Global distribution and conservation of marine mammals

Marine mammals are a polyphyletic group that comprises 129 species grouped in three orders, Cetacea, Sirenia, and Carnivora (Table 1). The smallest marine mammal is the sea otter (1.15 m, 4.5 kg), and the biggest is the blue whale (30 m, 190 tons). Marine mammals show very complex, heterogeneous distributions throughout the oceans and also are found in a few freshwater lakes and rivers. The average geographic range for all species is 52 million km² (Fig. 1A). The most widely distributed species, with ranges exceeding 350 million km², are Bryde's (Balaenoptera edeni) and humpback (Megaptera novaeangliae) baleen whales. The marine species with the most restricted range is the vaquita (42) (Phocoena sinus), a porpoise species endemic to 4 000 km² in the upper northern Gulf of California, Mexico. However, most of the species with very restricted ranges, such as the Baikal seal (Pusa sibirica), are freshwater species endemic to lakes. They probably have relict distributions, remnants of much larger ranges in geologic times (24). Both endemic and restricted-range species have high priority for conservation because they usually are more vulnerable to anthropogenic impacts (2, 10).

In terms of richness, the analysis of our 46,184-cell, ∼10,000-km² global geographic quadrant grid (Methods) showed that the number of species per cell varied from 1 to 38, with an average of 17 species, across vast regions of the oceans. Interestingly, latitudinal gradients of species richness of marine and land mammals are very different. Marine mammals have undergone considerable anatomical modifications during their evolution. The unique characteristics of the marine ecosystems have resulted in the many different physiological and ecological responses that marine mammals have experienced. These modifications undoubtedly have resulted in energetic constraints. One of several complex structures of the marine environment is a more-or-less unpredictable, patchy distribution of food over large spatial and temporal scales; this patchy distribution almost certainly has contributed to the evolution of marine mammal energetics, especially through its effect upon energy storage and expenditure strategies. Species richness of land mammals increases sharply from temperate latitudes toward the equator. In contrast, species richness in marine mammals has a more northerly temperate component, showing a higher concentration of species (24 species average) between 30° N and 40° S (Figs. 1B and 2A). Other factors contributing to this pattern in marine mammals remain to be evaluated; nevertheless, the results of our richness-distribution patterns are consistent with other approaches analyzing marine mammal distribution patterns (36, 43).

Regions especially rich in marine species (Fig. 2A) were found along the coasts of North and South America, Africa, Asia, and Australia. Such patterns apparently are correlated with ocean currents and their dynamics, especially with nutrient flows connected to upwellings. For example, along the Pacific coast of the American continent, the highest species richness was found along the California, Baja California, and Peruvian coasts, where large upwelling systems maintain very productive fish communities (44). Interestingly, among higher taxa, patterns of species distribution in marine mammals differed strongly (Fig. 3 A–C). Pinniped (seal and sea lion) species richness was concentrated at the poles, especially near Antarctica, whereas Mysticetes (baleen whales) exhibited high species richness at 30° S latitude, and Odontocetes (toothed whales) were concentrated near tropical coasts. There also was variation in distribution at the family level within and among orders; for example, the two families in Sirenia had contrasting distributions: The Trichechidae (manatees) were found exclusively in the North and South Atlantic, whereas the Dugongidae (dugong) were restricted to the North Pacific and Indo-Pacific.

Political endemic species [i.e., species found in only one country, a restriction that may increase their vulnerability (2)] included seven species; the Baikal seal (Pusa sibirica), the Australian sea lion (Neophoca cinerea), the Galapagos fur seal (Arctocephalus galapagoensis), the Galapagos sea lion (Zalophus wollebaeki), the New Zealand dolphin (Cephalorhynchus hectori), the Hawaiian monk seal (Monachus schauinslandi), and the vaquita (Phocoena sinus) (45). Seven species, among them the New Zealand sea lion (Phocarctos hookeri) and the Australian Snubfin dolphin (Orcaella heinsohni), had restricted ranges. In terms of extinction risk, 10% of all marine mammals are considered vulnerable, 11% endangered, and 3% critically endangered (SI Appendix). Species at risk were found throughout the oceans but were concentrated at higher latitudes, especially near the Aleutian Islands and the Kamchatka Peninsula, where extensive exploitation of whales and seals occurred in the past (Fig. 4).

To assess the conservation challenges to marine mammals, we determined the area (i.e., the number of cells) required to incorporate different percentages (i.e., 10%, 15%, 20%, and 25%) of the geographic ranges of all species, using the Marxan optimization algorithm (Methods). Conserving at least 10% of all of the species’ geographic range required ca. 45 million km² (5,700 grid cells), roughly equivalent to 12% of the world's ocean area (e.g., two times the extent of the Southern Ocean). This study provides grounds for future assessment of an area-explicit conservation parameter for marine mammals. The “target” of 10% was used so this work would be comparable to our previous papers on terrestrial mammals (15, 46); it also is one of the targets suggested by the Convention on Biological Diversity (47). This Convention has called for networks of protected areas, which, in addition to other conservation measures, are necessary components of sustainable use (39). Targeting 15%, 20%, and 25% of each marine mammal's distribution range considerably increased the area required to meet the targets (Fig. 5). Clearly, protecting larger targets must incorporate, by necessity, other conservation mechanisms in addition to reserves or MPA's (48, 49).

Our next step was to identify key conservation sites representing all marine mammal species in a geographically explicit way. We selected those sites using the grid cells with the greatest diversity followed by “irreplaceable” cells (i.e., cells with species represented nowhere else), using the Marxan optimization algorithm (Methods). We evaluated the representation of all marine mammal species in 1%, 2.5%, 5%, 7.5%, and 10% of the grid cells (Table 1). We chose 2.5% because these grid cells included 108 (84%) of all species; the missing species were all the 10 endemic species and 11 additional species that have restricted distributions, so they were dispersed in a very large area. Selecting the top 5% of the grid cells would include an additional 19 species but would require more than twice the area required in the 2.5% scenario. Therefore we used Marxan to select in a more effective way the cells containing all missing species, optimizing the area required so that all marine mammals would be represented in our network of key conservation sites. Also, the 2.5% cutoff has been used in studies with land mammals, so using that cutoff allows comparison of terrestrial and marine conservation issues (17, 18).

We identified 20 key conservation sites (Fig. 2 A and B). These key sites can be the basis for identifying a comprehensive conservation strategy with MPAs representing all marine mammals, their ecological roles, and some threats (39, 50). The nine key conservation sites selected because of their species richness were along the coasts of Baja California, Northeastern America, Peru, Argentina, Northwestern Africa, South Africa, Japan, Australia, and New Zealand. These sites represent 108 species (84% of all marine mammal species), including five endemic species (Fig. 2). They are located in all continental waters except Europe and are mostly in temperate latitudes; only the key conservation site off Peru is located in tropical waters. They occur in five of the seven ocean regions (24), being absent from the polar regions, and include 11 (25%) of the 44 marine ecoregions (25). Not surprisingly, these key sites seem to be located in upwelling oceanic areas, where there is a confluence of cold and warm currents. These oceanographic circumstances favor zones of high primary production, which are good feeding areas for marine mammals (43). As expected, the areas of concentration of specific orders vary strongly across space (Fig. 3 A–C). The 11 key conservation sites that were deemed irreplaceable because the presence of endemic species were the Hawaiian Islands, Galapagos Islands, Amazon River, San Felix and Juan Fernández Islands, Mediterranean Sea, Caspian Sea, Lake Baikal, Yang-Tze River, Indus River, Ganges River, and the Kerguelen Islands (Fig. 2B). These sites had unique species, such as the Galapagos fur seal (A. galapagoensis) and the Mediterranean monk seal (Monachus monachus). Interestingly, six irreplaceable sites were continental (rivers and lakes), and five were marine.

We understand that grid cells are not all equally important for conservation aside from their species richness or endemic species (51, 52). In marine mammals, breeding and feeding grounds and migratory routes are especially important for conservation. Therefore, to identify the key conservation sites, special weight was given in the Marxan optimization algorithm (Methods) to grid cells found in calving/breeding/feeding grounds and to known migratory routes of several species. For example, the locations of the breeding grounds for humpback and right whales are well known and often are relatively concentrated, as are all or part of the migratory corridors for some populations. However, such information is not available for many species. Giving more weight to breeding/feeding areas of migratory routes is very important for marine mammals that are highly mobile.

We analyzed the relationship of three human impacts—climate disruption, ocean-based pollution, and commercial shipping (53)—with grid-cell species richness, using a Spearman rank correlation. As we expected, the three impacts have a significant correlation with species richness (rs = 0.693, n = 46,164, P < 0.01 for climate disruption; rs = 0.666, n = 46,164, P < 0.01 for pollution; and rs =0.678, n = 46,164, P < 0.01 for shipping). Our results indicate the widespread impact of human activities on marine ecosystems and their potential for negatively impacting key marine mammal conservation sites. Around 70% of the highest values for the three impacts were located within or near one of our key conservation sites. Adding other human impacts such as commercial fishing probably will show even stronger impacts of human activities on marine mammal conservation.

Areas of overlap between fisheries and marine mammal groups are concentrated mostly in the Northern Hemisphere and appear to occur primarily between pinnipeds and fisheries. Partly because of the comparatively low total food intake of dolphins, the overlap between dolphins and fisheries is quite low and, again, is concentrated mostly in the Northern Hemisphere. Not surprisingly, the lowest overlap occurs between fisheries and deep-diving large-toothed whales, whose diets consist primarily of large squid species and mesopelagic fish not currently exploited by fisheries (54). Narrow coastal fringes are the location of nine of our key conservation sites identified by their species richness. The Japanese and Peruvian richness sites are located within the Northwest and Southeastern Pacific zones, respectively; these two zones have the highest fisheries catch of the major fishing areas in the world (55). The Australian key conservation site, the one with the highest species richness, is in the East Indian Ocean and the Southwest Pacific zones, which are ranked sixth and 18^th, respectively, by catch intake (55). The Japanese richness site also is located within Chinese waters (China is top fish-harvesting nation in the world), where 17 million tons of fish are captured annually and where at least 30 marine mammal species live (55). In addition, at least five of the key conservation sites overlap with highly impacted ocean areas where high bycatch fishing occurs (53).

Many marine species and populations [e.g., North Atlantic right whale (Eubalaena glacialis) and the Sei whale (Balaenoptera borealis)] are at the brink of extinction from overharvesting, pollution, bycatch, and exhaustion of prey-species populations (24, 25, 56–58), and their long-term survival depends on sound management that addresses the factors causing their decline. The baiji dolphin, once endemic to the Yang-Tze River in China, is a disturbing example of the plight of marine mammals impacted by human activities (9). The next candidate to become extinct if no solid conservation and management strategies are implemented is the Mexican vaquita. Endemic to the Gulf of Baja California, the species has been declining sharply for at least 2 decades; one fifth of the population is killed in gillnets every year, and there now are only an estimated 150–300 individuals (59). Indeed, more than 650,000 marine mammals die from entanglement in fishing nets each year (60), making bycatch the single largest cause of mortality for small cetaceans and pushing several species to the verge of extinction.

Conservation strategies also should take into account the possible impacts of anthropogenic climate disruption (61, 62) on the distribution of these mammals and its repercussions on the establishment of connective corridor systems between protected areas (61) and on management plans. Finally, management interventions must be evaluated critically with regard to ecological viability and benefits vs. costs (61).

By selecting the smallest area of reserves using an optimization algorithm, the opportunity conservation cost would be generally lower, but this approach will depend on the distribution of other potential economic activities (63). For instance, an evaluation of fisheries values could provide a feasible first cut at calculating those costs. Given the distribution patterns of marine mammals, the increasing pressures of human activities in the oceans, and the threat of climate disruption, the conservation of marine mammals is a daunting problem. Saving one or two populations of most species will not be enough (2) because of the role that such charismatic mammals play in the ecological dynamics of marine and freshwater ecosystems and in the provision of ecosystem services. As many scientists have emphasized in other forums, especially in connection with whaling (64), the complexity and scale of the problem requires an unprecedented international effort with the development of both new attitudes and institutions (65). The main objectives of selection criteria for MPAs are to identify potential MPAs for highly mobile and temporally variable pelagic species, including high-density areas, feeding or breeding grounds, and migratory routes; to provide a transparent and systematic approach to selection; and to help determine priorities for action (39).

Uncertainty will always be a factor in research on pelagic organisms and their environment. Empirical data point to dramatic declines and changes in marine systems, and ongoing research continues to provide techniques to incorporate and contend with uncertainty. The challenge is to produce timely and scientifically defensible research based on available data to address this conservation crisis now (56). The future of marine mammals in particular and biodiversity in general will depend on the actions we take.

We compiled and digitized the geographic range maps from published sources for all 129 species and created a Geographic Information System database for 46,184 1° x 1° grid-cells, ∼10,000-km². We then conducted a presence/absence analysis to determine the number of species in each grid cell and the number of cells in which each species was recorded. We created maps of global species richness, irreplaceable sites, endemism, and threatened species. Key conservation sites for species richness were determined either as the 2.5% of the cells with the highest species richness or as irreplaceable sites, defined as regions containing species not represented in any other part of the world (17, 18). Additionally, we used optimization algorithms, i.e., ResNet (66) and Marxan (67, 68 69), to determine the number of cells required to cover 10%, 15%, 20%, and 25% of the geographic ranges of all species and the area of the ocean covered by each percentage.

Marxan is software that delivers decision support for reserve system design intended to solve a particular class of reserve design problem in which the goal is to achieve some minimum representation of biodiversity features for the smallest possible cost. Given reasonably comprehensive data on species, habitats, and/or other relevant biodiversity features, Marxan aims to identify the reserve system (a combination of planning units) that will meet user-defined biodiversity targets for the minimum cost. In this particular case, Marxan selected planning units (here, grid cells) to meet the targets (10%, 15%, 20%, and 25% of the geographic ranges of all 129 species) and also considered the following factors. Each grid cell is assigned a “cost” depending on the target (e.g., area, number of species, threat), and Marxan minimizes the combined grid-cell cost of the conservation network, still selecting expensive grid cells if they are needed to meet the targets. This cost can be a measure of any aspect of the planning unit (25, 69, 70); in this case, it was species richness plus cells weighted for breeding/feeding ground or migratory route. We set Marxan to select adjacent planning units preferentially rather than a series of unconnected units, which would be less ecologically viable and more difficult to manage. Then Marxan identified a set of grid cells each time it was run: 100 runs generated 100 different grid-cell networks (67). Units that appeared in every network were considered irreplaceable, because they always would be needed to meet the targets, whereas other units could be swapped with similar units, and the targets still would be met. Fig. 5 was achieved using these methods.

We thank Gretchen C. Daily, Rurik List, Stuart Pimm, and Rob Pringle for insightful discussion on this topic and Irma Salazar, Antonio Iturbe, Jesús Sajama, and Pablo Ortega for help with the data analyses. This study was supported by grants from Dirección General de Asuntos del Personal Académico (Universidad Nacional Autónoma de México), Ecociencia Sociedad Civil, the National Council for Science and Technology (Mexico), and the Cetacean Society International.