The evidence that killer whale predation was the proximate cause of the decline of sea otters in the central Aleutian Islands is good (Estes et al. 1998), and it is at least reasonable to suppose that killer whales can play a major role in natural mortality for pinniped populations (Heise et al. 2003; Williams et al. 2004; Branch & Williams in press). However, reports of killer whales attacking other whales seem to have risen, rather than declined, since about 1970 (Reeves et al. in press), suggesting that any increase in attacks on marine mammals over the past 35 or so years has been the result of a scarcity of some food source other than living large whales.

The observations make it clear that the tongues of baleen whales were the main focus of the scavenging killer whales, although they also consumed blubber and skin from carcasses. This pattern of partial consumption, with the tongue regarded as a ‘delicacy’, is consistent with reports of consumption of whales killed by the killer whales themselves (Jefferson et al. 1991).

We estimated the availability of large cetaceans and minke whales (Balaenoptera acutorostrata and B. bonaerensis), both live and as carcasses, during the twentieth century in all oceans. We divided the whales into four classes:

From estimates in the literature, and our own extrapolations, we estimated global populations of each of these species groups at three points in the twentieth century: 1900, 1945, 2000 (table 2 of electronic supplementary material). Then, for the aggregate total population, and the population without minke whales, we filled in population trajectories between 1900–1945 and 1945–2000 (figure 1; see electronic supplementary material for methods). Also shown in figure 1 are the reported catches of the worldwide whaling industry indicating the availability of whaled carcasses, as well as approximate confidence intervals for the availability of floating carcasses resulting from natural mortality. In figure 2, the curves of figure 1 are translated into whale biomass, using the mean adult masses of each species.

There are many sources of uncertainty and inaccuracy in the curves plotted in figures 1 and ​and2.2. These include inaccurate catch figures because of mistakes, falsification and the omission of catches by open-boat whalers of the early twentieth century and pirate whalers of the mid twentieth century; poor population estimates for almost all species (especially poor for some); the approximate nature of the interpolation models; uncertainty of the trajectory of minke whale populations; uncertainty as to which carcasses float and thus would be available to scavenging killer whales; changes with time in methods of catching and processing whales, and thus in carcass availability to scavenging killer whales; and the lack of good data on natural mortality. Therefore, the details of these curves do not deserve any scrutiny. However, major, and real, features are:

Thus the scavenging killers were likely specialists to some degree. Scavenging on carcasses resulting from natural mortality, including possibly predation by other killer whales, may have been a part of the repertoire of some killer whales before whaling, but given the spatial and temporal dispersion of the carcasses, it would probably have been difficult to subsist completely on the carcasses of large whales that they had not killed. Scavenging likely became more important during the era of open-boat whaling (Dudley 1725). We hypothesize that it was with modern whaling that scavenging became widespread and commonplace. Numbers of carcasses were being produced in reasonably small areas, with loud ‘dinner-bells’. The large, fast animals of the genus Balaenoptera with their huge tongues and thin skin were suddenly available, already killed, inflated with air so that they did not sink, and often left floating unattended for several hours (Tønnessen & Johnsen 1982). Those killer whales for which scavenging was already a part of the feeding repertoire would have had a bonanza, and possibly increased their reproductive rates, although mortality from armed, irate whalers (Robertson 1954) would have had the opposite effect. Other groups of killer whales, probably naturally mammal-eaters (fish-eaters seem loathe to eat mammals; Ford et al. 1998), were likely recruited to the scavenging way of life either through innovation, or by social learning from other killer whales already scavenging. During the carcass-rich decades of the mid twentieth century, as the scavenging culture spread, other ways of securing nutrients and energy could have receded from the behavioural repertoire, as well as perhaps the knowledge base, of these animals.

In this scenario, by about 1980 these scavenging killer whales faced a problem: the carcasses were almost gone. It seems reasonable that these animals would have turned to attacking pinnipeds, living large whales, small cetaceans and sea otters more frequently.

Killer whales have developed different feeding specializations in different parts of the world, and we should expect the significance of carcass scavenging, like natural predation, to have varied geographically. With variation in the cultures of the killer whales interacting with environmental differences, the switch from whale-carcass scavenging would have taken different paths in different areas, including, perhaps, significantly increased predation on sea otters and pinnipeds off Alaska (Springer et al. 2003), pinnipeds and minke whales in the Southern Ocean (Branch & Williams in press), and large whales off western North America (Reeves et al. in press). The extent to which killer whale predation has contributed to apparent declines of some populations of pinnipeds and cetaceans is beyond the scope of this paper, and remains an open and controversial question.