Sensitivity and Vulnerability in Marine Environments: an Approach to Identifying Vulnerable Marine Areas MARK A. ZACHARIAS∗ AND EDWARD J. GREGR† Environmental Science and Resource Management Program, California State University–Channel Islands, 1 University Drive, Camarillo, CA 93012, U.S.A., email mark.zacharias@csuci.edu †SciTech Consulting, 1511 East 3rd Avenue, Vancouver, British Columbia V5N 1G8, Canada ∗ Abstract: Marine environments have suffered from a lack of quantitative methods for delineating areas that are sensitive or vulnerable to particular stresses, natural and anthropogenic. We define sensitivity as the degree to which marine features respond to stresses, which are deviations of environmental conditions beyond the expected range. Vulnerability can then be defined as the probability that a feature will be exposed to a stress to which it is sensitive. Using these definitions, we provide a quantitative methodology for identifying vulnerable marine areas based on valued ecological features, defined as biological or physical features, processes, or structures deemed by humans to have environmental, social, cultural, or economic significance. The vulnerability of the valued ecological features is a function of their sensitivity to particular stresses and their vulnerability to those stresses. We used the methodology to demonstrate how vulnerable marine areas for two groups of endangered whale species (inshore and offshore) could be identified with a predictive habitat model and acoustic stress surfaces. Acoustic stress surfaces were produced for ferry traffic, commercial shipping traffic, potential offshore oil production, and small-boat traffic. The vulnerabilities of the two whale groups to the four stressors considered in this example were relatively similar; however, inshore species were more sensitive to on-shelf, coastal activities such as offshore hydrocarbon production, ferry traffic, and small-boat traffic. Our approach demonstrates how valued features can be associated with stresses and the likelihood of encountering these stresses (vulnerability) in order to identify geographic areas for management and conservation purposes. The method can be applied to any combination of valued ecological features and stressors. Key Words: acoustic pollution, acoustic stress, cetaceans, marine protected areas, sensitive areas Sensibilidad y Vulnerabilidad en Ambientes Marinos: una Metodologı́a para Identificar Áreas Marinas Vulnerables Resumen: Los ambientes marinos han carecido de métodos cuantitativos para delinear áreas sensibles o vulnerables a ciertos estreses (naturales y antropogénicos). Definimos sensibilidad como el grado en que los atributos marinos responden a estreses, que son desviaciones de condiciones ambientales más allá de los lı́mites esperados. Vulnerabilidad, entonces, puede ser definida como la probabilidad de que un atributo esté expuesto a un estrés al que es sensible. Utilizando estas definiciones, aportamos una metodologı́a cuantitativa para identificar áreas marinas vulnerables. La metodologı́a identifica áreas marinas vulnerables con base en atributos ecológicos valiosos (definidos como caracterı́sticas, procesos o estructuras biológicas que tienen alguna significancia ambiental, social, cultural o económica para los humanos). La vulnerabilidad de los atributos ecológicos valiosos es una función de su sensibilidad a estreses particulares y su vulnerabilidad a esos estreses. Utilizamos la metodologı́a para demostrar como se pueden identificar áreas marinas vulnerables para dos grupos de especies de ballenas en peligro (cerca de la costa y lejos de la costa) mediante un modelo predictivo de hábitat y superficies de estrés acústico. Se produjeron superficies de estrés acústico para tráfico de transbordadores, tráfico de barcos comerciales, potencial producción de petróleo lejos de la costa y tráfico de embarcaciones pequeñas. La vulnerabilidad de los dos grupos de ballenas a los cuatro estreses considerados en este ejemplo fueron relativamente similares; sin embargo, las especies cercanas a la costa fueron más sensibles a actividades costeras como la producción de hidrocarburos, tráfico de transbordadores y de pequeñas embarcaciones. Nuestra metodologı́a demuestra como se pueden asociar atributos valiosos con estreses y la probabilidad de toparse con esos estreses (vulnerabilidad) para identificar áreas geográficas para su manejo y conservación. El método puede ser aplicado a cualquier combinación de atributos ecológicos valiosos y estreses. Palabras Clave: áreas marinas protegidas, áreas sensibles, cetáceos, contaminación acústica, estrés acústico Introduction The assumption that certain geographic areas are more sensitive or vulnerable—no matter how these terms are defined—than their surroundings has been a central tenant in terrestrial conservation and management since the 1920s (Golley 1993). The concepts of sensitivity and vulnerability have been commonly used as criteria (along with representativeness and uniqueness) in the identification of areas requiring special management or protection. However, efforts to identify sensitive or vulnerable areas are not unique to terrestrial environments. Nearly 30 years ago the International Convention for the Prevention of Pollution from Ships (MARPOL 73/78) defined certain high-traffic marine areas as “special areas” according to their ecological and oceanographic importance. Within these special areas the release of oil, ballast water, and plastics near coastlines and in environmentally sensitive areas (initially the Baltic, Mediterranean, Black and Red seas) was limited. The special-areas concept led the International Maritime Organization to establish particularly sensitive sea areas (PSSAs), which are recognized for their ecological, socioeconomic, or scientific importance and their vulnerability to damage by international maritime activities. Both special areas and PSSAs are implemented over large areas and encompass either regional seas or large portions of oceans (Van Bernem & Bluhm 2000; Johnson et al. 2001; Gjerde 2002). The earliest examples of identifying sensitive and vulnerable marine areas at higher resolution are found in the environmental sensitivity index (ESI) concept and procedures developed for responding to oil spills and planning countermeasures (Gundlach & Hayes 1978; Michel et al. 1978). These procedures are designed to prioritize response activities and specify appropriate clean-up techniques. Environmental sensitivity indexes are based on physical and biological habitat classifications from which models of sensitivity and vulnerability to oiling and cleanup are generated. Current shoreline classifications developed to support ESIs are hierarchical, encompass many types of resources (e.g., human use, birds), and have successfully mapped large areas at very high resolution (e.g., 1:5,000–1:10,000 scales) (Howes et al. 1994; International Petroleum Industry Environmental Conservation Association 1996; National Oceanic and Atmospheric Association 1996). There is increasing interest in identifying sensitive and vulnerable marine areas at spatial resolutions approximating those of marine management and conservation efforts (e.g., at a scale of 1:250,000, or between PSSAs and ESIs). Dutrieux et al. (2000) proposed a mapping-based methodology to identify vulnerable coastal nearshore areas in the Indian Ocean based on assessments of sensitivity and risk. Tyler-Walters and Jackson (1999) developed a methodology to assess the sensitivity and recovery of marine species and biotopes in the northeast Atlantic. Gordon et al. (1998) defined sensitive areas for European marine mammals based on areas important for multiple species. These efforts, however, suffer from a number of limitations and deficiencies. The terminology used in these studies lacks common, mutually agreed-upon definitions. This makes relating the various studies difficult. There is also a lack of systematic, quantitative methods for delineating sensitive and vulnerable areas (Emmett & Wainwright 2001). Areas are often selected based on an expert knowledge approach, which limits the predictive ability of these efforts. Our purpose was first to operationally define the terms sensitive and vulnerable in an explicit and quantifiable manner. We then integrated these terms into a methodology that quantitatively defines vulnerable marine areas (VMAs), allowing these areas to be predicted and mapped. We used a predictive habitat model and hypothetical acoustic pollution surfaces to apply the methodology to large whales. Our methodology will assist decision makers in assessing and comparing the ecological sensitivity and vulnerability of various types of marine areas. This ability to effectively prioritize marine areas according to their sensitivity, based on particular threats, will be a valuable tool in coastal and marine planning and management. Definitions of Sensitivity and Vulnerability Sensitivity and vulnerability are central concepts in the protection of marine ecosystems, yet the marine literature provides few explicit definitions (Holt et al. 1997; Tyler-Walters & Jackson 1999). Consequently, their meaning often relies on the context in which they are used. To avoid context-dependent definitions of these terms, we took the following value-neutral approach to defining sensitivity and vulnerability. It is axiomatic that all marine features have either evolved (in the case of biotic features) or been formed (in the case of abiotic features) within a certain range of environmental conditions. We define stress as a deviation of these environmental conditions beyond the expected range. Sensitivity is the degree to which marine features respond to such stress. Specifically, sensitivity is measured using one or more indicators (of species, communities, and habitats) that respond to one or more natural or anthropogenic stressors. These responses are potentially nonlinear and are likely to include interactions between stressors. In this context, sensitivity does not inherently assume the characteristics of fragility or intolerance with which it is often associated. There is no implied judgement that an increased association between the indicator and the stressor reduces a feature’s probability of persistence. Nevertheless, as exposure to a chronic perturbation or stress increases, the persistence of that feature is diminished. Vulnerability is the probability that a feature will be exposed to a stressor to which it is sensitive. In other words, vulnerability is the likelihood of exposure to a relevant external stress factor (sensu Tyler-Walters & Jackson 1999), combined in some way with the exposure (duration, magnitude, rate of change) to that stress. Subsumed under the concepts of sensitivity and vulnerability are the concepts of stability and fragility. Although these terms lack general consensus on their definitions, Holling (1986) states that stability is the tendency of a system to attain or retain an equilibrium condition of steady state or stable oscillation. Resilience is the ability of a system to maintain its structure and behavioral patterns when subjected to disturbance. A feature, therefore, that is stable or resilient in the presence of a stressor is not sensitive to that stress as we have defined it. Also, a feature that is sensitive to a stressor for which it has a low probability of exposure is not vulnerable. Our definitions of sensitivity and vulnerability are consistent with the ESI approach used for oil spill response and countermeasures (Gundlach & Hayes 1978). Under the ESI approach a resource is defined as sensitive to oil if it would be harmed by physical contact with oil or concentrations of oil in water. A resource is defined as vulnerable if it is likely that it would be exposed to oil or high concentrations of oil for long enough periods for the oil to affect it. Vulnerability in terms of oil spills on shorelines, therefore, is a function of duration of exposure, recognizing that certain resources are vulnerable to oiling regardless of duration. Our definition of a VMA, which incorporates the concepts of sensitivity and vulnerability, is a geographically definable area containing features that are sensitive to natural and/or anthropogenic stressors they are likely to encounter. Features may be biotic (species, communities) or abiotic (habitats) structures or processes. We caution against equating VMAs as we have defined them (and their sensitivity and vulnerability components) with terms such as priority areas, biodiversity hotspots, critical habitat, environmental significance, and areas of interest, which are often used to identify areas of special concern or areas requiring management attention. Although certain VMAs may also represent priority areas or hotspots, depending on how these terms are defined and applied, an area may be sensitive or vulnerable but not meet criteria commonly used (e.g., species diversity) to identify these areas. In addition, although our definitions of sensitivity and vulnerability may contribute to measures or assessments of ecological (or ecosystem) integrity, these concepts are again different. Lastly, identification of an area as sensitive or vulnerable does not suggest that an area should be recommended as a marine protected area (MPA) or marine reserve or that MPAs are the only management tool applicable to the management and conservation of VMAs. Viability and Efficacy of the VMA Concept in Marine Environments There are difficulties with implementing the VMA concept in marine environments. First, the spatial and temporal heterogeneity resulting from the dynamic nature of the ocean cannot be easily bounded by spatial referencing and treated as a feature with the terrestrial connotations of spatiotemporal homogeneity. This oceanographic heterogeneity is reflected in the ephemeral and patchy nature of marine biological communities, which are also, with the exception of benthic communities, difficult to spatially reference with any certainty to a geographic location at the spatial and temporal scales of interest. Associating VMAs with physiographic (i.e., benthic) features allows them to be geographically referenced, making them more stable in space and time, thereby facilitating the mapping and analysis of these features. Vulnerable marine areas will still be spatially variable (e.g., productivity associated with continental shelf breaks or interannual variation in kelp bed location), but this variability can be quantified with respect to a benthic feature, which may be biological (e.g., seagrass bed) or physiographic (e.g., seamount). Second, the marine environment is biogeochemically downstream from terrestrial environments; therefore, neritic (continental shelf ) VMAs may be structured or dependent to some degree on terrestrial inputs in the form of nutrients and energy. These terrestrial processes may operate independently of marine processes. Therefore, identification of VMAs in neritic environments will require an understanding of both marine and terrestrial processes and their interactions. Third, the characteristics (e.g., sources of nutrients, food, or recruits) that make a feature sensitive or vulnerable may be transported or advected to areas where they accumulate. Thus, a VMA may be identified in a particular location, but the mechanisms that structure it may be operating elsewhere. Using a recruitment analogy, there is a probability that sinks, rather than sources, will be identified (Dias 1996). Fourth, with the exception of certain well-studied nearshore communities, a lack of baseline data or incomplete knowledge of climax communities may result in VMAs that are either intermediate successional stages or identified incorrectly as a result of past human activities and practices. Lastly, many species of concern are migratory; therefore, some areas (e.g., staging areas) may only be vulnerable at certain times of the year. Clearly, designating sensitive or vulnerable areas in a dynamic, heterogeneous, three-dimensional fluid environment is challenging. Many of the marine features valued by humans, however, demonstrate site fidelity over space and time and can therefore be spatially referenced. Therefore, the VMA concept as we define it appears to have some application in marine environments. Methodology for Identification of VMAs Overview Our methodology for identifying VMAs is based on the a priori identification of valued ecological features (VEFs). These are biological or physical features, processes, or structures deemed by humans to have environmental, social, cultural, or economic significance. The incorporation of the term valued in the VEF concept reflects the social and political reality that if a feature were not valued in some way there would be few data with which to spatially bound and map it. From a pragmatic perspective, if a feature does not have value, there is little impetus to determine whether it is sensitive or vulnerable. Vulnerable marine areas are defined by evaluating VEFs in terms of their sensitivity and vulnerability to particular stressors. The VMA concept lends itself to spatial mapping and prediction because the distribution of both features and stressors can be mapped. Determining whether a VEF becomes a VMA then depends on whether the VEF is vulnerable to one or more stressors. For many features, this can be estimated in terms of spatial (or temporal) distance between the VEF and the stressor of interest. If VEFs can be spatially associated with particular biological, oceanographic, and physiographic conditions or habitats that have been systematically mapped or classified over larger regions, then the presence of VMAs can be predicted. This methodology, therefore, supports the identification of VMAs and allows their prediction over larger areas. The methodology can be applied at different scales (e.g., regional vs. local) and can be selectively applied for specific purposes (e.g., marine reserve identification, offshore hydrocarbon development). The methodology can also accommodate changes and updates such as the incorporation of traditional knowledge, new inventory data, or updated scientific results. Components of the Methodology The methodology includes three components: (1) identification and mapping of VEFs, (2) identification of VEF sensitivities and vulnerabilities, and (3) prediction of VMAs with ecological classifications. IDENTIFICATION AND MAPPING OF VALUED ECOLOGICAL FEATURES Valued ecological features are biological or physical features (e.g., kelp beds, seamounts), processes (e.g., upwellings), or structures (e.g., persistent fronts) that have environmental, social, cultural, or economic significance (Dale 1997). Valued ecological features may be species, habitats, communities, or processes that contribute to the formation and maintenance of populations, habitats, or communities. The scales of VEFs are determined by the spatial and temporal extent of the feature. There is therefore no single appropriate scale for VEFs. Valued ecological features may be identified according to a number of criteria, including the ecological importance of the VEF gained from direct observation (e.g., seagrass beds), inference from other surrogate measures (e.g., temperature gradients indicating fronts), or traditional knowledge (Dale 1997). Valued ecological features can also be examined in terms of the various combinations of structures and processes that contribute to their creation and maintenance. The associations between structure and process can be used to recognize potential VEFs or identify potential information sources of VEF characterization. Although individually these structures and processes might not warrant attention, when they co-occur they may form a new VEF or influence an existing one. Ten general combinations of structures and processes can be used to facilitate the identification of VEFs. These combinations, which we term compositional attributes, are listed in Table 1 along with types of impacts and potential indicators of stress. Valued ecological features must be able to be spatially delineated and mapped. Mapping certain VEFs may be as simple as identifying the extent of a particular biological or physical structure or may require the integration of multiple data sets that denote certain processes, implying the need for a temporal component in at least some cases. In the case of wide-ranging marine animals, it will likely be necessary to develop detailed species-habitat relationships. Most nations have implemented marine geographic Table 1. Valued ecological features (VEFs) can be identified, mapped, and predicted from various combinations of structures and processes that contribute to their creation and maintenance.∗ Compositional elements Code structure process CE1 fine sediments and shallow depth — CE2 CE3 fine sediments and shallow depth shoreline protected exposure or poor flushing high rate of (active) sediment transport CE4 — Ecological effect stress class impact consequence indicator class Examples of VEFs chemical contaminants; physiological biological stress contaminants; nutrient loading sedimentation; organic physical loading disturbance transport mechanisms erosion or disturbed or disrupted deposition community composition changed change in chemical water properties intertidal deltas, estuaries community diversity reduced nearshore habitat altered change in chemical water properties change in shoreline characteristics restricted circulation; chemical contaminants; reduced or stable water masses biological modified contaminants; physical circulation disturbance patterns characteristic benthic slow recovery rates physical disturbance habitat physical structures destruction community composition changed change in chemical or physical water properties mudflats, sandflats, estuaries spits, unconsolidated cliffs, channels with active sedimentation transport coastal fjords with sills, tidal lagoons habitat availability reduced amount of habitat disturbed CE6 characteristic benthic slow recovery rates biological structures habitat effectiveness amount of habitat reduced disturbed CE7 characteristic — physical landscape structures characteristic short-term, ephemeral physical regional-scale structures processes CE5 CE8 CE9 physical structures CE10 biological structures ∗ Structures used by a significant component of a population used by a significant component of a population chemical contaminants; reduced biological abundance contaminants; physical disturbance physical disturbance reduced abundance habitat diversity reduced transport mechanisms habitat ecosystem function disturbed or disrupted; destruction; reduced physical disturbance reduced abundance physical disturbance; behavioral change; population decline coastal structures habitat avoidance physical disturbance; behavior disturbance behavioral change; significant change habitat in use patterns avoidance; mortality amount of habitat disturbed change in physical water properties amount of habitat disturbed change in species distribution sponge reefs, geothermal vents, seamounts, coral reefs marshes, eelgrass beds, mangroves archipelagos, haulout rocks, bird cliffs, passages meso-scale eddies and gyres, fronts, upwelling zones, polynyas fish spawning areas, marine bird moulting and breeding areas, marine mammal mating areas rare and endangered species and habitats, fish holding and feeding areas, regionally significant populations (clams, oysters, marine mammals, marine birds) and processes can be used to recognize potential VEFs and identify potential information sources of VEF characterization. Ten general combinations of structures and processes (compositional elements) along with their associated types of impacts and potential indicators of stress can be used to facilitate the identification of VEFs. information systems (GIS) to capture some typical marine VEFs (Zacharias et al. 1998). In addition, many nations are beginning to map and predict VEFs with marine ecological classifications (Allee et al. 2000; Roff & Taylor 2000). Little work has been done, however, to predict VEFs beyond a few miles from shore or in depths of >50 m. IDENTIFICATION OF VEF SENSITIVITIES AND VULNERABILITIES Once the boundaries of a VEF are defined, a matrix is then developed listing the VEFs of interest and the potential stressors associated with each stress class (Table 2) (Tyler-Walters & Jackson 1999). In this value-stress matrix, each VEF may be sensitive to one or more of the potential stressors. This identifies a subset of cells in the row for which the VEF should be evaluated for sensitivity. The bottom of Table 2 lists stresses on various whale species, which form the basis of our example ( below). A key aspect of assessing sensitivity is the selection of appropriate indicators, which must be sensitive to the stressors being considered. In contrast to that of Tyler-Walters and Jackson (1999), our approach generates a sensitivity surface instead of a simple rank. The sensitivity surface represents sensitivity as a spatially distributed probability. Thus, each cell in the value-stress matrix is represented by a probability map in which a probability of 1.0 indicates certain overlap between the VEF and the stressor and in which a probability of 0.0 indicates no overlap. Once all the relevant stressors for a particular VEF are evaluated, the comprehensive spatial sensitivity of that VEF can be represented by the combined probability of the individual sensitivity surfaces. PREDICTION OF VMAS WITH ECOLOGICAL CLASSIFICATIONS A number of systematic, hierarchical marine ecological classifications have been developed (Zacharias & Howes 1998; Allee et al. 2000; Roff & Taylor 2000). Properly defined, these classifications could provide the ideal spatial unit for the definition and prediction of VMAs by acting as the building blocks of marine landscapes in an analogous manner to the “functional landscapes” in terrestrial ecology. Although a detailed discussion of these classifications is outside the scope of this paper, more well-known classifications include the large marine ecosystem concept (Sherman 1991), the Cowardin (1979) coastal and estuarine classification, and various intertidal classifications (e.g., Howes et al. 1994). Defining VMAs for Large Whale Species We combined a species-habitat model for large whale species with a hypothetical acoustic stress surface to pre- dict potential locations of pelagic VMAs. We chose large whale species for this example because their habitats represent the most complex type of VMAs—those comprised of ephemeral marine structures used by highly mobile marine animals. Our results represent an example of the application of the VMA methodology only. The stress surfaces are based on a number of untested assumptions regarding marine mammal sensitivity to acoustic pollution and on a combination of existing and potential acoustic sources anticipated from planned offshore hydrocarbon development on Canada’s western continental shelf. Additional research would therefore be required to make the VMAs presented definitive. We used two VEFs in this analysis: the humpback whale (Megaptera novaengliae) and the balaenopterid group, which includes sei (Balaenoptera borealis), fin (B. physalus), and blue (B. musculus) whales. The humpback whale occurs in both inshore coastal and offshore pelagic habitats, whereas the balaenopterids occur mainly in the offshore waters of the eastern North Pacific and regularly use the shelf break as a feeding area. We used species-habitat relationships developed by Gregr and Trites (2001) to define the spatial extent of these VEFs. The threats (stressors) we examined represent those most often associated with acoustic pollution in the marine environment. Acoustic pollution is produced at a range of frequencies and levels. Primary sources include shipping, seismic surveying, sonars, explosions, and industrial activity (Wenz 1962; Gordon & Moscrop 1996; National Research Council 2003). However, vessel traffic (small and large) is the most significant source of chronic acoustic pollution in the marine environment. Supertanker noise is one of the most powerful low-frequency human-made sounds. A 6.8-Hz tone from a supertanker has been detected from between 139 and 463 km away, with source levels estimated at 190 dB (all source levels referenced to 1 µPa) in the 40- to 70-Hz range (Gordon & Moscrop 1996). Small outboard engines have been estimated to produce noise levels of 140 dB at a range of 50 m and a frequency of 400–4000 Hz. Documented responses of cetaceans to acoustic pollution include avoidance, altered behavior, and avoidance (Gordon & Moscrop 1996; Moore & Clarke 2002; Williams et al. 2002). Some odontocetes occasionally approach vessels to bow ride, and some cetaceans habituated to vessel traffic have been observed to approach vessels, apparently to socialize ( Lusseau 2003; National Research Council 2003). The effects of noise on the physiology and psychology of marine mammals are poorly understood (Richardson et al. 1995). Experimental studies of hearing ability have only been conducted with a few odontocete and pinniped species (Gordon et al. 1998; National Research Council 2003). It is assumed that mysticetes hear over the same frequency range as they produce (approximately 10Hz–5 kHz) (Richardson et al. 1995). Thus, the Table 2. Examples of valued ecological features (VEFs) identified in the eastern North Pacific and their potential stress classes (modified from Dale 1997).a a The X X X X X vessel congestion X vessel strikes noise pollution X prey removal X indirect mortality X bottom trawl direct mortality X X thermal pollution X X marine platforms X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X sensitivity of each value feature is evaluated against all relevant stress classes (marked with an X). of cetacean VEFs discussed in this paper. b Examples change in freshwater input dams reclamation research activities dredging and infilling X X X dumping X X X dykes X X X Community disturbance breakwaters groins X X X X X X Physical disturbance industrial processing X X X X X X X X X log handling and storage X X X land-cover alteration dredging X X X X X upstream/upland development X X X X X X urban runoff, unprocessed X X X X urban runoff, processed X X X X X X X X X X X X X X X X X X agricultural runoff X species introductions, micro X X X bacterial and viral species introductions, macro X X X sediment mobilization non–point source, marine X X X sewage/outflow non–point source, terrestrial Community types estuaries X salt marshes X seagrass beds X kelp beds tidal flats X Other habitats rocky reefs seamounts sills and ridges anoxic basins offshore banks canyons tidal passages Processes localized upwellings cold seeps fronts plumes gyres currents productivity X Biogenic communities biogenic reefs X gorgonian corals sponges hydrothermal vents Focal species X fish or invertbrate habitat spawning areas X juvenile X nurseries seabirds X waterfowl X shorebirds X pinnipeds X (haul-outs) sea otters X X rare or endangered species Baleen migrationb Humpback migrationb Baleen feeding X areasb X Humpback feeding areasb oil spills Value ecological feature ( VEFs) chemical spills Chemical Biological Nutrient Sedimentation and contamination contamination loading organic loading mainly low-frequency noises from large ships are within the presumed acoustic sensitivity of balaenopterids. The higher frequencies from small boats may have a greater effect on small cetaceans (Richardson et al. 1995; Gordon & Moscrop 1996). Seismic surveys are usually conducted using air guns that generate primarily low-frequency sounds as short pulses lasting fractions of a second and repeated every 5–15 seconds. Although the arrays are normally pointed downward, significant amounts of sound energy are projected from the sides of the arrays. Source levels of well over 200 dB have been measured. Because of the frequency of the short, multiple pulses, received pulses from the air gun arrays generally increase in duration with distance from the source as a result of multipath propagation (Gordon & Moscrop 1996). Studies of responses by marine mammals to seismic surveys have documented behavioral change(s) at up to 10 km for blue whales (Macdonald et al. 1995) and 8 km for humpback whales (McCauley et al. 1998). Several studies have therefore suggested scheduling seismic exploration during periods when the potentially affected species are absent (LGL Consulting 1998; LGL Consulting 2000; Environment Australia 2001; Moore & Clarke 2002). This mitigation strategy may also apply to the VEFs considered in our example. We therefore did not consider noise associated with offshore hydrocarbon exploration. Oil drilling platforms affect the distribution of bowhead whales (Balaena mysticetus) at distances of up to 50 km (Schick & Urban 2000). Although no experiments have been conducted to conclusively demonstrate the causeand-effect relationship, levels of 143 dB in the 20- to 1000Hz range were recorded 1 km from an oil platform off California, suggesting that a wide range of frequencies are audible at significant distances from these platforms. Drilling as part of oil operations has been documented to generate sound with strong tonal components at low frequencies (<20Hz) (Gordon & Moscrop 1996). Figure 1. Humpback whale habitat (mean annual) as predicted by logistic regression and smoothed with a moving average window. Locations of historic whale harvests (Nichol et al. 2002) were related to bathymetry, depth, slope, and mean annual temperature and salinity. Probability of habitat is shown increasing from light (white = 0.0) to dark ( black = 1.0) (after Gregr & Trites 2001). Procedure The distributions of the VEFs in our example were based on habitat predictions developed by Gregr and Trites (2001) and represent mean annual habitat use for the component taxa. These predictions were based on a regression analysis that related the locations of historic whale harvests (Nichol et al. 2002) to bathymetry, depth, slope, and mean annual temperature and salinity. The model predictions resulting from the regression analysis were smoothed with a moving average window (Figs. 1 & 2). Although the propagation of sound through water is well studied, it is a complex phenomenon influenced by a range of factors including seabed geomorphology, water depth, water density, and sound frequency. We therefore produced potential stress surfaces for four sources of acoustic pollution: ferry traffic, commercial shipping traffic, offshore oil production, and small-boat traffic. The Figure 2. Offshore balaenopterid species habitat (mean annual habitat use for blue, fin, and sei whales) as predicted by logistic regression and smoothed with a moving average window. Locations of historic whale harvests (Nichol et al. 2002) were related to bathymetry, depth, slope, and mean annual temperature and salinity. Probability of habitat is shown increasing from light (white = 0.0) to dark ( black = 1.0) (after Gregr & Trites 2001). Figure 3. Stress surfaces for (a) ferry traffic, ( b) shipping routes, (c) oil production, and (d) small-boat traffic. Surfaces represent the probability of a response to the associated sound levels. Probability of stress is shown increasing from light (white = 0.0) to dark ( black = 1.0). surfaces represent a hypothesized likelihood of inducing a response in the humpback and balaenopterid VEFs. For each source, we defined the area in its immediate vicinity as representing certainty of disturbance (probability = 1.0). We then generated a simple, distance-based attenuation of the disturbance effect (given that propagation loss is a logarithmic function of distance) according to Pr (response) = n − log(r), where r represents the distance from the source in kilometers and n is a calibration parameter used to attenuate the probability of response to a presumed distance from the source. We defined shipping lanes in consultation with the Canadian Pacific Pilotage Authority (K. Obermeyer, personal communication). Ferry routes were obtained from the Province of British Columbia. We buffered the shipping lanes to 2.5 nm on either side and assumed the probability of disturbance within the buffer to be 1.0. Ferry routes were represented as lines. Probability of response for both surfaces was attenuated to 0.0 at 10 km (n = 1; Fig. 3a & 3b). Noise associated with hydrocarbon extraction is related to the location of the drilling platforms and the concomitant increase in vessel and air traffic. For simplicity, we considered only the effects of the drilling platforms, the location of which is currently a matter of speculation. We included them to demonstrate their potential effect on the VEFs under consideration. We placed six hypothetical platforms within the Queen Charlotte Basin east of the Queen Charlotte Islands. The basin is hypothesized to contain 9.8 billion barrels of oil and 25.9 trillion cubic feet of natural gas (Geological Survey of Canada, unpublished data). We attenuated the probability of disturbance to 0.0 at a radius of 50 km, based on the results reported by Schick and Urban (2000) (Fig. 3c). Any quantitative assessment of the actual impact of such platforms would require, at a minimum, an assessment of the acoustic properties of the study area. We defined small boats as planing craft powered by outboard engines. The high-frequency noise generated by this kind of traffic is believed to be the major source of acoustic pollution around developed areas (Gordon & Moscrop 1996). Small-boat traffic is composed primarily of private and commercial recreational vessels. Commercial traffic includes sport-fishing charters, whale watching, and water taxis. Private boats are used for sports fishing, sightseeing, and commuting. The majority of this traffic originates in coastal communities. For example, commercial whale watching is primarily centered in Victoria, Tofino, and the communities in the vicinity of Port McNeill. A key characteristic of this type of traffic is that it returns to its point of origin in the same day. The smallcraft “probability of disturbance” surface was therefore attenuated to 0.0 at a distance of 25 km (Fig. 3d). This represents the probability distribution of random round trips taken from the communities to a maximum distance of 25 km, a conservative distance that might be traveled away from dockside by sports fishers and whale watchers. Unlike sources of noise from shipping lanes and offshore platforms, the high-frequency noise from small boats is rapidly attenuated in the ocean. We created VMAs of humpback whales and the offshore balaenopterid group by combining the joint probability of the humpback whale and offshore balaenopterid group habitats with response surfaces for ferry traffic, shipping routes, oil production, and small-boat traffic (Fig. 4a & 4b). Surfaces represent the probability of sensitivity of the VEF to each of the stressors considered. To estimate sensitivity, the effects of the potential stressors would have to be evaluated on the basis of appropriate indicators, such as species abundance or distribution. Developing the baseline indicator values necessary to effectively evaluate sensitivity would require considerable research. In some cases, baseline data may no longer exist because the current species distribution may already reflect the stress. Results The vulnerabilities of the two VEFs to the four stressors considered were relatively similar, but, some clear differ- Figure 4. Vulnerable marine areas of the valued ecological features ( VEFs) for (a) humpback whale and ( b) offshore balaenopterid species in relation to acoustic pollution created by adding the joint probability of the VEFs with each of the stressors considered: ferry traffic, shipping routes, oil production, and small-boat traffic. Surfaces represent the probability of sensitivity of the VEF to the combination of the acoustic sources. Probability of sensitivity is shown increasing from light (white = 0.0) to dark ( black = 1.0). ences were observed. The humpback VEF was predicted to be more vulnerable to oil production, ferry routes, and small-boat traffic. Shipping lanes affected both VEFs, but at different locations. Greater vulnerability was exhibited offshore by the balaenopterid group VEF, whereas inshore the humpback VEF was more vulnerable. There are a number of simple refinements that would more accurately represent the vulnerabilities of the VEFs to the stressors. These include refinements to both the disturbance surfaces and the habitat predictions. First, the relative contributions of the different sources of acoustic pollution should be considered, and the individual valuestress surfaces could be weighted before being combined. Second, to more accurately reflect the potential for noise generation, the community surface could be weighted by population. Third, because VEFs are likely to respond differently to each stressor, the disturbance surfaces should be specific to each VEF. In this case, humpback whales may be more sensitive to higher frequencies than the offshore balaenopterid species, whereas the offshore balaenopterid species may be more sensitive to the lower frequencies generated by commercial shipping traffic. Lastly, the habitat models represent mean annual distributions. This simplification is unrealistic for migrating pelagic species such as those we considered. Similarly, the sources of acoustic pollution we considered were not constant and vary throughout the season. Therefore, both the acoustic and the habitat surfaces should reflect seasonal or even monthly changes. This introduces the potential for weighting by month, based on the ecological importance to the VEF. Although our example focused on pelagic habitats, VMAs are most likely to occur where the overlap between habitats and human activities is greatest. A critical step in this methodology is therefore the identification of the spatial (and temporal) distribution of the VEF under consideration. Establishing this distribution would then inform any future decisions about the introduction of potential stressors. Conclusions We demonstrated how VEFs can be associated with stresses and the likelihood of encountering these stresses (vulnerability) to identify geographic areas for management and conservation purposes. Our approach is a quantitative, repeatable, and defensible method of identifying areas of high ecological sensitivity in the marine environment and can be applied to any combination of VEF and stressor. Additionally, the social and economic values associated with VEFs and potential stressors could be included. Because it is probability based, it is suited to refinements with Bayesian methods. It can therefore support the incorporation of different types of knowledge (scientific, aboriginal, and observational) and quantitatively determine the adequacy of existing data or models. Finally, iterative refinement within this method is easily accomplished as new information becomes available. Ideally, VMA maps like those we created would be generated for all VEFs of interest. A combination of the resulting maps would represent the cumulative sensitivity of valued marine features to all possible human induced threats. The boundaries of the VMAs defined by this combination of maps would be based on the relative importance of the VEFs, the accuracy of the maps themselves, and the trade-offs with other societal values. All these objectives fit well into the methodology we have defined here. Acknowledgments We thank C. Ogborne, D. Howes, and H. Hofmeyr of the Province of British Columbia for providing funding and access to data. We also thank V. Deecke of the Marine Mammal Research Unit at the University of British Columbia and the staff of LGL Ltd. in Sidney, British Columbia, for their input and helpful reviews. Literature Cited Allee, R. J., et al. 2000. Marine and estuarine ecosystem and habitat classification. Technical memorandum NMFS-F/SPO-43. National Oceanic and Atmospheric Administration, Washington, D.C. Cowardin, L. M., V. Carter, F. C. Golet, and E. T. LaRoe. 1979. Classification of wetlands and deepwater habitats of the United States. FWS/OBS-79/31. U.S. Fish and Wildlife Service, Washington, D.C Dale, N. G. 1997. An overview of key conservation, recreation and cultural heritage values in British Columbia’s marine environment. ESSA Technologies, Vancouver, British Columbia. Dias, P. C. 1996. Sources and sinks in population biology. Trends in Ecology & Evolution 11:326–330. Dutrieux, E., S. Canovas, J. Denis, Y. Henocque, J. P. Quod, and L. Bigot. 2000. Guide methodologique pour l’elaboration de cartes de vulnerabilite des zones cotires de l’Ocean Indien Realise par Creocean. Ifremer et Arvan pour le compte de l’UNESCO/IOC et le PRE-COI/UE, Brest, France. Emmett, B., and P. Wainwright. 2001. Framework for identifying marine sensitive areas. Ministry of Sustainable Resource Management, Province of British Columbia, Victoria, Canada. Environment Australia. 2001. Guidelines on the application of the Environment Protection and Biodiversity Conservation Act to interactions between offshore seismic operations and larger cetaceans. Marine Species Section, Environment Australia, Canberra. Available from http://www.ea.gov.au/epbc/ assessmentsapprovals/guidelines/seismic/index.html (accessed January 2003). Gjerde, K. M. 2002. PSSAs: IMO guidelines. Sea Technology 43:40–46. Golley, F. B. 1993. A history of the ecosystem concept in ecology: more than the sum of the parts. Yale University Press, New Haven, Connecticut. Gordon, J., and A. Moscrop. 1996. Underwater noise pollution and its significance for whales and dolphins. Pages 281–320 in M. P. Simmonds and J. D. Hutchinson, editors. The conservation of whales and dolphins. Wiley, New York. Gordon, J., D. Gillespie, J. Potter, A. Frantzis, M. Simmonds, and R. Swift. 1998. The effects of seismic surveys on marine mammals. Proceedings of the seismic and marine mammals workshop, London, 23–25 June 1998. Sea Mammal Research Unit, University of St. Andrews, Fife, Scotland. Gregr, E. J., and A. W. Trites. 2001. Predictions of critical habitat for five whale species in the waters of coastal British Columbia. Canadian Journal of Fisheries and Aquatic Science 58:1265–1285. Gundlach, E. R., and M. Hayes. 1978. Classification of coastal environments in terms of potential vulnerability to oil spill damage. Marine Technology Society Journal 12(4):18–27. Holling, C. S. 1986. Resilience of ecosystems; local surprise and global change. Pages 292–317 in W. C. Clark and R. E. Munn, editors. Sustainable development of the biosphere. Cambridge University Press, Cambridge, United Kingdom. Holt, T. J., D. R Jones, S. J. Hawkins, and R. G. Hartnoll. 1995. The sensitivity of marine communities to man-induced change: a scoping report. Contract science report 65. Maes-y-Ffynnon, Penrhosgarnedd, Bangor, Wales. Howes, D. E., J. R. Harper, and E. Owens. 1994. British Columbia physical shore-zone mapping system. Resource Inventory Committee, Victoria, British Columbia, Canada. International Petroleum Industry Environmental Conservation Association (IPIECA). 1996. Sensitivity mapping for oil spill response. International Maritime Organization and IPIECA, London. Johnson, D., S. Lewey, P. Park, J. Hoar, M. Pourzanjani, S. Fletcher, and S. Tarver. 2001. Particularly sensitive sea areas (PSSA): the Wadden Sea feasibility study. Wadden Sea Newsletter 3:9–12. LGL Consulting. 1998. Environmental assessment of the seismic exploration on the Scotian Shelf. Canada/Nova Scotia Offshore Petroleum Board, Halifax. LGL Consulting. 2000. Environmental assessment of exploration drilling off Nova Scotia. Canada/Nova Scotia Offshore Petroleum Board, Halifax. Lusseau, D. 2003. Effects of tour boats on the behavior of bottlenose dolphins: using Markov chains to model anthropogenic impacts. Conservation Biology 17:1785–1793. McCauley, R. D., M. N. Jenner, C. Jenner, K. A. McCabe, and J. Murdoch. 1998. The response of humpback whales (Megaptera novaeangliae) to offshore seismic survey: preliminary results of observations about a working seismic vessel and experimental exposures. APPEA Journal 38:692–706. McDonald, M. A., J. A. Hildebrand, and S. C. Webb. 1995. Blue and fin whales observed on a seafloor array in the Northeast Pacific. Journal of the Acoustical Society of America 98:712–721. Michel, J., M. O. Hayes, and P. J. Brown. 1978. Application of the oil spill vulnerability index to the shorelines of Lower Cook Inlet, AK. Environmental Geology 1(2):107–117. Moore, S. E., and J. T. Clarke. 2002. Potential impacts of offshore human activities on gray whales. Journal of Cetacean Resource Management 4(1):19–25. National Oceanic and Atmospheric Administration (NOAA). 1996. ESI guidelines. Technical memorandum NOS ORCA 115. NOAA, Washington, D.C. National Research Council. 2003. Ocean noise and marine mammals. The National Academies Press, Washington, D.C. Nichol, L. M., E. J. Gregr, R. D. Flinn, J. K. B. Ford, R. Gurney, L. Michaluk, and A. Peacock. 2002. British Columbia commercial whaling catch data 1908 to 1967: a detailed description of the B.C. historic whaling database. Canadian technical report 2371. Fisheries and Aquatic Sciences,. Richardson, W. J., C. R. Greene, C. I. Malme, and D. H. Thompson. 1995. Marine mammals and noise. Academic Press, San Diego. Roff, J. C., and M. E. Taylor. 2000. National frameworks for marine conservation: a hierarchical geophysical approach. Aquatic Conservation: Marine and Freshwater Ecosystems 10:209–223. Schick, R. S., and D. L. Urban. 2000. Spatial components of bowhead whale (Balaena mysticetus) distribution in the Alaskan Beaufort Sea. Canadian Journal of Fisheries and Aquatic Science 57:2193–2200. Sherman, K. 1991. The large marine ecosystem concept: research and management strategy for living marine resources. Ecological Applications 1:349–360. Tyler-Walters, H., and A. Jackson. 1999. Assessing seabed species and ecosystems sensitivities; rationale and user guide. Report to English Nature, Scottish Natural Heritage and the Department of the Environment Transport, and the Regions from the Marine Life Information Network (MarLIN report 4). Marine Biological Association of the United Kingdom, Plymouth. Urick, R. J. 1983. Principles of underwater sound. Peninsula Publishing, Los Altos, California. Van Bernem, K. H., B. Bluhm, and H. Krasemann. 2000. Sensitivity mapping of particular sensitive areas. Pages 229–238 in G. R. Rodriguez and C. A. Brebbia, editors. Water studies. 8. Oil and hydrocarbon spills, modelling, analysis and control II. WIT Press, Southampton, United Kingdom. Wenz, G. M. 1962. Acoustic ambient noise in the ocean: spectra and sources. Journal of the Acoustic Society of America 34:1936–1956. Williams, R., D. E. Bain, J.K. B. Ford, and A. W. Trites. 2002. Behavioural responses of male killer whales to a “leapfrogging” vessel. Journal of Cetacean Research and Management 4:305–310. Zacharias, M. A., and D. E. Howes. 1998. An analysis of marine protected areas in British Columbia using a marine ecological classification. Natural Areas Journal 18(1):4–13. Zacharias, M. A., D. E. Howes, J. R. Harper, and P. Wainwright. 1998. The British Columbia marine ecosystem classification: rationale, development and verification. Coastal Management 26:105–124.