Effects of offshore wind farms on marine wildlife—a generalized impact assessment

Marine management plans over the world express high expectations to the development of offshore wind energy. This would obviously contribute to renewable energy production, but potential conflicts with other usages of the marine landscape, as well as conservation interests, are evident. The present study synthesizes the current state of understanding on the effects of offshore wind farms on marine wildlife, in order to identify general versus local conclusions in published studies. The results were translated into a generalized impact assessment for coastal waters in Sweden, which covers a range of salinity conditions from marine to nearly fresh waters. Hence, the conclusions are potentially applicable to marine planning situations in various aquatic ecosystems. The assessment considered impact with respect to temporal and spatial extent of the pressure, effect within each ecosystem component, and level of certainty. Research on the environmental effects of offshore wind farms has gone through a rapid maturation and learning process, with the bulk of knowledge being developed within the past ten years. The studies showed a high level of consensus with respect to the construction phase, indicating that potential impacts on marine life should be carefully considered in marine spatial planning. Potential impacts during the operational phase were more locally variable, and could be either negative or positive depending on biological conditions as well as prevailing management goals. There was paucity in studies on cumulative impacts and long-term effects on the food web, as well as on combined effects with other human activities, such as the fisheries. These aspects remain key open issues for a sustainable marine spatial planning.


Introduction
Global demand for renewable energy is increasing, as motivated by our challenge to reduce carbon dioxide loading and mitigate global warming, as well as to reduce risks for radioactive pollution, the dispersal of harmful substances, and the depletion of non-renewable resources. Wind energy is often described as a suitable alternative from all of these perspectives (Martínez et al 2009a, 2009b, Saidur et al 2010, Esteban et al 2011, Leung and Yang 2012. Expectations on the offshore areas are particularly high, as wind conditions are often stronger and more stable over sea. Further, OWF can allow for larger units and a higher total level of energy production, and large units may be transported and constructed more easily (EC 2008, EWEA 2012. Importantly, potential conflicts of interest with other sectors of society are also less pronounced in the marine landscape than on land (Pedersen et al 2010). Although visual disturbance can be decisive for the consenting processes in many cases (Zoellner et al 2008, Ladenburg 2009), conflicts of interest further away from land are often related to conservation and fisheries issues (OSPAR 2004, HELCOM 2010. Whereas the effects on fisheries' distributions and landings can be assessed by vessel monitoring systems (VMS) and spatially explicit landing reporting, the effects on marine biodiversity are harder to encompass. Assessing effects on biodiversity is limited both by information on natural distribution patterns of species and habitats, and by ecological understanding of the sensitivity of species to the presence of an OWF.
Our understanding on the potential effects of offshore wind farms (OWF) on the function of marine ecosystems, as well as marine biodiversity, is steadily improving as empirical evidence from operational wind farms is accumulating (Leonhard et al 2011, Lindeboom et al 2011, Mann and Teilmann 2013. However, the potential risk this may entail to marine ecosystem structure and functioning are only rarely assessed systematically, as may be required in order to inform ecosystem-based marine management and spatial planning efforts. In the present study, we have synthesized the current state of knowledge on the effects of OWF on marine and aquatic wildlife. We based the study on published records of empirical observations at global level, and translated the findings into a generalized impact assessment for Swedish waters. The Swedish coastline covers aquatic ecosystems of various salinity, gradually changing from marine (30 ) to nearly fresh waters (2 ). By this, the assessment covered various types of ecosystems, and made it possible to compare local and general impacts. Based on the conclusions, we highlight future key issues for the OWF sector from the conservation perspective.

Methods
Information for the synthesis was obtained from empirical studies addressing either the effects of OWF directly, or addressing some pressure identified as potentially influential on marine wildlife during OWF construction or operation. These were identified by searches in scientific databases, but also by directed searches over the internet for reports produced by consultant agencies or governmental authorities in connection to monitoring programmes of existing OWFs, as these encompass a significant part of the total written volume on the topic. An important background material in this respect was provided by the compiled literature review by Wilhelmsson et al (2010). The initial material included over 600 reports and publications. These were screened for relevance in relation to the delineations of this study (main pressures and ecosystem components, as outlined below). Key papers referred to in the text were selected to as far as possible represent peer-reviewed publications, or reports summarizing main findings from a specific topic or surveillance programme.

Generalized impact assessment
The assessment was made separately for different main pressures, identified as the ones most frequently mentioned in the studied literature. For the construction phase, the main pressures included were; acoustic disturbances and increased sediment dispersal, and for the operational phase; habitat gain, fisheries exclusion, acoustic disturbance, and electromagnetic fields (figure 1). In order to compare general versus more local effects, the assessment was made separately for three geographical subareas (see below). Effects were assessed separately for three different ecosystem components in each subarea. Impacts during the third stage of an OWF life cycle, decommissioning, were not assessed, as little or no research has hitherto been directly dedicated to evaluating this stage. However, available studies indicate that impacts during decommissioning are likely to be similar to those of the construction phase.
Probable impact on marine species was assessed with respect to the following aspects; (i) temporal extent, (ii) spatial extent, and (iii) sensitivity of species within each ecosystem component. The magnitude of impact was valued by scores from 1 to 3, where higher scores implied higher impact, using the categorization criteria described in table 1. In order to facilitate comparisons across pressures and geographical areas, the sum of all scores was calculated as an indicator of overall impact. A total sum of 3-4 indicated low overall impact (mainly low scores and no high scores for any of the specific aspects), whereas a total sum of 5-6 indicated moderate overall impact (predominantly moderate scores, or high scores for one aspect combined with at least one low score for the other aspects). A total sum of 7-9 indicated high overall impact (moderate to high scores for all aspects, or high scores for more than one aspect). In addition, the level of certainty in the assessment was evaluated based on how well the conclusions were supported by the peer-reviewed part of literature (table 1,

Geographical area
The assessed area ranged from the Skagerrak in the North Sea region to the inner Baltic Sea (figure 2). This was divided into three subareas with more similar species richness and species composition; the Skagerrak-Kattegat coast with near marine conditions (salinity 15-30 , hereafter SK), the Baltic Sea proper with brackish conditions (6-12 , hereafter BP) and the Gulf of Bothnia with near freshwater conditions (2-5 , hereafter GB). The sounds connecting SK and BP are characterized by fluctuating salinity (8-15 ) due Overview of main pressures from OWF during the operational phase. Expected effect on the local abundance of marine organisms is indicated as (+) aggregation/increase, (−) avoidance/decrease. Table 1. Criteria for assessing the probability of impact on marine life from pressures associated with offshore wind farms. The evaluation was made separately for each pressure (acoustic disturbance during construction, sediment dispersal during construction, habitat gain, fisheries exclusion, acoustic disturbance during operation, and electromagnetic fields). Spatial extent was defined as the expected dispersal of the pressure from its source, temporal extent as its expected duration. Sensitivity was assessed in relation separately for each ecosystem component (marine mammals, fish, and benthic species) and geographical area (Skagerrak-Kattegat, Baltic Proper, and Bothnian Sea). The level of certainty was assessed based on the level of documentation in peer-reviewed literature.

Score
Spatial extent Temporal extent Sensitivity Certainty to irregular mixing of marine and brackish water masses, and were included in SK because it holds several marine species that do not extend into the BP (HELCOM 2012). The subareas were defined based on salinity, as the salinity gradient is clearly correlated with changes in species richness and species composition (Ojaveer et al 2010), as seen for various species groups (HELCOM 1996, Snoeijs 1999, Bonsdorff 2006, Nohrén et al 2009, HELCOM 2012, Olsson et al 2012. Although there are also other environmental differences among subareas, for example changes in climate and topographical variation, these aspects are not decisive for differences in species composition at the current scale of study (Ojaveer et al 2010). The Swedish Energy Authority has identified suitable locations for OWF in all subareas. The results were typically representative for offshore areas with a depth of 5-40 m, reflecting the hitherto prevailing sites for OWF establishment.

Ecosystem components assessed
The assessment was made separately for the following species groups: marine mammals, fish, and benthic species. These were represented by different dominant species in the different subareas, and also varied in conservation status. The study was delimited to underwater pressures. Hence, it did not include impacts on seabirds and bats, which are also of high concern for the planning of OWF. These are mainly affected by pressures relating to above-surface properties of the OWF, and including them would have unduly increased the scope of the study. Marine mammals were represented by four species. The harbour porpoise (Phocaea phocaea) and harbour seal (Phoca vitulina) are found mainly in SK and to some extent in BP. Grey seal (Halichoerus grypus) is found in all subareas but mainly in BP and GB. The ringed seal (Pusa hispida) is mainly found in the GB and parts of BP (Gulf of  Benthic species also decrease in diversity from SK to GB (HELCOM 2012). Large crustaceans, as well as many attached invertebrates, such as ascidians, sponges, corals, echinoderms and many molluscs are only common in SK. A dominant invertebrate in offshore areas of BP is the blue mussel, Mytilus edulis. In the GB, attached invertebrates are scarce and mainly represented by barnacles, bryozoans and hydroids (Balanus spp., Electra spp., Cordylophora spp.). A similar pattern is seen for macroalgae and submerged aquatic plants, with a decreasing structural complexity and species richness from SK to GB. Error bars show maximum score in any subarea (for details see tables 2-4) A minus (−) sign indicates negative impact, a plus (+) sign predominantly positive impact. Level of certainty in the assessment is indicated by the colour of each bar; black = high, striped = moderate, grey = low certainty, based on criteria shown in table 1.

Characterization of available studies
Empirical research on environmental effects of OWF has hitherto primarily been carried out in northern European marine waters (Belgium, Denmark, Germany, The Netherlands, UK and Sweden). Minor part of the studies has been conducted in the brackish Baltic Sea. The research field has gone through a rapid maturation and learning process, starting around year 2000. The initial years were characterized by broad monitoring programmes with relatively low precision, aiming at identifying or excluding impacts of OWF on marine species. Many early efforts were also aimed at developing survey methods. In later years, studies have become more targeted. Both the amount of studies, their topics and geographical coverage has increased rapidly. One remaining limitation is that the studies are typically restricted in spatial and temporal scope. Additionally, they have mainly focused on responses in single species, with little elaboration to ecosystem and seascape scales. Moreover, there is a considerable paucity of ecological baseline data for existing OWF areas, although this aspect seems to be gradually improving as more targeted monitoring programmes are being formed.

Construction phase
Studies on impacts during the construction phase were strongly focused on marine mammals, and to some extent fish. Very few studies addressed effects on sessile species, and none highlighted particular risks to these. The generalized assessment indicated a high impact from noise on marine mammals in all subareas, and on fish in SK. This was due to weak populations of many fish species that depend on shallow areas for recruitment, e.g. cod (Hammar et al 2014). The impact on fish in the other subareas was rated moderate, and for benthos low with low certainty, as strong differences among species may be expected.
The high scores were associated to extreme noise from pile-driving, which is mainly used in the deployment of OWF based on monopiles or jacket foundations. Pile-driving has been observed to cause significant avoidance behaviour in marine mammals (Richardson et al 1995, Carstensen et al 2006, Tougaard et al 2008, Bailey et al 2010, Brandt et al 2011, Dähne et al 2013, and is highly likely to cause mortality and tissue damage in fish (Popper et al 2003, Nedwell and Howell 2004, Popper and Hastings 2009. A considerably lower acoustic impact can be expected for OWF based on gravity foundations, which do not involve pile-driving (Hammar et al 2008). In these cases, acoustic disturbance is mainly expected from sea floor preparing activities, such as drilling or dredging, as well as an intensified vessel traffic (expected for all construction types). Available studies indicate that fish and marine mammals react to low intensity noise from these sources ( Wahlberg and Westerberg 2005), and may respond by leaving the area. However, the intensity Table 2. Synthesis of potential impact on marine life from main pressures during OWF construction. Values give scores for probable impact (1 = low, 2 = moderate, to 3 = high) in relation to each of the criteria spatial extent, temporal extent and sensitivity. 'Total' denotes their sum. 'Certainty' indicates the level of literature documentation to support the evaluation. For definitions, see table 1. SK = Skagerrak-Kattegat, BP = Baltic Proper, GB = Gulf of Bothnia. The scores are based on a subjective evaluation of the cited literature and should be updated as new relevant results become available. Total scores are colour coded as: 3-4 = low, 5-6 = mod, 7-9 = high. of disturbance is low, and animals are likely to return soon after exposure has ended. Hence, low impact can be expected, provided that significant habitats and seasons are avoided.
On the other hand, gravity foundations involve higher impact from sediment dispersal, due to dredging. Although organisms inhabiting wave-exposed sites typical for OWF establishment can generally be expected to be tolerant of turbidity, some studies indicate that elevated turbidity may harm sensitive organisms, such as juvenile fish (Auld and Schubel 1978, Lake and Hinch 1999, Partridge and Michael 2010. The impact of sediment dispersal was rated low to moderate, with good to moderate certainty (table 2).
In summary, available studies suggest that construction activities should not take place in important recruitment areas for marine mammals and fish, and that actions to reduce exposure to damaging noise levels should always be undertaken. For migrating species, this could potentially be solved by timing construction activities outside of biologically sensitive periods of the year. Ways to induce avoidance behaviour in fish and marine mammals have been addressed in some studies (Nedwell and Howell 2004, Mueller-Blenkle et al 2010, Andersson 2011. However, the ability to avoid harmful noise levels is probably reduced in young life stages with more limited mobility (Knudsen et al 1992, Wahlberg and Westerberg 2005).

Operational phase
In contrast to the construction phase, pressures during the operational phase entailed both positive and negative impact ( figure 3). In addition, many studies emphasized the importance of local environmental conditions. This infers that the valuation of a certain pressure into causing either positive or negative impact is dependent on existing values and prevailing management goals. At the generalized level, the probability of negative impact during the operational phase was rated low to moderate, whereas potential positive impact was rated low to high (tables 3 and 4). The level of certainty was low to moderate, due to the high dependency on local conditions (variation within subareas). The result indicates a need for systematic studies across OWFs in different settings, in order to improve the scope for estimating outcomes under different environmental conditions. Studies on the operational phase were early focused on the effects of habitat gain (Petersen and Malm 2006 (Scheidat et al 2011, Teilmann andCarstensen 2012). Research has to no or little extent investigated physiological effects on marine species, in response to e.g. elevated noise and EMF, or the effects of habitat gain on population fitness or reproductive success (Reubens et al 2014). Obviously, empirical studies in OWFs are bound to study combined effects to various extent, as the   . Clearly, regular fishery by an added habitat could have a strong effect on local fish densities, by increasing local mortality (Pickering andWhitmarsh 1997, Seaman 2007), but created habitats are also used as feeding areas by natural predators (Mikkelsen et al 2013). Such impacts, involving food-web interactions, can only be addressed by coordinated studies on different ecosystem components.
The extent of impact also depends on the relative increase in habitat complexity, in comparison to the original substrate (Charton and Ruzafa 1998, Hunter and Sayer 2009). The use of a scour protection, which increases habitat complexity (Hammar et al 2008, Wilson and Elliott 2009), was probably decisive in many cases. For invertebrates, the type of construction material may influence succession patterns. Benthic communities have been observed to be less diverse on foundations made of steel than of concrete (Qvarfordt et al 2006, Wilhelmsson andMalm 2008), although total abundances and biomasses were not necessarily affected. The level of colonization of species onto the new substrate is also related to the local species pool, in particular the presence of species with motile juvenile stages. In the generalized assessment, this translated into lower scores in the less diverse GB and BP, compared to SK, for fish and benthic species (table 3).
The hitherto observed impacts were primarily related to increased aggregation by the turbines, reflecting behavioural preferences of the species. As studies have only been conducted during the first few years of operation, it remains to be seen if an increase in habitat or food availability will lead to increased productivity of resource-limited populations of time (Bohnsack 1989, Pickering and Whitmarsh 1997, Reubens et al 2014. For this to occur, conditions within the OWF would probably have to be significantly more benign than in surrounding areas, and any negative pressures on the species of small magnitude within their full migration distance (Bohnsack 1989, Palumbi 2004).
3.3.2. Fisheries exclusion. Fisheries are not routinely excluded from OWF, but may be restricted as a consequence of excluding shipping for safety reasons (other than that related to maintenance). Fisheries exclusion is likely to increase local species abundances by reduced mortality rates of both target species and by-catch (Leonhard et al 2011, Lindeboom et al 2011, Wilhelmsson and Langhamer 2014, whereas increases in overall productivity and potential spill over effects to adjacent areas are more uncertain (Gell and Roberts 2003). In areas where bottom-trawling were previously conducted, beneficial effects on local benthic species can be expected (Thrush and Dayton 2002). However, empirical evidence from existing OWF is limited, due to restrictions in study design, as available references areas have generally not allowed separating effects of fisheries exclusion from other effects, such as habitat gain. Hence, the probability of a positive impact from fisheries exclusion was rated moderate to high with low certainty (table 3). In contrast, combining OWF with fisheries may be expected to increase local mortality rates of fish, if an increased aggregation close to the foundations serves to enhance catch rates (Polovina 1989, Grossman et al 1997, Pickering and Whitmarsh 1997, Reubens et al 2014. A particular challenge for marine spatial planning is to assess the effects of trade-offs at a larger geographical scale. If the fisheries is reallocated to other geographical areas when an OWF is established, the new fishing area could be either less or more resilient to fishing. 3.3.3. Acoustic disturbances. Vibrations in the turbine towers generated by the gearbox mesh and the generator typically cause underwater noise of 80-150 dB re 1 µPa, at wavelengths that are within in hearing range of both fish and mammals. The tower will also transmit vibrations through the sea floor but this effect is in most cases highly local and therefore considered of minor importance (Nedwell et al 2003, Andersson 2011). In addition, acoustic disturbance may increase due to increased boat traffic for service and maintenance.
Impacts of acoustic disturbances from OWFs were evaluated early (Nedwell et al 2003, Wahlberg and Westerberg 2005, Madsen et al 2006, Tougaard et al 2009, but no empirical studies have hitherto revealed clear negative effects of turbine-generated noise on marine species (Mueller 2008, Båmstedt et al 2009, Andersson 2011, Scheidat et al 2011. However, effects on behaviour are likely, as evident from studies indicating avoidance of the OWF area by harbour porpoise, and possibly a habituation over time (Teilmann and Carstensen 2012). The hearing and processing of sound can be expected to differ strongly among species (Popper and Hastings 2009), many of which have not been studied, and a knowledge gap remains regarding the nature and detection levels of noise from wind turbines and OWF associated boat traffic (Mueller 2008 andAndersson 2011). Also, the extent of the pressure may vary depending on local conditions. Stronger impacts might be expected in pristine areas compared to areas where ambient noise is already high (Scheidat et al 2011). On the other hand, the impact of cumulative effects in such areas remains unclear (Slabbekoorn et al 2010). Hence, probable impact was rated as moderate, with low to moderate certainty (table 4).
3.3.4. Electromagnetic fields. Shielded electric transmission cables do not directly emit electric fields, but are surrounded by magnetic fields that can cause induced electric fields in moving water (Gill et al 2012). Probable negative impact from electromagnetic fields (EMF) was generally rated low, but the level of certainty varied among ecosystem components (Gill et al 2012). A higher score was given for fish in SK, due to the presence of cartilaginous fish, which use electromagnetic signals in detecting prey (Gill 2005, Kimber et al 2011. EMF could also disturb fish migration patterns by interfering with their capacity to orientate in relation to the geomagnetic field, as indicated by empirical studies on eel (Westerberg and Begout-Anras 2000, Westerberg and Lagenfelt 2008, Gill et al 2012. The extent of EMF can potentially be mitigated by adequate cable design. Only few studies have addressed electroreception in marine mammals (Czech-Damal et al 2012) or invertebrates (Karlsen and Aristharkhov 1985, Aristharkhov et al 1988, Bochert and Zettler 2004, and no significant effects have been shown to date (table 4).

Conclusions
Whereas the construction phase was consistently associated with negative impact, pressures during the operational phase may impose both negative and positive effects, depending on local environmental conditions as well as prevailing management targets.
The assessment was made in three subareas with clear differences in species composition and abundance, but revealed similar general results. Thus, we conclude that the results may also facilitate initial impact assessments in other aquatic systems. The matrix in which the results are presented is highly simplistic, but transparent and adjustable to a finer geographical scale where local biodiversity patterns are well known, as well as to knowledge increase. It may also be used for comparing pressures, if combined with similar impact assessments for other marine activities, such as oil and gas extractions, fisheries, aquaculture, or other options for energy provision.
The strongest remaining uncertainties were seen for acoustic disturbances during the operational phase and effects of fisheries exclusion. As most empirical information today is from short-term studies in relatively small-scale OWF's, it is likely that conclusions made today will change when information accumulates from larger OWFs, over longer time scales, or when techniques to diminish negative impacts are developed. Current studies have to no or limited extent addressed combined effects, such as the effects of several marine activities within the same area, or long-term effects on the food web.
Many potential negative effects of OWF can be reduced within the planning process, by avoiding important recruitment habitats and by timing construction activities outside of important breeding seasons. Obviously, such measures should be based on real knowledge on the distribution and population status of local species and habitats. Given the high dependency of the obtained conclusion on local environmental conditions, a fundamental issue for the sustainable development of OWF is the availability of reliable seafloor and habitat maps and information on population connectivity.
The synthesis revealed a clear scope for research to identify holistic targets for marine management. In some cases it was not possible to value the anticipated impact into being either positive or negative, as this would depend on prevailing management goals. As there are obvious overall limits to human utilization of marine landscapes, it is clear that such comprehensive approaches are key to ensuring their sustainable management. OWF constitute a relatively new mode of usage of marine resources, and knowledge on its impacts has accumulated only in recent years. In this time, however, the development has provided a significant incentive for efforts to improve integrated coastal management strategies and marine spatial planning, and raised issues on long-term risks of human activities on marine habitats and species that are highly applicable also to other marine sectors.