Vulnerability of Arctic marine mammals to vessel traffic in the increasingly ice-free Northwest Passage and Northern Sea Route

Forty-two (53%) of the 80 AMM subpopulations were exposed to portions of either one of or both the NWP and NSR, with exposure scores ranging from 1.01 to 3.0 (SI Appendix, Table S1). Sensitivity scores for these subpopulations ranged from 1.63 to 2.50, resulting in subpopulation-specific vulnerability scores ranging from 1.76 to 7.50 (Figs. 2 and 3). Of the subpopulations that overlap with the sea routes, the Eclipse Sound narwhal subpopulation was most vulnerable to vessel traffic largely because of a combination of high exposure to the NWP and biological (species-specific) traits that increased vulnerability. The Hudson Bay–James Bay ringed seal subpopulation was least vulnerable. We found an intermediate vulnerability level among all AMM species (mean score = 4.20) exposed to Arctic sea route traffic, reflecting relatively high vulnerability among narwhal, walrus, bowhead, and beluga subpopulations (mean scores = 5.59, 5.34, 5.16, and 5.06, respectively) compared with intermediate vulnerability of bearded seals (mean score = 4.01) and relatively low vulnerability of polar bears and ringed seals (mean scores = 2.95, 3.52, respectively; Table 1). However, uncertainty was generally greater for the least vulnerable subpopulations, particularly among ice seals and polar bears (Fig. 3). Walruses were the outlier among pinnipeds, with relatively high vulnerability and low uncertainty.

The spatial distribution of AMM vulnerability and uncertainty varied regionally and by taxonomic group (Figs. 4 and 5). Regions of the Pacific Arctic (i.e., Bering Strait, Chukchi, Beaufort, and East Siberian seas), Russian Arctic (Laptev Sea), and eastern Canadian Arctic (Lancaster Sound, eastern Baffin Island, Barrow Strait, Gulf of Boothia) were characterized by high vulnerability scores, which corresponded to cetacean and pinniped subpopulations with high vulnerability. Several of these regions were also characterized by particularly high uncertainty (i.e., Russian Arctic, eastern Canadian Arctic). In contrast, subpopulations distributed north and south of the NWP in the central Canadian Arctic and near Baffin Bay, Greenland, and the Barents Sea generally had lower vulnerability scores, although the number of exposed subpopulations was also low in these regions compared with up to 10 subpopulations using the Bering Strait-Chukchi Sea region (SI Appendix, Fig. S2). Polar bears had the lowest vulnerability scores, with the exception of the Northern Beaufort Sea subpopulation. Cetaceans were characterized by relatively low uncertainty despite their high vulnerability in comparison with pinnipeds, which had high uncertainty across the Arctic sea routes.

A suite of ecological impacts are emerging as a result of unprecedented sea ice loss across Arctic marine systems (23, 24), but the consequences of increasing human access in the face of environmental change have not been comprehensively considered across AMMs. Here, we have performed an analysis of the vulnerability of some of the most sensitive Arctic species to an anthropogenic risk factor that is primed to expand in the absence of sea ice. We used the extensive literature available about vessel effects on marine mammals in more temperate regions, and ultimately we provide a comprehensive assessment of the combined effects of vessel exposure and sensitivity across all populations of AMMs under increasingly navigable Arctic sea routes during open-water conditions. Understanding the heterogeneity of risk across this remote seascape provides useful insights for developing best practices for maritime industries, as well as species and regions to prioritize.

Planning for risk avoidance is complicated by the highly mobile nature of AMMs and the fact that vessels also move (10). However, sea ice loss is forecasted to continue for several decades even if aggressive mitigation is immediately implemented (25), so prudent marine spatial planning would aim to implement risk abatement measures in advance of the development of extensive commercial shipping. We found greater variability in exposure than sensitivity, but cetaceans were particularly vulnerable to vessel effects based on relatively high exposure, as well as sensitivity scores. Arctic cetaceans are migratory, often following genetically based migration routes and exhibiting site fidelity to productive regions with extensive summer foraging opportunities (e.g., refs. 26 and 27) that are experiencing variable impacts to changing sea ice habitat (e.g., refs. 28–30). The NWP and NSR have extensive overlap with the same foraging areas and fall migration routes of these cetacean species. In addition, we found that cetaceans are particularly sensitive to vessel disturbance, acoustic effects, and in the case of bowhead whales, ship strikes. Cetacean-oriented mitigation strategies developed elsewhere suggest that avoiding key habitats by routing, detecting, and deviating from whales visible at the surface and minimizing sound production could be effective ship-based measures, all of which can be further enhanced by restricting speed (11, 31–34). Specific maximum noise level thresholds, or noise budgets, have also been proposed as a solution for highly sensitive areas and species (35). Another approach called dynamic spatial management, based on a fluid spatial and temporal management framework, has been proposed in some Arctic regions and has shown promise elsewhere (36, 37).

Understanding where knowledge gaps exist can help target future research and monitoring efforts to minimize vessel threats, and we found that uncertainty varied among regions, species, and subpopulations. Some of the species and regions identified as the least vulnerable were also characterized by high uncertainty, highlighting areas that need additional data and monitoring. The highest uncertainty occurred throughout Russia in the NSR and in some portions of the Atlantic Arctic. There are large data gaps on AMM subpopulation status, trends, and distribution (3), which contributed to much of the uncertainty in species exposure and sensitivity. We found a disproportionate focus on cetacean sensitivity to vessel effects, whereas few studies examined vessel effects on ice seals or polar bears. Walruses also scored relatively high vulnerability compared with the other pinniped species, largely because of their higher climate change sensitivity and several small subpopulations that had high sea route exposure in comparison with the ice seals, which are characterized by high uncertainty in subpopulation size and structure. Ice seals are presumably focused on foraging during the open-water period, and many data gaps exist on the summer ecology, behavior, and in turn, impacts of vessels on ringed or bearded seals in particular (38, 39). For many of these subpopulations, there was a paucity of information, or we relied on studies of sister species. Our focus on the open-water period precluded examination of ice-breaking vessel activities during the ice-covered season, which would potentially alter vulnerability scores for some species, especially those that rely on sea ice as a platform for reproduction, molting, and foraging (i.e., seals and polar bears). For example, icebreakers can negatively affect ice-breeding seals during pupping and lactation periods by direct collision, as well by separating mothers and pups (40). Examination of potential vessel impacts to sub-Arctic species was also beyond the scope of this analysis, even though more temperate marine mammal species appear to increasingly use Arctic regions in summer. Additional research is warranted to understand the seasonal distribution, abundance, and potential vulnerability of sub-Arctic species to vessels.

Our analysis focused on the NWP and NSR, where recent vessel transit data were available, but there is important AMM habitat in other regions with vessel exposure (e.g., Svalbard, Sea of Okhotsk). Although our study used open data on full transits of the two sea routes, vessel traffic in much of the Arctic is not well-documented (7, 10). Nearly 65% of the Arctic marine environment experienced vessel traffic in a recent snapshot of 2015 ship tracks, although most was focused in the Barents, Bering, and Norwegian seas during open-water periods (41). Broader data on vessel type and tracks would enable more quantitative analyses of exposure and associated impacts, such as probabilistic modeling of vessel noise propagation (42), individual exposure estimates (43), and overlap with ecologically significant areas (32). Regions such as the Bering Strait and the Chukchi Sea, which we identify as areas with high AMM vulnerability, are at the crossroads of the NWP and NSR and are also of substantial biological and cultural importance; as such, they are ripe for more focused research on vessel impacts (11).

A population’s resilience or capacity to adapt affects vulnerability, yet this is often excluded from vulnerability analyses because of the inherent challenges associated with scoring adaptive capacity (14) and limited information (e.g., refs. 19, 20, and 44). Slow intrinsic growth rates, long generation time, habitat specificity, several foraging specializations, and small population sizes suggest that most AMM species are susceptible and have limited ability to respond to large changes in their environment (3, 45). Our inclusion of variables corresponding to sensitivity to sea ice loss, as well as subpopulation size and trends, provide some assessment of potential adaptive capacity, yet we recommend additional research to assess AMM resilience, to vessels or otherwise, under climate change scenarios.

We have provided the framework and initial assessment of the developing threats associated with increasing navigability of the Arctic environment for key wildlife species that range widely across international boundaries, use regional habitats, are critical and traditional resources for indigenous communities, and are at the forefront of climate change impacts. Our direct measures of vulnerability may inform policy decisions, as well as guide future research, regarding upcoming economic and conservation challenges associated with vessel presence in this newly accessible, rapidly changing, and data-poor ecosystem.

Thanks to Peter Westley for reviewing an earlier draft, Richard McGovern for assisting in mapping sea routes, and Sheila Ocoma and Kathryn Koo for entering vessel transit data. Two anonymous reviewers improved the paper. This work was primarily funded by NASA Grant NNX16AG33G (to H.L.S. and K.L.L.), with additional support for D.D.W.H. from the Collaborative Alaskan Arctic Studies Program.