Applicability of the "Frame of Reference" approach for environmental monitoring of offshore renewable energy projects.

This paper assesses the applicability of the Frame of Reference (FoR) approach for the environmental monitoring of large-scale offshore Marine Renewable Energy (MRE) projects. The focus is on projects harvesting energy from winds, waves and currents. Environmental concerns induced by MRE projects are reported based on a classification scheme identifying stressors, receptors, effects and impacts. Although the potential effects of stressors on most receptors are identified, there are large knowledge gaps regarding the corresponding (positive and negative) impacts. In that context, the development of offshore MRE requires the implementation of fit-for-purpose monitoring activities aimed at environmental protection and knowledge development. Taking European legislation as an example, it is suggested to adopt standardized monitoring protocols for the enhanced usage and utility of environmental indicators. Towards this objective, the use of the FoR approach is advocated since it provides guidance for the definition and use of coherent set of environmental state indicators. After a description of this framework, various examples of applications are provided considering a virtual MRE project located in European waters. Finally, some conclusions and recommendations are provided for the successful implementation of the FoR approach and for future studies.


Introduction
1 Offshore winds, waves and currents have a large poten-2 tial for long-term electricity generation world wide (Pelc 3 and Fujita, 2002;Thresher and Musial, 2010). The wind 4 industry is leading the way, whilst devices to harvest off-5 shore wave and current energy are still under development 6 (Sutherland et al., 2008;Inger et al., 2009;Bedard et al., Multi-platform or not, MRE projects are also expected to Indicators are commonly defined and organized in frame-89 works that facilitate their understanding and interpret-90 ation ensuring at the same time the appropriate match 91 between end-users and scientists (Gabrielsen and Bosch, 92 2003;Gubbay, 2004). Frameworks can also help to 93 understand the inter-relations between various indicat-94 ors (Stegnestam, 1999). Several environmental frame-  (Borja et al., 2006). This framework is useful 101 as a descriptive method reporting the environmental im-102 pacts of a particular sector through the use of indicators; 103 as such, it is largely used to report indicators set at na-104 tional levels and is able to provide a link between the socio-105 economic aspects of an activity and the induced environ-106 mental changes. DPSIR may be therefore well-adapted for 107 the strategic development of the offshore MRE industry 108 (Elliott, 2002). However, this type of framework might 109 not be relevant -or difficult to implement-if the focus is 110 on environmental monitoring of specific projects, where 111 guidance is required to select specific indicators. In this 112 case, other prescriptive and fully quantitative frameworks 113 that explicitly link objectives and quantitative parameters 114 are more adequate. 115 This paper assesses the applicability of the Frame of Refer-116 ence (FoR) approach for the environmental monitoring of 117 offshore MRE projects. Even though the proposed method 118 is applicable to any type of offshore large-scale project, the 119 focus is upon projects harvesting energy from winds, waves 120 and currents (multi-platform or not). Given the complexity of the marine environment and the 125 multiplicity of technologies to harvest MRE, it is conveni-126 ent to classify environmental effects within a framework. 127 The framework used in the present paper is based on the 128 one proposed by McMurray (2008) for wave converters, 129 subsequently modified by Boehlert and Gill (2010) and 130 by Polagye et al. (2011) for application to various MRE 131 devices.

132
The framework describes environmental concerns in terms 133 of stressors, receptors, effects and impacts. Stressors are 134 features that may induce environmental changes.  ors are elements of the ecosystem that may (or may not) 136 respond to the stressor. Effects describe how the receptor 137 is affected by the stressor, but do not indicate magnitude 138 or significance. Impacts deal with severity, intensity or 139 duration of the effect, and also with its direction (i.e., pos-140 itive or negative). Impacts are generally recognized when 141 the effects induce changes in specific variables that are 142 used to define the status of the concerned receptor. These 143 impacts can be either direct or indirect (the latter obvi-144 ously being more difficult to evaluate). Indicators can be 145 used to determine if the effects are strong enough to induce 146 impacts and if a response is required.   cope with the changes (e.g., Larsen and Guillemette, 2007;248 Masden et al., 2009). In general, the greatest risks faced 249 by marine mammals are hearing injuries and habitat loss 250 due to the production of sounds during the construction 251 phase (Bald et al., 2010;Wilhelmsson et al., 2010), even 252 though strikes by the blades of current devices may also be 253 of concern in some cases (Boehlert and Gill, 2010). Fur-254 thermore, the production of noise during operation may 255 mask bio-acoustics for communication and navigation of 256 long-distance migrating whales and sea turtles (Samuel 257 et al., 2005;Wilhelmsson et al., 2010). The newly con-258 structed structures may serve as steppingstones for in-259 vasive species (dispersal effect), which might pose as a 260 threat for local benthic and pelagic communities (Bulleri 261 and Airoldi, 2005;Glasby et al., 2007;Wilhelmsson et al., 262 2010). The production of magnetic fields by cables may 263 also modify the behaviour of resident or migratory species 264 that use geomagnetic field for localisation and orientation 265 (CMACS, 2003;Gill, 2005;Gill et al., 2009;Normandeau 266 et al., 2011;Wilhelmsson et al., 2010). Overall, oil slick 267 resulting from aircraft or ship accident is considered to 268 have the largest potential negative impact upon all recept-269 ors in terms of duration, spatial extent and intensity (Bald 270 et al., 2010).

271
The main potential positive impacts are associated with 272 the physical presence of MRE devices. The exclusion of 273 fishing activities, including trawling, within the project 274 area should act as a "no take zone", with positive impacts 275 for pelagic species (e.g., increase of fish stocks) and benthic 276 communities with a more favourable environment for long-277 lived rather than opportunistic species (Defew et al., 2012;278 Fayram and De Risi, 2007;Wilhelmsson et al., 2010). Fur-279 thermore, the introduction of hard structures (e.g., piles, 280 foundations, scouring protection, buoys) will provide ad-281 ditional (or new) settlement surface/habitat for benthic 282 organisms and fishes (Langhamer et al., 2009), thus act-283 ing as an "artificial reef" (Langhamer et al., 2009); ob-284 servations at artificial reefs and wind farms suggest that 285 this effect generally results in positive impacts in terms 286 of ecosystems and biodiversity (Petersen and Malm, 2006;287 Seaman, 2007; but see also Inger et al., 2009). In par-288 ticular, the new settled communities may attract pelagic 289 and nektonic organisms, forming a so-called "fish aggreg-290 ation device" (Wilhelmsson et al., 2006). The resulting 291 modification of pelagic and benthic habitats, communities 292 and prey distributions may in turn enhance feeding op-293 portunities for certain species of seabirds, cetaceans and 294 pinnipeds (Wilhelmsson et al., 2010). The key environmental regulations of offshore MRE activ-299 ities are similar in principles worldwide, as they derive 300 from various international agreements and conventions. In particular, an Environmental Impact Assessment (EIA) is generally required prior to project consent, in order to en-303 sure that the responsible authority makes a decision with 304 the full knowledge of any significant effects (cumulative, 305 positive and negative) on the environment. In this paper,

306
Europe is taken as an example since it is where the off- areas, standardized monitoring protocols should be adop-388 ted to enhance their usage and utility (see Johnson, 2008;389 Degraer and Brabant, 2009). In that context, the use of 390 tools such as the FoR approach may be useful for the defin-391 ition and use of coherent sets of environmental indicators 392 at offshore MRE projects. An additional benefit of these 393 tools is the possibility to compare between different applic-394 ations of the same indicator, in a process of gradual im-395 provement. These approaches also help to evaluate if the 396 cost of measuring the indicator is justified by the expected 397 gain (increased level of environmental protection). whether or not these objectives are met (Figure 1).

408
The FoR framework has been used so far for the imple-  The first step of the FoR approach is the formulation of 421 "strategic objectives" based on the long-term vision about 422 the desired status of the system (Figure 1). In a second 423 step, the means of satisfying (at least partly) each strategic 424 objective at the short-term are made explicit through the 425 definition of one or several "operational objectives". Fol-426 lowing Marchand et al. (2011) and Mulder et al. (2011), it 427 might be more adequate to qualify these objectives as "tac-428 tical" -rather than "operational"-because this step implies 429 a choice between distinct expedients to realise the corres-430 Obviously, the strategic objectives should derive from the 489 key environmental issues identified through the EIA pro-490 cess. They might follow from (national or international) 491 legislation, conventions or treaties. Strategic objectives 492 might as well derive from binding conditions set by envir-493 onmental agencies and local authorities for project con-494 sent, or by informal commitment of project managers. To 495 ensure that all separate elements of the generally long last-496 ing EIA process still fit together at the end of the pro-497 cess, a common framework for impacts classification, such 498 as the stressor-receptor framework described in Section 2 499 (Boehlert and Gill, 2010), should be adopted during both 500 the EIA and FoR procedures. This facilitates the com-501 munication between the various parties involved at any 502 stage of the EMP. For example, stressor-receptor matrixes 503 can be drawn to represent impacts severity, the temporal 504 and spatial scales and uncertainties (see Polagye et al., 505 2011).

506
The operational phase should rely on the specific actions 507 proposed in the EMP regarding the identification, follow-508 up and mitigation of impacts. At offshore wind farms 509 (and presumably at any type of future MRE projects), the 510 EMP commonly follows the Before-After Control-Impact 511 (BACI) approach (Green, 1979), where the current state 512 of the site is compared to previous and/or pristine environ-513 mental conditions known from the baseline study and from 514 concurrent measurements at "reference areas". The defin-515 ition of thresholds representing the "desired state" might 516 be one of the most difficult tasks of impacts evaluation, 517 since natural temporal and spatial variability of paramet-518 ers must be considered. In some cases, indicator thresholds 519 are fixed by legal requirements. In most cases, however, 520 they must be established based on robust expertise to-521 gether with a good knowledge of the natural environmental 522 conditions at various spatial and temporal scales. Simil-523 arly, monitoring surveys alone might not be enough to em-524 brace the natural variability (both spatial and temporal) 525 of the measured parameters. The establishment of the cur-526 rent state may also be based on statistical and numerical 527 models. Not only do these tools allow the interpretation of 528 a limited number of measurements over broader areas and 529 longer time-scales, but they can also be useful in defining 530 environmental policies.    Water quality: The water quality may be affected as a 575 result of oil spilling from components (e.g., gear boxes, 576 hydraulic pumps) of MRE devices, and also from ves-577 sels and helicopters supporting the activity.

578
Regarding harbour porpoise protection against underwa-579 ter sound, the long-term strategic objective of the FoR 580 could be 'to preserve the regional population given the 581 planned activity' (Table 3). Studies have shown that wind 582 farm related sound, for example, has the potential to af-583 fect the behaviour and physiology of harbour porpoises at 584 considerable distances. Physiological effects include Tem-585 porary and Permanent Hearing Threshold Shifts and more 586 severe injuries up to death, depending of the distance of 587 the individual to the source. Hence, one tactical object-588 ive could be that 'no porpoise should suffer from sound 589 related to the activity' (Table 3). More specifically this 590 objective could be achieved by either reducing the sound 591 at the source or by physically keeping the porpoise away 592 from areas where sound levels are potentially harmful. The 593 underwater sound hazard is greatest during the construc-594 tion phase, when porpoises are present in the area. Past 595 experience has shown that porpoises avoid areas where pil-596 ing activities take place; lethal hearing injuries may occur 597 if they are located too close to the source when hammer-598 ing starts. In this example we will focus on keeping the 599 porpoises at a safe distance from the sound source. Re-600 cent studies have indicated that during piling, severe in-601 juries are estimated to occur in a radius of 1.8 km from 602 the source (Thomsen et al., 2006). The ESI in this case 603 might be derived from the observation of the 'number of 604 Figure 2: Example of field set up for an operational phase designed to scare porpoises during piling (see tactical objective). Eight pingers (dots) with 1 km range each (see the dark grey area, represented for one pinger only) are used to scare porpoises before piling starts to allow them to escape the area. Marine mammal observers, one at each pinger location plus one at the piling site, establish the current state (benchmarking), by checking for the presence of porpoise in the 2 km radius area of potential injuries by sound (light grey area). Piling is allowed to start if no porpoise is observed in this area during the preceding 2 hr (ESI).
individuals within 2 km from the source, after the deploy-  ing the proposed procedure it seems likely that this FoR 620 will contribute to its strategic objective. The tactical ob-621 jective is vulnerable to marine mammals observers missing 622 porpoises that still are present despite the period of har-623 assment. Put in practice procedures should be optimized 624 to minimize this risk.

626
The proposed project will undoubtedly induce local phys-627 ical changes of habitat (if only for the introduction of hard 628 structures in the water). The development of organisms 629 such as mussels on the structures may create locally "hot 630 spots" of biological activity (e.g., Norling and Kautsky, 631 2008) that could be beneficial to the ecosystem (including 632 the shoal benthic community). Thus, one strategic ob-633 jective could be to enhance biodiversity and productivity, 634 providing there is no negative impact -for example due to 635    Table 3). This approach 644 supposes that it has been previously evidenced that reduc-645 tion of this habitat induces negative impacts (at present, 646 this effect is generally not a major concern, but it could 647 become substantial in the case of farms with several hun-648 dreds of devices operating for decades). It is technically 649 difficult and costly to inhibit the colonisation of organisms 650 on newly introduced material. Thus, the tactical object-651 ive could be that 'shell beds should not cover more than 652 25% of the project area during operation' (Table 3). The 653 QSC stage may define the parameter to quantify the ex-654 tension of shell beds as the 'relative percentage of surface 655 area which is covered by shell deposits within a subset 656 region (selected randomly) corresponding to 10% in sur-657 face area of the proposed project' (Table 3). This requires 658 that the distinction between sandy and mussel bed hab-659 itats is clearly defined at the benchmarking procedure, as 660 it depends on the method used to establish the current 661 state. For example, if mechanical sediment sampling is 662 involved, the classification of sand mixed with mussels as 663 "sandy" or "shell" habitats requires the definition of limit 664 values, e.g., in terms of relative weight of shells or grain size 665 parameters. Our example contemplates (ground-truthed) 666 side-scan sonar images since they generally allow a clear 667 distinction of hard and soft beds based on their tonal con-668 trast ( Table 3). The desired state corresponds to 'less than 669 25% of shell beds within the subset area'. More complex 670 proactive approaches could rely on the outputs from mod-671 els of mussels growth and deposition (e.g., Maar et al., 672 2009). The intervention procedure could encompass envir-673 Figure 3: Implementation of the operational phase designed to control the development of mussel beds within the project area (Tactical objective). The total area (plan view) is 100 km 2 and includes 176 MRE devices (dots). Ground-truthed side-scan sonar surveys are conducted in a 10 km 2 area (dashed line) to distinguish sandy bed (white) from shell beds (dark grey). If shell beds represents > 25% of the survey area, environmental dredging is performed along parallel corridors over the entire project area to remove the mussel layer (examples of these corridors are shown in light grey).  Any impact of the project on migrating Common Eider 694 ducks must be analysed at a population level. As a stra-695 tegic objective, the project activities should 'preserve the 696 population of Common Eider ducks passing over the re-697 gion' (Table 3), where the extension of the "region" is 698 clearly defined. One way of meeting this objective could be 699 'to prevent any increase of their mortality rate related to 700 the project activities' (Tactical objective, Table 3). At the 701 QSC stage, this objective may lead to the development of 702 a parameter representing the percentage of the duck pop-703 ulation risking collision with the structures. For selected 704 periods, the current state can be predicted with a level of 705 certainty (for example, 95% confidence interval) based on 706 stochastic models built from compilations of observations 707 (Figure 4; Table 3). In particular, reliable model predic-708 tions require estimates of the number of individual Eider 709 duck collisions within the project area and of their fluxes 710 throughout the project area (e.g., Band, 2000;Petersen 711 et al., 2006;Troost, 2008). Collision estimates can be ob-712 tained using non-contact sensors (e.g., acoustic sensors, 713 microphones) deployed on a number of turbines, especially 714 during periods of heavy migration (spring and autumn). 715 Likewise, surveillance radars are useful to measure the 716 volume of bird movement and to track their altitude and 717 trajectories through the area (visual observations are also 718 necessary to calibrate the radar signal for species distinc-719 Figure 4: Illustration of the definition of the current state for selected periods of Eider duck migration. Observations (radar and collision monitoring data) are compiled to build a stochastic model. The current state is derived from model prediction of the percentage of bird risking collision with the turbines at a 95% confidence interval. tion). The desired state (Table 3), for example 'no more 720 than 5% of the population predicted to collide', should be 721 fixed considering the effects of the increased mortality on 722 the population over longer time periods (Fox et al., 2006).

723
As a proactive intervention measure, it could be possible to 724 shut down turbines during the periods when the indicator 725 threshold is exceeded (Table 3).

726
With respect to water pollution, one obvious strategic ob-727 jective is 'to preserve favourable water quality for local 728 flora and fauna' ( Table 3). One of the various tactics that 729 can contribute to meet this objective is to ensure a 'timely 730 proactive maintenance of the operating devices containing 731 oil', e.g., gear boxes, hoses, in order to prevent oil leaks from happening (Table 3). In this case, the QSC can make impacts. Thorough long-term (years) EMPs should be 755 implemented in order to enhance scientific knowledge re-756 garding impacts. The implementation of environmental 757 indicators within these programs is recommended as they 758 generally describe in a convenient (simplified) manner the 759 status of (complex) systems and thus may facilitate man-760 agement decisions.

761
For the implementation of indicators within EMPs, the 762 FoR approach is advocated over other frameworks due 763 to its prescriptive nature. The FoR provides clear guid-764 ance for the selection of indicators that are linked dir-765 ectly to specific management issues. This framework also 766 makes sure that predefined intervention procedures will be 767 implemented if mitigation or remediation actions are re-768 quired.

769
The examples presented in this contribution describe the 770 use of FoR as a remediation tool. However, the most effect-771 ive options to mitigate environmental impacts are gener-772 ally available during the design phase of the project, i.e., 773 during the selection of the site, of the technology to be 774 used, and of the project layout. It is recommended to im-775 plement the FoR approach at various phases of the lifecycle 776 of a project.

777
The development of a FoR framework is recommended 778 for each of the potential environmental issues. The pos-779 sible interaction between management issues from differ-780 ent FoRs must be addressed. In particular, future re-781 search should seek to optimise the collaborative effort not 782 only between scientists of distinct disciplines, but also 783 between all the parties at stake (non-governmental organ-784 isations, nature conservation organisations, stakeholders, managers, policy makers). Furthermore, the development 786 of several FoRs may require the integration of various time 787 and space scales. It is therefore also recommended to 788 investigate the articulation between the distinct manage-789 ment scales (see Mulder et al., 2006).

790
The occurrence of many environmental issues may also and international level (e.g., the "OpenEarth" approach;  Oceanography Society 23 (2)