How to Deter Seals? A Manually Triggered Acoustic Deterrent Study

Seals interact with numerous human activities in different ways around the world (Bonner 1982). Many of these interactions are viewed negatively, and methods are often sought to mitigate undesirable impacts from seals (Cummings et al. 2019). One approach uses Acoustic Deterrent Devices (ADDs), often known as “seal scarers”, to deter seals away from areas where they are unwanted. ADDs typically produce sounds loud enough to be unpleasant or potentially cause pain through high intensity sound emitted at regular or randomized intervals. Since they were first trialed in 1985 (Hawkins 1985), they have been widely used in attempts to deter seals from fish farms, and to reduce marine mammal bycatch and depredation during fishing activities (Barlow and Cameron 2003; Tixier et al. 2021). They have also been used to exclude marine mammals from areas of potentially harmful effects such as in the vicinity of pile driving during construction (Kastelein et al. 2010). Atlantic salmon (Salmo salar) are a species said to be at a crisis point due to gradual but continuous declines in abundance, which likely stem from a range of factors acting on their survival, including overexploitation, migration barriers, pollution, climate change, and predation (Dadswell et al. 2022; Gillson et al. 2022). Salmon are vulnerable to a wide range of predators in both their marine and riverine habitats. Seals, which are known to follow them upriver, may also affect populations in other ways, including disturbance during spawning. To help mitigate seal predation on salmon, continuing to investigate, develop, and test non-lethal methods of seal control, thereby improving our understanding of the effectiveness of these measures, has been identified as a priority action in the Scottish Government's Wild Salmon Strategy Implementation Plan. Various methods have been trialed or discussed (see review by Thompson et al. 2021), including the use of ADDs in rivers, yet the efficacy of that approach is not clear. A number of scientific studies have demonstrated success in rivers (Yurk and Trites 2000; Graham et al. 2009; Harris 2011; McKeegan et al. 2024). However, as seals are regularly observed to ignore devices in Scottish rivers (and elsewhere), anecdotal evidence from local stakeholders suggests that deterrents deployed in this manner may offer little or no benefit. Furthermore, criticism exists over the potential impacts of ADDs on non-target species, such as otters (Lutra lutra), beavers (Castor fiber), or, when closer to the coast, cetaceans (Benjamins et al. 2018; Findlay et al. 2021). There are many hypotheses as to why ADDs may perform poorly and why some seals appear to ignore them. These include: the presence of deaf seals, individual variation in responsiveness (e.g., determined individuals), habituation to the acoustic signal and consequent ignoring, poorly installed or maintained equipment, dinner bell effects (where seals learn to associate the sound of the ADD with an easily accessible food source, see Bordino et al. 2002), and individual system characteristics. Götz and Janik (2010) showed that captive harbor seals (Phoca vitulina) habituated quickly to aversive sounds when food was available, while Jacobs and Terhune (2002) observed harbor seals already familiar with ADDs showing no behavioral response when one was switched on in their vicinity. Harris (2011) suggested that individual seals either quickly learned to ignore deterrents in rivers (in a matter of hours or days) or consistently appeared, over successive years, not to pass upriver of acoustic deterrents. Therefore, populations may consist of those that do habituate and those that do not, and ultimately the proportion of individuals of both persuasion in the local population might determine the level of success of deterrents. Acoustic signal characteristics, specifically those which trigger the startle reflex, have been highlighted as a potentially important factor in seal deterrence (Götz and Janik 2011). Acoustic stimuli with a rapid onset (15–20 ms) and received amplitude above a certain threshold can elicit the startle reflex in seals (Koch and Schnitzler 1997; Fleshler 1965; Götz and Janik 2013). Experiments on captive gray seals (Halichoerus grypus) found that the average pure tone startle threshold at 1 kHz is about 159 dB re 1 μPa, reflecting a sensation level approximately 93 dB above the hearing threshold, and that, in contrast to other deterrent sounds, startle sounds elicit an increasing avoidance response in seals with repeated exposures (Götz and Janik 2011). Harbor and gray seal hearing thresholds are similar (Götz and Janik 2010) and therefore ADD signals can be tailored to target their peak sensitivity and maximize the likely response. One method that has been suggested that may improve efficacy is an effective triggering system to target ADD use only when seals are present and at close range (Thompson et al. 2021). A triggered deterrent should limit any impacts on non-target species, reduce power requirements, and lessen the likelihood of habituation by only emitting sound when seals are at a range likely to induce the startle reflex. Research until now has focused primarily on the use and effectiveness of devices that are designed to transmit continuously (albeit on a duty cycle). More information is therefore required to compare the efficacy of triggered vs. continuous ADDs. In this study, we aimed to determine whether seals that are traveling upriver change their behavior or stop traveling upriver when exposed to a triggered acoustic deterrent at close range, and ultimately to determine whether they pass the location of the deterrent when it is active. A behavioral response study was carried out at a site on the River North Esk, Angus, Scotland, whereby acoustic deterrents were manually triggered when seals reached a specific location. The deterrents were triggered at ranges anticipated to elicit a startle reflex in seals (determined by sound measurements made before the start of the study) and cause them to move away from the deterrents and return downriver. Land-based observations were made to detect seals moving upriver, and a sonar was used to provide verification of any seals passing the deterrent. Reactions of seals were observed, and behavior was recorded by an experienced observer. The study location was chosen for its characteristics; shallow water and good line of sight along an ~850 m stretch of the river (the visible area—Figure 1) that make for easy seal detection by riverbank observers. Based on previous data recorded by Harris et al. (2022) the sighting probability for seals in the area defined as the study area, distinct from the visible area (Figure 1), is close to 1 during daylight hours for seals moving upriver. In the study area, the width of the river ranges between 33 m and 45 m, and the depth fluctuates with precipitation runoff and spring tides. A ford in the river close to the downstream margin of the study area (Figure 1) is uncovered at low tides, reducing the likelihood of seals traversing upriver outside spring high tide periods (Kinnaber Fishery staff, pers. comm.). At typical river levels during spring high tides, water depth at most locations in the study area is < 2 m. Figure 1 shows the location of the study area and the larger area visible to the observer relative to the beach and river mouth. The observation location was behind riverbank foliage that provided shelter and kept the observer largely hidden from view by seals. Six visits were made to the study area coinciding with spring tides. These visits occurred roughly every 2 weeks and lasted between four and 6 days each, except for Christmas, where the visit lasted just 2 days. All equipment was installed at the start and removed at the end of each visit. An experienced observer monitored the river for the duration of daytime high tides (approximately 4.5 h), during which the river is accessible to seals. Binoculars with 10x magnification and a Canon 7D mark 2 camera with a 600 mm IS USM f4 lens were used to collect photo identification data of seals, providing information on the frequency of attempts by identified individuals to pass the deterrents. Treatments were conducted on seals in the trigger zone as they moved upriver toward the deterrents. Existing data collected during the previous winter field season were judged to be sufficient as a control (SMRU unpublished data, see below), allowing the number of exposure treatments to be maximized given time constraints and a relatively low expected encounter rate. The control data from winter 2021/22 consisted of daytime observations of seal behavior from the same location. All seals (n = 14) that started to ascend the river through the study area continued upriver, passing the proposed deterrent site. In addition, sonar data that were collected both day and night showed that 51 out of 53 seals that entered the trigger zone from downriver continued up the river, passing the proposed deterrent location, i.e., 96% continued upriver after reaching the trigger zone. The seal deterrent used in this study was produced by GenusWave Ltd. and included four Lubel underwater speakers mounted in pairs, located upstream of the triggering zone (Figure 1), with one pair on each opposite bank. The system produced sounds with a peak frequency of 1 kHz and emitted short 0.2 s signals on a pseudo-randomized duty cycle. The deterrent was manually triggered via radio link once seals were in the trigger zone. In addition to daytime visual observations, a sonar was used to validate seal movements underwater and during the final month of the study, it was also used to conduct treatments during nighttime high tides. A Gemini 1200ik was installed in the river, operating in low frequency mode (720 kHz) and was able to monitor the river's full width (~45 m). Power was provided by a portable generator (Honda i10) and a Wi-Fi bridge was used to transmit sonar data to a laptop for storage. The sonar's wide field-of-view (120°) meant that detection probability for adult seals moving upriver was likely close to 1. Adult seals could be confidently distinguished from otters or salmon based on their body size as well as their swimming behavior. Responses to the acoustic deterrent were characterized by a termination of upriver progress. The two speakers and sonar were mounted on frames made from steel tubing to provide the weight necessary to maintain the equipment's position and orientation in the river under normal flow conditions (Figures 2 and 3). Ahead of the trials, sound measurements of the deterrent were made in the river using a Soundtrap acoustic recorder (Ocean Instruments NZ—serial 806109211). A representative waveform and spectrogram are provided to illustrate the GenusWave signal characteristics (Figure 4). Signal energy was centered around 1 kHz, and signal reverberations were evident in the recordings, likely due to riverbanks and the shallow river environment. Sound recordings were made at 10 locations during a spring high tide; the Soundtrap was suspended at 0.5 m depth in water that was approximately 1.3 m deep (Figure 5). Five measurements were made at each location to produce a mean received sound pressure level (SPL) for both the north and south speakers, except for location 1, where a measurement was only made from the south bank speakers. Recordings at location 1 were made at a range of approximately 1 m to provide an indication of device source level. Measurements provided a means to predict the potential sound levels that a seal may perceive on approach, allowing the distance at which the device should be triggered to be determined. Received levels at each location were found to differ depending on which speaker (north bank or south bank) was producing sounds. Sound traveled better across the river compared with down the adjacent riverbank as sound transmission was likely better through the deeper water away from the riverbanks (Figure 6). For example, there was little difference in the received level for locations 2–5 for the signal emitted from the speaker on the north bank (Figure 6). As river conditions were expected to affect sound transmission, and these may fluctuate greatly within and between tidal cycles, these measurements only provide an indication of the possible received levels that seals may experience. Furthermore, large boulders within the river may create localized sound shadows that cause localized drops in received levels (Shapiro et al. 2009). These measurements indicated that seals close to the riverbank may be largely shielded from one deterrent but not the other (Figure 6). Therefore, for seals close to the riverbanks, to maximize the chances of the first signal being above the threshold likely to provoke the startle reflex, devices were only triggered once seals were within 30 m. For seals in the middle of the river then devices were triggered when seals were within 60 m. During the six visits to the river, observations were carried out over a total of 34 spring high tides for approximately 4.5 h each tide. These included land-based observations during 26 daylight high tides (December 2022—February 2023) and sonar images were monitored for eight nighttime tides (February). Sixteen exposure treatments were conducted and were 100% effective (all 16 seals terminated upstream progress and returned downriver). Fifteen treatments were conducted during the day and one at night. The first visit collected the majority of treatments (14) over the initial 4 days of the study, and seals subsequently stopped swimming upriver into the trigger zone during the daylight observation periods. All treatments (n = 16) resulted in seals immediately changing direction and swimming away from deterrents. Most treatments produced a strong visible response from seals, typically including a large disturbance on the water's surface as they changed direction followed by either bow waves or sequences of surfacings, indicating rapid directed swimming until they were outside of the study area or occasionally leaving the entire visible area (Figure 1). Consequently, the duration of each treatment was short and deterrents were typically only engaged briefly (~20 s) resulting in the generation of ~5 acoustic signals. However, there were four exceptions where seals did not immediately leave the study area (treatments 7, 10, 11, and 12) where the seal initially changed direction but then remained between approximately 60–100 m downriver of the acoustic deterrents for a short time before eventually returning downriver of the study area. In these cases, the deterrent was engaged for longer and until the seal demonstrated sustained downriver movement. No treatment exceeded 5 min. During daylight periods, photo ID allowed seals in treatments to be identified to the individual level (Figure 7). Four different seals were identified in 12 treatments, and known seals were numbered 1–4 with the prefix Pv used to indicate seal species, Phoca vitulina. No photographs were taken during four treatments due to a camera fault in three treatments, and the fourth was conducted at night using sonar only. Seal Pv.3 was identified in seven treatments showing persistence in attempting to travel upriver (Table 1). Photo ID data showed that out of four treatments where the seal did not immediately leave the study area, three were attributed to the same individual, Pv.3. No photographs were taken during the fourth treatment; therefore, it was not possible to determine the ID of the seal. Interestingly, during daylight observations, seal presence in the entire visible area (Figure 1) also declined by approximately 50% when compared with January 2022 (Table 1). Seals that did venture up the river following these initial treatments typically turned around just before reaching the trigger zone (n = 10). This behavior had not been observed in the control data from January 2022, when all seals (n = 14) that entered the defined study area (Figure 1) continued swimming upriver through the trigger zone (Table 1). Treatment 15 was the only treatment to be conducted at night using the live image from the sonar. Based on sonar target size and swimming behavior, a seal was identified traveling upriver at 03:28 at a speed of 1 m/s; the deterrents were instructed to start when the seal was within 30 m of the deterrents. The seal demonstrated a strong response by traveling back downriver at a speed of 5 m/s. The decline in the frequency of seals traveling upriver into the trigger zone could be a direct effect of seals learning to avoid the deterrents and changing their behavior to either forage downriver or elsewhere. Seal sightings in rivers typically follow cyclic patterns within and between years, with seals returning to target salmonids at specific times (Graham et al. 2011). Although little data were available for other months, information from January 2022 acted as control data. Seal activity in the study area was high during this month, with some 72 seal events registered as seals moved up or down through the study area. Seals were absent on only three out of 15 spring tides observed in winter 2021/22. Comparison with these control data suggests that the observed lack of seals at the deterrent site following repeated treatments, coupled with the observations of seals turning around shortly before reaching the trigger zone, could be attributed to the fact that seals may have quickly learned to avoid the triggering location in 2022/2023. Four different seals were individually identified throughout the study period, and although not all seal sightings were captured by photo ID, it is unlikely that many seals were not identified. The number of seals that specialize in using Scottish rivers is generally low (Graham et al. 2011; Harris and Northridge 2019) and four individual seals aligned with expectations prior to the study. The response of these seals to the deterrent may or may not be typical of seals in general, and it is not possible to say if the deterrent was redeployed in a different location with different seals whether they would respond in the same way. In particular, the North Esk is a relatively small river, and the responses of seals in larger rivers may be different. Gray seals were not observed during the study period, and therefore we cannot make any assumptions about their potential responses. Further testing and development is required in order to fully understand the potential effectiveness of triggered acoustic deterrents in discouraging seals from rivers in the future. Although manually triggering acoustic deterrents are unlikely to be a long-term solution, due to the considerable resources required to provide visual monitoring, these results indicate that acoustic deterrents may form part of an effective “detect and deter” system if an efficient automation of the real-time detection of seals can be developed. The value of such systems to fisheries might be further increased if fish numeration abilities were incorporated into any automated detection function. the relatively small size and short of this study, it a responsiveness of seals to acoustic exposure treatments as well as an change in behavior that until at the end of the study and given results from winter field In seals were present in the river on most observation days during and and then only at night identified from sonar during January and with one on (SMRU unpublished Therefore, it is possible the of seals during daylight hours in is an indication of long-term behavioral change as a direct of this behavioral response study, carried out during the previous where these known individuals have learned to avoid the river at these data review and with from the Scottish under the Seal deterrent equipment was provided by are to and the Salmon Fishery for our The no of The data that the of this study are available from the