likely total extent of exclusion and disturbance of ADDs on different species.
57 Population level significance of potential exclusion of cetaceans by ADDs.
58 Population level significance of potential disturbance of cetaceans by ADDs.

For coastal locations, such as most current aquaculture sites, another situation in which displacement might be particularly harmful would be if animals were excluded from 'movement corridors' required to access large areas of suitable habitat, for example exclusion from the mouth of a sea loch might exclude access to the whole of the loch itself. Similarly, disruption in channels or at headlands might make it difficult for animals to move between habitats at either side of them.

The Habitats Directive prohibits reckless disturbance of individuals of Annex II species (which includes all cetaceans). However, derogation can be granted provided this disturbance at the individual level does not affect the status of the species concerned, does not affect local populations and it can be shown that there are no feasible alternatives to the activity of concern.


Back to top