The fifth deadly attack in 10 months by a White Shark (Carcharodon carcharias) has released all alarms among Western Australians. The fatalities in Western Australia have been said to exceed, by a factor of 5, the average number of deaths caused by shark attacks in the entire nation. Something unusual and alarming seems to be happening, and the public demands that action be taken.
But, is it really unusual? Do we know why? Do we know what to do? Who should the public turn for answers? At this point I must declare that although a marine ecologist I have no particular expertise on sharks and have never published on the subject. However, we can apply scientific logic to try to narrow down the questions above before we do require expert opinion.
Statistics is all about testing the likelihood that a particular observation or result departs from that expected by chance alone. How unlikely an event should be to consider that it is not likely to be explained by change depends on a convention around an arbitrary number that identifies a level where invoking chance is no longer convincing. For most applications scientists chose a probability of 0.05 as the threshold (technically alpha value) separating a result possibly obtained by chance from a significant one. This is equivalent to consider that when the probability that an observation occurs by change is less than one in twenty times, it is unlikely that this observation is indeed the result of chance. For instance, it will take that the flip of a coin returns a head five times on a row ( probability that occurs by chance = 0.031) before we could consider that the coin is faulty or not flipped with equal chance of either outcome; or we need to roll a dice more than twice with the same outcome to suspect that the dice maybe loaded. The higher the number of observations around a particular event, the more likely we are to be able to detect a departure from a result attributable to chance alone. Hence, statistics are highly unreliable when dealing with small numbers.
Shark attacks are counted on single digits, so that our capacity to infer that the larger number of shark attacks reported in the past 10 months in Western Australia relative to the average cannot be explained by chance is limited. Indeed, the added mortality due to Shark attacks is such a minimal component of the annual mortality in Australia (143,500 deaths in 2010), that it is not visible in these statistics. The international shark attach file maintained at the Florida Museum of Natural History reports fatal shark attacks in Australia to range from 0 to 3 per natural year, with the total number, fatal and non-fatal, of attacks ranging from 7 to 21 per natural year. Hence, the 3 fatal attacks in 2011 and the 2 attacks thus far in 2012 are toward the high side but do not represent a significant departure from those observed in Australia in previous years. Yet, so much is at stake that we should make sure that the possibility that we erroneously assign these events to chance when there maybe a cause – an error known in statistics as Type II error – is ruled out.
Hence, the quest for an explanation of why there may have been more attacks this year. Three elements need be assessed: (a) has the population size of white sharks in Western Australia increased significantly over the past 10 months?; (b) has the number of people exposed to sharks increased?; or © has the likelihood of any one Shark attacking a swimmer increased?. Questions (a) and (b) affect the encounter probability between a shark and a human, which is affected by increases in either the number of sharks, humans or both. Question © affect to a change in behavior of either sharks or humans. An explanation that requires no change in behavior, i.e. questions (a) and/or (b), is more parsimonious that one that requires that sharks (or humans) behave differently, and according to the principle of parsimony, or Ockham’s razor, the longest-lived principle in science, would provide a stronger explanation in the case of equal evidence in support of questions (a) and (b) or ©.
Questions (a) and (b) require the existence of a baseline against which we can assess unusual increases in shark abundance or that of swimmers. Despite efforts to produce fisheries statistics for Western Australia, our baselines on shark abundance are probably very thin and too weak to detect a significant increase in shark abundance even if this may have occurred, so we are in a rather weak position to resolve question (a) with confidence. Question (b), however, can be resolved. Western Australia is the state in Australia with the steepest population growth, and the beach remains an environment of choice for recreational activities, so it is likely that the number of people on the beach and in the water any one day has increased over the past years. Whereas I do not have access to those data, I am sure they do exist, so we should quickly assemble the data required to test the merit of question (b). To resolve whether the encounter probability of sharks and humans has increased as a possibly cause for the (apparent) increase in shark-derived casualties also requires high-quality data on the abundance of regular preys of white sharks, for it is not just the probability of encounter between humans and sharks that matter but the proportion of encounters with potential preys that end up being humans. So, even if the abundance of sharks and that of humans in their environment may be the same, a decline in the abundance or availability of shark preys would result in a higher encounter probability of sharks and human-preys. Hence, our search for causes need include also possible trends in populations dynamics of seals and sea lions, the preferred preys of white sharks.
Question © is more difficult to assess as it requires detailed analyses of behavior over time for sufficient numbers of white sharks as to detect significant departures from a baseline or reference behavior. Whereas a number of white sharks are fitted with GPS and accelerometers to better understand their movement (GPS can track the animals over space) and behavior (accelerometers in their guts can provide a record of swimming behavior and attacks) for individual sharks, the number of sharks tagged and the observations over time are too thin to possibly provide statistically robust questions to ©. However, an increase in the likelihood of an attach in the absence of behavioral changes in sharks may also derive from changes in human behavior. Behavioral changes, such as an increase use of the marine environment, including more time and more distant swimming per capita or increase in the percent of the population that practices surf, which involves long-periods of time in the water, can also affect the encounter probability. Possibly these data do also exist and can be queried for evidence of behavioral changes in humans possibly conducive to a higher chance of attacks.
Knowing the why’s is important to formulate effective responses. Responses can act, again, over questions (a) through ©. The population size of sharks can be controlled through hunting; obviously thinning down the population of sharks can reduce encounter probabilities. However, in a context of a global decline of sharks due to overfishing it is unlikely that the population of white sharks be increasing. Moreover, sharks, also Great Whites, play a role in the ecosystem, and doing away with Grey Whites will create changes cascading down to the entire ecosystem. We also know that the number of road accidents depend on the number of cars on the road, so, applying the same logics, we could also do away with cars to reduce their vastly larger mortality stats. Indeed, thinning of the population of white sharks is not the answer.
We could also act on (b) and restrict human use of the marine environment. But, will this be acceptable to a society of beachcombers such as that in Western Australia? Likely not.
Hence, the most effective response is arguably acting on question ©, the likelihood any one shark attacking a swimmer. We can reduce this by acting on the sharks, the swimmer or both. Effective monitoring programs that trigger a response on the users of the marine environment to avoid encounters is an excellent strategy. These can involve fixed barriers between sharks and humans, such as those deployed in some areas in the East coast, or detection systems.
Detection systems include acoustic curtains that detect acoustically-tagged sharks transiting in the area, such as that deployed between Rottenest island and Cottesloe beach in Perth. Once a shark is detected a message is issued that alerts of the presence, triggering beach closure among other responses. Surveys with small airplanes, a traditional system in WA, is, however, costly and not very effective, as the airplanes surveyed the beach area only in the early morning. New technologies, such as oceanic gliders fitted with acoustic and video capabilities interfaced with target recognition software would represent an effective option. Australia is leading the world in the use of gliders for oceanographic purposes, but new innovations are required to produce the “biological sentinels” required for this application. These are in development in the UWA Oceans Institute. Also drones, which have been developed for war and intelligence purposes, could be developed to survey for sharks at a much lower cost than the use of airplanes. Both sets of autonomous vehicles would allow a sustained patrolling of beaches where humans are most likely to interact with sharks.
Systems to affect shark behavior upon a contact, such as the use of shark repellents, is also an effective option. These are also in development by the Marine Neurobiology group at the UWA Oceans Institute.
Lastly, human behavior can be managed through simple advice and best practices can help reduce the risk of encounter and attacks. For instance, Warren “Starry” Starr, Boat and Diving Officers the UWA Oceans Institute has provided our members with a number of rules, including, among other recommendations, the adhesion to any beach closure by local authorities, no in water activity to be carried out within a 10 nautical mile radius of a great white incident or sighting for a minimum of 72 hrs, and the correct use by all in-water parties of shark shield or equivalent deterrent devices. Additional advice is provided at the File Attack File at the Florida Museum of Natural History.
There is, however, an additional element that need be managed, the social alarm that has been released by the sequence of attacks. This requires the involvement of social psychologists and psychologists specialized in the management of fear. Indeed, as indicated above the deaths causes by these attacks are, even in the case of the current five-fold increased rates, insubstantial compared to death statistics in Western Australia. However, neither one citizen is satisfied by statistical detail once his or her own life is at risk. The reason for our alarm with shark attacks to be far greater than that associated with car accidents, causing far greater mortality rates, has to do with fear. This fear is, in turn, associated with a sentiment that we lack control on shark attacks. Citizens feel, correctly, that we can reduce our chances of suffering a car accident by implementing a number of safety measures and prudent practices which confer confidence and dissipate fear.
Hence, the effective communication of the actions adopted to manage these risks and ensuring that citizens are aware of best practices to avoid shark attacks should be a top priority. The mass media, which in some cases play a role in escalating fear and alarm by choosing extreme headlines, can chose to play a more positive role by communicating the measures above. This communication strategy need be complemented with the capacity to respond to citizens concerns. Moreover, reports and insights from the public can also provide guidance to scientists in assessing the likely causes for the perceived increase in shark attacks. A “town hall” meeting format, which the UWA Oceans Institute is currently organizing, where scientists and the public meet and engage in a dialogue about the issues above, will be effective as a step towards such effective communication.
In summary, whereas the perceived rise in shark attack may not be statistically significant, the case remains one causing high concern among the public, as the public is not interested in statistics but on their individual well being. Assessing the possible drivers of the increased shark attacks require effective and sustained monitoring efforts of the sharks and their natural prey, to detect departures from baselines possibly contributing to the attacks. It also requires information on the increase in the presence, attitudes and exposure of humans to shark habitat. Effective management measures should contain effective shark detection methods, triggering appropriate responses on the citizens, and the use of deterrents and protective curtains. Effective communication remains central to manage the fears the movie “Jaws” seeded in generations of citizens.