Identifying predictors of species diversity to guide designation of marine protected areas
Funding information: Natural Resources Defense Council
Abstract
Marine Protected Areas (MPAs) are a widely-used tool for conserving biodiversity. Features that support marine mammal foraging have been suggested as important components to include in MPAs, but research is needed to understand the relationship between these features and diversity. For example, the Northeast Canyons and Seamounts Marine National Monument represents an area known to support marine mammal foraging and was designated to protect an area of high marine mammal diversity. However, no comparisons have been made between marine mammal diversity in the Monument and other areas. We used 3,174,167 km of survey effort and 189,175 sightings to assess alpha and beta diversity in the Monument and 500 randomly selected sites along the east coast of the United States. We used linear models to relate diversity to variables that represent marine mammal foraging areas. Our analyses showed a gradient of higher to lower diversity from north to south and that the shelf-edge, canyons, and areas of likely upwelling support high diversity. We also found that the Monument protects a diverse and unique marine mammal community. Our analyses contribute to efforts to designate MPAs to conserve habitat that is important for protecting species by identifying drivers of biodiversity and potential sites for protecting 30% of the planet by 2030.
1 INTRODUCTION
Marine Protected Areas (MPAs) are a widely-used tool for conserving biodiversity. While the effectiveness of MPAs has been debated (Hillborn, 2015; Lubchenco & Grorud-Colvert, 2015), studies have shown that well designed and effectively managed MPAs can lead to conservation successes (Edgar et al., 2014; Gill et al., 2017; Halpern et al., 2009). These successes have continued to place MPAs at the forefront of global conservation initiatives. Currently, the goal of protecting at least 10% of representative and well-connected coastal and marine areas of particular importance to biodiversity, adopted by the Conference of the Parties to the Convention on Biological Diversity, is being re-evaluated and there is pressure to increase the goal to 30% (Visalli et al., 2020).
The criteria used to identify candidate MPAs continue to evolve. Global criteria have been established by the Convention on Biological Diversity (Ecologically or Biologically Significant Areas) and the International Union for the Conservation of Nature (Key Biodiversity Areas). These global criteria focus on: importance for species life history or ecological processes; uniqueness or rarity; threatened species, habitats, or biodiversity; vulnerability, fragility, sensitivity, slow recovery; irreplaceability; and naturalness (see review in Corrigan et al., 2014). Criteria have also been developed for specific taxonomic groups. For example, criteria for identifying important areas for marine mammals include biological processes (reproductive, foraging, and migration), population indicators (abundance, residency, structure, vulnerability, and resilience), and habitat features (Corrigan et al., 2014).
Marine mammals have long been a taxon of interest in marine research because of their impact on marine ecosystems and their societal importance (Hammerschlag et al., 2019; Kiszka et al., 2015). More recently, increasing attention has been paid to the threats facing marine mammals from the growth in human use of the marine environment. Marine mammals have been found to forage preferentially in areas containing fronts, offshore banks, internal waves, oceanographic stratification, shelf-edge upwelling, and complex topography that is subject to strong tidal flow (Cox et al., 2018). Submarine canyons have also been found to be important foraging areas for cetaceans (Moors-Murphy, 2014).
Features that support marine mammal foraging have been suggested as important components to include in MPAs (Cox et al., 2018; Moors-Murphy, 2014). Foraging and consumption of prey is necessary not only for an individual's growth and survival, but also for the survival of a species through the process of reproduction (Cox et al., 2018). However, more research is needed to understand the relationship between these features and marine mammal diversity. For example, our understanding of the relationship between submarine canyons and cetacean distribution and abundance is based on limited data and biased towards a small number of well-studied canyons that are known to have high cetacean abundance (Moors-Murphy, 2014).
On the United States (U.S.) East Coast, the Northeast Canyons and Seamounts Marine National Monument (hereafter, Monument) was designated to protect an area of ecological connectivity and high diversity for marine mammals and seafloor communities (Auster et al., 2020). Protections include the exclusion of commercial scale extraction activities (e.g., fishing and mineral extraction) and the designation of the Monument under the Antiquities Act of 1906 establishes these protections in perpetuity (Auster et al., 2020). The Monument includes three submarine canyons and four seamounts (Figure 1); consequently, it is representative of areas known to support marine mammal foraging. While the diversity of marine mammals in the Monument was found to be high (Auster et al., 2020), no comparisons were made to other areas along the U.S. East Coast.
We quantified marine mammal diversity in the Monument using a suite of species diversity indices. Although the Monument consists of two units, a Canyons Unit and a Seamounts Unit (Figure 1), our analyses focused exclusively on the Canyons Unit (hereafter, CU) because survey effort was limited in the Seamounts Unit. We compared the diversity within the CU to diversity in randomly selected sites along the east coast of the United States (Figure 1a) to assess broader patterns of marine mammal diversity along the U.S. East Coast. The randomly selected sites had the same area as the CU and we ensured that the same amount of survey effort was used to calculate diversity at all sites. We used alpha diversity indices to quantify diversity within sites and beta diversity indices to quantify differences in species composition between sites. Finally, we related the alpha diversity indices to geomorphic features and oceanographic variables to assess the importance of these potential predictors of marine mammal diversity on the U.S. East Coast.
2 MATERIAL AND METHODS
2.1 Study area
Our study area is part of the Northwest Atlantic Shelves Biogeographical Province (Longhurst, 1998), which represents shelf ecosystems of the western boundary of the Atlantic basin, and extends from the Gulf of St. Lawrence off Canada to the tip of Florida off the U.S. East Coast. The geomorphology of this area is complex and contains multiple canyons, deep basins, and an elevated bank. Characteristics of the shelf edge change from north to south in our study area. From the northern boundary of the study area to the Outer Banks, North Carolina, the shelf edge contains many canyons and the slope off the shelf edge is steeper and reaches the abyssal plain over a relatively short distance. In contrast, from South Carolina to the southern boundary of the study area, the slope off the shelf edge is more gradual, contains the Blake Plateau (a wide flat portion of the slope), and reaches the abyssal plain at a much greater distance.
2.2 Marine mammal data
We acquired aerial and shipboard marine mammal survey effort and sightings data from 1979–2020 from the North Atlantic Right Whale Consortium Sightings Database (NARWC, 2020). From those data, we extracted data from line-transect surveys and platforms-of-opportunity for our analyses. We created on-effort survey tracklines (Figure 1a) from survey effort point data. Survey trackline was defined as on-effort if it was not over land, was coded as on-watch, and met the following sightings conditions: beaufort sea state ≤ 3, visibility ≥ 2 nm, and altitude (if aerial) ≤ 366 m. We omitted surveys that had limited data recording (e.g., surveys that only recorded sightings of large whales or a single species). The dataset contained a total of 3,174,167 km of on-effort survey trackline.
We used only NARWC sightings that had an identification reliability of probable or definite. We excluded sightings made off-effort, using automated photography, or by anyone other than an on-duty observer. We also excluded pinnipeds because of the difficulty of distinguishing between species and inconsistent recording of pinnipeds throughout the years of our study. Many unidentified species groups (e.g., unidentified large whale, unidentified dolphin or porpoise) were omitted and some species were grouped together due to the difficulty of differentiating between species (e.g., all beaked whales were aggregated to a general beaked whale category). This process resulted in 189,175 sightings of 1,021,965 animals from 30 unique species or species groups (see Table S1 in Supporting Information).
2.3 Comparison sites
To compare marine mammal diversity within the CU to diversity within randomly selected sites, and to characterize species diversity across the entire U.S. East Coast, we created five hundred polygons (sites) within our study area (Figure 1a). Each site was identical to the CU in size (2435 km2) and shape, but the orientation (rotation) was randomly selected. To account for potential bias associated with survey effort (e.g., higher diversity could be associated with areas that have more effort), species diversity in the CU and at all comparison sites was calculated using the same amount of survey effort, 3358.6 km, which was half of the total amount of on-effort survey trackline in the CU. This level of effort was chosen to ensure that polygons in the mid-Atlantic, an area with comparatively low survey effort, met the minimum effort threshold and were included in the analyses.
Comparison sites were only drawn in areas that had at least 3358.6 km effort. For each site, effort from each survey occurred in chronological order and effort from all surveys was ordered by survey identification number. The survey identification numbers do not group data by months, seasons, or years, which helped to eliminate temporal bias in the resulting data sets. The survey identification numbers do correspond to the type of survey conducted (i.e., aerial survey or shipboard survey); consequently, data were grouped by survey type. If effort in a site exceeded 3358.6 km, a random survey identification number was selected and a random point along the trackline for that survey was used as a starting point. Survey effort was summed from that point forward until 3358.6 km of survey effort was obtained. If the end of the data set, containing all surveys for a given site, was reached before 3358.6 km of survey effort was obtained, the process resumed at the start of the data set. This process was repeated 50 times, resulting in 50 sets of tracklines, to ensure that the variability created by randomly selecting tracklines was incorporated in our analyses. Requiring the entire site to occur over water resulted in a gap in nearshore waters that was not covered by any sites. To remove this gap, 20% of a site was allowed to cover land (i.e., 80% of the site had to occur over water). The 20% threshold was selected to ensure that an adequate sample size of nearshore sites could meet the 3358.6 km effort requirement.
2.4 Species diversity calculations
- Species richness (S) = number of unique species found within a site.
- Shannon-Wiener Index (H′) = , where pi is the proportion of individuals belonging to species i and n is the number of species such that (Hill, 1973).
- Simpson's Diversity Index (D1) = , where pi is the proportion of individuals belonging to species i and n is the number of species such that (Hill, 1973).
- Sørensen Dissimilarity Index = , where a is the number of shared species between sites, b is the number of unique species at the site with the smallest number of species, and c is the number of unique species at the site with the highest number of species (Jost, 2007).
- Jaccard Dissimilarity Index = , where the definitions of a, b, and c are the same as for the Sørensen Dissimilarity Index (Jost, 2007).
2.5 Predictors of species diversity
We used linear models to relate species richness to ecological production units (Friedland et al., 2015; Lucey & Fogarty, 2013), geomorphic variables, and oceanographic variables. We modeled species richness, rather than the other alpha diversity indices, to facilitate comparisons between our results and previous studies of predictors of species diversity (e.g., Tittensor et al., 2010; Kaschner et al., 2011; Pompa et al., 2011). The northern section of our study area was divided into five ecological production units (EPUs) used by Friedland et al. (2015).These units were generated based on several physiographic, oceanographic, and biological factors including primary production and biological resource extraction (Lucey & Fogarty, 2013) as well as spring bloom characteristics (Friedland et al., 2015). We used the boundaries of these five EPUs: Georges Bank, Gulf of Maine East, Gulf of Maine West, Middle Atlantic Bight North, and Middle Atlantic Bight South (Figure 1b). We treated waters from the mid-Atlantic Bight to Florida as an additional single EPU (hereafter, South) (Figure 1b). We assigned each of the 500 comparison sites and the CU to one of the six EPUs using their geographic centerpoint. Any site whose centerpoint fell outside of the EPUs was attributed to the closest EPU.
The geomorphic and oceanographic variables used in our models were selected to represent preferred marine mammal foraging areas, including distance to the shelf edge (Cox et al., 2018), canyons (Moors-Murphy, 2014), upwelling and tidal-mixing (Cox et al., 2018), and oceanographic stratification (Cox et al., 2018). The distance to the shelf edge was represented by the shortest distance between the shelf edge and the centerpoint of each site because the shelf edge has been identified as an important feature for marine mammals on the U.S. East Coast (Hain et al., 1985; Kenney et al., 1997). Canyons were represented by the percentage of a site's area that contained canyons. Canyons were derived from a global, seafloor geomorphic features map (Harris et al., 2014). We used sea temperature and salinity to indicate areas of potential upwelling (Knauss, 1996) and mixed layer depth to indicate oceanographic stratification. Our analyses were conducted at an annual timescale. Although dynamic ocean variables and processes change on a seasonal basis, the annual time scale used in our analyses is consistent with the year-round designation of many MPAs. We obtained 1/4° resolution, 1981–2010 objectively analyzed climatologies for sea temperature (Locarnini et al., 2018), salinity (Zweng et al., 2018), and mixed layer depth from the World Ocean Atlas 2018 (Boyer et al., 2018). For annual sea temperature and salinity, we computed the water-column averaged values. We associated all variable values to our sites using the minimum distance between the centerpoint of the World Ocean Atlas grid cell and the site centerpoint. Sea temperature, salinity, and mixed layer depth data was unavailable for five sites; therefore, those sites were not included in the models. Additionally, two EPUs (Gulf of Maine East and Gulf of Maine West) did not contain any canyons, resulting in problems with model convergence. Consequently, we combined the Georges Bank, Gulf of Maine East, and Gulf of Maine West EPUs for models that included canyons.
We developed four linear models and used them to perform two sets of model comparisons: (1) to determine whether distance to the shelf was a better predictor of species richness than canyons; (2) to determine whether the relationship between species diversity and the geomorphic and oceanographic variables varied among EPUs. We used sea temperature, salinity, and mixed layer depth in all four models. Models considered in the first comparison used different geomorphic features (i.e., distance to the shelf or canyons). Models considered in the second comparison used an interaction between the geomorphic features and the EPUs. We fit the models in a Bayesian framework with flat priors. We ran each model with four chains of 8000 iterations. We used the R package “brms”, which fits models with the Hamiltonian Monte Carlo algorithm (Bürkner, 2017). We visually inspected trace plots and used Gelman-Rubin statistics to assess convergence. We assessed model fit by visually inspecting posterior predictive checks and by comparing the relationship between predicted and observed model estimates. Posterior distributions of coefficients that did not include 0 in the 95% credible intervals were considered to be predictors that are extremely likely to have an effect on species richness. We compared models using the Pareto smoothed importance-sampling leave-one-out cross-validation (LOOIC).
3 RESULTS
We found that marine mammal species richness was generally higher in the northern EPUs and lower in the southern EPUs (Figures 2a and 3). The sites with the highest richness fell within the Georges Bank unit, on or near the shelf edge, and contained submarine canyons. Higher values of the Shannon-Wiener index (Figure 2b) and Simpson's diversity index (Figure 2c) generally occurred along the continental shelf edge in the Georges Bank, Middle Atlantic Bight North, and Middle Atlantic Bight South regions. Particularly high values of the Shannon-Wiener index and Simpson's diversity index, and relatively high richness values, occurred off the shelf edge near the Outer Banks of North Carolina. Lower values of all three alpha diversity indices occurred closer to shore, particularly in the Middle Atlantic Bight North, Middle Atlantic Bight South, and South units (Figure 2a–c).
All alpha diversity indices indicate that the CU has high marine mammal species diversity (Figures 2a–c and 3). Specifically, the CU had the ninth highest species richness value (top 1.8%) compared to the 500 random sites, the sixth highest value of the Shannon-Wiener index (top 1.2%), and the fifth highest value of the Simpson's diversity index (top 1%). Of the 15 sites with the highest richness values, 14 fell within close proximity to the CU: nine intersected the CU and five occurred within 90 km of the CU.
Beta diversity was used to compare marine mammal species compositions between the CU and the 500 random sites. Beta diversity values are represented on a 0–1 scale, where 0 indicates identical species composition and 1 indicates no shared species between sites. Both the Sørensen dissimilarity index (Figure 2d) and the Jaccard dissimilarity index (Figure 2e) suggest that sites closer to the CU had more similar species compositions than sites farther from the CU. In particular, the beta diversity indices suggested increased similarity between species compositions in the CU and portions of the Georges Bank, Gulf of Maine East, and Gulf of Maine West EPUs. Species composition in the CU was also more similar to sites along the shelf edge from the northern boundary of the study area to the Outer Banks of North Carolina (Figure 2d,e).
To understand the factors influencing species richness, we developed four models and used them to perform two sets of model comparisons: (1) to determine whether distance to the shelf was a better predictor of species richness than canyons; (2) to determine whether the relationship between species diversity and the geomorphic and oceanographic variables varied across EPUs. The model that used distance to the shelf had more support than the model with the percentage of a site's area that contained canyons (Figure 4). The model that contained an interaction between distance to the shelf and the EPUs also had more support than the model that contained an interaction between the percentage of a site's area that contained canyons and the EPUs. Consequently, the models with and without an EPU interaction suggest that distance to the shelf was a better predictor of species richness than the percentage of a site's area that contained canyons.
Models that included an interaction between the geomorphic variables and the EPUs had more support than models that only included the geomorphic variables. The model with the most support included an interaction between EPU and the distance to the shelf edge, salinity, mixed layer depth, and sea temperature (LOOIC = 1704; Bayesian R2 = .85, 95% Bayesian R2 credible interval = .84–.86; Figure 4). This model suggested that species richness is higher in the northern EPUs than the southern EPUs (Figure 5a) when averaged over the distance to the shelf and that the relationship between the distance to the shelf edge and species richness varied by EPU (Figure 5b). It also suggested an extremely likely positive effect of salinity on richness (βsalinity = 1.31, 95% credible interval for βsalinity = 0.99–1.63; Figure 5c) and an extremely likely negative effect of sea temperature on richness (βseatemperature = −0.49, 95% credible interval for βseatemperature = −0.59 to −0.39; Figure 5d). This model suggested that there was no effect of mixed layer depth on richness at the 95% credible interval (βmixedlayerdepth = −0.02, 95% credible interval for βmixedlayerdepth = −0.05 to 0.01; Figure 5e). For this model, inspection of residual plots indicated the relationship between observed and predicted estimates was similar for each EPU (Figure 6).
4 DISCUSSION
Designating MPAs to conserve habitat and ecosystems that are unique, sensitive, threatened, or important for protecting species requires an understanding of the drivers of biodiversity. Our multi-decadal, continental-scale assessment of marine mammal species diversity in the Northwest Atlantic Shelves Biogeographical Province showed distinct geographical gradients that are consistent with results from studies of global marine mammal species diversity. Specifically, we found the highest values of three alpha diversity indices for the northern EPUs in our study area (i.e., Georges Bank, Gulf of Maine East, and Gulf of Maine West) and lower values for the southern EPUs. Multiple global studies (e.g., Tittensor et al., 2010; marine mammal species richness within 0.5° × 0.5° grid cells by Kaschner et al., 2011; Pompa et al., 2011) have shown that marine mammal species diversity is higher at the higher latitudes and lower at the lower latitudes in our study area, which ranges from approximately 28° N–50° N. Our species diversity metrics represent minimum values because some species (e.g., all species of beaked whales) were grouped due to the difficulty of differentiating between individual species in the field.
To assess ecological drivers of diversity, our analyses focused on features that have been suggested to support marine mammal foraging in ecosystems across the world: the shelf edge (Cox et al., 2018), canyons (Moors-Murphy, 2014), upwelling and tidal-mixing (Cox et al., 2018), and oceanographic stratification (Cox et al., 2018). We found that distance to the shelf edge was a stronger predictor of marine mammal species diversity than the percentage of a site's area that contained canyons. The shelf edge has been shown to support increased species diversity through shelf-edge fronts and wind-driven upwelling fronts that create and can sustain high levels of primary and secondary productivity (Cox et al., 2018; Townsend et al., 2006). In our study area, the physical characteristics of the shelf edge change from north to south. In the north, there is a steeper slope from the shelf edge to the abyssal plain. North of Cape Hatteras (see Outer Banks, Figure 1b) there is a prominent shelf-break front, persistent throughout the year, which separates relatively cool, fresh shelf water from warmer, more saline slope waters. Cross-shelf flow over this boundary promotes biological productivity (Townsend et al., 2006). In contrast, the southern shelf edge has a more gradual slope and contains a large plateau that reaches the abyssal plain at a greater distance. These physical distinctions may influence biophysical processes, such as the degree of upwelling (one would predict more upwelling in the north compared to the south) (Cox et al., 2018; Townsend et al., 2006), and contribute to higher species diversity on the shelf edge in the northern and middle latitudes of our study area.
Globally, there is evidence that cetaceans are associated with submarine canyons (Moors-Murphy, 2014). When topographic breaks in the shelf edge, such as canyons, are located down-current from upwelling centers, they may provide large marine mammals with an opportunity for high energy gains (Croll et al., 2005; Redfern et al., 2017). While the models with the most support in our study included distance to shelf, rather than canyons, it is important to note that the majority of the sites with the highest marine mammal species richness did contain canyons. Specifically, 92% of the 25 sites with the highest richness values contained canyons and only 2% of the 250 sites with the lowest richness values contained canyons. Sites with high species richness and canyons tended to occur on the shelf edge, suggesting that these shelf-edge canyons are features that support high marine mammal species diversity. The models may not have been able to identify the importance of canyons because most sites (approximately 84%) did not contain canyons. Additionally, species richness in canyons may have been reduced because we grouped all beaked whale species, which are associated with canyon habitat (Moors-Murphy, 2014; Waring et al., 2001), in our analyses because individual beaked whale species are difficult to differentiate in the field.
Our results also indicated that areas of relatively high salinity and low sea temperatures were associated with higher species richness. These conditions are generally consistent with upwelling regions that bring nutrients from below the pycnocline into the photic zone, leading to increased primary productivity and supporting trophic interactions (Cox et al., 2018; Moors-Murphy, 2014). More productive waters can support a broader suite of species than less productive waters (Moors-Murphy, 2014). Cross-shelf flow and tidal mixing fronts are common in the northern region of our study area (Townsend et al., 2006) and they may promote upwelling and produce the cooler temperature and higher salinity signals that were important in our models. The relationship between species richness, cooler temperatures, and higher salinities, could also be driven by latitude, with northern regions having lower temperatures and higher salinities than southern regions. In addition, we used annual averaged, and depth averaged, salinity and temperature data. Thus, seasonal variation and its resultant changes in productivity were not discernable in our analysis. García-Barón et al. (2020) also found that sea surface temperature, chlorophyll-a, and distance to the shelf-break were good predictors of marine mammal diversity calculated using predictions of marine mammal density in north and northwestern Spanish waters. More broadly, Cox et al. (2018) and Díaz López & Methion (2019) identified upwelling as an oceanographic feature of preferred marine mammal habitat.
We also found that an interaction between distance to the shelf edge and EPUs was a predictor of marine mammal diversity. The EPUs summarize a number of oceanographic variables and capture patterns of commercial fishing (Lucey & Fogarty, 2013) and spring bloom characteristics (Friedland et al., 2015). Northern EPUs (Georges Bank, Gulf of Maine East, and Gulf of Maine West) had a larger effect on species richness than southern EPUs. The spring bloom is stronger in the northern EPUs than in both Middle Atlantic EPUs, where it is not a prominent feature (Friedland et al., 2015). Consequently, areas with a strong spring bloom may support a more diverse community of marine mammal species than less productive areas. Additional research is needed to assess the effect of spring bloom on species diversity. We did not find a relationship between stratification, as indicated by mixed layer depth, and species richness. Mixed-layer depth varies with latitude and season. Consequently, the utility of mixed layer depth as an indicator of stratification and a predictor of species richness may need to be assessed on seasonal timescales. In particular, seasonal timescales may better capture the interannual variability that affects productivity, which may be particularly important for migratory marine mammals that use different habitats on a seasonal basis.
Some of the polygons used in our analyses overlapped and the species diversity metrics for these overlapping polygons may have been calculated using some of same survey tracklines. It is unlikely that any species diversity metrics were calculated using identical survey trackline because the trackline was randomly selected for each polygon. However, this overlap violates an assumption of strict independence between the species diversity metrics. Additionally, spatial autocorrelation may have artificially inflated the statistical significance of the relationship between species richness and the environmental variables. The violation of independence and the spatial autocorrelation are the same in all models and should not influence comparisons among models. However, these issues can limit the interpretation of our results and restrict the transferability of our models (Dormann et al., 2007). Consequently, our models should not be used to infer mechanistic relationships between species richness and the variables used in our models. Additionally, our models should not be used to make predictions outside the study area, where spatial autocorrelation may differ.
The CU of the Monument and the sites surrounding it had some of the highest values of all three alpha diversity indices compared to randomly selected sites along the U.S. East Coast. Two beta diversity indices also indicated that the CU of the Monument and the sites surrounding it support a distinct assemblage of species. The Monument was designated to protect a high diversity of marine mammals and includes features (e.g., submarine canyons) known to support marine mammal foraging (Auster et al., 2020). However, no comparisons had been made between diversity in the Monument and other areas in the Northwest Atlantic Ocean. Our comparison used standardized survey effort across sites in the Northwest Atlantic Shelves Biogeographical Province to assess both alpha and beta diversity. Our results show that the Monument was well sited and fulfills its designated purpose to protect a diverse and unique marine mammal community on the U.S. East Coast.
Multiple studies (Cox et al., 2018; Moors-Murphy, 2014) suggest that it is important to identify features that support marine mammal foraging and ensure these features are included in MPAs. Our analyses show that higher marine mammal species diversity in the Northwest Atlantic Ocean is associated with areas that contain these features, including shelf-edge canyons and shelf-edge areas of increased salinity and colder temperatures, often indicative of upwelling. Our analyses also identify a suite of sites, indicated by high marine mammal diversity, that could be considered as candidates for meeting the goal of protecting areas of particular importance to biodiversity adopted by the Conference of the Parties to the Convention on Biological Diversity. However, further research is needed within these sites to more fully characterize marine mammal species diversity. For example, Williams et al. (2014) found that marine mammal species richness hotspots included marginal habitat of multiple species and excluded core habitat (i.e., highest density areas) of individual species. It may also be valuable to include species extinction risk (O'Hara et al., 2019) and species roles in ecosystem functioning (Pimiento et al., 2020) in biodiversity assessments. Further research is also needed within these sites to assess the biodiversity of the full wildlife community and the habitat. For example, the Monument contains exemplars of offshore Northwest Atlantic Ocean wildlife communities and habitats, including seabirds, shelf-edge cetaceans, shelf fish, chemosynthetic communities, deep-shelf invertebrates, and deep-sea corals, sponges, fish, and soft sediment (Auster et al., 2020).
Identifying MPAs in our study area is critical because U.S. East Coast waters face intensive human use (e.g., fishing, shipping, planned wind energy development) and features that support marine mammal foraging are subject to specific threats. For example, submarine canyons are not well represented in existing MPAs (Fischer et al., 2019) and are facing increasing fishing, oil and gas extraction, and dumping of land-based mining tailings (Fernandez-Arcaya et al., 2017). The Gulf of Maine is one of the most rapidly warming marine ecosystems in the world (Mills et al., 2013) and evidence of changes in productivity have already been observed (Staudinger et al., 2019). Designating MPAs and establishing effective management measures to meet the goal of protecting areas of particular importance to biodiversity adopted by the Conference of the Parties to the Convention on Biological Diversity is needed to protect marine mammals and the ecosystems on which they depend in the Northwest Atlantic Shelves Biogeographical Province.
ACKNOWLEDGMENTS
In total, 49 organizations and groups contributed marine mammal data for this study. We thank the New England Aquarium and NOAA's Northeast Fisheries Science Center vessel and aerial survey teams for contributing a substantial proportion of the marine mammal data that we analyzed. We recognize that the present study could not have been undertaken without those dedicated observers, their pilots, captains, support staff, and collaborators. We thank the North Atlantic Right Whale Consortium (NARWC) for curation and dissemination of marine mammal survey data used in our study. Aerial surveys of the Northeast Canyons and Seamounts Marine National Monument conducted by the New England Aquarium were funded by the Conservation Law Foundation, National Ocean Protection Coalition, and Natural Resources Defense Council. We thank Julia Olson for helpful discussions on methods for measuring biodiversity. We thank the National Centers for Environmental Information for the World Ocean Atlas datasets. The Natural Resources Defense Council provided financial support for this study.
CONFLICT OF INTEREST
SK and OO are standing declarants for groups who have filed complaints in Federal court regarding the Northeast Canyons and Seamounts Marine National Monument.
AUTHOR CONTRIBUTIONS
BH, DP, LG, OO, SK, and JR contributed to the conceptualization. OO, EQ-R, and SK contributed to the data curation. BH, DP, LG, and JR contributed to the analyses. All authors contributed to the writing of the manuscript.
ETHICS STATEMENT
This research was conducted following all ethical guidelines. All data used came from existing datasets. Since no new data were collected, no research permits were required and no animal or human subjects were used.
Open Research
DATA AVAILABILITY STATEMENT
Marine mammal sightings and survey effort data were obtained from the North Atlantic Right Whale Consortium: (https://www.narwc.org/accessing-narwc-data.html). Geomorphology data, including shelf edge and canyons, were obtained from Blue Habitats (https://www.bluehabitats.org/). World Ocean Atlas 2018 data were obtained from the NOAA National Centers for Environmental Information (https://www.ncei.noaa.gov/).
