Current state of knowledge of effects of offshore renewable energy generation devices on marine mammals and research requirements

4 Potential Proximate Impacts

Three main classes of effects of OREG devices have been repeatedly identified in reviews of their potential impacts on marine mammals. These are noise, risks of collision, and changes in the availability of the animals' habitats. These are each discussed in turn, and followed by another short section on effects due to electric fields.

4.1 Noise

All developments create noise - albeit at varying levels. Some of this is common to all technologies and stages of operation, such as that from increased shipping traffic associated with the construction, maintenance and dismantling of devices. There are also noisy activities associated with individual stages in the lifecycles of some projects. Pile-driving has been identified as being of particular concern. There is also noise resulting from the operation of both wind- OREG and tidal- OREG, though this is much quieter and studies to date suggest it may be less important (Tougaard et al. 2009) Each of these three are discussed in turn, after a brief introductory description of the hearing capabilities of important species.

4.1.1 Construction

Like oil and gas production platforms, current designs for OREG devices are based on prefabrication. The main noise, apart from the vessels required for installation, will be associated with the formation of foundations. Wave and some tidal devices can be tethered rather than directly sitting on their foundations, but they still require some anchoring to the seabed. Most tidal energy devices require substantial foundations. Four basic technologies are available: piling, gravity footings, anchors or drilled and grouted attachment. The noisiest of these is piling, which has been widely used for offshore wind farms in shallow parts of the North Sea. It is unlikely to be appropriate for many tidal devices because the strong current s that those devices utilise prevent the accumulation of the soft sediment into which piles are most easily driven.

4.1.1.1 Piling

Estimates of received levels of piling noise vary widely but there is a general consensus that, in some conditions, they have the potential to cause hearing damage to a wide range of marine mammal species over considerable areas. The use of piles in wind farm construction has been an issue since the earliest stages of the planning process (Thomsen et al. 2006). As a direct consequence, the effects of pile driving noise on marine mammal behaviour is probably the most intensively studied aspect of the environmental impacts of marine renewables industries. Most of the available information comes from monopole wind turbines, though the SeaGen tidal turbine in Strangford Lough was eventually fixed to the seabed with four pin-piles.

4.1.1.1.1 Noise characteristics

Thomsen et al. (2006) measured pile-driving noise from a jacket-pile construction in the German Bight. They reported peak sound pressure levels and sound exposure levels in 1/3 octave bands. Operation noise was also measured at a range of 110m from a 1.5MW wind turbine in Sweden. Sound levels at various distances from the source were calculated and zones of noise influences were assessed based on published data.

The broadband peak sound pressure level during pile-driving was 189 dB _0-p re 1 µPa ( SEL = 166 dB re 1 µPa ²s) at 400 m distance, resulting in a peak broadband source level of 228 dB _0-p re 1 µPa at 1 m ( SEL = 206 dB re 1 µPa ²s at 1 m). The 1/3 octave sound pressure level was highest at 315 Hz (peak = 218 dB _0-p re 1 µPa at 1 m) but as found in other studies, there was considerable sound energy at higher frequencies above 2 kHz. Source levels were estimated for larger pile-diameters by extrapolating from these results. Operational noise levels were much lower in both amplitude and frequency, with 1/3 octave sound pressure levels between < 90 and 142 dB _Leq re 1 µPa at 1m with most energy at 50, 160 and 200 Hz, at wind-speeds of 12 m/s.

Tougaard et al. (2009) reported measurements from the installation of 4 m diameter steel monopile foundations driven into hard sand in shallow water at Horns Reef, Denmark. The impulsive sounds generated had high sound pressures [source level 235 dB re 1 µPa(pp) at 1 m], and measurements at different ranges from the pile driving indicated an 18 log(R) transmission loss function. The sound profile had most of its energy at low frequencies, but they detected significant energy up to 100 kHz.

Bailey et al. (2010) measured pile driving sounds from installation of two 5 MW wind turbines in relatively deep water (>40 m) water off the NE coast of Scotland. Received levels were recorded at distances of 0.1km producing an estimated maximum broadband peak to peak sound level 205 dB re 1 µPa. Received levels were measured at distances up to 80 km before pile driving noise fell below background noise levels.

Nedwell et al. (2007a) reported the results of a substantial recording programme during pile driving operations at five wind farms North Hoyle, Scroby Sands, Kentish Flats, Barrow and Burbo Bank. Estimated source levels at the five wind farms varied between 243 and 257 dB re 1µPa at 1m, having an average value of 250 dB re 1µPa at 1m. The transmission losses were characterised by values of geometric loss factor N of 17 to 21, and absorption factor of 0.0003 to 0.0047 dB/m. Under some conditions pile driving noise was detectable above background to ranges of between 10 and 25 km or more depending on local background noise levels.

4.1.1.1.2 Zones of audibility and potential hearing damage

Bailey et al. (2010) related the sound levels from installation of 5 MW turbines to noise exposure criteria for marine mammals to assess possible effects. They estimated that bottlenose dolphins could suffer auditory injury but only within 100 m of the pile-driving. They also estimated that behavioural disturbance, defined as any modifications in behaviour, could have occurred up to 50 km away.

Thomsen et al. (2006) estimated that both harbour porpoises and harbour seals are likely to be able to hear pile driving blows at ranges of more than 80 km. They concluded that behavioural responses are possible over many kilometres, perhaps up to ranges of 20 km and that masking might occur in harbour seals at least up to 80 km. Using potential hearing damage criteria of 180 dB _rms re 1 µPa for cetaceans and 190 dB _rms re 1 µPa for seals they estimated that hearing loss might be a concern, at 1.8 km in porpoises and 400 m in seals. Thomsen et al. (2006) also concluded that severe injuries in the immediate vicinity of piling activities cannot be ruled out.

Nedwell et al. (2007a) used a metric of 90 dB above hearing threshold (referred to as dB _ht) to assess likely reaction and damage ranges for fish and marine mammals and predicted strong avoidance within ranges of a few kilometres by "sensitive species" such as harbour porpoise. This would suggest that phocid seals could be expected to react at significantly greater ranges. They also adapted metrics used to model the cumulative effects of noise on humans and suggested that exposure at a level of 90 dB above threshold for eight hours, or exceeding a peak level of 130 dB above threshold for 3 seconds is likely to cause hearing damage. Using the 90dB _ht L _eq criteria ( i.e. sound exposure weighted for both hearing sensitivity and signal duration) they estimated that a harbour porpoise could be exposed to the noise during an entire pile driving operation at a typical range of 250 metres without harm. However, their results also indicated that peak levels may exceed 130 dB above threshold at larger ranges and that injury ranges indicated by the measurements using this criterion may be several hundred metres.

4.1.1.1.3 Masking vocalisations

David (2006) estimated that pile-driving sound would be capable of masking vocalisations by bottlenose dolphins within 10-15 km and weak vocalisations up to 40 km. For operational installations, Lucke et al. (2007) have suggested that there is potential masking of low frequency hearing. Conversely Tougaard et al. (2008) state that it is unlikely that the low frequency tonal noise would mask the high frequency signals of porpoises at any range. There is insufficient information on the extent to which pile-driving or seismic pulses mask biologically significant sounds for marine mammals (Bailey et al. 2010). The better low frequency hearing of seals could mean that noise from operational installations would be able to mask biologically significant sounds.

4.1.1.1.4 Behavioural responses by harbour porpoises

These have been investigated during the construction of two wind farms at Horns Reef and Nysted.

4.1.1.1.4.1 Horns Reef offshore wind farm

Direct disturbance effects from piling during construction were reported at Horns Reef (Tougaard & Teilmann 2006; Tougaard, Madsen & Wahlberg 2008). Individual 4m diameter piles took approximately 70 mins to drive with source levels of 235 dB re 1 µPa (pp) at 1 m. Data from passive acoustic monitoring (T- PODs) indicated that porpoise detections fell throughout the area during piling operations. The effect was widespread, with similar declines in porpoise activity apparently out to ranges in excess of 25 km from the piling. This lack of a detectable spatial gradient in response means that it is not possible to extrapolate reactions to estimate the extent of the area affected.

Porpoise acoustic activity apparently returned to levels typical of the overall construction period within 6- 8 hours of the cessation of piling. Tougaard et al. (2009) suggested that although this may indicate that porpoises return to the area shortly after the disturbance, it could also indicate that there is a high natural turnover in porpoises in the area and that the recovery is due to undisturbed animals coming through. Without a method to identify and record responses of individual porpoises this cannot be resolved and the actual disturbance effects on individuals cannot be assessed.

Tougaard et al. (2009) estimated that during piling operations at Horns Reef, porpoises were significantly disturbed and may have been excluded from the construction area for up to 17% of the time over a 5 month period during which 80 foundations were piled. Between piling events there was no apparent decline in porpoise acoustic activity suggesting that other construction activities did not have a significant disturbing effect. Visual observations of surface behaviour of harbour porpoises was compared between days with pile-driving and days without. On days without pile-driving, the dominant behaviour was non-directional swimming (presumably associated with feeding), whereas the dominant activity on days with pile-driving was directional swimming (presumably associated with travelling) (Tougaard et al. 2003 ). Both acoustic and visual observations demonstrated significant effects at ranges up to 15 km from the construction site during pile-driving.

In a follow up study Brandt et al. (2011) monitored porpoise vocalisations during construction of the Horns Rev II offshore wind farm in summer 2008. Porpoise acoustic activity fell to zero for 1hr after pile driving and stayed below normal levels for up to 72hr at a distance of 2.6 km from the construction site. A negative effect was detectable out to a mean distance of 17.8 km and within 4.7 km the recovery time exceeded the interval between pile driving bouts. The longer recovery periods meant that porpoise activity was reduced over the entire 5 month construction period.

4.1.1.1.4.2 Nysted Offshore Wind Farm

At Nysted, the main noise generating activities during construction were dredging and backfilling of gravity foundations. However some piling activity (1.5 to 10 hours per day over a 25 day period) occurred for installation of sheet piles around one turbine foundation (Carstensen, Henriksen & Teilmann 2006). Harbour porpoise acoustic activity was monitored by acoustic data loggers (T- PODs) in a structured Before-After Control Impact ( BACI) experiment. A significant decrease in detection of porpoise clicks relative to the pre-exposure baseline period was seen in response to general construction noise (Henriksen, Teilmann & Carstensen 2003; Carstensen, Henriksen & Teilmann 2006; Tougaard et al. 2005). Mean waiting times, defined as the period between two consecutive encounters of echolocation activity, increased from 6 hr in the baseline period to three days in the wind farm area during the construction period with an apparently greater increase in waiting times (4hr to 41hr greater) during piling operations compared to general construction activities. The effect was apparently widespread although the increase within the wind farm was six times larger than changes observed in a reference area 10 km away (Carstensen, Henriksen & Teilmann 2006; Tougaard et al. 2005). Activity apparently returned to normal levels compared with the overall construction period some days after the pile-driving ceased.

4.1.1.1.5 Behavioural responses by seals

Tougaard & Teilmann (2006) used satellite telemetry and visual surveys to monitor harbour seal movements and behaviour during construction and operation of Nysted Offshore Wind Farm. Results suggested that the wind farm area was an important foraging site, but was not of greater importance than surrounding areas. The location accuracy of the telemetry system was not sufficient to allow estimation of the effects of construction activity. At least one seal was active inside the wind farm during operation and visual observations confirmed that seals were present. However during construction activities very few seals were observed either within or close to the construction site. Tougaard & Teilmann (2006) concluded that this was a response to pile driving noise.

Brasseur et al. (2010b) tagged grey seals in the Netherlands to investigate the effects of wind farm construction and operation. Their sample size at the time of pile driving activity was too small to assess the effects, but movement patterns of individual seals suggested that they may have moved towards the wind farm area after pile driving stopped.

4.1.1.1.6 Changes in local haulout counts of seals

A mixed haulout of harbour and grey seals is situated less than 2 km from the Scroby Sands wind farm (Skeate, Perrow & Gilroy 2012). Monthly surveys of the haulout showed a decline in harbour seal numbers during construction and an apparent failure to recover in the two subsequent years. During the annual moult monitoring surveys ( SCOS 2011) numbers of harbour seals recorded at Scroby has increased continuously since 2003 suggesting that wind farm operation has not depressed haulout numbers. The numbers of grey seals increased year on year throughout the construction and early operational periods.

The temporary decline in harbour seal numbers seen at Scroby may indicate an effect of construction activity with some persistence in that effect. However, the Scroby counts represent approximately 5% of the East Anglian population and the observed changes may simply reflect similar changes in the harbour seal population in East Anglia ( SCOS 2011).

A similar temporary reduction in numbers of seals using haulout sites close to Horns Reef and Nysted (Edren et al. 2010) was recorded during construction phases.

Recent piling activity in the Wash has presented an opportunity for two on-going studies: Using a combination of high resolution GPS telemetry and direct measurement of hearing sensitivity of seals using auditory evoked potential ( AEP) methods SMRU are currently investigating the responses of seals to pile driving activity off the English east coast. In addition to allowing estimation of the detectability of signals from any specific piling blow for each individual seal the AEP data will also be used to assess hearing of wild seals in an area with a history of pile driving activity.

Research gap
Title	Code	Details	Status	Reporting date
Do harbour seals exhibit auditory permanent threshold shift in the presence of piling activity?	DECC1	Audiograms for all harbour seals captured as part of DECC2 will be obtained using standard auditory evoked potential measurements during capture events (Wolski et al. 2003). These will be used to: 1. Identify the hearing thresholds of individual seals to assess the sensation level at which reactions occur. 2· Assess the variability of audiograms within the sample of telemetry tagged harbour seals 3· Identify evidence of hearing damage that may be attributable to exposure to piling noise.	Funded by DECC; on-going	2013

Research gap
Title	Code	Details	Status	Reporting date
Harbour seals behavioural responses to the presence of piling activity.	DECC2	25 harbour seals will be fitted with GPS/ GSM tags in the vicinity of piling operations in the Wash in February 2012. These data will permit: 1. Parameterising the dose-response of piling activity (source energy, range, received and perceived energy) to changes in behaviour ( e.g. movement and dive patterns). 2. assessment of change in at-sea usage, comparing pre- and during- pilling operations	Funded by DECC; on-going	2013

4.1.1.1.7 Mitigation

Pile driving is known to affect seal and porpoise distribution and behaviour. Current pile driving mitigation measures involve visual and/or passive acoustics monitoring surveys before piling starts to reduce the risk that sensitive animals are within a dangerous range of the pile when piling starts. These mitigation operations require dedicated marine mammal monitoring teams and are therefore expensive. The low probability of detecting animals at sea in a wide range of operational conditions means that current mitigation activities may not be effective in avoiding risks to marine mammals in sub-optimal conditions. Other than reducing piling noise or detecting animals and delaying operations until they move away, the only potential mitigation method seems to be some form of acoustic deterrence. Acoustic Deterrent Devices ( ADDs) are widely used at fish farms to keep seals away from direct contact with fish cages. To be useful in the pile driving situation they need to also apply to cetaceans and, have the ability to move animals away over very large distances, at least several hundred metres.

Gordon et al. (2007) argued that aversive signals that cause animals to temporarily move away from an area where they would be at risk could underpin mitigation procedures that are both more effective in protecting wildlife and less expensive and onerous to industry. To be effective, such a method will require acoustic signals that move seals and cetaceans several hundred metres away from a sound source without contributing significant additional acoustic energy to the environment and thereby increase the risk of hearing damage. Work on ADDs at fish farms indicated that harbour porpoises may move away from, and be excluded from, large areas around ADDs (Olesiuk et al. 2002). Indeed, disturbance effects for harbour porpoise have been observed at ranges up to 3km. If a similar effect can be shown with seals, it may be possible to develop an effective mitigation measure for high energy pile driving that will be cheaper to operate and potentially more reliable and effective than current observation based methods.

Development and testing of such mitigation methods have been identified as a major research requirement under the ORJIP programme and are covered by Project 4: Improvements to standard underwater noise mitigation measures during piling.

Wet renewables are still at a relatively early stage of development but there are concerns about the potential for collision, especially with tidal turbines (Wilson et al. 2007; Wilson & Gordon 2011). Alerting marine mammals to help them detect and avoid structures such as tidal turbines could reduce collision risk if a collision risk is identified. Such alerting signals would need to be more or less permanent features and would therefore need to have very different characteristics to the long range disturbance signals suggested for piling mitigation.

Research gap
Title	Code	Details	Status	Reporting date
Acoustic deterrence for mitigation of pile driving activities	AcMit	1. Identify potential mitigation signals 2. Conduct behavioural response trials with telemetry tagged seals. 3. Conduct behavioural response trials with harbour porpoises using 3D passive acoustic array and visual observations.	Funded by MS	2015

4.1.1.2 Gravity footings and anchors

Both these approaches involve the lowering of suitable objects to the seabed. The difference between them is that gravity footings simply rely on their weight to maintain their position while anchors dig into or snag on the seabed. Although gravity footings may require extensive dredging to flatten the seabed the noise generated will be unlikely to approach the levels generated during pile driving operations.

4.1.1.3 Drilling

The original design for the SeaGen device in Strangford Lough was supported on a single foundation grouted into a large hole drilled into the seabed. Difficulties in obtaining a suitable barge to carry out this work led to four pin-piles eventually being used instead. Drilling noise was measured during installation of one of the footings for the SeaGen device in Strangford. Measurements indicated that at ranges between 28 m and 2130 m, the dBht (level above hearing threshold) for harbour seals varied from 59 to 30 dBht and fell below the minimum background levels at a range of 300m from the drill. Although audible to seals at close range, it seems unlikely that the drilling of such foundations would cause substantial disturbance. Harbour porpoises have less sensitive hearing at low frequencies so that the range of detection and potential disturbance would be lower (Nedwell & Brooker 2008).

4.1.2 Operation

There will be noise associated with both the moving parts of these devices and flows over their structures. The frequencies and intensity of these noises is likely to vary between the technologies, so wind, tidal and wave devices are each considered in turn below.

4.1.2.1 Wind- OREG operation

4.1.2.1.1 Noise characteristics

The noise characteristics of operational offshore wind farms have been reviewed by Madsen et al. (2006). In comparison to the loud impulsive sounds of pile driving, the underwater noise from the operating turbines is generally low intensity (Madsen et al. 2006; Tougaard, Madsen & Wahlberg 2008; Tougaard, Henriksen & Miller 2009). Low frequency sounds generated in the turbine are transmitted through the tower to the foundations and radiated into the water column and the substrate. Sound levels from a range of turbines measured approximately 100 m from the foundations lay in the range of 100-120 dB re 1 µPa (1/3 Octave band levels) (Tougaard, Henriksen & Miller 2009).

Marine Scotland has let a contract to model underwater turbine operation noise. It is likely that this will feed into the NERC EBAO project into optimising array designs in terms of the acoustic effects of large arrays.

Wahlberg & Westerberg (2005) reviewed underwater noise measurements from operating wind turbines. They reported considerable variation in the noise levels from wind turbines related to different wind speeds and recording conditions but also noted major device-specific differences in noise output and sound radiation patterns. There are nevertheless strong indications that some wind turbines make more underwater noise than others. For example, intensities reported from the Utgrunden wind farm in the Baltic Sea were approximately 10 dB higher than elsewhere (Wahlberg & Westerberg 2005).

The underwater noise produced by wind turbines is dominated by low frequency pure tone signals below 1 kHz and mostly below 750 Hz. The strongest tonal component in Ingemansson Technology's (2003) recordings was around 180 Hz at a wind speed of 13m/s. The frequency content of the signals does not seem to vary with wind speed. Early studies indicated that sound intensity is not closely related to the size of the turbine, but this contention may not be valid for large turbines of several megawatts.

Ingemansson Technology (2003) reported that sound level increased with the number of active wind turbines in a wind farm. The measured sound intensity at any point will therefore be a composite of noise from several devices and the resulting interference patterns will create a complex sound field.

The received level will also depend on the transmission characteristics. Transmission in deep open water can be approximated by a spherical spreading model where received sound intensity will decrease by approximately 20log(r), where r is the distance in metres, (at the low frequencies generated by turbines absorption is trivial). However, sound may be channelled through reflection at the surface and bottom in shallow seas, or through refraction in stratified, water. The degree of channelling will depend on the surface conditions (wave structure) and the topography and sediment type of the sea bed. The site specific modelling of propagation or transmission loss can produce accurate estimates of received levels, but extrapolation of such models to greater ranges or to other apparently similar sites and areas may be problematic.

Tougaard et al. (2008) suggested that although operational noise levels are relatively low, the fact that they will be produced almost continuously for long periods means that they could significantly increase the local ambient noise level. If background noise levels are low the turbine noise may be audible to seals and odontocetes at distances of several kilometres from the turbines.

Tougaard et al. (2009) used recorded noise from three different operating turbines to assess the zone of influence on both harbour seals and harbour porpoises. Signals were only detectable above background levels at frequencies below 500Hz. They estimated that harbour porpoises would only be able to hear the sound at ranges of 20-70 m from the foundations. The better low frequency hearing of harbour seals meant that they would be able to detect the signals at ranges of between 60 m and 6.4 km depending on the specific measurement conditions and the choice of cylindrical or spherical spreading loss models.

In addition to spreading and absorption effects, the structure of the footings or foundations of the turbines will influence the transmission of sound and the ranges at which different effects will occur. Marine Scotland has recently funded a modelling study of operational sound propagation from wind turbines with different types of footings/foundations. The results should provide more accurate estimates of noise exposures, but are unlikely to dramatically change these conclusions.

4.1.2.1.2 Operational noise effects on small cetaceans

Harbour porpoises and bottlenose dolphins have relatively poor hearing at the low frequencies generated by wind farms. For example, the estimated received levels at 83m from a single device at Utgrunden were around 125 dB re 1µPa at around 180 Hz. and between 100 and 110 dB at frequencies up to 1 kHz. Hearing thresholds for both species are around 100dB at 500 Hz and increase rapidly for lower frequencies (Johnson 1967; Kastelein et al. 2002). The sound levels recorded at Utgrunden would not cause hearing damage to porpoises or bottlenose dolphins even at very short ranges. It is also unlikely that the low frequency tonal noise would mask the high frequency signals in porpoise vocalisations at any range (Tougaard, Madsen & Wahlberg 2008) although there is potential masking of low frequency hearing (Lucke et al. 2007).

4.1.2.1.3 Operational noise effects on phocid seals

Phocid seals have better low frequency hearing than either porpoises or bottlenose dolphins, e.g. harbour seal hearing thresholds at around 180 Hz have been reported to be around 80 to 85 dB although Kastelein et al. (2008) suggest that in unmasked conditions harbour seals may be 5 to 10 dB more sensitive at these low frequencies. The recorded source levels at Utgrunden would be approximately 70dB above threshold at a range of 10 m from the source. Kastak and Southall (2005) reported temporary threshold shifts ( TTS - reversible hearing loss, see section 5.1.1. for definition) of between 2.9 and 12.2 dB resulting from 20 to 50 minutes of exposure to 2.5 kHz noise at received levels 80 to 95 dB above hearing threshold in a harbour seal. All animals recovered from the exposure within 24 hr and usually much earlier. The degree of TTS was related to received level and duration. They obtained similar results from a northern elephant seal and a California sea lion, suggesting that the results may be applied across pinnipeds and therefore apply to both harbour and grey seals.

If TTS is related to sound intensity in the same way at lower frequencies, harbour seals may be susceptible to TTS only at very short ranges, less than 5 m from a turbine and only if they remained this close for several seconds. This suggests that although the turbine noise may be perceived as a loud sound it is unlikely that it would cause TTS in any realistic field conditions and is therefore unlikely to cause permanent hearing damage in phocid seals.

4.1.2.1.4 Operational noise effects on large cetacea

There is little information on the hearing capabilities of large cetaceans although their predominantly low frequency vocalisations would suggest that they have good low frequency hearing. It is likely that large cetaceans will be able to hear the noise from wind turbines at least as well as seals. Future developments of wind farms in the central and northern North Sea and other waters around Scotland mean that larger numbers of large cetaceans such as minke whales have the potential to come into contact with wind farms.

4.1.2.2 Porpoise distribution and behaviour

Koschinski et al. (2003) modified recordings of a smaller turbine to simulate a 2 MW turbine and played the noise to harbour porpoises. They documented a clear reaction, with closest approach distance increasing from 120 to 182m and increasing vocalisations. This implies that harbour porpoises can detect the sounds produced by wind turbines. However the playbacks may have contained higher frequency artefacts due to the signal enhancement method used. It is not clear whether the porpoises were responding to the turbine noise or these higher frequency components.

There are few offshore wind farms old enough to have produced useable data on marine mammal responses. Consequently there are few published reports on empirical studies. Three published reports describing the effects of wind farm operations on distribution and local abundance of harbour porpoise are available for wind farm developments.

4.1.2.2.1 Horns Reef offshore wind farm

This study entailed seven years of surveys and five years of acoustic recordings of harbour porpoises between 1999 and 2006 covering the pre-construction, construction and operation phases (Tougaard, Henriksen & Miller 2009). Acoustic activity monitoring and visual surveys were carried out at the wind farm site and a reference site.

The results showed a clear effect of pile driving. The T- POD acoustic data indicate that porpoises left the entire Horns Reef area in response to the loud impulse sound generated by the pile driving operation. After a period of 6-8 hours, activity returned to levels normal for the construction period as a whole. Overall the level of porpoise acoustic activity was not significantly lower during construction, but was lower during a period described as "semi-operation" when large amounts of boat and other maintenance activity seems to have reduced porpoise activity within the wind farm. Ship survey data indicated a reduction in porpoise activity within the farm during construction. Overall the authors considered there to have been a weak negative and local effect of the wind farm during construction.

Porpoise acoustic activity and ship based sightings surveys indicated an increase in porpoises in the area as a whole during the operational period compared to the baseline. This is consistent with the general increase in porpoise numbers in the Southern North Sea. Overall the study found no significant changes in the distribution of porpoises between wind farm and reference areas in the operational phase compared to the baseline period.

4.1.2.2.2 Egmond aan Zee wind farm

This study entailed two periods of monitoring acoustic activity at the wind farm site and at two reference sites (Scheidat et al. 2011). The study covered the preconstruction/baseline period (2003-2004) and an operational period (2007-2009). Porpoise acoustic activity increased during the operational period when compared to the pre-construction baseline. However, there was an overall increase in porpoise abundance in Dutch waters over the last decade. Porpoise activity was significantly higher inside the wind farm than in the reference site. The authors suggest that this apparent increase in porpoise activity within the operating wind farm may indicate an attraction effect due to increased food availability inside the wind farm (reef effect) and/or a sheltering effect with reduced levels of disturbance from vessels within the wind farm compared to the heavy ship traffic in adjacent areas of the southern North Sea.

4.1.2.2.3 Nysted Wind Farm

Porpoise acoustic activity was monitored before, during and for two years after construction of the wind farm by deploying three T- PODs within the wind farm site and three at remote reference sites 10km away. Porpoise activity declined significantly in the wind farm during and for two years after construction. A smaller but significant decrease in activity was recorded in the reference area. This may indicate a more widespread disturbance effect due to construction activities. The levels in the reference sites had returned to pre-construction levels by the second year of operation.

4.1.2.2.4 Seal distribution and behaviour

Koschinski et al. (2003) modified recordings of a smaller turbine to simulate a 2 MW turbine and played the noise to harbour seals. They documented reduced surface activity of harbour seals within 200m of the playback system implying that the seals could clearly hear the sounds and moved away from the source. However, as mentioned above, the playbacks may have contained higher frequency artefacts due to the signal enhancement method used. It is not clear whether the seals were responding to the turbine noise or these higher frequency components.

4.1.2.2.4.1 Nysted and Rødsand II

McConnell et al. (2012) used high resolution GPS telemetry tags to study movements of harbour and grey seals in southern Denmark. Seals were tagged at haul out sites within 10 km of two wind farms: Nysted and Rødsand II. The results were compared with similar data collected in 2009. Both species frequently transited from the haulout sites through the two nearby wind farms. Visually, there was no obvious interruption of travel at the wind farms' boundaries. Interactions with wind farms were assessed using residence times within wind farm zones, comparison of path speed and tortuosity inside and outside the wind farms and the proximity of individual locations to individual turbines. No significant effect of the wind farms on seal behaviour was detected. This is in accord with another local study (Edren et al. 2010) of haulout counts that concluded that the wind farms had no long term effect on the local seal population trends.

4.1.2.2.4.2 Egmond aan Zee

Brasseur et al. (2010a) used similar GPS tags and older ARGOS satellite tags to track 12 harbour seals before and 24 seals after the construction of the Egmond aan Zee wind farm in the Netherlands. The satellite telemetry data indicate that seals tended to avoid shipping activity in the major shipping routes. The large distance between the wind farm and the haul-out areas meant that there were limited data to assess interactions. Their results indicated that seals avoided the area during construction, but were observed to use the wind farm areas after construction activities ceased and seals from another study were also recorded inside the operational wind farm (Lindeboom et al. 2011). The authors concluded that although seals have been observed in the wind farm, minor effects on behaviour cannot be ruled out.

4.1.2.2.4.3 Horns Reef

The movements of seals from haulout sites adjacent to Horns Reef wind farm site were studied using similar telemetry devices (Tougaard & Teilmann 2006; Tougaard, Henriksen & Miller 2009). They deployed 21 simple location only satellite transmitters. The results showed that seal foraged over a wide area that incorporated the Horns Reef wind farm area. The results did not indicate a major effect of either construction or operation but the study animals spent little time inside the wind farm site either before or after construction and the study therefore had limited power to detect effects. Tagged seals were recorded in or close to the wind farm during operational periods and concurrent visual surveys indicated reduced seal activity in the area during construction but showed that seals were present within the operating wind farm.

4.1.2.2.4.4 Scroby Sands

Monitoring surveys during the annual moult ( SCOS 2011) indicate that the numbers of harbour seals recorded at Scroby have increased continuously since 2003 suggesting that wind farm operation has not depressed haulout numbers despite the disturbance associated with construction. The numbers of grey seals increased year on year throughout both the construction and early operational periods.

The Department of Energy and Climate Change ( DECC) is currently funding a project run by SMRU to investigate the impact of piling activity on the distribution and behaviour of telemetry tagged harbour seals off the south east coast of England.

Research gap
Title	Code	Details	Status	Reporting date
Harbour seals behavioural responses to the presence of piling activity.	DECC3	New data from tagged harbour seals in the Wash (see DECC2) and Thames will be compared with historic data and periods of non-operation within the current study to assess dose-response of movement and behaviour in relation to wind farm operation.	Funded by DECC; on-going	2013

4.2 Physical contact

Collisions between OREG devices and marine mammals are a cause for concern. These are most likely to occur and result in serious injuries when marine animals come into contact with parts that are moving rapidly relative to the water. This problem is likely to primarily be associated with tidal- OREG devices and shipping associated with the construction and operation of OREG devices. A related issue is entanglement with cables tethering devices to the seabed. That is most likely to occur with some designs of wave OREG device.

4.2.1 Shipping

Ship strikes are a common cause of death for cetaceans (Laist et al. 2001). Until recently ship strikes were not considered to be an important issue for phocid seals. However, recent events in UK waters suggest that seals may be killed in substantial numbers by collision with ships (Thompson et al. 2010a; Bexton et al. 2012). Although the circumstances and conditions under which such fatal interactions occur are as yet unknown, it is likely that OREG activity will increase the amount of shipping activity in coastal waters close to seal haulout sites and in offshore areas that may be important foraging sites for seals and cetaceans. Therefore there is potential that any such harmful interactions could increase. This is primarily a shipping issue, and is not restricted to marine renewable developments. The risks posed by shipping may be similar to those experienced by other marine industries such as the oil and gas, transportation, fishing and fish farming. However one major incident in Norfolk in 2010 was probably linked directly to the increase in near-shore shipping activity associated with the construction of Sheringham Shoal wind-farm (report Unexplained Seal Deaths ( USD) tasks 1 & 2 of Theme 2 of MMSS/001/11).

Fatal interactions between seals and ships probably occur when boats are manoeuvring slowly or maintaining position in areas of high seal density. The construction and maintenance of tidal energy devices in strong tidal flows often in constricted water ways and channels will add an additional level of complexity to interactions. The constrained channel will necessarily increase the chance of any marine mammal using the channel being involved in a close encounter with a vessel. For example, a vessel (using dynamic positioning or simply motoring) holding station in a tidal stream will effectively be moving rapidly with respect to the water but stationary with respect to the bottom. Since ship/boat strikes appear to be a relatively common cause of anthropogenic marine mammal mortality, the additional complexity of shipping operating in strong tidal currents may pose some greater risk of harmful strikes. To date there is insufficient information to be able to estimate the scale of the problem or identify when or where these problems will arise.

There is a need to understand the nature of collisions and to suggest mitigation measures. This is the focus of current Unexplained Seal Deaths Tasks under the MMSS/001/11 Research Project being carried out by SMRU:

Research gap
Title	Code	Details	Status	Reporting date
Unexplained seals deaths	USD	1 Testing the hypothetical link between shipping and unexplained seal deaths through a series of controlled tests of candidate mechanisms using model testing and full scale carcass tests with candidate mechanisms. 2. Testing the hypothetical reasons for lethal interactions through a series of behavioural response trials using both captive and wild grey and harbour seals 3. Examining the distribution of observed carcasses to identify biological and oceanographic patterns and distribution of potential causes to assess the patterns of risk associated with these unexplained seal deaths. 4 Assessing the impact of the observed and estimated levels of mortality on seal populations at a local, national and international level. 5 Identify and evaluate practical management and mitigation measures that could be developed in the short, medium and long term.	Funded by MS, current	2013

4.2.2 Tidal- OREG

The most obvious, and probably the most important interaction in terms of public perception, is the potential for injuries or fatalities resulting from direct contact with moving parts of tidal power devices (Linley et al. 2009; Wilson & Gordon 2011).

Devices and marine mammals must coincide in both space and time in order for any such effects to occur. Currently we lack any hard information on the behaviour of marine mammals during such proximate interactions so we can only estimate the potential for collisions. How animals act in terms of avoidance or attraction towards devices and their ability to evade collisions will scale the potential collision risk assessment. Understanding behavioural response to an operating tidal- OREG is a priority.

Research gap
Title	Code	Details	Status	Reporting date
Avoidance and evasion behaviour by marine mammals in close proximity to tidal turbines.	Avoid	1 Using high resolution telemetry to observe the behaviour of seals in close proximity to marine renewable devices, concentrating on tidal turbines. 2 Using high resolution 3D hydrophone arrays to monitor porpoise behaviour in close proximity to marine renewable devices, concentrating on tidal turbines. 3 Using high resolution 3D hydrophone arrays and ultra-sonic pinger tags to monitor seal behaviour in close proximity to tidal turbines.	Not funded	N.A.

4.2.2.1 Collision risk models

Two models have been proposed for estimating the risk of collisions between marine mammals and tidal turbines in UK waters:

a) The Scottish Natural Heritage ( SNH) Collision Risk Model, also known as the Band model, was developed to estimate the number of birds that could be expected to collide with onshore wind farms (Band, Madders & Whitfield 2007). A modified version of it has been used to predict the rate of collisions between seals and the demonstrator tidal array that is planned for the Sound of Islay.

b) An alternative model (Wilson et al. 2007) was based on a movements and interactions model developed to investigate predation by zooplankton (Gerritsen & Strickler 1977).

Both models are based on two assumptions. The first is that any collision between a marine mammal and a marine renewable device will result in death. The second is that the patterns of movement of marine mammals will be the same in a particular place irrespective of the presence or absence of a marine renewable energy device. That is, marine mammals show neither attraction nor avoidance behaviour, and make no attempt to evade the moving parts. Under this assumption the number of marine mammals impacted can be derived from an estimate of how many will pass through the footprint of a device scaled by the likelihood of being hit by a blade based on the transit time of the animal and the rotation rate and number of the blades.

The Band model then applies a correction factor, assuming that 95% of potential collisions will be avoided, based on data on birds within terrestrial wind farms. A similar modification would also need to be applied to convert the output of the other model to estimates of actual risks to animals. Clearly, as accepted in the Band model, the assumption (not based on data) of no reaction is unlikely to be true. In addition, several factors are likely to influence both the likelihood and severity of such contacts (Wilson et al. 2007). To assess the probabilities of such occurrences we need information on:

• The characteristics of the device, e.g. rotation speed, blade length and number, and its position in water column.
• The short term and seasonal movement patterns of animals
• The size of the population at risk
• The dive patterns, depth usage and small scale movement patterns of individuals
• Reactions to presence of devices
• • Avoidance/ Attraction of animals to the turbines.
  • Evasion behaviour in close proximity to devices.

4.2.2.2 Data collection methods

In most cases, direct observation of collision is unlikely to be achievable, but there are available technologies that may be employed directly or modified to allow either direct (photography, sonar imagery etc.) or indirect (high resolution telemetry, acoustic localisation of natural vocalisations or attached high frequency pingers) observation of fine scale behaviour close to devices. The available methods have been reviewed and assessed under task MR3 of the Marine Mammal Scientific Support Research Programme MMSS/001/11 ^[2] . Briefly they are:

4.2.2.2.1 Active sonar

Several active sonar systems have been developed for tracking marine mammals (Hastie 2012). In good conditions, modern acoustic imaging devices can provide reasonable quality images of marine mammals in real time. It has been proven to successfully detect and track marine mammals in the vicinity of underwater turbines at sufficient spatial and temporal scales to identify potential collisions. Such systems have been deployed as part of the research, monitoring and mitigation system at the Marine Current Turbines' Sea Gen device in Strangford Narrows (Keenan et al. 2011). Partly due to the safety shut-down procedures when marine mammals approach the turbine, there have been no recordings of close encounters in this study.

4.2.2.2.2 Passive acoustic monitoring ( PAM)

PAM using an array of hydrophones that can detect (to species level) and track vocalising animals (primarily toothed whales - especially porpoises and dolphins) could be used to track fine resolution behaviour around tidal turbines. A static, accurately positioned array of hydrophones around a turbine, connected to a central processor should be capable of achieving the required level of precision. It is an unsatisfactory system for baleen whales that vocalise unpredictably. Whilst seals also do not regularly vocalise, they could be captured and fitted with individually coded acoustic 'pinger' tags. Such individuals would thus be capable of being tracked by a PAM system.

4.2.2.2.3 High resolution tracking

Here we refer to high resolution tracking using animal-borne telemetry devices. Due to the difficulties of catching small cetaceans this technique is currently limited to seals in the UK. The current best available tracking systems are based on GPS locations and detailed dive depth profiles that are sent either through the mobile phone system (the GPS/ GSM tag) or through the ARGOS satellite system. These devices provide GPS quality locations for individual surfacings but positions during submergence can only be interpolated assuming some form of movement model. The accuracy of such estimates is determined by the rate at which position fixes are calculated and by the predictability of movements between fixes. Information of sufficient accuracy to determine proximity to turbine blades may be possible with the incorporation of 3D accelerometers and magnetometers to allow dead reckoning ( DR) so that the 3-D underwater track can be accurately generated in between surface GPS locations. However for DR to be practically useful the local water current needs to be known to a degree of (spatial and temporal) accuracy that is not readily available.

As part of the monitoring programme at the Marine Current Turbines SeaGen site in Strangford Narrows (Keenan et al. 2011), GPS/ GSM tags on harbour seals provided very accurate locations for a large proportion of surfacing events. Whilst the lack of usable DR precluded accurate underwater track recreation, it was possible to estimate the number of times seals passed the device as they transited through Strangford Narrows (approximately 1 km wide). Thus the method provides some information on avoidance behaviour.

4.2.2.2.4 Mechanical sensing

Turbine blades are routinely equipped with strain and accelerometry sensors to monitor mechanical performance. It has been suggested that information from such sensors could be used to detect collisions. Developers in Scotland and Wales are in the process of assessing the effectiveness of such systems for detecting collisions with marine mammals. As far as we are aware no results have been published and there have not been any direct collision impact tests of such systems. Such a test could be conducted using animal carcasses introduced upstream of an operating turbine.

Research gap
Title	Code	Details	Status	Reporting date
Assessment of mechanical sensing of impact with tidal turbine blades.	MECH	In cooperation with the Operator a turbine device will be instrumentation with appropriate strain and accelerometer sensors. A series of carcasses (resembling the size and mass of a seal or porpoise) will be presented to the rotating blades to determine whether the turbine sensors provide sufficient data to enable automated strike detection.	Not funded	NA

4.2.2.2.5 Video surveillance

Animal-borne video cameras have been widely used to identify prey types (for example Davis, Hagey & Horning 2004). However as these all rely on retrieving the camera and the low probability of successfully recovering the device means that few samples would be obtained. Thus, although this may be an effective research tool under particular conditions, it is unlikely to be useful as a monitoring method for any UK marine mammal species.

Static video surveillance cameras can be used to monitor the local underwater activity of marine mammals (for example Simila & Ugarte 1993; Herzing 1996). In addition, direct video monitoring of a functioning underwater device has been conducted at the Open Hydro test site at the European Marine Energy Centre ( EMEC) ^[3] . However two issues limit their ability to observe marine mammal interactions at turbines. First, underwater visibility is often limiting. Second, video surveillance would require an artificial light source at night. The responses of marine mammals to lights underwater have not been studied and they may be either attracted or repelled by artificial lights at night. In good conditions (sufficient ambient light and good visibility) video surveillance has the potential for detecting impacts and is perhaps the only method with the capacity to allow assessment of the immediate consequences of impacts.

Research gap
Title	Code	Details	Status	Reporting date
Marine mammal responses to artificial lights	Light	investigate responses to different light sources to identify possible illumination for night-time video surveillance.	Not funded	N.A.

4.2.2.2.6 Stranding surveys

As part of the monitoring programme at the Marine Current Turbines SeaGen site in Strangford Narrows (Keenan et al. 2011) targeted standings scheme with post mortem evaluations of any stranded marine mammal carcasses was established to look for signs of turbine impact. Over a three year period of operation no such signs have been reported. At present this provides little information on the likelihood of collisions having occurred as there are no estimates of the likelihood of an injured animal or damaged carcass washing ashore and being found.

4.2.2.3 Data availability

Medium scale information about the movements of harbour seals in the vicinity of an operating tidal turbine have been obtained using GPS/ GSM tags in Strangford Lough. In addition, similar information will be available in 2013 from a study of harbour seals movements close to operating turbines at the EMEC site in Orkney ( NERC "RESPONSE" study). However there is still a pressing need for fine scale interaction studies for both seals and cetaceans. The potential to conduct such a study could exist at the Sound of Islay.

Research gap
Title	Code	Details	Status	Reporting date
Fine scale marine mammal behaviour in the vicinity of a working tidal array.	ARRY	1. Building on the recommendations of the Marine Scotland project, Hastie (2009) and Hastie (2012), suggest active sonar systems that would be appropriate for trialling at the Sound of Islay. 2. Consider the capability of developing Passive Acoustic Monitoring ( PAM) systems to track vocalising cetaceans around tidal turbines. Develop and test systems for possible trials in the Sound of Islay, taking account of the use of acoustic tags for seals. 3. Evaluate the ability of the above, or other, technologies to monitor potential actual impact detection. 4. Trial the feasibility of these technologies for direct observation of marine mammal movements.	Not funded.	NA

In addition to studies at operating tidal- OREG sites there is the potential to examine the response to playback of recorded turbine noise to wild and captive seals:

Research gap
Title	Code	Details	Status	Reporting date
Investigation of marine mammals' responses to playback of turbine noise	PLAY	1. Use high resolution telemetry and active sonar to track free ranging seals and observe responses during controlled exposure to turbine noise. 2. Monitor the behavioural responses of captive seals to controlled exposure to turbine noise 3. Use high resolution passive 3D acoustic array to track free ranging porpoises and observe responses during controlled exposure to turbine noise	Funded by NERC RESPONSE; current	2013

4.2.2.4 Effects of collision

Other than the assumption that there will be some form of trauma there is in fact little evidence of the likely consequences of a collision between any marine mammal species and a turbine blade. Analogies with ship strikes and boat propeller strikes are speculative and may be misleading because of the differences in shape and collision speeds involved. Experimental exposure of marine mammals to tidal turbines has not been attempted and to date there have been no direct tests of the effects of turbine blade impacts on marine mammal carcasses. There are therefore currently no data on the levels of damage caused by collisions. A targeted research project to examine likely damage patterns is thus required (see 4.2.2.4.2).

4.2.2.4.1 Computer simulation.

An impact damage model has been developed to assess the likely consequences of the impact of a specific tidal turbine blade on a killer whale ( US Department of Energy 2012). The model was developed in response to concern about the potential severity of encounters between killer whales and an OpenHydro tidal turbine in Puget Sound. The model estimated the forces developed during a head-on collision with an adult male killer whale assuming the strike occurred on the head. The consequence of those forces on the skin and underlying tissues of whale were assessed. In the absence of data on the biomechanical properties of whale tissue the characteristics of a number of alternative natural and synthetic materials were substituted as surrogates for killer whale tissue.

Although testing the appropriateness of the surrogate tissues needs further work, the results of the finite element models suggest that the maximum forces generated in such a collision would not have been sufficient to tear killer whale skin or cause skeletal damage. Thus it is unlikely that such a collision would kill or seriously injure an adult killer whale.

The authors point out that the results are not likely to be applicable to other species and other turbine designs, for example the hollow centred OpenHydro device is unlike most other designs being tested for deployment in UK and Scottish waters. However, a similar approach could be applied to each design and each species of potential interest.

4.2.2.4.2 Carcass field experiments.

A practical and viable model of collision effects requires information on a number of aspects of marine mammal behaviour, movements and distribution. However a major constraint on the utility of these models for management is the fact that they currently take the precautionary view that any collision will be fatal. This is unlikely to be the case and in some potential device-animal collisions there may not even be any significant injury ( US Department of Energy 2012).

Blade speed increases with distance from the hub and varies, often in a nonlinear fashion, with current speed. The levels of injury sustained during a collision will therefore depend critically on where along the blade and when in the tidal cycle they occur as well as on the current velocity-blade rotation speed relationship. Information on the levels of and types of injury inflicted by collisions at different speeds is required to assess the likelihood of serious injury. The same information is required to interpret damage observed on stranded and by-caught carcasses.

Research gap
Title	Code	Details	Status	Reporting date
Collision damage assessment	COLL	A series of collision damage assessment trials with carcasses of seals and/or other species when available using a purpose built test rig. A section of turbine blade will be dropped onto seal carcasses at a range of speeds. The seal carcasses will be positioned just below the surface of a 2.5m deep pool so that the speed of impact is known and the carcass is coupled to the water and will therefore resist the impact in the same way as a free swimming seal. Carcasses will then be examined both visually and by x-ray/ultrasound to assess damage.	Funded by SNH	2014

4.2.2.4.3 Fish studies.

Laboratory flume tank experiments have been carried out on live fish and these results may provide some insight into the levels of damage that may be suffered during collisions. However caution should be used extrapolating from such studies due to taxon, size, anatomy, behaviour, and turbine type differences.

Amaral et al. ( 2011) report on a study to determine behaviour, injury and survival rates. Rainbow trout ( Oncorhynchus mykiss) and largemouth bass ( Micropterus salmoides) juveniles were released directly upstream from two types of operating turbine (Darrieus open helical cross flow and Welka UPG axial-flow) to observe their behaviour and to record injury and survival rates from turbine impact. Two size classes of fish (100-150mm and 225-275mm) were used in flow rates of 1.5 and 2.1 m/s. Few fish passed through the area swept by the turbines, some swam upstream and others swam around the turbine. Of those assessed to have passed through the turbines over 98% survived in each case (98.4 ± 1.10), not significantly different to control groups with no turbine present, and few injuries were seen in the experimental fish other than damage due to handling.

Additional experiments with Atlantic salmon ( Salmo salar) smolts and American shad ( Alosa sapidissima) adults using a vertical axis Encurrent turbine produced similar results with no significant injuries observed with either species.

Fish behaviour around an array of six tidal turbines in the East River of New York was studied as part of a demonstration for the Roosevelt Island Tidal Energy (RITE) ^[4] . Fish movements were monitored using an array of acoustic cameras (Split Beam Transducers) and DIDSON sonar cameras. Results showed that resident and migratory fish generally avoided the areas in which the turbines were located and tended to prefer inshore, slower moving waters. Fish were not recorded in the array at flow velocities greater than 0.8 m/s. Limited observations showed fish passing by the rotating turbines, following the hydrodynamics of the system. These data indicated that fish were able to detect and successfully pass around the operating turbines.

Observations of fish were recorded around a barge mounted tidal turbine in Cobscook Bay, Maine, USA ( http://www.orpc.co). An Ocean Renewable Power demonstration turbine was mounted on a barge that allowed the turbine to be lowered into the water for testing. Two acoustic ( DIDSON) cameras were mounted in front of and behind the turbine, and data were collected over a 24 hour period. The study indicated that fish did not entirely avoid, and regularly approached, the turbine and barge. More fish interacted with the turbine when it was still than when it was rotating. The majority of the fish detected by the cameras were already located above or below the turbine when they entered the field of view, which may indicate that they were able to detect the turbine at greater distances than could be monitored (> 2.5 m upstream) by the DIDSON cameras. Results of day night behaviour comparisons suggested that vision was an important component of turbine detection.

Operating turbines at the EMEC site in Orkney were monitored using ambient light video monitoring. When turbidity conditions allowed, concentrations of fish were observed on the downstream side of the device when the turbine blades were stationary. However, during strong flows they were absent and no fish were seen to pass through the rotors.

4.2.3 Entanglement

Marine mammals are known to become entangled or entrapped in a wide variety of man-made structures especially fishing gear (Read, Drinker & Northridge 2006). In terms of marine renewables, most/all tidal- OREG devices will be fixed directly to the seabed or held in place by robust tethers that will be unlikely to pose an entanglement risk. Because of the severe drag on cables in high tidal flows it is also unlikely that there will be any freely hanging cables associated with tidal turbines.

Most perceived entanglement issues are related to wave- OREG devices. Very little is currently known about the risk of entanglement to whales and other marine wildlife in such devices. In general any rope or cable in the sea can pose a finite risk of entanglement or collision to whale or other cetacean. The facts that sperm whales have been reported caught in undersea telegraph cables and dolphins get caught in crab pot lines make it clear that such animals are vulnerable. Most work on baleen whale entanglement has previously been focused on interactions with fishing gear since most ropes and cables in the oceans are put there by fishery related activities. In Scottish waters it is generally assumed that most baleen whale carcasses that are recovered with evidence of rope marks have been the result of entanglement in lobster creel lines. Previous work has estimated that there may be over 7000 km or rope deployed in the sea around Scotland at any one time, and indeed it is therefore likely that this single source poses the greatest risk (Northridge et al. 2010). A fin whale that was stranded at Stoer in the Highlands in October 2007 was diagnosed as having died due to entanglement, where the marks on the animal's body were consistent with a thick cable rather than a creel line. Whale entanglements have also been reported in aquaculture mooring lines in Australia and Iceland - which are likely to be much thicker and more taught than the loosely set polypropylene line used for creel fishing. Sometimes cetaceans will actively rub against taught cables and can then get entangled.

The advent of moored structures such as wave machines on a large scale could potentially open up a new area of concern for wildlife entanglement. The issue has been addressed in at least one workshop in the US, where the risk of entanglement was identified but not quantified. Mitigation measures to minimise whale interactions are already being deployed around at least one wave power development site in Oregon.

We assume that slack lines will pose a greater risk of entanglement to large cetaceans than taut cables, but it is difficult to predict how cables will behave during operation and how whales will react to them. However, small the risk of entanglement posed by mooring lines, the greater their number, the greater the probability of entanglement.

4.3 Exclusion and barrier effects

If habitat exclusion and/or barrier effects do occur it is highly likely that the disturbance would be due to marine mammals responding to the acoustic signatures of devices or arrays.

To date the only study to have addressed the issue of barrier effects or habitat exclusion has been the environmental impact study of the single SeaGen turbine in Strangford Narrows. A combination of GPS/ GSM tag telemetry studies of harbour seals and visual and passive acoustic monitoring ( TPODs) of porpoises showed that both species continued to swim past the SeaGen device and moved into and out of Strangford Lough while the turbine was operating. There was some indication that harbour seals avoided the centre of the channel when the turbine was operating. It was also noted that during the operational phase the rates of transit past the device were lower when the turbine was active compared to periods when it was stationary. The TPOD data did not allow precise estimation of locations so it was not possible to assess fine scale avoidance. There was a reduction in porpoise activity during construction, but this was temporary and porpoise acoustic activity returned to baseline after construction was complete.

Porpoise activity has also been monitored in the vicinity of an OpenHydro tidal turbine device in the Minas Passage in the Bay of Fundy (Tollit et al. 2011). Long term data were collected from two C- PODs placed at 150m and 700m from the turbine. Harbour porpoise presence was detected on most days (93%), but usage of the site was typically low. Activity was significantly higher at night. There was no significant difference in porpoise activity levels between the turbine (11%) and control (12%) site although there appeared to be a difference in click train structure. However, the study reported that the device was not operational during the study so the porpoise activity relates simply to the presence and absence of devices and not the noise associated with operation (Tollit et al. 2011).

Long term data from an on-going wildlife observer programme at the EMEC site in Orkney has been collected before and during operation of several wave devices (that effectively form a small, ad hoc array of tidal devices). To date there are no published analyses of the effects of device operations on the observations and it is not clear that the data are sufficient to allow an assessment of effects on marine mammals at either site. An analysis of the entire dataset is currently underway.

Research gap
Title	Code	Details	Status	Reporting date
Analysis of visual observation data	VisOb	A detailed analysis of the long term data set from the EMEC visual observation programme should be carried out to assess the likelihood of being able to detect changes in distribution or fine scale habitat use in the vicinity of turbines	Part Funded by SNH	2014

4.3.1 Wave- OREG operation

Most of the wave generators in a relatively advanced stage of development are floating platforms of some sort and also have minimal contact with the seabed. Even bottom mounted systems such as Oyster ^[5] are unlikely to reduce overall foraging habitat availability. Although wave generators will have mooring and or anchor systems they are unlikely to have a major impact on the available habitat in comparison with the scale of foraging area used by marine mammals.

4.3.2 Tidal- OREG operation

Individual tidal turbines are relatively small and many designs have only minimal structures in contact with the sea bed.

There may be some downstream changes in sedimentation or benthic communities as a result of disruption of tidal flow patterns and there may be changes in shorelines due to changes in wave patterns, but again, on the scale of marine mammal foraging ranges these would not be expected to significantly reduce foraging habitat availability and would, at most, have a small effect on several animals or a larger effect on a small number.

Changes to local oceanographic features such as tidal eddies may also alter foraging habitat quality. To date we have little information on foraging behaviour for any species at the fine spatial and temporal scales needed to identify such effects. Recent studies of harbour porpoise activity suggest that they may exploit eddies in tidal rapids (Gordon et al. 2011) but it is not clear if this is an important habitat requirement at an individual or population level. Nor is it clear how the changes in tide streams would alter foraging behaviour and/or alter foraging habitat quality.

Research gap
Title	Code	Details	Status	Reporting date
Fine scale habitat use by porpoises in tidal rapids	EDDIE	Combinations of towed arrays and static 3D arrays may be used to monitor fine scale movements of porpoises within tidal rapids to investigate their use of small scale and/or transient eddies.	Not Funded

Any large stable structure, either floating or fixed to the sea bed, is likely to attract prey species. Several offshore fisheries exploit this by deploying Fish Aggregating Devices ( FADs) (Buckley, Itano & Buckley 1989). It is likely that large arrays of any marine renewable device will have some sort of reef effect. The magnitude of any such effect, however, is likely to be linked to the size of the structure and is likely to be strongly dependent on the site characteristics and the device architecture.

This could potentially provide enhanced foraging habitat and improved foraging success for marine mammals in the vicinity of the devices/arrays. Conversely it may attract marine mammals to the vicinity of devices and increase the potential for direct potentially harmful interactions with devices. To date there have been no studies of marine mammal foraging behaviour in the vicinity of multiple tide or wave devices. Studies in Netherlands may indicate increased use of wind farm sites by harbour porpoises and possibly also harbour seals, after initial disturbance during construction. This has been speculatively associated with either reef effects or reduced boat traffic.

4.4 Electric fields

Recent reports of tests of electric fields as seal deterrents appear to show that both phocid and otariid seals are extremely and unexpectedly sensitive to electric fields. This is being exploited as a means of deterring seals from predating on fish in freshwater systems. Forrest et al. (2009) showed that seals were deterred from swimming though a 200 microsecond pulse electrical field with gradient of between 0.1 - 0.32 V/cm. These levels did not seem to affect the behaviour of salmonid fish and catch rates of salmon were higher at nets protected by an electric field.

Recent trials using a similar system in sea water, Milne et al. (2012) demonstrated similar deterrence effects but also showed that the electric field intensities eliciting responses in seals are similar to those eliciting similar voluntary and involuntary responses in humans and dogs. They conclude that there is no evidence for higher sensitivity in seals than in large terrestrial mammals. This is consistent with the fact that seals do not appear to have specially adapted electrically sensitive organs.

Wilson & Gordon (2011) point out that the seal exclusion trials used short pulse length electrical fields and that seal sensitivity increased as pulses lengthened. However, Milne et al. (2012) show that this response asymptotes at pulse duration around 1 ms, indicating that seals are unlikely to be more sensitive to a continuous electrical field.

Estimates of the electrical fields that will be generated in seawater from buried power cables from offshore renewable energy devices are around 4 orders of magnitude lower. The maximum electrical field in the sea for buried power cables was estimated to be 0.9μV/cm and even lower, between 0.015-0.025 μ V/m in a later study (Gill et al. 2005). It is therefore unlikely that seals would be able to detect these signals and extremely unlikely that any avoidance behaviour would result from such exposures.

There have been no attempts to assess the sensitivity to electrical fields in any cetacean species found in UK waters, it has been shown that at least one dolphin species, the Guiana dolphin ( Sotalia guianensis), has specialised electro receptor sense organs. These modified vibrissae appear to be sensitive to voltage gradients several orders of magnitude higher than the apparent sensitivity of seals (Czech-Damal et al. 2012).

Research gap
Title	Code	Details	Status	Reporting date
Electrical sensitivity of small cetaceans	Elect	A series of carefully controlled tests of sensitivity of small cetaceans (porpoises and bottlenose dolphins in the first instance) to electric fields similar to those generated by OREG devices and export cables. These will necessarily be carried out in captive animal facilities. As there are no captive cetaceans in the UK such studies will require an international collaboration.	Not Funded

4.5 Changes in distribution

In this section we only consider the effects of operating wind farms and ignore the effects of piling noise and other construction disturbance.

4.5.1 Porpoises

The underwater noise levels from operational wind farms are considered to be too low to pose any realistic risk of physical damage to porpoises. The frequency range of the underwater noise from wind turbines also makes it unlikely that there are any masking effects.

A summary of information on displacement of marine mammals around operational wind farms is available from Scottish Government at http://www.scotland.gov.uk/Resource/0040/00404921.pdf. Controlled exposure experiments indicate that porpoises may be able to detect wind turbine noise at ranges of tens to hundreds of metres. However, based on the results of acoustic and visual monitoring of porpoises at operational wind farms, there is no clear evidence of significant displacement of animals from the wind farm sites due to turbine noise. Studies at Nysted and Horns Reef suggest that construction effects may be detectable tens of kilometres from the wind farm sites. At Nysted the observations suggested that the decrease in activity due to construction carried over into the two subsequent years although there were reportedly indications that the effect was decreasing. Activity levels in the reference area were depressed during construction but had returned to normal by the second year. It therefore seems that the underwater noise generated by operational wind farms is unlikely to cause significant disturbance to harbour porpoises within the wind farm area. If there are no significant local effects it seems highly unlikely that there are significant wider scale effects.

We are not aware of any reports of studies of other cetaceans around wind farms. Other small cetaceans likely to occur in UK waters will have similar hearing capacities to harbour porpoises at the low frequencies produced by tidal turbines. Some may be several dB more sensitive, but the same arguments about lack of damage risk and lack of masking effects will apply.

As there have been no reported studies of reactions of other species to wind farms it is not possible to predict their responses to wind turbine noise. However Marine Scotland have a contract to model underwater turbine noise and make a judgement on likely effects on various species (these include bottlenose dolphins, minke whales, harbour porpoises and harbour seals). The report will be available in September at http://www.scotland.gov.uk/publications/scottish-marine-freshwater-science-volume-4-number-5-modelling-noise//downloads.

4.5.2 Seals

The underwater noise levels from operational wind farms are probably too low to pose any realistic risk of physical damage to seals at ranges of more than 10m. Even within this range, seals would need to remain within the sound field for considerable periods before suffering any TTS effects. It therefore seems unlikely that seals will suffer hearing damage from wind turbine noise.

Seals are more sensitive to low frequency sound than are small cetaceans. If they use this low frequency band for passive prey detection or predator detection, it is possible that wind farm noise may cause some masking of biologically significant sounds. If such effects occur and are biologically meaningful they will probably be restricted to the close vicinity of turbines.

The only study of seal movements with sufficient power to detect effects of an operating wind farm was McConnell et al. (2012). They found no effect on any of the movement and distribution metrics that they could test. In addition, other studies with lower power are broadly in agreement.

A study of haulout behaviour at Scroby Sands within 2km of the wind farm indicated that counts of both harbour and grey seals have continued to increase during the 6 years of operation after a possible temporary effect of construction activity on harbour seals (Skeate, Perrow & Gilroy 2012).

There is at present only a limited amount of information, but these preliminary results do not indicate a major change in distribution of either grey or harbour seals as a result of current wind farm operations. It is also therefore unlikely that there have been larger scale redistributions as a result of wind farm operations