The Springer et al. hypothesis has generated considerable controversy (e.g., DeMaster et al. 2006, Mizroch and Rice 2006, Trites et al., in press), and has significant ramifications scientifically, as well as in the realms of management and politics. However, is it plausible and supported by existing data? Our analyses of a broad range of information lead us to conclude that the hypothesis—that killer whale predation induced declines in populations of pinnipeds and sea otters—is based upon a simplistic and highly selective analysis of trend data. We also conclude that the prey-switching hypothesis rests upon a poorly supported assumption regarding the importance of large whales as prey items for killer whales, notably in the high-latitude areas of the North Pacific (the focus of Springer et al.) where most commercial whaling occurred. Below, we discuss in detail the available data in each region in the context of the Springer et al. hypothesis.
Gulf of Alaska
A substantial number of whales were removed from the GOA region, and catches of whales had essentially collapsed by ∼1968. Harbor seals and Steller sea lions were declining by the late 1970s, and although spatial variability exists in the timing of sea lion declines, overall the declines of those two species were not sequential as reported by Springer et al. (2003). When the pinniped declines were reanalyzed, a statistical test concluded that these declines (along with northern fur seals and sea otters from the BSAI region) were not sequential (DeMaster et al. 2006), as can be seen in Figures 5 and 6. However, declines have also been verbally reported for the 1990s for sea otters in the GOA, and a substantial population of transient killer whales occurs there. Do these declines provide support for the Springer et al. hypothesis?
The cause of the decline of Steller sea lions has been the subject of prolonged debate, but several plausible explanations exist that have nothing to do with killer whale predation (see below). The decline of harbor seals in this region has not received as much scientific scrutiny, but plausible explanations also exist that do not involve killer whales. The cause of the recent decline of sea otters in part of this region does not yet have a satisfactory explanation. Substantial observations of transient killer whales in Prince William Sound show that predation occurs on harbor seals, Dall's and harbor porpoise, and Steller sea lions, with no predation observed on the abundant humpback whales found there (Saulitis et al. 2000). Humpback and fin whales are both thought to have been increasing since at least 1987 and were likely increasing earlier. These species are currently abundant enough in many areas that, if they were preferred prey, much more predation on these species would be seen. The possibility that killer whales contributed to the declines of harbor seals and sea otters in this region cannot be excluded, but the declines were not sequential and therefore cannot be explained by the sequential prey-switching hypothesis.
Bering Sea and Aleutian Islands
The removal of large whale biomass in the BSAI region took place over a longer time period but collapsed at approximately the same time (∼1968) as it did in the Gulf of Alaska. Northern fur seals were declining at the same time, during the late 1950s and 1960s. Immediately following the collapse of whale removals, the fur seal population actually stabilized during the 1970s before declining again in the late 1970s and early 1980s, while Steller sea lions were likely declining. The decline of fur seals and Steller sea lions was initially simultaneous, but Steller sea lions declined severely while the fur seal population stabilized for most of the 1980s and early 1990s. Other populations were stable or increasing in the 1970s and 1980s, such as gray whales, walrus, and Bristol Bay harbor seals. Humpback and fin whales appear to have increased since the late 1980s. As in the other regions, substantial large whale biomass was available throughout the 1980s and 1990s. Thus, the declines of the species considered by Springer et al. were not sequential in this region, with the exception of the decline of sea otters in the 1990s.
Most telling is that fur seal biomass at its lowest value was still two orders of magnitude larger than sea otter biomass at its highest value in this region (Fig. 5). Therefore, although the killer whales in this region may well have caused or contributed to the decline of sea otters (Estes et al. 1998), it cannot be argued that they did this because of a lack of fur seal biomass. Although there is little disagreement that fur seals have declined in the Bering Sea, they cannot be said to exist in low numbers: indeed, the most recent estimate of abundance puts the population at 888,120 animals (Angliss and Lodge 2004). Thus, the available biomass of fur seals remains so high in this region that it is inconceivable that killer whales would have needed to switch to sea otters as their primary prey. This appears to strongly refute the hypothesis that prey switching would have occurred by killer whales for the reasons proposed by Springer et al. (i.e., that killer whales were “fishing-down” the marine food web, sequentially switching to less desirable prey as more desirable prey became unavailable).
Overall, the major point here is that all three of the regions as well as the Commander Islands include areas, ignored by Springer et al., in which industrial whaling depleted populations of large whales, and where transient-type killer whales are known to occur but where populations of pinnipeds and/or sea otters are stable or increasing rather than in decline. That this is so despite a history of whaling and killer whale presence in these areas argues strongly against the validity of prey switching as a general pattern. Although this does not exclude the possibility that killer whale predation was partly responsible for some of the declines noted by Springer et al., it greatly weakens the case that those declines were the result of a prey-switching event initiated by industrial whaling.
It is worth reiterating the key point that Springer et al. support their hypothesis using trend data from only four populations at five sites: Tugidak Island (harbor seals), St. George and St. Paul Islands in the Pribilofs (northern fur seals), western Alaska (Steller sea lions), and the Aleutian Islands (sea otters). As we have documented here, selection of these sites ignores other populations of the same species that are either stable or increasing, or which remain large despite a recent decline. For example, the northern fur seal and sea otter data that Springer et al. used are from the BSAI region, but the harbor seal data are from near Kodiak Island in the GOA. Springer et al. have also inappropriately extrapolated trends from these sites to the entire stock, or even to the entire ocean basin. Additionally, the use of relative abundance as the scale in Figure 2 of Springer et al. does not show the dramatically large differences in biomass between some species (e.g., see Fig. 6).
The Western U.S. stock of Steller sea lions certainly declined substantially in the central and western Gulf of Alaska and in the Bering Sea and Aleutian Islands (NMFS 1992). However, Springer et al. oversimplify the temporal and geographic pattern of the population decline as well as its possible causes, and they ignore substantial literature that gives alternative explanations for the decline of Steller sea lions. The decline was first observed in the eastern Aleutian Islands and may have begun as early as the late 1960s or early 1970s (Braham et al. 1980). The decline spread west through the Aleutian Islands and east throughout the central and western Gulf of Alaska, reaching its maximum rate of decline between 1985 and 1989 at approximately 15% per year (Loughlin et al. 1992; York 1994). Through the 1990s, the decline slowed across the range of the Western stock to approximately 5% per year (Sease and Gudmundson 2002), and may have nearly abated since 2000 (Fritz and Stinchcomb 2005).
During the 30 yr of population decline, both top-down and bottom-up forces likely affected the Steller sea lion population. Over 20,000 Steller sea lions were killed between the 1960s and 1980s as a result of being accidentally caught during groundfish fishing operations (Loughlin and Nelson 1986; Perez and Loughlin 1991); but, in the 1990s, incidental catches totaled less than 300 (Perez 2003). In addition, approximately 45,000 pups were killed in the eastern Aleutian Islands and Gulf of Alaska between 1963 and 1972 (Pascual and Adkison 1994). Numbers of Steller sea lions shot illegally may also have been high in the 1980s (Trites and Larkin 1992). However, direct mortality sources alone were not responsible for the decline experienced by the Steller sea lion population in the 1970s and 1980s (Pascual and Adkison 1994), suggesting that other factors were also implicated.
The primary bottom-up hypotheses for the Steller sea lion decline both involve a reduction in prey biomass and quality caused by either environmental variability (Trites and Donnelly 2003, Trites et al., in press) or commercial fisheries (Braham et al. 1980, NMFS 2000). In apparent response to reduced prey availability, lower growth and pregnancy rates were observed in the Western Steller sea lion population in the 1980s than in the 1970s (Calkins and Goodwin 1988; Pitcher et al. 1998). Thus, during and following a period when direct sources of Steller sea lion mortality were at high levels, the carrying capacity was likely declining as well. This apparently continued through the 1990s as evidenced by the persistent decline in Steller sea lion counts (Sease and Gudmundson 2002) as well as a possible decline in fecundity (Holmes and York 2003), even as the rates of human-related direct mortality were greatly reduced. These shifts in life history parameters during the declines argue against killer whale predation as a main cause of the decline, as, for example, there is no direct reason why increased killer whale predation would lead to a decline in fecundity.
For northern fur seals, York and Hartley (1981) estimated that known direct kills of females alone explained approximately 70% of the decline in the Pribilof northern fur seal population from 1956 to 1980. From 1956 to 1968, approximately 315,000 female fur seals were killed on land at the Pribilof Islands in an attempt to increase the productivity of the stock (York and Hartley 1981, Gentry 1998). In addition, approximately 40,000 northern fur seals (roughly three-fourths females) were killed as part of U.S., Canadian, and Japanese scientific pelagic collections in the North Pacific Ocean from 1958 to 1974.6 Interestingly, instead of increasing the productivity of the stock, pregnancy rates declined and the mean age at first reproduction increased (Trites and York 1993). The remaining 30% of the decline is unexplained, but York and Hartley (1981), Fowler (1987), and Gentry (1998) attributed it largely to one or all of a variety of factors: (1) methodological problems associated with pup production estimation, (2) changes in oceanic conditions, (3) entanglement in marine debris such as packing bands or discarded trawl netting, or (4) competition with groundfish fisheries whose catches in the eastern Bering Sea increased considerably in the early 1970s and have remained at approximately 2 million mt per year (NPFMC 2004). Furthermore, an unknown number of female northern fur seals and pups were killed by Russia during both on-land and pelagic collections. With the major portion of the population decline from 1956 to 1980 due to direct kills of females by humans, and the population being relatively stable thereafter until 1998, it is unlikely that killer whale predation contributed significantly to the population dynamics of Pribilof northern fur seals during this period.
Not all northern fur seal breeding colonies in the North Pacific had similar population dynamics through the late 20th century. Interestingly, with respect to the Springer et al. (2003) hypothesis, pup production increased at the Commander and Kuril Island fur seal rookeries from the mid-1960s through the mid-1980s while the Robben and Pribilof Island populations were declining by over 50% (Gentry 1998). Furthermore, rookeries became reestablished on San Miguel Island, California (in the mid-1970s), and on Bogoslof Island, Alaska (in the late 1980s) (Gentry 1998, Ream et al. 1999).
In the harbor seal example, counts from a single haulout on Tugidak Island are taken by Springer et al. to be representative of harbor seals throughout the GOA and BSAI; they also noted declines in counts at Otter Island (Pribilof Islands) from 1974 to 1978, and from 1978 to 1995. However, the largest concentration of harbor seals in the BSAI region (southern Bristol Bay) appears to have remained relatively stable during the time period concerned (Hoover-Miller 1994). Counts are not available to estimate population trend for the greater Gulf of Alaska region; specifically, for the south side of the Alaska Peninsula, Cook Inlet, the western Kodiak Archipelago, or the Kenai Peninsula. Counts from the eastern Kodiak Archipelago, primarily Tugidak Island, indicate that a substantial decline, which began in 1976 or earlier, continued through the late 1980s prior to an increase that began in the early 1990s. The trend in counts from south-central Prince William Sound has decreased since the mid-1980s. Of interest are the increasing trend in the Kodiak region and the continued decreasing trend in Prince William Sound, which suggests disparate population dynamics within separate stocks, a view of population structure that is supported by genetic and movement data (O'Corry-Crowe et al. 2003).
The amount of research on harbor seals has been substantially less than for Steller sea lions or for northern fur seals. Unfortunately, the information that would be required to fully assess the possible causes of harbor seal declines in the 1970s and 1980s was not obtained during the period of decline. In particular, unlike that of Steller sea lions, there are no data to investigate whether reduced growth or pregnancy rates occurred for harbor seals and, thus, evidence for reduced survival or reproduction due to reductions in prey biomass or quality does not exist. There is some indirect evidence, based on the timing of pupping and haul-out behavior, that harbor seals may have been nutritionally limited in the late 1970s in the Kodiak area ( Jemison and Kelly 2001); however, this evidence is not definitive. Thus, although predation could have been a factor in the decline of harbor seals, numerous other factors could also have been the cause, including contaminants, disease, parasites, subsistence hunts, disturbance, illegal shooting, incidental take, and reduction in prey biomass and quality.
Availability of Large Whales and Other Potential Prey
If large whales represented an important prey item for killer whales, there are at least two lines of evidence that argue against any need to prey switch to pinnipeds and sea otters because of an insufficient supply of such fare. First, the decline of large whale biomass was not as dramatic as suggested by Springer et al. and most of it occurred in the 1800s and first half of the 1900s, well before the start of the decline of pinniped populations. Instead of showing the trend in available biomass of large whales alongside the biomass trend curves for pinnipeds and sea otters, Figure 2 of Springer et al. uses the decreasing numbers of whales caught by whalers to make the case for a decline. This provides a neat set of sequentially declining curves. However, if one adopts instead the former metric (available biomass) for whales, a very different and far less elegant picture emerges. There probably remained a significant standing biomass of large whale prey, even for some species that were subject to intensive whaling. Although Figure 3 in Springer et al. suggests there has been a dramatic decline in cetacean biomass in the Bering Sea and Aleutian Islands region, this figure is based on a major error of calculation. The reduction in sperm whale biomass was calculated by comparing an estimate of current abundance of adult males for the Bering Sea and Aleutian Islands (15,000) to an estimate of historic abundance of adult males for the entire North Pacific (195,000) (see Pfister 2004: p. 5). This obviously greatly overestimates the decline in sperm whale abundance. Instead, a current estimate of 172,000 should be compared to the historic estimate of 195,000 for the entire Pacific; this equates to comparing a current estimate of 15,000 to a historic estimate of 17,006 in the Bering Sea and Aleutian Islands. Using those latter numbers for sperm whales combined with the estimates in Pfister (2004) for other large whales leads to a recalculated drop in whale biomass from historic to current levels of only 45%, not the 82% reported in Figure 3 of Springer et al.
Additionally, the sperm whale abundance estimates are considered unreliable because they were based on methods that have been discredited (Perry et al. 1999, Whitehead 2002, Pfister 2004); therefore, we do not recommend giving those numbers any credence. The overall reduction of cetacean biomass excluding sperm whales was calculated to be 54%, and most of this decline was noted to have occurred 50–100 yr ago (Pfister 2004), well before the declines of pinnipeds began in the northern North Pacific.
It is also important to note that the two large whale species (gray and humpback), which killer whales are regularly (gray) or occasionally (humpback) reported to attack, at least as calves, did not have a major drop in biomass in the 1950s and 1960s. Relatively few humpbacks were caught because they were already depleted from pre-World War II takes, and gray whales were increasing during that time period.
Furthermore, other cetacean prey was available for killer whales. Minke whales—known to be taken as adults by killer whales—were never subject to whaling in Alaska, nor in many other areas of the North Pacific. Accordingly, they have likely been an abundant potential prey item for killer whales throughout the entire period. In addition, although available data suggest that other cetaceans (e.g., various small odontocetes, notably Dall's porpoise) have represented an alternative food source of substantial abundance and biomass during the period concerned, and are some of the most frequently observed prey of killer whales in most areas, this information was not considered by Springer et al. It is noteworthy that in some areas of Southeast Alaska, killer whales are known to prey on Dall's and harbor porpoise despite the local abundance of large whales, notably humpbacks.
In short, even if large whales are or were a significant prey item for killer whales, the evidence indicates that there has been no lack of this and other potential prey in higher latitudes. Thus, a switch by killer whales to pinnipeds and sea otters as primary prey because of a lack of cetacean prey seems unlikely. Indeed, if cetaceans (large or small) were once the primary prey for killer whales, the increasing biomass of this taxon in the areas concerned would suggest that they should by now have switched back to this food source. If they have, two predictions from the Springer et al. hypothesis would be that (1) observed attacks on large whales should increase over the next decade, and (2) depleted populations of pinnipeds should increase in response to the reduction in predation pressure.
Prey Selection by Killer Whales: Does This Commonly Involve Large Whales?
Springer et al. suggested that depletion of large whales by industrial whaling forced killer whales to switch to other prey species. The key question here is therefore whether killer whales regularly attack large whales in the high-latitude areas that are the focus of the prey-switching hypothesis. In all three regions, the majority of observations of predation by mammal-eating killer whales involve pinnipeds and small odontocetes. The prey observations are remarkably consistent from region to region, with pinnipeds and small odontocetes (in that order) taken most frequently in all areas (although it should be remembered that the majority of these observations take place in summer and in coastal waters). Killer whales have been observed scavenging carcasses of large whales killed by whalers (Scammon 1874, Mitchell and Reeves 1988, Whitehead and Reeves 2005), events in which the killer whales have usually attempted to consume the tongue (Heptner et al. 1996). However, there are relatively few reports of predation on living large whales, particularly in the northern North Pacific (Mizroch and Rice 2006). It is also noteworthy that logbooks and journals from hundreds of North Pacific whaling voyages in the 1800s (notably to the Gulf of Alaska) almost never reported killer whales attacking large whales.
Although there are not many observations of killer whales preying on marine mammals in the North Pacific prior to 1950, those that were reported included harbor seals (Moran 1924, Scheffer and Slipp 1948), northern fur seals (Hanna 1922, Zenkovich 1938, Tomilin 1957), Steller sea lions (Scammon 1874, Zenkovich 1938), and Pacific walrus (Bailey and Hendee 1926, Zenkovich 1938, Tomilin 1957). These records suggest that pinnipeds have always been a primary prey of mammal-eating killer whales, including before the depletion of many large whale species in the 1950s and 1960s.
Observations of attacks on large whales are not common, and for those species seen attacked most frequently, the attacks are often on calves (Weller 2002). There are very few observations of lethal attacks on large whales in the high-latitude areas that were the site of industrial whaling, and still fewer of fin and sperm whales (the two species hunted most intensively). An isolated exception may be the Aleutian Islands, where in spring a population of transients has been observed feeding primarily on carcasses of gray whales (C. Matkin and L. Barrett-Lennard, unpublished data). If the carcasses result from predatory attacks, this may represent a strategy in which the killer whales intercept northbound migrants. However, even in this case it remains unclear whether age classes other than calves are commonly targeted. Killer whale attacks on young gray whales have also been reported from areas farther north than we considered, including the Russian Chukotka Peninsula7 and the Bering Strait (Lowry et al. 1987). Reported killer whale attacks from the BSAI region in summer include Steller sea lions, Dall's porpoise, fur seals, minke whales, and (in Bristol Bay) harbor seals and beluga. These observations are consistent with predation described from well-studied areas such as Southeast Alaska and British Columbia, with some modifications that reflect regional differences (e.g., fur seals are abundantly available in the Bering Sea but not in British Columbia). This could be interpreted as suggesting that killer whales preferentially pursue smaller prey that can be captured with a minimum of energy expenditure and risk.
In epistemological terms, one can never conclusively rebut the argument that lethal attacks on large whales are common but not observed: in the absence of perfect observational data, one can never prove that an event does not occur. However, it is very hard to accept the plausibility of this argument given that some populations of large whales have been under study for years or decades in areas where they co-exist with killer whales, yet where serious attacks are rarely, if ever, seen. This is all the more surprising given the obvious and dramatic nature of a killer whale attack on a large whale, which stands in contrast to the much more cryptic acts of predation on species far less conspicuous to observers, such as harbor seals (Ford et al. 2000).
Two prominent examples in this regard are Southeast Alaska and Prince William Sound. These are both well-studied areas where large whales (notably humpbacks) and killer whales are abundant, yet where a lethal attack has yet to be witnessed (Dolphin 1987; D. Matkin et al., in press; M. Dahlheim, unpublished data). In contrast, in both of these areas numerous killer whale attacks have been documented on smaller cetaceans and pinnipeds, including Dall's and harbor porpoise, harbor seals, Steller sea lions, Pacific white-sided dolphins, and (less commonly) minke whales. It makes little sense to hold that killer whale attacks on large whales occur regularly, but are somehow never witnessed, while at the same time other species are commonly observed being taken, often in exactly the same areas. In some species, such as humpback whales, there are many individuals with killer whale tooth-mark scars on the tails and bodies (Weller 2002). However, further analysis of tooth-mark scars supports the view that attacks on adult large whales in high latitudes are uncommon (Mehta 2004, Mehta et al., in review, Steiger et al., submitted). The timing of acquisition of scars originating from killer whales is key here: the majority of scarred whales tracked by long-term individual identification studies have such scars on their first sighting, rather than acquiring them from year to year. This strongly argues that most attacks occur during an animal's natal year, and in lower latitudes (rather than in the areas subject to the Springer et al. hypothesis). This conclusion is consistent with observations in the Archipiélago Revillagigedo in Mexico between 1996 and 2001 where researchers reported several humpback whale calves with fresh rake marks on their flukes and/or bodies, including chewed fluke tips that were still bleeding (J. Jacobsen and S. Cerchio, unpublished data), whereas fresh rake marks on humpback whales are not reported from high latitudes. Furthermore, killer whale scars are quite rare on some of the species reduced by commercial whaling, notably fin and sei whales. This argues that these species—which are a focus of the Springer et al. hypothesis—are of little importance as potential killer whale prey in the regions concerned.
It is certainly possible that killer whales made a significant living on whale calves. However, the evidence (from scar acquisition and direct observations) that calf attacks occur primarily in lower latitudes presents a serious problem for the Springer et al. hypothesis. For the hypothesis to remain tenable, killer whales would have had to be distributed primarily in offshore or tropical/subtropical waters during the period before whaling depleted whale populations, then have shifted their distribution substantially to high-latitude areas to exploit populations of pinnipeds and sea otters (all the while ignoring increasing populations of some large whale species in the latter regions). In fact, killer whales from Mexico and California are genetically distinct from killer whales in the northern North Pacific and are not part of the same population.