Contributions from the Mote Marine Laboratory Volume 1, Number 2

Shark Attack: A Program of Data Reduction and Analysis H. David Baldridge

Sarasota, Florida 1974 ANNOUNCEMENT

The Mote Marine Laboratory is an outgrowth of the Cape Haze Marine Laboratory founded in 1955 by William H. and Alfred G. Vanderbilt and ably directed by Dr. Eugenie Clark from 1955 to 1966. In 1961 the Cape Haze Marine Laboratory was moved from its original location in Placida, Florida to Sarasota, Florida and in 1967 its name was changed to honor its President, William R. Mote. The Mote Marine Laboratory is an independent, nonprofit institution, staffed by resident scientists and visiting investigators. Major areas of interest include continuing programs in Neurobiology and Behavior, Biomedical Studies, Estuarine Ecology, Microbiology and Environ­ mental Health, and the Biology of Elasmobranch Fishes. Volume 1, Number 2 of Contributions from the Mote Marine Laboratory represents the second of a series based on work entirely or chiefly undertaken at this laboratory. Number 1 of the series appeared in 1972 and the mono­ graph entitled "On the relationship of teleost scales to pigment patterns" was authored by Dr. C. M. Breder, Jr. It is with pleasure that we continue the high standards of excellence Dr. Breder's monograph set for the series and publish "Shark attack: A program of data reduction and analysis" by Dr. H. David Baldridge, Senior Research Associate at MML. The publication of the second number in this series is made possible by a generous financial contribution from Elizabeth Mote Rose of Tampa, Florida and from funds provided in Contract N00014-73-C-0252 between the Mote Marine Laboratory and the Office of Naval Research, United States Navy.

PERRY W. GILBERT, PHD Director

Sarasota, Florida January, 1974 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY

Volume 1, Number 2

Shark Attack: A Program of Data Reduction and Analysis

H. David Baldridge

Sarasota, Florida 1974 Additional copies may be obtained from

MOTE MARINE LABORATORY 9501 Blind Pass Road Sarasota, Florida 33581 $4.00 postpaid

Publication costs were partially defrayed by the Office of Naval Research, Department of the Navy, under Contract NOOOI4-73-C-0252 (NR 104-148).

PRINTED IN U .S.A. BY ALLEN PRESS, INC., LAWRENCE, KANSAS 66044 ABSTRACT

Data from 1165 case histories of shark attack against man were reduced to a form handleable by automatic data retrieval systems. Numerous probes by computer were made to (a) provide statistical significance to the existence or absence of common factors associated with known instances of predaceous shark behavior, and (b) evaluate the present approach to gathering meaning­ ful data on shark attack and thereby determine requirements for maintaining such an effort in the future. Although information was carefully screened for correlations between occurrence of shark attack and a number of environ­ mental and behavioral parameters, it was considered of greater importance that patterns or relationships be identified that distinguished victims from non-victims among exposed populations. Points of interest were examined in the light of previously accepted correlations and popular beliefs. Wound characteristics and other considerations suggested that perhaps 50-75% of shark attacks on humans have no direct relationship to feeding. A number of popular concepts linking shark attack to environmental parameters, in­ cluding water temperature, appear to be casual relationships having to do more with determining bathing pressure at beaches. Contrary to current ideas, divers appear attack prone, with a strong relationship to spearfishing. Trends indicate that, in the 1970's, attacks upon divers will average at least one-third of all reported cases. Yet, divers show fewer, less damaging injuries than swimmers and enjoy a far lower mortality rate. Effectiveness of a variety of weapons and diversionary actions were examined. Updated advisories were developed for bathers and swimmers, divers, and attack victims. The question of shark motivation in human attack was examined. A case was made strongly favoring continuation and expansion of the Shark Attack File. CONTENTS BACKGROUND 1 DEVELOPMENT OF DATA REDUCTION PROCEDURES 2 DATA ASSIMILATION PROGRAM . . 5 PRELIMINARY COMPUTER PROGRAMS 6 RESULTS REFLECTING UPON DATA CREDIBILITY 6 Documentation lag time 7 Primary data source 7 Analytical approach 7 DATE AND TIME . 8 Trends in mortality rates 8 Human population dynamics 10 Day of the week . 10 Time of day. . . 10 Shark population dynamics 11 Conclusions regarding time factors 11 GEOGRAPHICAL CONSIDERATIONS 11 METEROLOGICAL AND HYDROGRAPHIC FACTORS 13 Sunlight conditions . . . 13 General weather conditions 13 Underwater visibility 14 Sea conditions 14 TEMPERATURES OF AIR AND WATER 14 Water temperature . . 14 Feeding behavior of sharks versus temperature 17 Human physiology temperature considerations 18 Conclusions regarding water temperature. . 19 Water temperature affecting species of attackers 20 FEATURES OF THE SITE OF ATTACK 21 Proximity to deep water . 21 Depth of water at attack site 23 Relative water depth . 24 Distance from shore . . . 26 Depth of victim in water at time of attack 28 GENERAL CHARACTERISTICS OF VICTIMS 29 Race ...... 29

VII CON TEN T S (Continued) Sex of victims 30 Age of victims 31 Physical condition of victims prior to attack 31 Skin pigmentation 32 Degree of clothing . 32 Wetsuits . 33 Colors of clothing and gear 34 Laboratory studies on color effects under attack conditions 35 Strikes directed at colored bathing suits . 37 Conclusions regarding importance of color 37 Sudden presentation of unfamiliar color 37 Patterns and special features of clothing 38 General visual presentation . 38

PRESENCES OF POSSIBLE SIGNIFICANCE AT A ITACK SITE 39 Artificial lighting . 39 Animals. 39 Porpoises (dolphins) 39 Other people . 40 Fish. 41 Large floating objects 41 Blood, human 42 Blood, fish . 42 Human waste 43 Garbage or refuse 43 Other chemical stimuli 44 ACTIVITIES OF VICTIMS AND OTHERS IMMEDIATELY PRIOR TO ATTACK 44 Fishing . 44 Provocative acts 46 Specialized gear 46 General activities of others than victims 46 Pre-attack shark behavior . 47 Pre-attack activities of victims . 48 Length of time victim was in water prior to attack 51 Headings of victims in water 51 THEATTACK. 52 Number of sharks directly involved 52 viii CON TEN T S (Continued) Direction of original strike received by victim 53 Nature of initial strike . 53 Aggressive close passes without contact 53 Number of strikes by shark. 53 Number of bites or wound-groups 54 Attacks as single-event encounters 54 Shark behavior during attack . 55 High mortality rate of victims of frenzied sharks 55 Use of weapons against attacking sharks . 55 Chemical and physical anti-shark devices 56 Shark billy . 57 Spears, spearguns, etc. 57 Powerhead or bangstick 57 Diversionary actions by victims and others 57

OUTCOME OF A ITACK 59 Mortality rate . 60 Injury to rescuers 60 Injury to others than primary victims and rescuers 61 Wound characteristics 62 Recovery of tooth fragments from wounds 62 Body parts injured 62 Nature of wounds 64 A ITACKER SHARKS 65 Identity of attackers 65 Basis for identification of attacker sharks 65 Lengths of attacker sharks . 65 Basis for length assignments . 69 Evaluation of abilities to estimate shark length 70 Post-attack shark behavior . 71 Capture of attacker shark . 71 Human remains in sharks' stomachs 71 DIVERS AS VICTIMS OF SHARK A ITACK 72 Trends with time . 72 Distance from shore 73 Depths at which attacks occurred on divers 74 Provocation by divers 74 General activities of divers . 75 ix CON TEN T S (Continued) Direction of initial strike against divers 75 Close passes, strikes, and bites . 75 Body parts injured . 76 Nature of wounds received by divers 77 Mortality rate among diver-victims 78 Lengths of sharks attacking divers 78 Pre-attack shark behavior . 79 Nature of initial strike . 80 Shark behavior during attacks on divers 80 Post-attack shark behavior 81 Identity of sharks . 81 Water temperature . 81 Rescuers of diver-victims 82 Frequent availability of speargun as weapon 82 Summary of observations on attacks involving divers 82 CONCLUSIONS AND RECOMMEND A TIONS 83 Advice to bathers and swimmers 84 Advice to divers . 85 Advice to victims . 86 THE QUESTION OF SHARK MOTIVATION 86 REMARKS CONCERNING STATISTICAL TREATMENTS 90 THE CASE FOR CONTINUING AND EXPANDING THE SAP 91 ACKNOWLEDGMENTS 92 LITERATURE CITED . 93 APPENDIX. Summary listing of localities of 1165 shark attacks for which data were coded and analyzed . 96

x Shark Attack: A Program of Data Reduction and Analysis

H. DAVID BALDRIDGE Mote Marine Laboratory

BACKGROUND

This program had its beginnings along with that of the Shark Research Panel of the American Institute of Biological Sciences (AIBS), which was established on 25 June 1958 by the Office of Naval Research (Gilbert, 1963a). Dr. Sydney Galler, who then headed the Biology Branch of ONR, assembled in April of that year in New Orleans, Louisiana a group of 34 scientists internationally recognized for their knowledge of sharks. The pur­ pose of the conference, which was jointly supported by ONR and the Navy's Bureau of Aeronautics, was to consider basic research approaches to the development of better means for protecting naval personnel from predaceous shark activity. It was generally agreed that to accomplish this goal a much better understanding of the shark attack problem would have to be gained through intensified basic research in such areas as shark behavior, taxonomy, and functional anatomy. The Shark Research Panel was then charged with responsibility to the Navy for serving as a clearing house for information related to elasmobranch biology in general and the shark hazard problem in particular. It would also coordinate Navy-supported research in these and related areas, acting through the Hydrobiology Committee of the AIBS. Dr. Perry Gilbert, present Director of the Mote Marine Laboratory, was chairman of the Panel during its almost 12 years of existence. Other members were Dr. Sydney Galler (ONR), Dr. Albert Tester (Univ. of Hawaii), Dr. John Olive (AIBS) , Dr. Leonard Schultz (Smithsonian Institution), and Stewart Springer (u. S. Fish and Wildlife Service). The Office of Naval Research was later represented by Helen Hayes and then by Deane Holt. My election to membership on the Panel followed Dr. Schultz's retirement from the Smithsonian in 1968. One of the earliest major activities of the Panel was the establishment by Dr. Perry Gilbert of a world-wide Shark Attack File (SAF). With funds sub­ sequently made available by ONR for its maintenance, the SAP effort was undertaken by the Smithsonian Institution. For approximately nine years, data were gathered on all known shark attacks which occurred from time to time in various parts of the world. Dr. Leonard Schultz of the Smithsonian's Department of Vertebrate Zoology provided scientific guidance as Principal Investigator during the entire period of ONR support. In this effort, he was very ably assisted by Mrs. Marilyn Malin. Information was accumulated on 2 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No.2 approximately 1500 shark attacks (Schultz and Malin, 1963). Contents of individual case histories ranged from single newspaper accounts to detailed scientific observations. The Panel's chairman, Dr. Gilbert, continued to maintain a similar, but less extensive, file at Cornell University. Even though Navy funding for this data-collecting effort ceased in 1968, some subsequent additions have been made to the SAF, bringing the total number of cases available for analysis up to 1652. In 1967, attention of the Shark Research Panel was directed to the need for thorough objective (computer) analysis of the data on hand. It was the general consensus of Panel members that sufficient data were then available to (a) provide statistical significance to the existence or absence of common factors associated with known instances of predaceous shark behavior towards man, and (b) evaluate the present approach to the gathering of meaningful data on shark attack and thereby determine the requirement for main­ taining such an effort in the future. To provide for such an analysis, data of significance first had to be gleaned from present files of newspaper clippings, data forms, letters, photographs, etc. and organized (i.e. coded) so as to facilitate their transfer to a suitable automatic retrieval system.

DEVELOPMENT OF DATA REDUCTION PROCEDURES

As early as January 1966, I, as a naval officer, arranged for statisticians and programmers from the Navy's Bureau of Medicine and Surgery (BUMED) to assist Dr. Schultz in developing means for organizing the data held in the SAF. A series of conferences followed in which the mathematicians were made familiar with the nature of information on hand. Under the close supervision of Dr. Schultz, there was developed an impressive check-the­ appropriate-box type of questionnaire for use in assimilating the various types of data existing in the case histories. Numerous modifications followed reviews by members of the Shark Research Panel and statisticians from the Smith­ sonian Institution and Catholic University in Washington, D. C. The Panel met in December 1966 in response to notice from ONR that support of the Shark Attack File (SAF) would be terminated after one additional year of funding. A decision was then made to place highest priority on assimilating data already in the SAF in a form suitable for later analysis, with only very limited effort devoted to keeping the SAF up to date by in­ clusion of new attack records. Although made primarily for economic reasons, the Navy's decision also made some sense scientifically. The heavy investment of ONR funds in the SAF was basically to gain some insight into causative 1974] BALDRIDGE: SHARK ATTACK 3 factors of shark attack with the idea of possibly doing something to counter aggressive shark behavior. And if there was not sufficient statistically valid information for such analyses in 1500 files, then it would not likely be found in 2000, or 2500, or any other reasonable number of similarly compiled records. Furthermore, computer analyses were perhaps long overdue. Even though earlier analyses might not have yielded what was wanted in terms of information on causative factors, they might well have indicated constructive changes in data collecting procedures. By discontinuing support for keeping the SAF up to date, the Navy sacri­ ficed a secondary, yet very important, reason for originating the file. That was, to establish a means for providing to the general public information on the shark hazard problem of such a nature and degree of accuracy that the matter would more likely be retained in the proper perspective. The Navy fully appreciated the importance of this objective in terms of its ultimate effects upon morale of servicemen at sea. On the other hand, perhaps this responsibility more properly belonged to some government agency other than the Navy, since by far the majority of attacks recorded in the SAF occurred on civilians involved in recreational activities rather than on military sailors and airmen. In any event, due to lack of funding, the traditional informational services associated with the SAF were phased out beginning in 1967. Because of the impending retirement of Dr. Schultz from the Smithsonian Institution, I assumed primary responsibility in late 1967 for completing the task of organizing data in the SAF and then subjecting it to a thorough anal­ ysis, making use of computers where indicated. This decision was predicated upon the fact that for approximately two years I had already been working closely with Dr. Schultz and other personnel of the Smithsonian and the Navy in the development of procedures for coding information in the attack files . The Mote Marine Laboratory (MML) agreed to provide material, admin­ istrative, and technical support for the data reduction phase of the program under subcontract to the Smithsonian's ONR project. Utilization of the services of MML was indicated since I was at that time located there while attached to the Naval Aerospace Medical Center, Pensacola, Florida, in con­ nection with my own independent ONR-supported program of research on sharks and shark repellents. The SAF was microfilmed for security purposes before shipment to MML. My assistant, MML Data Analyst Joy Williams, and I began in the interim a review of a similar, but less extensive, shark attack file maintained at Cornell University and later at MML by Dr. Perry Gilbert. By late 1968, we were ready for trial coding operations using the nearly completed Smithsonian data reduction questionnaire. It was soon recognized that radical modification of the Smithsonian format would be required to make it more compatible with the data as it actually 4 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No.2

existed in the files. The proposed format would have served well (with a number of additions) as a questionnaire for gathering information on recent attacks where the opportunity existed for going into great detail while memories were still fresh. However, we were faced with the problem of codifying existing, often very limited, and usually very general information on attacks of the past, for which there was essentially no chance to collect amplifying data. Many of the codes available as multiple choice responses to questions had to be modified and expanded. Additional subject areas had to be covered, including some reflecting upon the overall value of the information being analyzed, both as to statistical significance and to credibility. To make for easier (and probably more accurate) subsequent handling by card punch operators, we prepared a series of coded answer sheets for each case using a master questionnaire instead of completing a multi-page, check-the-appropriate­ box type of form for each file. Each code sheet indicated, in an unambiguous manner directly usable by a card punch operator, all the entries to be made on a single IBM card. My final format evolved from the Smithsonian ques­ tionnaire during the intense review and test coding of about 250 randomly selected shark attack case histories. Other than the inclusion of a number of totally new points of information and the use of code sheets, my format differed from the Smithsonian's primarily in the phrasing and increased availability of coded responses. Codes were designed to reflect as accurately as possible the information as it existed in the SAF; no better and no worse, very general in some instances and relatively specific and quantitative in others. The Smithsonian format dealt with 157 separate responses making use of 247 of the 320 card columns available on four IBM cards. This was expanded to cover 211 responses in 299 card columns just on the matter of attacks on humans. The Smithsonian format dealt also with strikes by sharks on boats as well as shark activity associated with air and sea disasters. I found it more convenient to use separate (equiv­ alent to one IBM card each) questionnaires for each of the latter two problem areas, since the data available in those cases were so inherently different from that involving direct attacks on swimmers and divers. Analysis of information on boat attacks and air/ sea disasters will be the subject of a later, separate study. Where appropriate, care was taken in the data reduction program to utilize codes that were consistent with those already in use by the Navy for similar purposes; for example, the BUMED set of worldwide geographical codes used in its accounting systems for hospital inpatients. Finally, the format was criti­ cally reviewed by the Data Processing Department of the Naval Aerospace Medical Center, Pensacola, Florida in order to minimize the chance that problems might arise later during card-punching operations and in the 1974] BALDRIDGE: SHARK ATTACK 5 preparation of computer programs. For further details on the data reduction format, including procedures for air/sea disasters and boat attacks and listings of various codes, see Baldridge (l969a).

DATA ASSIMILATION PROGRAM

The data reduction operation was a formidable one; i.e. to carefully review all records of shark attack on humans, to ask of each of 1165 cases where data were sufficient a series of 87 questions, and to accurately record on code sheets the 299 numerals or letters for each case which indicated the responses (multiple choice in most cases) that best fitted the case histories. To complete the 348,335 data entries on 4660 code sheets took well over a year of con­ centrated effort by MML Data Analyst Joy Williams. Information in over 200 additional files was so limited (single line in a newspaper clipping with no particulars, etc.) or so diffuse that coding of them was not considered worthwhile. There were also 168 cases dealing with strikes by sharks against boats and 105 records concerned with shark involvement in air and sea disasters. Because of the inherent nature of this task and the quality of the information available for its performance, questions at times had to be answered (coded) with a degree of subjectivity not normally acceptable in a scientific study. We were well aware of this hazard, and a conscientious effort was made to keep such subjectivity at a minimum. It was always tempting to extrapolate in our minds the incomplete information found in the files and thereby contrive a completed mental picture of what had happened. However, the primary watchwords of our whole data assimilation program were 00 NOT ASSUME ANYTlllNG. Our job, as we saw it, was to assign codes which accurately re­ flected the data as they were actually found to exist in the shark attack records and not as we visualized the attack on the basis of what we read. This is where we differed greatly from most earlier analysts. Where conflicting accounts were found in the files describing certain aspects of the same incident, a code (the letter Y) was provided to indicate to the computer that information was available on a certain point, but that no clear evaluation could be made. The completed code sheets were thoroughly reviewed to eliminate illegible letters and numbers which might later confuse card punch operators. Then, as a security measure, they were carefully copied before shipment to the Naval Aerospace Medical Center at Pensacola for preparation of IBM cards and magnetic tapes. Since my program was Navy(ONR)-funded, the Data Processing Depart­ ment of NAV AEROMEDCEN readily agreed to handle, with no cost to my limited research budget, the operations of IBM card preparation, transfer of 6 CONTRmUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1. No. 2 data to magnetic tapes, program writing (in response to my instructions), runs on the computer, and unlimited consultation. For this, I am deeply indebted to the Officer-in-Charge of the department, Commander Harry Boone (and later LCDR Tom MacConnell), Medical Service Corps, U. S. Navy. The actual card and computer operations at Pensacola were handled primarily by Betsy White and her able assistants, particularly Kathie Bevins. Without their generous cooperation this study would not have been possible within the limits of available funding.

PRELIMINARY COMPUTER PROGRAMS

Checks were first made to detect and correct illegal codes. It was a satisfying verification of the care with which the code sheets were prepared to rectify only 39 illegal codes out of a total of 348,335 entries. The next step was to verify that questions in our data reduction format adequately reflected information that was actually available in a significant number of files. It would have been ideal had information been available on all 211 points of interest in every one of the 1165 coded cases of attacks on individuals. The alphabetic code R was used in response to questions where information was either unknown, not available in the file, or was of such form as to be insufficient for a clear-cut answer. The degree to which the R code was used was then taken as a measure of compatibility between the question­ naire and available data. Frequency analysis showed a median of 50 uses of R out of a possible 211. Thus, there was information available in half of the files to provide at least 161 or 76% of the 211 responses sought from each case history. This high level of availability of information supported our con­ tention that the coded responses which we provided for our questions did in fact adequately reflect the data as they existed in the files. To indicate better the variation among individual files, the average (as opposed to median) use of R was 60.8 with a standard deviation of 39.1.

RESULTS REFLECTING UPON DATA CREDIBILITY

In 17.6% of the 1165 coded case histories, the available information was insufficient to support a conclusion that the incident truly happened, much less whether or not sharks were actually involved. There was serious doubt in an additional 3.9% of the cases that an attack actually happened. Another 3.5% apparently dealt with an actual incident, but there was serious doubt about the involvement of sharks. In the remaining 874 (75.0%) cases, the information on hand fully supported the incident as a true happening, and sharks were considered as directly responsible for them. 1974] BALDRIDGE: SHARK ATTACK 7

DOCUMENTA TION LAG TIME

In 414 (35.5%) of the files, the earliest documentation was prepared within days of the attack and was taken to reflect relatively fresh memories of what had happened. A minimum lag of weeks occurred in 44 (3.8 % ) cases and months in another 78 (6.7%). It should be considered seriously in interpreting all results of the analyses to follow that in 600 (51.5 % of those coded) attacks the earliest documentation was not prepared until at least a YEAR after the attack occurred.

PRIMARY DATA SOURCE

One source of information in each file was determined as that providing the major portion of coded data. Authoritative scientific reports served as the primary data sources in 3.0% of the cases. Official U.S. Navy letter reports were not required by Navy order until 1965, so they were of primary value in only 0.9% of all cases. Letters or similar correspondence provided the major information in 10.5% of the files. Unfortunately, newspaper accounts had to be relied upon for most coded information in 28.9% of the attacks, with other popular accounts (books, magazines, etc.) providing the best data in 38.5% of the coded cases. Smithsonian attack reporting forms served as the primary data sources in only 15.3 % of analyzed case histories. Surely the weakest point in the whole program of data analysis is the fact that the primary coded data were supplied directly by the victims in only 86 (7.4 %) attacks, by eyewitnesses in 28 (2.4 %) others, and by a combination of victims' and witnesses' statements in another 4 (0.3 %). It was sobering to find that in 89.9% of the files on human shark attack held in the SAF, accounts of what happened were based primarily upon information supplied by persons who were neither the objects of the attacks nor were they even there at the time to actually see what happened. To be completely realistic, therefore, it must be conceded that the SAF is made up largely of hearsay evidence, mostly documented long after the event happened.

ANALYTICAL APPROACH

Before moving on to a discussion of specific analyses, our goal should be clearly stated and understood. Even though we were certainly interested during our analysis in uncovering relationships between a host of factors and the occurrence of shark attack, we were much more concerned with whether or not we could differentiate between those relationships and corresponding ones holding for non-victims generally. Our analytical approach was to try 8 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [VO\. I , No. 2 at all times to maintain some semblance of statistical control and thereby to identify and examine factors which distinguished shark attack situations and shark attack victims from ordinary nice days at the beach and ordinary people enjoying uneventful outings.

DATE AND TIME

Numbers of documented shark attacks are given in Fig. 1 for each year from 1941 thru 1968. Almost two-thirds (730 or 62.7%) of the 1165 total coded cases happened during that 28-year period. The average of about 26 cases per year is far below the approximately 100 attacks per year often stated in the literature as an estimate of worldwide incidence of shark attack. The likelihood of an attack being documented would surely depend upon the level of injury received by the victim, with deaths in recent years probably recorded to a high degree except possibly for wartime cases and those occurring in the most remote regions of the world. Particularly noteworthy points in Fig. 1 are the cyclic natures of the plots and the peaks observed in 1959 for both fatal and non-fatal reported cases. The 1959-peaks could be real or they could possibly be reflections of establishment of the Shark Research Panel in 1958. There is a strong possibility that reporting efficiency greatly improved in response to publicity incident to setting up the SAF shortly thereafter. On the other hand, vigorous efforts to collect information were maintained at least thru 1967, and still the reported numbers of attacks decreased each year fol­ lowing the 1959-peak. Furthermore, cyclic patterns and 1959-peaks were seen for both fatal and non-fatal cases. Since there is little basis for ques­ tioning the cyclic natures of these patterns, the possibility cannot be denied that the short-term cycles themselves may be superimposed upon long-term cycles leading to the observed 1959-peaks. If this be factual, we may now still be in a period of decreasing incidence of shark attack for which there is no ready explanation. For the period 1941-1958, reported attacks increased at an average rate of about 1.2 cases per year. Following the peak of 56 cases in 1959, the incidence of documented attacks has fallen off at the average rate of 2.0 per year, with only about 23 expected for 1973 (calculated from regression lines in Figure 1).

TRENDS IN MORTALITY RATES

There were 225 (29%) fatalities reported in 730 cases during the period 1941-1968, inclusive. Regression of percent fatalities on time indicated that mortality rates have generally decreased (an average of almost 1 % per year) from about 46% in 1940 to an estimated 16% for 1973 (from regression line 1974] BALDRIDGE: SHARK ArrACK 9

60 Mortality rates >- ~ 50 ..,., % : 82.9 - 0.912(year-1900) co N : 28, r : 0.644 : 40 c: u'"

Total cases

'" 30 -'".,u .., co ." '" 'C 2 Fatalities ~ :l U o ." /\Ji .... \ .,o ~ 10 "il :l Z \ ....

1940 Year 1945 1950 1955 1960 1965 Fig. 1. Total numbers of reported shark attacks, numbers of fatalities, and mortality rates as functions of time. in Fig. 1). These rates are considerably lower than the average death rate of about 50% often reported in the literature. To ascribe significance to these numbers, one must assume an equal reporting efficiency for fatal and non-fatal attacks-something that might well be very far from true. As with total number of cases, fatalities were also more numerous in 1959, with a high of 16 deaths by shark attack reported for that year alone (Fig. 1). Mortality rates dropped much more slowly (only about 0.52% per year) 10 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2

20.------~------~------r------_.------_,r_------._----__.

Myrtle Beach, South Carolina

15 \. '\ /' ..... _/ // \ I _------. I \ I \ I \ .... 10 I u.. I o u I .... I o I u I ~ 5 ..u ) '""

OL-______--JI' -______L- ______+- ______~ ______L ______~ ______~ Hours 0800 1000 1200 1400 1600 1800 Local time of day Fig. 2. Time of day related to occurrence of shark attack compared with time-distri­ bution of 1018 bathers at Myrtle Beach, South Carolina on 29 July 1972. during the post-peak period 1959-1968. Here again, this might only be a reflection of increased efficiency in reporting fatal attacks.

HUMAN POPULATION DYNAMICS

There is no evidence to link probability of shark attack at a particular time and place to any general turn by sharks towards more aggressive behavior. Instead, it is more likely related to human population dynamics as man expands his use of the sea for recreational purposes. The term "human population dynamics" is taken to mean variations on a time scale in populations of humans exposed to shark attack.

DAY OF THE WEEK. My son, David, pointed out that more shark attacks should occur on holidays and days of the weekend as opposed to weekdays if the rate of attack depended heavily upon human populations at beaches. His check of 729 available dates since 1900 by means of a perpetual calendar strongly supported this contention by showing that 65 % more attacks occurred on days of the weekend than on normal workdays of the week: Mondays, 97 cases; Tuesdays, 82; Wednesdays, 100; Thursdays, 90; Fridays, 70; Saturdays, 131; and Sundays, 159.

TIME OF DAY. Fig. 2 shows the relationship between relative incidence of attack and the local times of day when they occurred. For comparison, my 1974] BALDRIDGE: SHARK ATTACK 11 associate, Edward Broedel, conducted an hourly bather count at Myrtle Beach, South Carolina, on a day of heavy attendance. Both plots rise in early morning to a peak at about 1100 hours, falling off at about lunchtime, followed by a larger peak at midafternoon, and finally falling again to low numbers at about nightfall. These similarities, coupled with the observations regarding higher incidence on weekends, indicate that the rate of occurrence of shark attack is strongly associated with numbers of people in the water.

SHARK POPULATION DYNAMICS

That the feeding population of large sharks in a sparsely fished area does not vary widely with time is evidenced by the catch records of the Cape Haze Marine Laboratory (predecessor of the Mote Marine Laboratory). Collection rates are expressed as catches per unit effort, where catch was the number of sharks caught and a unit of effort was the use of one line of 15 to 18 hooks for an average set time of 24 hours (Clark and von Schmidt, 1965). Catches per unit effort off Placida and Boca Grande, Florida were as follows: year 1955, 1.8 ; 1957,2.2; 1958, 1.3; and 1959, 1.7. Those off Sarasota, Florida were: 1960, 0.76; 1961, 0.98; 1962, 0.95; and 1963, 0.63. The catches, of course, involved a number of different species, the relative abundance of which varied considerably with time of year.

CONCLUSIONS REGARDING TIME FACTORS

On a longtime scale, the incidence of attack appears strongly related to those periods of time when potential victims are most abundant. However, lack of control data prevents any inference concerning danger to any par­ ticular individual swimmer. Nevertheless, the chance of at least some un­ specified person being attacked at a certain time at any particular beach would be expected to increase as the concentration of people in the water increases. The time periods for such correlations are long, and the above considerations do not in any way deny the possibility of short-time rashes of attacks at times of low beach population. Neither do they mitigate against periods of calm even though the beaches are crowded with people.

GEOGRAPHICAL CONSIDERATIONS

All but four reported attacks occurred between latitudes of 47 degrees South and 46 degrees North. The four exceptions were all by captured sharks against fishermen and were considered to have been provoked. The most northerly unprovoked attack occurred not in oceanic waters, but in the upper reaches of the Adriatic Sea. The most southerly attacks happened off South 12 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

96 Sea Temperatures 70 128

1 26

60 24

22

20

18

40 .. .>/ .,u Shark Attacks !:!., 30 V' ~., .<::...... o ..... 2C ~ il :I :z;

ANY 50 40 30 20 10 o 10 20 30 40 50 I Degrees South Latitude Degrees North Latitude Equator Fig. 3. Geographical distribution of documented shark attacks and average sea (sur­ face) temperatures.

Island, New Zealand, almost directly opposite the Adriatic Sea on the surface of the earth. Attacks occurring at extreme latitudes, both north and south, should not be taken as being necessarily associated with the lowest water temperatures. Slightly more (54% of 976 cases) attacks occurred in the Southern Hemi­ sphere. The distributions within each hemisphere follow the same general pattern (Fig. 3): i.e. very few attacks near the equator, rising to a peak at 1974] BALDRIDGE: SHARK ATTACK 13 the middle latitudes, and falling off rapidly at higher latitudes. This pattern of attacks versus latitude follows generally what would be expected in terms of worldwide population distribution of man. A summary listing of incidence of shark attack in specific areas of the world is given in the appendix. The heavy predominance of attacks reported from English-speaking countries is highly suggestive of a language barrier in the procedures for gathering information on such happenings in other localities.

METEROLOGICAL AND HYDROGRAPHIC FACTORS

The extent to which people make recreational use of the sea is surely a function of weather. Furthermore, incidence of shark attack appears to be highly dependent upon the degree to which man presents himself to sharks thru the wide variety of uses made of the oceans. It should not be surprising, therefore, that there are a number of apparent correlations between frequency of attack and any number of meterological and hydrographic factors. Keep in mind, however, that the mere existence of a correlation between two such parameters does not necessarily mean that the occurrence of a particular con­ dition has any direct bearing at all upon why the second thing happened. The relationship could just as well be a casual one in that the parameters might be independently related to another factor or factors which actually govern the occurrence of both correlated happenings.

SUNLIGHT CONDITIONS

There were 707 cases in the SAF where sunlight conditions were reported or deduced from time-of-day with reasonable certainty: 1.3 % happened near dawn and 3.8% at dusk, 3.1 % in the darkness of night, and 91.8% during daylight. The necessary information is not available which would permit development of statistically valid conclusions from these data regarding rela­ tive levels of hazard. For this, information would be needed on numbers of people exposed during the same periods but who were not attacked. Never­ theless, logic dictates that the ratio between daylight-swimmers and night­ swimmers would be far higher than 91.8 %/3.8% = 51. To this extent, these data support the contention of greater danger of shark attack at night. There are in further support repeated field observations on increased feeding activity by large sharks as darkness approaches.

GENERAL WEATHER CONDITIONS

No inference at all regarding predatory tendencies of sharks can be drawn from data on general weather conditions. Of 298 files where data were avail- 14 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 able, 61 % of attacks occurred in clear weather, 32% on cloudy days, 6% in rainy weather, and 1 % during storms. Such conditions obviously so directly affect the degree of usage of beaches by people that to separate out variations in shark behavior would be impossible. So, of course, most shark attacks on bathers occur when most people go to the beaches; i.e. on clear and, to a lesser extent, on cloudy days.

UNDERWATER VISIBILITY

Essentially the same number of attacks happened under conditions of good underwater visibility as occurred when visibility was limited due to material being suspended in the water; i.e. 51.5 % of 336 cases in clear water and 48.5% in turbid, murky, or muddy water. These data do not take into account those times when there was good visibility in an area and a shark was seen in time to prevent an attack from occurring. Furthermore, if some shark attacks resulted from mistaken identity (black suited diver possibly appearing as a seal, etc.), such would surely have been less likely in clear water. So, available data do not necessarily detract from the general recommendations against swimming in water of low visibility when the presence of sharks is even remotely suspected.

SEA CONDITIONS

Sea states, as with general weather conditions, have more to do with the degree of man's use of the sea than with any detectable effect upon likelihood of shark attack. As expected, 69 % of 631 cases occurred in calm water, 19% in association with surf, 7% in choppy water, 3% in swells, and only 2% where seas were considered violent.

TEMPERATURES OF AIR AND WATER

It is difficult to visualize how air temperature per se could effect behavior of a submerged shark other than possibly indirectly thru its relationship with water temperature. The median air temperature for 172 cases where data were available was 79°F, with all but four attacks occurring between 55 and 104 degrees. These data again very likely reflect requirements for human comfort rather than any direct effects upon predaceous shark behavior.

WATER TEMPERATURE

Water temperature is supposedly related very strongly to occurrence of shark attack (Coppleson, 1962, 1963 and Davies, 1965), with attack being 1974] BALDRIDGE: SHARK ATTACK 15

50 "F 55 60 65 70 75 80 .'85 o 160 - • o • •• • 140 ~ • •• • • Weekend or holiday o o o Days of work week ~ .. B o 120 - o o o~. Ii> o • o .0 o. 0.0 o o • 100 - o o 00<1> 8 o 8 fJJ 0 o • 0 o 80 o -

... 0 0 GI o o 4.1; • 0 o l! o ~o o 4.1 60 - d o o .... o CD o ... o l! 4.1 .. . cC o .J:J .0 • ... 40 o .0 • ... o • o i o~ 0 :i ..... E> 0 ..... S> 0 8 20 o. 0 .0 "'0.0 .0 o .~cP. 0 000 • 00 00 0 o ,~~ ~ 00 0 o o~B ffofJ' B 0 o • (lea> Q:> 0 0 • 00 oel> 8 • 00 . ... 10 14 16 18 20 22 24 26 28 30 32 Water temperature Fig. 4. Counts of bathers in water of various temperatures at Myrtle Beach, South Carolina made by Edward Broedel at about 1500 hours during the period August 1971 thru October 1972. 16 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2

so S5 &0 65 70 75 80 85

160 *

140 * 0

• Weekend or holiday * Christmas holiday 120 o Days of work week - 0

0 0

100 -

". III 0 ~ .. 0 :J QJ 0 ..c:: 80 ~ 0 0• 0 ....c:: 0 0 0 OIl 0 ". QJ 0 ..c:: 0 0 ~ III .&l 60 .... • 0 0 0 • 0 f) ". 0 QJ ~ 0 0 ~ 0 0 =' 0 0 z 0 0 ~ 00 0 40 e 0 0 0 0 oCO 0 0 0 00 0

0 o. 0 0 0 20 0"• @ 8~ 0 0 0 0 "0 0 0 64 I» 10 12 14 16 18 20 22 24 26 28 30 °c Water temperature Fig. 5. Counts of bathers in water of various temperatures at Siesta Beach, Sarasota, Florida made by Beth Arthur at about 1500-1600 hours during the period February 1971 thru January 1972. 1974] BALDRIDGE: SHARK ArrACK 17 very unlikely below a "critical" temperature of 68-70°F. It is my contention that observed correlations between water temperatures and incidence of shark attack are only causal relationships having to do more with the physiological and comfort requirements of man rather than the inducement of predatory or feeding behavior in sharks. To provide direct evidence for a relationship between sea temperature and man's use of the sea for recreational purposes, two of my colleagues counted on essentially a daily basis (at about 1500 hours) the numbers of persons actually in selected regions of water of known temperature at two widely separated resort beaches. At Myrtle Beach, South Carolina, Edward Broedel counted 11,684 bathers on 240 days spaced between 9 August 1971 and 31 October 1972. Water temperatures ranged from 53 to 85 °F. The strikingly strong relationship between water temperature and bather popUlation at Myrtle Beach is shown in Fig. 4. Based upon Broedel's observations, it was estimated that during a full season of beach use from April thru November only about 4 % of bathers at Myrtle Beach would be in water colder than 70°F. Swimming continues all year round at Siesta Key near Sarasota, Florida where Beth Arthur counted 9,328 bathers from February 1971 thru January 1972 in waters of temperatures ranging from 59-89°F. The added bathing pressure provided by winter tourists is apparent in Fig. 5, especially that resulting from very heavy beach loads over the Christmas holidays. Even with such year round use of the beaches at this popular winter resort, it was estimated that only about 14% of the bathers enter water colder than 70°F. Based upon data in the SAF, the temperature (water) frequency distri­ bution for attacks upon humans (Fig. 6) exhibits a peak temperature range of 21-24°C (70-75°F). About 42% of 197 attacks where temperature data were available occurred within the range of 70-75°F, with 21 % happening in water colder than 70°F. Water temperatures ranged from 1°C (34°F) to 34°C (93°F), with a median of 22.rC (72.9°F). Although it would not appear so at first glance, the peak range of 70-75°F seen in Fig. 6 is compatible with observations plotted in Figs. 4 and 5 showing greatly increased use of beach waters at higher temperatures. To account for the shape of the plot in Fig. 6, one would have to have detailed data on fre­ quency of daily occurrence of water temperatures on a worldwide basis. Such data simply are not available. However, such a peak range well below max­ imum observed sea temperatures becomes understandable when the geographic distribution of shark attacks (Fig. 3) is re-examined in terms of mean sea tem­ peratures occurring at various latitudes.

FEEDING BEHAVIOR OF SHARKS VERSUS TEMPERATURE. In further contra­ diction of Coppleson's hypothesis, shark fishing experience during 1960- 18 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

35 "F 40 45 50 55 60 65 70 75 80 85 90 95 24 .-.---~---.---.--~.---~--~--~--~--~r---T---~,~~r-1

20

16 18 20 22 24 26 28 30 32 34 Water temperature Fig. 6. Incidence of shark attack as a function of water temperature.

1963 at the Cape Haze Marine Laboratory (predecessor of MML) suggests a negative rather than a positive correlation between water temperatures and feeding behavior of sharks (Clark and von Schmidt, 1965). Fishing efficiency expressed as catch per unit effort was found to fall off rapidly with increasing sea temperature over the range of 60-90°F (Fig. 7). Furthermore, Springer (1963) found in his extensive experience as a shark fisherman that warm water seemed to provide more rigid limits for sharks than cold water and led him to conclude that "shark fishing in water warmer than 85 °F was useless." Broadly speaking, it could then perhaps be that for any particular locality the real "shark attack season" would be that time when the local waters are warm enough for man (above 70°F) and not too hot for sharks (less than 85°F).

HUMAN PHYSIOLOGY TEMPERATURE CONSIDERATIONS. The relationships be­ tween water temperature and bather population observed by Broedel and Arthur (Figs. 4 and 5) are surely linked to the findings of Spealman (1945) and Molnar (1946) ; i.e. in water colder than about 68 OF, cooling of a human body exceeds heat production, and body temperature will continue to fall. The Navy's manual for Sea Survival (NAVWEPS-OO-80T-56, revised 1961) teaches that an unprotected man can endure essentially unlimited immersion in water at 70°F without danger of intolerably lowering his body temperature, 1974] BALDRIDGE: SHARK ArrACK 19

1.2 •

1.0 Y = 2.10 - 0.018lX ...... N : 10, r = 0.698 ° '" • 1>0" .5 0.8 ....c.. • '" • '"o • ...... § 0.6 ... "Q, .c ...u u.. .>:.. 0.4 .c • '" •

60 "F 64 68 72 76 80 84 88 Estimated average water temperature Fig. 7. Shark fishing efficiency as a function of water temperature observed by Clark and von Schmidt (1965) at the Cape Haze (now Mote) Marine Laboratory, Sarasota, Florida. Regression of fishing efficiency upon water temperature indicates an inverse dependency at a significance probability of better than 98 percent. while a stay of about two hours in water only 10 degrees colder (i.e. 60°F) would lead to about 50% expectancy of unconsciousness and probable drowning. All this is to say that there are perfectly good, well understood physiological reasons for man's reluctance to spend much time in water colder than 70°F, unless protected by a wetsuit. If man is relatively unavailable to sharks in waters of such temperatures where he is unable to maintain satisfactory levels of body heat, then it follows that shark attacks on man would be relatively rare at such temperatures, i.e. below about 70°F.

CONCLUSIONS REGARDING WATER TEMPERATURE. Accordingly, I see no jus­ tification at all in shark attack data for assuming any cause-effect relationship with water temperature or any inference regarding increased danger to indi­ viduals with rising sea temperatures above 70°F or any other arbitrarily chosen level. In fact, considering the observed degree of reluctance of people to swim 20 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [VO\' 1, No.2 in cold water, it is surprising that as high at 21 % of recorded attacks have happened in waters colder than 700 P. Very likely, the wearing of wetsuits has been a factor leading to higher attack potential in cold water than would be expected for unprotected bathers. On the other hand, Coppleson's hy­ pothesis is no longer in conflict with the above considerations when it is modi­ fied slightly and then stated somewhat in reverse, i.e. in the event of a shark attack upon a man (unprotected by a wet suit ) anyplace in the world, it is highly probable that the temperature of the water in which it happened would be above 68-70o P.

WATER TEMPERATURE AFFECTING SPECIES OF ATTACKERS. There appears to be another loose correlation involving water temperature that should not have been totally unexpected. Water temperatures were available in 82 cases where identity of the shark was also determined. Por those sharks where there were reasonable numbers of temperature measurements, it appeared that the species responsible for the greater numbers of attacks on humans were also those where the attacks spanned a wider range of water temperatures. Great white sharks were most often identified with 32 attacks to their record, and the water temperatures associated with 18 of these attacks spanned a range of 16°C; i.e. 1O.5-26.5°C or 51-79°P, inclusive. Tiger sharks were the next most frequently cited species with 27 human attacks. Six of these were in waters of 15.5 thru 31.5°C (60-88°P), for a range also of 16°C. It is inter­ esting to note that 14 (78%) of the 18 attacks by great white sharks where water temperatures were known happened in waters 700 P or colder. Sharks called mako in various parts of the world were known to have attacked 17 humans. Water temperatures in 5 of these cases ranged over 8°C; 18.5 thru 26.5°C (66-7rp). Bull sharks were found in 10 cases to attack over a narrow range of 9°C; 18.3-27.5°C (65-81°P). Even though this might infer that bull sharks should then be found responsible for few attacks because of limited geographical distribution, a number of very closely related sharks found throughout the warm waters of the world are loosely called bull sharks, and collectively they have been identified in 21 cases. Hammerhead sharks have been held responsible for 12 attacks, and 9 of them ranged over 12°C (17.5- 29.5°C or 64-85°P). There were insufficient temperature measurements associated with shark identifications to carry this association further. Even though the correlation may be a fairly loose one, it does seem logical that the finding of a wide operating range of water temperatures for a certain species of dangerous shark would imply a broader geographic distribution and that this in turn would provide it greater opportunity for attack on humans. Even though the probability of shark attack per se may not depend heavily upon water temperature other than indirectly thru its effect upon man's use of the sea, the chance of an attack by a particular species of shark may very 1974] BALDRIDGE: SHARK ATTACK 21

well be highly temperature dependent. In limited areas of the world where the primary hazard may be due to either a single or only a very few species of migratory sharks, there may very well develop in that area a strong relationship between water temperature and incidence of shark attack. It is my contention, nevertheless, that such a temperature range of higher attack probability would still be that which would put the greatest numbers of men and sharks in the water together and would not very likely result from an increase in predatory behavior of sharks which were present in reasonable numbers over a tem­ perature range significantly broader than the range of high attack probability.

FEATURES OF THE SITE OF ATTACK Sharks attack people in essentially every location where it is possible to get a man and a shark together. It isn't even necessary that the two be together in water. However, it is stretching the meaning of 'attack' when one includes a bite inflicted upon a fisherman by a wounded shark that has been dragged into a boat or up onto a dock, or the torn hand received by someone in reaching over the side of a boat to manipulate a shark caught on a line or snared in a net. In 51 cases held in the SAF, the 'victim' was not in the water with the shark. The majority of such 'attacks' were associated with some form of fishing, with the bite most often resulting from efforts either to boat a hooked shark, remove it from a net, or to otherwise provoke the shark after having cap­ tured it. Summing up the more usual situation where both the victim and the shark were in water together, Table 1 lists the reported numbers of cases on file for a variety of water types. As expected, the waters heading the list are those more often used by people for recreational purposes. It is easy enough in reviewing shark attack case histories to have it register on one's memory that a certain type of water keeps cropping up, such as breaker lines, river mouths, between sandbars, alongside channels, etc., and at the same time ignore the large number of attacks that have happened off ordinary beaches having no notable characteristics. In considering objectively all the attacks on file, however, there is very little that can be deduced from the listing in Table 1 other than that there seems to be no immunity from shark attack in any type of water where attack is possible. Without comparable control data on non­ victims, it simply is not possible to conclude from the order of this listing that anyone 'shark-infested' water is more or less hazardous than any other.

PROXIMITY TO DEEP WATER

It is generally accepted that it probably isn't a good idea to swim or dive just at the edge of a dropoff to deep water or immediately alongside a channel 22 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No.2

TABLE 1. General description ot waters ot attack.

No. cases in SAF Near shore waters/ beaches (no specifics) 407 Offshore reef! bar I bank 179 Open sea 88 Harbor/bay 89 River, etc.; other than mouth 73 Alongside breakwater I jetty I dockl wharf 41 Inside breakerI surf line 21 Just beyond breakerlsurf line 21 Mouth of river, creek, etc. 21 Waters between sandbar and shore or other bar 15 Shark penl tank, etc. 9 Lake 4 Inside faulty total enclosure 2 Lagoon 2 Tidal pooll rock pooll etc. 2 Inside meshed area 2

TOTAL 976 or trough. This supposedly provides the shark with easy access to its victim without, at the same time, exposing itself to danger by having to cross an expanse of shallow water. It surely makes sense not to make oneself any more available to sharks than is necessary while using the sea for whatever purpose. So, in an area where sharks have easy passage via channels or up over the edge of a dropoff, it would appear prudent not to make oneself readily available by spending a great deal of time along the edges of such accesses. This recommendation is an intuitive one, for there is very little information in the SAP on this point. Judgments were made in only a rela­ tively few cases, and even then they were at times admittedly highly subjective. A channel or trough was considered to have been or was very likely to have been within 50 feet of the attack site in at least 23 cases. Deep water (30 feet or over in depth) was thought to have been equally close in another 27 cases. In three other cases, deep water was considered to have been within 51-100 feet; another with deep water 101-150 feet away, and 2 cases where channels or troughs were present at distances of 151-200 feet. Attacks happened directly in what was reported to have been channels or troughs in 43 cases. 19741 BALDRIDGE: SHARK ArrACK 23

TABLE 2. Total depth oj water at the attack site.

Total depth No. cases in SAF 0-5 feet 290 6-10 56 11-20 54 21-30 24 31-40 16 41-50 11 51-60 3 61-70 5 71-80 3 81-90 1 91-100 101-120 121-150 1 Over 150 4

TOTAL 470

Thus, 99 attacks were thought to have been associated to varying degrees with a nearby channel or other ready access to deeper water. This low number can­ not be taken to mean necessarily that such accesses were not present with many other attacks. It means only that information is sorely lacking on this point. In the absence of statistical significance, logic still suggests that the probability of attack by a shark could increase if encounter were made more likely as in the situation where access to a victim was facilitated by the presence nearby of a channel or dropoff to deep water.

DEPTH OF WATER AT ATTACK SITE

On the subject of total water depth at the attack site, there were data avail­ able in 470 cases (Table 2). Sixty-two percent of these occurred in water no deeper than 5 feet, with an expected fall-off in incidence of attack as water depth increased. There is evidently no minimum depth below which shark attack is ruled out. The data in Table 2 do not necessarily mean that shark attack is more likely in very shallow water, for the relationship between the numbers of cases re­ ported for diverse depths of water could reflect no more than the normal dis- 24 CONTRmUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 tribution pattern of people in their recreational usage of the sea. In the absence of control data on non-victims, very little else can be said.

RELATIVE WATER DEPTH. Water depth can be examined from another angle for those cases which occurred in very shallow water. Of the 282 attacks

TABLE 3. Comparative data on shark attack conditions and observations at a "typical" resort beach (Myrtle Beach, South Carolina).

Relative depths of people in water SAP Myrtle Beach Total SAP (Beach activities)

~ / ~ % Total ~ / ~ % Total ~ / ~ % Total Knee deep or less 17/15 17% 31/3 16% 20/2 15% Knee to waist deep, including waist 37/26 33% 88/13 47% 56/10 46% Waist to neck deep 50/26 40% 59/ 8 31 % 43/5 34% Neck deep to over victim's head 18/ 2 10% 13/1 6% 5/1 4%

Total number 191 216 142

Headings of people in water SAP Myrtle Beach Total SAP (Beach activities) No. % Total No. 0/0 Total No. % Total Facing shoreward 42 23 % 24 45% 14 41 % Facing seaward 96 53 % 20 38% 14 41% Facing parallel to shore 42 23% 5 9% 3 9% Random activities 2 1% 4 8% 3 9%

Total number 182 53 34

General patterns of swimsuits SAF Myrtle Beach Total SAP (Beach activities) No. % Total No. % Total No. % Total No pattern or stripe 89 49% 154 79% 59 69% Patterned 49 27% 20 10% 12 14% Striped 43 24% 22 11 % 15 17%

Total number 181 196 86 1974] BALDRIDGE: SHARK ATTACK 25

TABLE 3. (Continued.)

Distance from shore SAP Myrtle Beach TotalSAF (Beach activities)

~ / 9. % Total ~ / 9. % Total ~ / 9. % Total 0-50 feet ------33/44 37% 149/27 31 % 93/17 39% 51-100 ------62/ 31 44% 5717 11 'Yo 36/ 5 15% 101-150 ______3017 18% 35 / 2 6% 19/2 8% 151-200 ------3/ 0 1% 17/1 3% 11/1 4% 201-250 ------13/ 0 2% 9/ 0 3% 251-300 ------32/ 3 6% 19/2 8% 301-350 ------5/ 0 1% 4/ 0 1% 351-400 ------7/ 0 1% 4/ 0 1% 401-500 ------15/0 3% 13 / 0 5% 501-1000 ------36/0 6% 25/ 0 9% 1/ 5-1 mile ______38/2 7% 15 /1 6% Over 1 mile/ open sea ______122/ 2 22% 3/ 0 1%

Total number 128/ 82 526/ 44 251/28 210 570 279 which were believed to have happened in very shallow water, relative depths in terms of the heights of the victims were determined in 216 cases. Of these, 16% of the victims were struck in water which was knee deep or less ; 47 % from knee up to and including waist deep; 31 %, waist to neck deep; and only 6% in water that was from neck deep to just over the head of the victim (Table 3). As an aid to interpreting these data, let us examine some of the control information collected by Ed Broedel at Myrtle Beach, South Carolina. Of 191 people actually counted in shallow water at that resort beach, 17 % were in water knee deep or less as compared to 16% of the victims of shallow-water attack; from knee deep up to and including waist deep, 33 % of beach bathers compared to 47 % of attack victims; waist-to-neck deep, 40% compared to 31 %; with 10% of the control bathers at neck deep to over the person's head compared to 6% of victims of shallow-water shark attack. Except for the reversed positions of the intermediate relative depths, the patterns are strikingly similar. The reversal at mid-depths could perhaps be due to a likely tendency to report a "waist-deep" depth for any attack which occurred in water reaching from upper-thigh to mid-chest. The control data would also, of course, be highly influenced by the contour of the bottom at Myrtle Beach. In any event, 26 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2 the relative depths at which victims were attacked by sharks appear to reflect the general distribution expected for people in shallow water at beaches and cannot be taken to indicate that any relative depth is more or less hazardous than any other. The correlation is meaningful, however, if stated in the right way (as was also noted with correlations having to do with water temperature). Thus, the data in the SAF, correctly interpreted, indicate that whenever a shark attack occurs in shallow water, it is more likely to involve a victim in knee-to-waist deep water, with a somewhat lesser chance of occurrence at depths reaching from waist-to-neck, a much lower probability for water knee­ deep or less, and less than 10% chance that the victim would have been in water at least neck deep. Stated this way, no inference is made at all con­ cerning the level of danger to any particular individual in shallow water at these respective depths.

DISTANCE FROM SHORE

Considering that there have been cases where charging sharks actually beached themselves in their eagerness to reach their victims, there is evidently no minimum distance from shore within which there is immunity from shark attack. Looking at Table 3, we see that over half (54%) the shark attacks at beaches which were thought to have happened within a mile from shore actually took place within 100 feet of the water's edge. This was somewhat expected, since, as shown by counts of bathers at a typical resort beach, by far the majority of bathers are found within 100 feet of shore. At increasing dis­ tances out from the beach, the incidence of shark attack did not fall off any­ where near as rapidly as did the population of bathers at a control beach. Beyond 300 feet from shore, the attack incidence at beaches remained at an average of about 7 cases per 100 feet of distance up to a range of about 1000 feet. The popUlation of bathers at beaches in general would be expected to fall off much more rapidly than that, with only a very small percentage of bathers expected to be out beyond about 200 feet. Yet, about 34% of the 279 attacks which involved beach activities took place more than 200 feet from shore. Surely, 34% of all the people present in the water, and presumably potential victims of shark attack, were not out that far. All this seems to suggest that there may be a very real increase in chance of shark attack as a bather moves out from shore. The observed data would be consistent with an interplay of two factors. First, a high percentage of shark attacks would be expected very close to shore (within 100 feet) simply because that is where the greater number of people would be encountered by any shark that came in close to a bathing beach. The hazard to any particular individual would be low because of the very high people-to-shark ratio expected very close to shore. The incidence of attack would begin to fall off rapidly as distance from 1974] BALDRIDGE: SHARK ATTACK 27 shore increased, thus reflecting the drop in bather population expected with increasing shore distance. Now a second factor could come into play in the form of increased chance of attack on any person moving out beyond about 200-300 feet from shore. This appears reasonable if we accept the assumption that the person-to-shark ratio might then be quite low, thus favoring the shark (i.e. a relatively high shark-to-person ratio). Consequently, incidence of shark attack as a function of distance from shore would not fall off as rapidly as would bather population. We may also be looking here at something that would at least partially explain the very high (over 1O-to-l) male-to-female ratio among shark attack victims. Looking at the control data in Table 3 collected by Broedel, we see that males tend to move out considerably further from shore at beaches than do females. If by doing so, they become more likely candidates for shark attack than people closer in, then perhaps there is no need to propose any fancy chemical or olfactory difference between human sexes, as sensed by a shark, to explain the high male-to-female ratio among attack victims. Of the 44 SAF cases of shark attack on female victims where distance from shore was noted, only 16% (7 attacks) occurred beyond 200 feet from shore, while almost 51 % of 526 male victims were at least that far out. When only attacks during beach-type activities were considered, attacks on females beyond 200 feet remained about the same percentage of the total as before (11 %-3 cases--{)f 28 attacks). However, the corresponding percentage of male attacks at over 200 feet from beaches dropped to 37 % of 251. This could mean that a significantly higher percentage of males than females take part in non­ beach-type activities (diving, surfing, etc.) which in turn routinely take them more than 200 feet from shore. This appears to be a reasonable idea to accept. The male-to-female ratio of total attacks happening beyond 200 feet from shore is 3 8-to-l, and remains as high as 31-to-l when only beach activities are considered. So, in a number of ways, the SAF data suggest that the chance of shark attack increases as one moves further out into the water away from a beach, and that the high ratio of men to women among attack victims may be due at least in part simply to women in general not venturing as far away from shore as do men. The term "as far away from shore" is used here in a relative sense. Thus, males would not necessarily have to move far away from shore in an absolute sense (i.e. in terms of actual distance) to become more likely attack victims than females. It would only be necessary that in any group of males and females, the outermost fringe would be most likely predominantly male. This is exactly what our control data at Myrtle Beach shows us to be true. The seven cases where females were attacked beyond 200 feet from shore involved one woman fishing while standing in waist deep water on an offshore reef, one who fell overboard from a boat, one thrown into the sea when a small 28 CONTRmUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 boat swamped, one reported to have been attacked 100 yards from shore with no amplifying data, one who was body surfing, and two who were swimming well offshore perhaps alongside a deep channel. Details of case histories clearly indicate that female victims are attacked equally as viciously as men. In our discussions of the relationship between level of hazard and distance from the beach, keep in mind that there is nothing in our considerations that precludes the possibility of any particular shark ignoring bathers further from shore and moving in to take someone, male or female, closer to the beach. It is not very often that a shark enters a well populated beach area to select a victim from among a group of people. On the other hand, quite often the victim is the person left alone and farther out from shore than others, as when one person 'misses a wave' and is left behind by other shore-bound surfers.

DEPTH OF VICTIM IN WATER AT TIME OF ATTACK

In popular writings, it has been hypothesized that shark attack is primarily a hazard of surface swimmers, and that divers, being beneath the surface, are in a less hazardous position. In support of this, it is generally stated that a diver disturbs the environment far less than a swimmer and he also is in a better position to see an aggressive shark in time to counter its charge. There are 881 cases in the SAF where there were data on how deep the victims were in the water when the attacks occurred. By far the majority (797 cases or 90%) happened either at the surface or no deeper than five feet from the surface. Sixty-two attacks occurred within 6-30 feet of the surface (23 at 6-10 feet, 27 at 11-20 feet, and 12 at 21-30 feet). Deeper attacks were as follows: 8 at 31-40 feet, 2 at 41-50 feet, 2 at 51-60 feet, 3 at 61-70 feet, 4 at 71-80 feet, one at 101-110 feet, one at 111-120 feet, and one at over 150 feet. Great care must be taken in interpreting these data in the absence of control information on non-victims. It is true that only about 10% of the shark attacks on file were directed against subsurface victims. These figures appear to confirm the idea of relative immunity for divers. Yet, intuition tells us that it is surely not likely for any period of time on a worldwide basis that 10% of the people in the water, either at beaches or offshore, are to be found more than 5 feet below the surface. Considering the great masses of people who swim at beaches in the warm water regions of the world, divers of all categories would be expected to constitute only a very small percentage (cer­ tainly far less than 10%) of persons exposed to shark attack. And consider further the fact that skindiving, SCUBA diving, spearfishing, and other related subsurface water sports have only recently become immensely popular. While the data in the SAF cannot be taken to conclusively indicate a greater hazard for divers as compared to surface swimmers, it certainly does not appear to support any status of immunity for divers. That divers are actually more likely 1974) BALDRIDGE: SHARK ATTACK 29 to be attacked than surface swimmers would be consistent with earlier con­ clusions regarding distance from shore. In order to dive to any appreciable depth, one would generally have to move farther out from shore than for ordinary surface swimming. To move farther out from shore is to make encounter with a shark possibly more likely. So, the fact that the person is either at the surface or well below it may not matter at all. There simply is not enough information to separate the effects of attack depth and distance from shore. It is extremely interesting that of the 65 cases of attacks on females where depths of the attack sites were reported, NOT A SINGLE CASE IS ON RECORD OF A FEMALE BEING A TT ACKED BY A SHARK BELOW 0-5 FEET FROM THE SURFACE. Thus, it may be that not only do females not venture away from shore as far as males, but when they do, as in the case of female skindivers, perhaps they also do not do some of the things done by male skindivers which are likely to invite shark attack. Could it be that female skindivers swim with a basically different movement than males-a movement far less exciting to sharks? Is it possible that sharks differentiate chemically between male and female humans even when the divers are encased in full wetsuits? Or maybe female skindivers are not as taken to such activities as spearfishing, or wearing black wetsuits where sharks feed on seals, or prying abalone from rocks in waters inhabited by large sharks, or tying bleeding fish to their belts, or any number of other things reported in association with shark attacks on male skindivers.

GENERAL CHARACTERISTICS OF VICTIMS

There are for shark attack victims a series of considerations which at best can be described as open-ended or totally uncontrolled. That is, in the absence of corresponding data on non-victims, there are very few conclusions that can be drawn which have any true significance, statistical or otherwise. This, however, has not stilled the hands of an assemblage of earlier writers, both popular and scientific, from drawing a number of often-quoted associations and assigning to them the status of cause-effect relationships.

RACE

Shark attack victims come from all races of men. It was not always possible to identify victims as to race, and admittedly in more than a few cases assign­ ments were simply educated guesses. Rather than classical racial groupings, we were more concerned with levels of skin pigmentation and so utilized only the broad categories of white, black, and brown/ yellow. Of the 1006 cases in the SAF which could be so classified, 725 (72.1 % ) were considered as 30 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 white, 176 (17.5%) were black, and 105 (10.4%) were brown/ yellow. It was very tempting to try to draw conclusions from these data as to the sharks' relative preferences for victims of one race or another based upon patterns of world population. This, however, would not have been meaningful in the total absence of control data. Thus, these numbers evidence only that no race of man is totally immune to shark attack and that there is insufficient information in the SAF to draw any statistically valid conclusions as to the relative preferences of sharks for the flesh of men of different races.

SEX OF VICTIMS

Almost all of the victims of coded attacks were identifiable as to sex. A surprising 93.1 % (1080 out of 1160) were male. Perhaps even more mean­ ingful is the fact that this predominance of male victims held true when race was also considered: white, 93.0% of 725 victims were male; black, 94.3 % of 176; brown/yellow, 92.4% of 105; and those of unknown race, 92.9% of 154. On the average, there were 13.5 male victims for each female. The first thought that came to mind was that more men than women are engaged in occupations which offer greater opportunity for shark attack, i.e. fishermen, mariners, etc. So, the computer was restricted to only those cases associated with recreational activities at or near beaches. Again, shark attack victims were found to be very predominantly male: white, 89.7% of 341; black, 91.4% of 58; yellow/brown, 93.9% of 33; and race unknown, 87.5% of 32. The overall ratio of male to female victims at beaches was 9.1 to 1. Counts of people in the water at typical bathing beaches showed a slight predominance of man, but nothing consistent with an attack ratio of over 9-to-1. Our observations, however, did confirm suspicions that males, in general, are more active when in the water. Activity, and the associated sonic and visual stimuli provided to sharks, has long been recognized as one of the prime triggers of aggressive shark behavior. In studies (Baldridge, 1966) where laboratory-bred rats were used to simulate survivors at sea, sharks repeatedly struck struggling, splashing rodents while showing little or no interest in either the same animals or others nearby when they were relatively motionless in the water. Perhaps males present to sharks significantly different olfactory profiles than do females. It has been often suggested that the danger of shark attack to a female might be affected by the stage of her menstrual cycle, but, in the absence of definitive data, this remains but conjecture. Very little is known about the chemical interchange which occurs between the body of a man or woman and the water in which they may be totally immersed. Maybe there is in this interchange some substance, hormonal perhaps, that is more peculiar to males and which sharks interpret as indicating some form of threat to which they respond with aggressive behavior. This begins to 1974] BALDRIDGE: SHARK ArrACK 31 make sense when, as discussed later, it is realized that a considerable per­ centage of shark attacks do not appear to have been motivated by hunger. In any event, we have in this about 10-to-l male-to-female ratio in shark attack victims something that cannot be explained off as being consistent with observed patterns in beach populations. Clearly, there is here a need for further basic research.

AGE OF VICTIMS The age distribution of shark attack victims indicates a heavy bias towards the teenager and young adult. This is consistent with so many of the recorded attacks having happened in association with recreational activities such as are engaged in principally by the young. Thus, the predominance of teenagers and young adults among shark attack victims is very likely a result of increased availability of these age groupings to sharks rather than anything to do with preferential selection on the part of the attackers. A total of 681 victims could be identified as to age. The actual spread in ages was from two victims in the 2-3 year bracket to two over 70 years of age. The median age was about 22 years. The 315 cases which involved beach-type activities showed only a slightly lower median age of 21 years. The median age for victims in United States waters during the period 1941-1966 was also approximately 22 years. Correspondingly, Bureau of Census data showed the median age of all people in the United States to have been between 28 and 29 years of age during that period. Unfortunately, we have no real data on the ages of actual beach popu­ lations, but we would be very surprised not to find the median age there to be several years less than that of the population in general, i.e. closer to that observed for shark attack victims. Hence, minimal restraint is felt in con­ cluding that there is little in victim-age data (median as well as spread of ages about the median) to indicate that sharks select victims of any particular age grouping. Instead, victims appear to have been selected from all ages making up the population of people normally expected to be involved in recreational use of waters at or near beaches.

PHYSICAL CONDITION OF VICTIMS PRIOR TO ATTACK Not too many years ago, recognized authorities on sharks were still willing to go on record with the belief that chances were very much against a shark attacking a healthy, uninjured live human being. Of the 1115 coded SAF cases where a judgment could be made, 1106 (99.2%) of the victims were considered to have been alive at the time of the attack. In the other nine cases, the evidence indicated a high probability that death (most likely by drowning) preceded mutilation by sharks. It is interesting to note that none of the live victims of the cases coded were considered to have been seriously 32 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 wounded, and therefore perhaps bleeding profusely, prior to the attack. Remember that the cases being considered in this analysis are those involving attacks on individuals not associated with air or sea disasters. It is with these latter cases where the chances would be greater for sharks to be drawn to injured, bleeding people in the water. There were, however, 19 cases where it was considered highly probable that the victims were bleeding at least to some minor extent at the time of or immediately prior to the attacks; i.e. from coral or shell cuts, open sores, etc. There are, then, no data in the SAF to support the position that immunity from attack would be bestowed upon a man simply because he was alive and not wounded.

SKIN PIGMENTATION

Skin color and contrasting areas of skin pigmentation have been considered to influence shark attack to the point that the lighter palms of the hands and soles of the feet are, by some divers, artificially darkened so as to provide a minimum of contrast with adjacent skin areas. Others would provide them­ selves protection by placing the lighter palms of the hands under their armpits at the first sighting of a shark. Our assignments of relative darkness of skin color were admittedly highly subjective, even where the victim was reported to have been of the black race. This information was not routinely recorded as such in most attack reports. Of the 482 cases where assignments were made, 163 (33.8 % ) victims were considered of dark colored skin; 63 (13.1 % ) of light color, and 251 (52.1 % ) were thought to have been either tanned or equivalently pigmented naturally. Here again, the total absence of similar data on persons who were in the water but were not attacked prevents us from drawing any conclusions whatsoever about the importance of skin color in marking a person as a likely candidate for shark attack. In at least five cases, on the other hand, it was considered that the locus of initial strike on the bodies of the victims may have been selected because of uneven tanning of the skin in those areas.

DEGREE OF CLOTillNG

The relative importance of skin pigmentation and uneven shading of skin color would be heavily dependent upon the degree to which the body of the victim was clothed at the time of the attack. This was not always reported as such in the case histories, but with 703 attacks, assignments were made with reasonable certainty. A few victims (15 or 2.1 % ) were thought to have been fully clothed, with an additional 9 (1.3 % ) fully clothed except for shoes. Included in these groups would be those persons attacked after having fallen overboard from a ship or pier, etc. As expected, the majority of victims (566 1974) BALDRIDGE: SHARK ArrACK 33 or 80.5%) were partially clothed, i.e. swimsuits, etc. Another 53 (7.5%) were either nude or essentially so.

WETSUITS. With the relatively recent advent of skin diving, increasing numbers of people are entering the sea wearing protective clothing, usually in the form of 'full wetsuits' completely covering the body except for the face or 'half wetsuits' covering only the upper torso and perhaps the arms. Of the 703 cases where degree of clothing was judged, 50 (7.1 %) victims wore full wetsuits with an additional 10 (1.4%) being partially covered by half wet­ suits. Now it was more likely that such details as degree of clothing would have been noted more often after the time of establishment of the Shark Research Panel in 1958. By restricting the computer to attack data later than 27 June 1958, the following results on 332 cases were obtained: fully clothed with shoes, 8 (2.4%); fully clothed except for shoes, 5 (1.5%); partially clothed/ swimsuit, 257 (77.4%); nude or essentially so, 7 (2.1 %); full wetsuit, 44 (13.3%) ; and half wetsuit, 11 (3.3%). Once more we are faced with the problem of drawing conclusions in the total absence of control information concerning the degrees to which non-victims were clothed. However, there appears to be something sufficiently lopsided in the above data to merit attention. Almost 17% of the victims (at least those where degree of clothing was considered) of attacks occurring since 1958 were wearing some form of wetsuit. It is seriously doubted, even with the great popularity of skin diving, that wetsuits were worn by almost 17 % of all the people who may have been exposed to the hazard of shark attack during that period of time. Perhaps the sense of it lies in the fact that wetsuits are primarily worn by skin divers, and that, by the very nature of their activities, skin divers expose themselves to shark attack to higher degree than would individuals who populate the beaches and seashores in general. Yet it might also have to do with the wetsuit itself. Particularly, since, until permitted by recent developments in the processing of pigments into rubber, wetsuits were almost exclusively black in color. It is certainly not beyond reason that a person so clad might appear to a shark to resemble a seal or other marine animals upon which the shark might be feeding in a particular area, especially if the water were murky and the range of vision limited. Perhaps, this would explain why, especially in a number of attacks by great white sharks on men in wetsuits, sharks occasionally have quickly released and deserted their victims after initial sudden assaults. It must be of little comfort to the victims, however, that they were perhaps as much objects of mistaken identity as they were intentional subjects of ag­ gressive shark behavior. Yet, it would appear prudent for divers to consider such as a real possibility when they dress in a way that might resemble natural, customary prey to an animal whose eyesight is not geared to fine detail. 34 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

COLORS OF CLOTlllNG AND GEAR

There remains considerable doubt among researchers as to the role of color in the life and behavior of sharks in general. Even though some anatomists and behaviorists (Gruber, Hamasaki, and Bridges, 1963; Hamasaki and Bridges, 1965; Gruber, 1969; and Myrberg, 1970) disagree on the point of functional capability of the shark's eye structures for perceiving color as such, there is general agreement with findings (Gilbert, 1963b) indicating that the rod-rich, cone-poor retina of the shark eye provides the animal with low visual acuity, but high sensitivity for distinguishing an object from a contrasting background even in very dim light. It should also be kept in mind that there are factors of color which would be available for informational purposes to even a color-blind animal, i.e. brightness, reflectivity, shading, contrast, etc. Thus, the importance of bathing suit colors in inviting shark attack would not necessarily depend at all upon any assumption that the animal sees the colors as we do. There are few points in these analyses that have less control data, or even the chance of ever getting any, than the matter of colors or patterns of bathing suits and other gear and equipment either carried or worn by shark attack victims. Consider for a moment the infinite variations in colors and styles of bathing suits that have occurred during the period of time from which the coded shark attacks have been drawn. Add to this the continually changing individual preferences of different peoples and cultures on a worldwide basis. Factors in attack reports such as color and pattern of swimsuits can only be shown to be meaningful if, by comparison with control data on non-victims, those factors characterize victims at a higher level than can be explained by the laws of chance. In the total absence of corresponding control tabulations for non-victims at the same times and places, there is just no way to make such a determination. Without it, sweeping conclusions regarding colors and patterns worn by shark attack victims simply cannot be justified on the basis of information held in the SAF alone. There are many ways to present the data on color as reported in case his­ tories. Schultz (1967) chose a block diagram wherein the occurrences of contrasting colors or shades were listed against background colors reported for gear or suits used by the victims. No control data were listed for com­ paring these color combinations with those worn by people who were not shark attack victims. Schultz found that, of 151 attacks where color com­ binations were reported, 72% of the victims displayed dark base colors (blue, brown, green, red, etc.) with lighter contrasting shades (white, yellow, tan, etc.) as opposed to light background colors with darker contrasting shades or patterns. Whether this is significant or whether it merely reflects the color preferences of beach patrons in general cannot be decided on the basis of 1974] BALDRIDGE: SHARK ATTACK 35

TABLE 4. Colors reported in association with known instances of shark attack on man.

No. of cases where color was Color Present Absent Black 82 148 Blue 50 171 White 45 177 Yellow 25 199 Red 24 197 Green 19 202 Brown 13 206 Tan 12 211 Orange 5 219 Pink 2 222 Other misc. colors 15 210 these data alone. The color data presented in Table 4 are only intended to indicate the number of times certain colors were reported to have been asso­ ciated with shark attack. Nothing in regard to relative hazard levels is neces­ sarily implied by the magnitudes of these numbers. The darker colors pre­ dominate, including the traditional black for wetsuits. Considering that, until a relatively few years ago, bathing suits were primarily made of wool, the higher incidence of darker colors would be no surprise, for the more colorfast earlier dyes for wool were of the darker shades. In those days, bleaching was the primary alternative to using relatively unstable dyes for most fabrics, hence the relatively high occurrence found by Schultz of white contrasting with blue or some other dark color should also be no surprise.

LABORATORY STUDIES ON COLOR EFFECTS UNDER ATTACK CONDITIONS. Everything so far has been to say that, in the absence of control data, it is not possible to draw meaningful correlations between incidence of shark attack and colors reportedly worn by victims. But, there are considerations other than attack case histories that definitely do point to the probable importance of colors and patterns in affecting shark behavior. To shark researchers, the term "Yum Yum Yellow" stands for International Orange and the related bright yellow and orange-yellow colors most often employed in connection with such sea survival equipment as life jackets, rafts, etc. These pigments are used for such purposes primarily because they con­ trast strongly with the background color of the surface of the sea, especially when viewed from the air. It should be kept in mind that the primary hazard 36 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 to a man adrift at sea is not shark attack, but is instead the very high proba­ bility of not being sighted by search aircraft and surface vessels. Thus, it is extremely important that measures be taken to make the man in the water as conspicuous as possible, at least above the surface of the water. Unfor­ tunately, conspicuity is not a selective quality, and to be easily seen by one searcher is to be easily seen by all, be they rescuers or predators. A number of tests involving survival gear of different colors (including variations in hue, brightness, contrast, and reflectivity) have clearly demonstrated the sharks' predilection for attacking objects of bright, contrasting, and/ or reflec­ tive appearance. Dr. Scott Johnson of the Navy and Ernest McFadden of the Federal Avi­ ation Administration (1971) observed such effects during their studies on color and reflectivity of sea survival equipment as related to shark attack. A standard yellow life vest occupied by a child dummy was repeatedly attacked at the surface by blue sharks, Prionace glauca. Strikes on a red infant flotation device were few, while a similar black flotation device suffered only two strikes. Mako sharks, [surus oxyrinchus, circled at the limits of visibility and occasionally made high speed attacks from below with teeth bared and snap­ ping jaw motions just prior to contact with their target, and with none of the preliminary surface behavior of the blue sharks. In one instance, an arm was torn from the yellow-jacketed dummy and apparently was swallowed by one of the mako sharks. No attacks by mako sharks were observed against either the red or black infant flotation devices or a dummy in a black life vest. Only the dummy wearing the yellow jacket was struck by these highly dangerous sharks. In view of such observations, one needs no further inducement to agree with the recommendations of McFadden and Johnson in that (1) methods for rendering life vests and reversible life rafts a less conspicuous and attractive visual target for shark attack should be explored, (2) highly reflective and attractive hardware such as chrome-plated carbon dioxide inflation cylinders, buckles, and snaps normally found on life vests should be of black non-reflective material, and (3) the submerged portion of life rafts or slide rafts considered non-reversible should be of a black non-reflective material. It would appear prudent for designers of diving and other aquatic sportswear to also take note of these recommendations. Drs. Albert Tester, G. Nelson, and C. Daniels (1968) observed effects of reflectivity on aggressive behavior by sharks during their testing of Scott Johnson's Shark Screen in Hawaii and in the lagoon of Eniwetok Atoll. The Shark Screen is a large plastic bag, filled with water, and buoyed by flotation rings attached to its top. A survivor at sea would await rescue floating motion­ less in the confines of such a bag, all the while putting no blood or any other olfactory stimulants into the water and presenting to curious predators only a gross uninteresting shape with no dangling arms and legs. While test results 1974] BALDRIDGE: SHARK ATTACK 37 indicated the color per se of the bag to be relatively unimportant in influencing shark behavior, reflectivity was considered to have been very important. Bag colors of low reflectivity (i.e. black) were recommended. The incidence of approaches and contacts by sharks were both found to rise with increasing bag reflectivity, reaching maximum values in tests on a highly reflective bag which had been covered externally with a silvery foil. Shark screens of a variety of colors were also tested by Dr. Perry Gilbert (assisted by myself, Dr. Johnson, and Dr. Karl Kuchnow) in the shark pens of the Lerner Marine Laboratory, Bimini, Bahamas. The results were similar to those found in the Pacific tests in that dark blue or black bags appeared to be the least attractive to sharks, while white and silver bags were the most attractive (Gilbert, 1968). Here again, it was recommended that colors having a high degree of reflectivity be avoided.

STRIKES DIRECTED AT COLORED BATffiNG SUITS. If bathing suit colors per se were of very great importance in shark attack, then it might appear reason­ able to expect that those parts of the victims' bodies covered by the suits would be preferred objects of attack. As will be discussed in detail later, wound charactertistics do not support this view, for the arms and legs are struck far more often than any part of the torso. Ready availability of the appendages, however, might be an overriding factor here.

CONCLUSIONS REGARDING IMPORTANCE OF COLOR. Without further bela­ boring the matter of color as it might affect shark behavior, some general con­ clusions appear to be justified on the basis of field observations if not by data held in the SAP. The bright, highly reflective dyes and pigments employed in some swimwear and diving gear for attracting the attention of humans out of the water can be expected to do much the same with sharks when submerged. This would also be true of bright metallic buckles and fittings of diving gear and survival equipment. It is highly probable that a shark near enough to be attracted or excited by any of the qualities (reflectivity, hue, etc.) of color worn by a swimmer would already be aware of the person's presence thru acoustic and/ or olfactory cues. Vision comes into play primarily during the final approach of a shark to its prey (Gilbert, 1966). It seems logical, therefore, that an aware shark, patrolling at the limits of visibility, might be transformed into a curious shark, close-at-hand, as a result of visual stimu­ lation provided by highly reflective dyes and pigments. A curious shark close-at-hand has all the makings of a potential attacker. Such factors might well come into play in the shark's selection of a particular victim among a group of bathers.

SUDDEN PRESENTATION OF UNFAMILIAR COLOR. Undoubtedly, the shark responds to the whole presentation made by the victim (i.e. visual, acoustic, 38 CONTRmUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I , No. 2 olfactory, etc.) in some complex, integrated fashion. When one or more facets of the total picture are not that to which a shark has become accustomed, behavior evidently can be affected markedly. Clark (1963) observed an interesting response to an unexpected presentation of color when a yellow target was substituted for a white one in visual discrimination studies of sharks. A large male lemon shark, Negaprion brevirostris, made its usual, fast and deliberate approach to the target, the touching of which the shark had been conditioned to associate with receipt of food. The shark suddenly stopped a few inches before its snout touched the target of unfamiliar color and did a backflip out of the water. A general fright reaction seemed to follow, and all seven of the sharks in the pen started swimming in a nervous manner. The subject lemon shark never fed again, after that test, and died several months later, apparently of starvation.

PATTERNS AND SPECIAL FEATURES OF CLOTIllNG

Information on the general pattern of clothing was available in only 196 cases in the SAF. Of these, 154 victims (79%) were believed to have been wearing clothing not characterized by any distinct pattern or stripe. Even where there was information regarding color, it was rare that mention was made of patterns or other use of the colors mentioned. Patterned clothing was noted in only 20 cases, while stripes were reported in 22 other attacks. Even less information was available on the matter of special features of clothing, gear, etc. Of the 127 cases where judgment was possible on the basis of evidence at hand, shiny items were noted in 12 cases, highly con­ trasting colors in 22, and very bright (such as International Orange) colors in 9 others. No special features of clothing or gear as to color, brightness, etc. were considered present in the remaining 82 (65%) cases.

GENERAL VISUAL PRESENTATION

If any recommendations at all can be made in regard to visual presentations made by swimmers and divers, they would appear to be along two general lines : ( 1) unnecessary conspicuity beneath the surface of the water should be minimized by attention to bright (reflective) dyes and pigments, shiny metallic fittings, strongly contrasting colors and patterns (including unevenly tanned skin areas), and any other factor that might lead to an interesting visual presentation from the shark's viewpoint; while at the same time, (2) great care should be taken not to appear in any way to resemble visually any natural prey upon which sharks, even relatively small ones, in that particular area are known to feed. Both these considerations should be expected to increase in importance when operating under conditions of limited visibility, i.e. murky water or restricted lighting. 1974) BALDRIDGE: SHARK ATTACK 39

PRESENCES OF POSSIDLE SIGNIFICANCE AT ATTACK SITE

ARTIFICIAL LIGHTING

Lights, both submerged and just above the surface of the water, are used often in underwater photography and diving operations. Such lights might well attract sharks, either directly or perhaps indirectly by first attracting smaller fish. Very little information is available on this point, however. Subsurface artificial lighting of a constant nature was reported in only two cases, with underwater photo flash noted in two others. Lights above the surface were mentioned in 10 others; five involving lights constantly on, two reports of intermittent or flashing lights, and three cases where lights were mentioned with no specifics. In all but perhaps two of the cases where artificial lighting was mentioned, the presence of such lighting did not appear to be directly linked to the shark's behavior.

ANIMALS

It is important to consider the possible effect of animals other than humans being present in the water in close proximity to the victim at or immediately prior to the time of attack. Only limited information is available on this point, perhaps due to unawareness of the possible importance of noting such things in association with attacks. Thus, even though such presences were reported in only a very few (7) cases, their absence in all other cases cannot be taken with any level of certainty. Sharks have been known to attack and at times devour a wide variety of land animals, including dogs, cats, cattle, horses, etc. Racehorses have been attacked a number of times in Australia, where they are routinely exercised in the surf. McCormack, Allen, and Young (1963) told of a thirst-crazed elephant which, in 1959, stampeded into the sea off Kenya, Africa, evidently in search of water on a nearby island. It never made it, for huge sharks fell upon it and tore it to shreds.

PORPOISES (OOLPmNS)

According to legend, shark attack is very unlikely when porpoises (dolphins) are nearby. It is popularly believed that dolphins use their snouts to ram the less agile sharks in some vital spot, usually thought to be the gills or liver, and thereby inflict very rapidly fatal injuries. Occasionally, people exposed to possible shark attack, whether real or imagined, have credited their relief from harm to the nearby presence of dolphins. There can be no satisfactory argu­ ment against such testimonals. It is difficult enough to identify some of the reasons why shark attacks happen without attempting to delve into the matter of why in other situations they do not happen. It can only be pointed out that 40 CONTRmUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2 studies under controlled conditions (Mathewson and Gilbert (1967); Gilbert (1972); Gilbert, Irvine, and Martini (1971); Irvine, Wells, and Gilbert (1973); and Wood, Caldwell, and Caldwell (1970», have shown no such high-level natural animosity between sharks and dolphins. In fact, Gilbert found that a very intensive period of patient training was needed to get a large Atlantic bottlenosed dolphin to act aggressively towards first a small dead shark and finally a larger live one. Furthermore, dolphins, both whole and in parts, are at times found in the stomachs of sharks. Granted, the dolphins might have been in a weakened condition or even dead when devoured by the sharks. Shark fishermen have reported dolphin flesh to be a very attractive shark bait. And finally, there were only two cases in the SAF where dolphins (porpoises) were known to have been present immediately prior to or during the attacks, apparently without significantly altering the course of events.

OTHER PEOPLE

Are sharks more apt to strike at people who are alone or are they likely to be attracted to a group of several people from which a victim is then selected? Again, conclusions should not be drawn which cannot be supported by con­ sideration of similar information on non-victims. Of 637 attacks where judg­ ment was made, it was considered that in 217 (34%) cases there was no person in the general area of the attack other than the victim. This is con­ sistent with the findings of Schultz (1967) where 31 % of 275 selected cases involved attacks on victims with no companions closer than 101 feet. It was very difficult indeed to determine from newspaper accounts, etc. the exact distances separating the victim from companions, if any. The accounts of 726 cases wherein at least a minimum of information was available indicated that 62 (9%) attacks occurred on victims having one companion within to feet, with 10 (1 %) others having more than one person that close. Indicative of the extreme difficulty and high level of subjectivity unavoidably involved in interpreting case histories on the matter of proximity of others, Schultz con­ sidered 122 (44%) victims among 275 selected cases to have had companions within 10 feet at the time of attack. Agreement was better at greater distances. Schultz reported 54 (20% of 275) attacks where the nearest companion was 11-50 feet away, while we interpreted the data on 576 attacks to indicate 157 (27%) instances where 1-5 persons were within to-50 feet of the victim, with 17 (3 %) additional cases where others were considered present within these distances but where no evaluation of exact number of companions was possible. Although our interpretations of the data may differ quantitatively from that of Schultz, there is consistency in the general conclusions that shark attacks occur on both accompanied and unaccompanied victims and that no real significance can be attributed to the actual numbers involved in the 1974] BALDRIDGE: SHARK ArrACK 41 absence of information on the distribution of companions about individuals under non-attack conditions. The above data may well represent nothing more than the "normal" distribution of individuals making up a group of people in the water. Of course, the more important, and as yet unanswered, question is related to the selection by a shark of one particular individual as its victim from among a seemingly uniform group of people. Unless we accept the selection as being a completely random one, then there must be something different about the chosen victim either in those things which are nearby and in some way associated with him or in the olfactory, visual, or auditory profiles which he himself presents to the attacking shark.

FISH

Fish, either in unusual numbers or behaving in an abnormal or otherwise noteworthy fashion, were reported present in 213 cases and presumed with reasonable certainty to have been present in 39 additional cases. Even though information was not available on this point in 68 % of the applicable cases on file, the total of 252 cases associated with fishing or fish clearly supports the very reasonable idea that sharks are attracted to fish in a disturbed state, either wounded or otherwise distressed. A number of recent studies have shown that recorded sounds of struggling fish serve as very effective shark attractants and excitants. It certainly should be no surprise that sharks would find very appealing the olfactory stimulation provided by large numbers of fish or even just one wounded, bleeding fish.

LARGE FLOATING OBJECTS

The possible influence of large floating objects at the attack site is suggested by the reported or presumed presences of boats in 281 cases. Surfboards or other similar floats (air mattresses, etc.) were noted in 66 cases. On the other hand, keep in mind that most recorded shark attacks have happened in waters which are well suited to recreational activities. Accordingly, it is in such waters that boats, surfboards, floats, etc., would be expected to be rather commonplace. Still, the possibility cannot be ignored that the very presence of such floating objects might be attractive to curious sharks, by virtue of either their sizes, shapes, colors, or perhaps even the vibrations put into the water as a result of their movements. Surfboard skegs, for example, often produce audible vibrations as the surfer moves over the water's surface. Serving as further evidence for the possible existence of an association between shark attack incidence and the presence at the attack site of large floating objects are the approximately 168 records in the SAF which deal with strikes by sharks not against humans, but against boats. 42 CONTRffiUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No.2

BLOOD, HUMAN

Human blood has long been considered a primary motivator for shark attack, yet in only 19 cases were the victims thought to have been appreciably bleeding into the water prior to the attack. Blood from persons other than the victims was considered likely in 12 other cases. Dr. Albert Tester (1963) at the University of Hawaii found fresh human blood to be strongly attractive to sharks, whereas aged hemolyzed blood produced an alarm or avoidance response. Interestingly, human sweat produced variable repellent effects. In other studies (Baldridge, 1966), fresh mammalian blood released from sur­ gically wounded rats swimming in shark pens, produced no discemable at­ tractant or excitant effects upon captive juvenile and adult lemon, dusky, and tiger sharks. Rat blood, presumably, should be no more foreign or less attrac­ tive to sharks than human blood. As an aside, let me point out that rat peritoneal fluid (the small amount of liquid which bathes the viscera within the body cavity) was found to be extremely attractive to the same sharks. It is difficult to accept the concept that human blood is highly attractive and exciting to sharks in general when so many shark attack victims have been struck a single blow and then left without further assault even though they were then bleeding profusely from massive wounds. Granted, for every two persons bitten only once, there was one person who received multiple bites. Yet, there was not a single case in the SAF (other than those associated with group feeding and with air-sea disasters or other situations where bodies were in the water for extended periods of time) where blood from an attack victim was known to have subsequently attracted a second shark and excited it to the point of also attacking the man. Yet, a number of victims remained in the water for periods of time where attack by a second shark would not have been surprising. All this tends to support my contention that many shark attacks on man are not motivated by hunger. In any event, observations in the laboratory and those made in association with shark attacks largely do not support the generally accepted belief that, as was stated in a recent U. S. Air Force training film on survival at sea, a "single molecule" of human blood will drive a shark into a frenzy.

BLOOD, FISH

There is no doubt that sharks are strongly attracted to fish blood and what­ ever other fluids or chemicals that might be given off by wounded or frightened fish. That is as it should be, for such olfactory cues are directly related to normal feeding habits of sharks. Tester (1963) showed that grey and blacktip sharks responded strongly to water siphoned from a container in which living 1974] BALDRIDGE: SHARK ATTACK 43 fish had been stressed or excited by threatening or probing them with a stick. Sharks will strike essentially anything that has been treated with fish "juice." In some of his studies on olfactory responses, P. W. Gilbert (unpublished data) observed that lemon sharks would repeatedly strike a cellulose sponge which had been dipped in fish body fluids. Dead rats were quickly taken after their fur had been wetted with fish blood, although the rodents' own fresh blood elicited no such response (Baldridge, 1966). Such observations, especially when considered along with the 255 shark attack cases associated with the presence of wounded fish and/ or fish blood and body juices, strongly evidence the lack of wisdom shown by those spearfishermen who keep captured fish close at hand on a stringer, in a nearby float, or even tied to their belts. Spearfishing in a limited area for an extended period of time might be equally unwise, for the appetizing aromas in the water may be very difficult for a cruising shark to ignore.

HUMAN WASTE

Human waste has been questioned as possibly attractive to sharks, but in the SAF there were only 11 cases where it was reported to have been present. In 59 other cases, it was likely there because of the length of time the victim was in the water prior to the attack. However, the data are not sufficiently definitive to permit the drawing of any meaningful conclusions. It can be realistically assumed, considering the numbers of people who populate the waters at resorts, that more than just a little human urine finds its way into the sea in the immediate vicinity of beaches. In laboratory studies conducted by Tester (1963), one species of shark (blacktip) showed only an awareness of the presence of human urine in the water, being neither attracted nor repelled by it. The responses of other species have not been investigated. There is, of course, the possibility that awareness may be in some cases but the first link in a chain of responses that terminates in attack. Once it takes notice of the presence of a potential victim, a shark could easily be led by curiosity to more exciting stimuli to which it responds with aggressive behavior.

GARBAGE OR REFUSE

It is well known that sharks at times follow ships in order to feed on dis­ carded foodstuff. They also abound along shorelines adjacent to dumping sites. It is therefore reasonable to probe the presence of such material in association with known attacks. Information was available for judgement in 463 cases. Garbage or refuse was reported in the immediate vicinity of 34 (7.3 %) attacks and was presumed present with reasonable certainty in 74 44 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

(16%) others. The noted or strongly suspected presence of such material in almost one-fourth of the cases considered clearly supports the very reason­ able contention that sharks are attracted and/ or excited by such waste matter and that to swim in nearby waters is possibly to court disaster.

OTHER CHEMICAL STIMULI

Considering the well developed means by which a shark can detect very low concentrations of waterborne substances, it seems reasonable that such an animal might respond to essentially any form of chemical discord occurring in its relatively stable environment. Even a chemical that might be strongly repellent at high concentrations could conceivably at very low levels elicit iIi a shark response of awareness, curiosity, and an urge to investigate the source of the strange substance.

ACTIVITIES OF VICTIMS AND OTHERS IMMEDIATELY PRIOR TO ATTACK

Thinking in terms of aggressive behavior by a shark should by no means be limited to attempts to satisfy a hunger drive. Predation for the purpose of seeking food is quite separate from aggression in response to a threat, either real or imagined, and evidence suggests that shark attack on man probably results from both of these drives as well as perhaps other as yet unrecognized factors. That a shark generally uses its teeth as its weapon of choice in striking out at the objects of its aggression may be more a matter of its limited offensive capability than any indication that the fish is always trying to devour its victim. The "man-eating" shark in some cases could be merely striking out with the only weapon at its command in an effort to either drive the victim away from where he is or to make him stop whatever it is that he is doing.

FISHING

It should be no surprise to find a strong relationship between fishing activ­ ities and the occurrence of shark attack, for fishing would logically have a high potential for bringing man and shark into close contact, and under con­ ditions where the shark would likely be excited or stimulated to the point of aggression by the presence of fish in distress. It is even thinkable that attack against a man under such conditions might be an accident, or, if deliberate, might be an attempt on the part of the shark to drive away what might appear to be a competitor for available food rather than any attempt to actually feed upon the man. When spearfishermen are thoughtless enough to attach cap­ tured fish to their belts and are then subsequently injured by a shark, it often 1974] BALDRIDGE: SHARK ArrACK 45 appears as if the strike was directed at the tethered fish and that contact with the man was incidental. In 95 cases in the SAF it was considered that fish were being hooked by the victim at the time of or immediately prior to the attack, and by other persons near the victim in 80 instances. Netting of fish by the victim was either noted or presumed with reasonable certainty in 62 attacks and by other persons nearby in 44 cases. Spearfishing has a particularly high potential for association with shark attack because it, unlike polefishing and/ or netting, more often puts the fisherman directly into the water. In 191 cases, the victims were considered to have been directly involved in spearfishing, with 107 cases noting spearfishing by other persons nearby. Spearfishermen were more likely to have been among the 140 victims (39 others presumed with reasonable certainty) who were carrying or holding captured fish at the time of the attack. In 97 cases, others in the vicinity of the victims were thought to have had captured fish in their possession. Other miscellaneous forms of fishing were engaged in by the victims in 52 attacks and by others nearby in 31 more. Of the 941 case histories where judgment could be made on the basis of available information, about one out of every five (19%) victims was thought to have had captured fish in his possession. Now, consider the expected very low ratio of saltwater fishermen of all types to the great numbers of people who utilize the sea for other purposes, and then go a step further and visualize the small fraction of those fishermen who would be holding or carrying cap­ tured fish at any point in time. The fact that such persons make up almost a fifth of all known victims of shark attack points strongly to the foolishness of keeping a bleeding, struggling, or dead fish anywhere near one's person in waters where the presence of a shark is even a remote possibility. The same logic applies to the sport of spearfishing with its high level of incidence among the activities of shark attack victims. Again, spearfishermen made up almost one-fifth (18.4%) of the objects of 1037 attacks where sufficient information was available to judge the activities of the victims at the times of the attacks. Spearfishing is a relatively recent development in water sports and has become immensely popular only in the last few years. For spearfishermen to constitute already nearly 20 percent of all known victims of shark attack would seem to indicate a rather high potential for disaster among devotees of this sport as compared to other forms of recreational use of the sea. When the computer was asked to readout only those attacks which have occurred subsequent to June 1958 (date of establishment of the Shark Research Panel and relatively early in the advent of spearfishing popularity), we found that victims were spearfishing in over 27 percent (122 attacks) of the 449 cases where judgement of activity was possible, i.e. better than one out of every four shark attacks recorded since 1958 have been directed towards spearfishermen. 46 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

PROVOCATIVE ACTS

In a sense, fishing, and especially spearfishing, could be taken as a pro­ vocative act on the part of some attack victims. Most (86%) shark attacks were not associated with any known provocative acts on the part of the victims. Nevertheless, there were records of sharks responding aggressively to being seized in 23 cases, kicked or struck 17 times, hooked 42 times, netted 18 times, and speared 31 times by 'victims' of what could be better described as subsequent acts of elicited resentment on the part of the sharks. There were also the rare (12) cases where persons were struck by sharks during the rescue of another person, or (4) in the conduct of research involving captive sharks, or (6 cases) where an overly eager shark in an aquarium bit the hand that fed it.

SPECIALIZED GEAR

Closely associated with the activities of victims immediately prior to the attack would be the presence of specialized gear, either worn or carried. Only 12 victims were known to have been wearing a life jacket or some other form of individual flotation gear. Keep in mind that the cases analyzed here did not include air/ sea disasters, and it is with these situations where the wearing of flotation gear would be more likely. Weapons of some sort were carried in 210 cases. The actual use of these weapons and resulting benefits, if any, will be discussed in detail later. That a large percentage of these weapons were spears or spearguns is consistent with the wearing of swimfins by 177 victims and a faceplate, with or without a snorkel, in 221 cases. SCUBA gear was being used by 20 victims at the times they were attacked. There seems to be no way to evaluate the effect of the presence of such gear about the person of victims except to point out that the numbers involved and the natures of the weapons again seem to suggest a comparatively higher hazard potential for divers in general and spearfishermen in particular.

GENERAL ACTIVITIES OF OTHERS THAN VICTIMS

Nothing particularly striking is presented by analysis of the general activities of other people in the water in the vicinity of the victims. In Table 5, the numbers in parentheses refer to those cases where the activity was not specifi­ cally reported in the files but was presumed to have been going on with reasonable certainty. It is interesting to note that those activities by nearby persons which are characterized by violent movements, erratic splashing, etc. did not appear to be associated with an inordinate percentage of shark attacks. Observations on activities of bathers made for this study by the San Diego Lifeguard Service at La Jolla, California and by my associate Ed Broedel at 1974] BALDRIDGE: SHARK ArrACK 47

TABLE 5. General activities of others in or about the water near the victim at or immediately prior to the time of the attack.

Percentage of total No. cases Myrtle in SAP SAP La Jolla Beach Normal bathing/ swimming 240(29) 34 22 29 Wading 98(59) 20 22 29 Diving/ underwater activities 172(10) 23 22 19 Splashing/horseplay, etc. 33(23) 7 14 10 Surfing/with or without board 54(10) 8 5 10 Thrashing/ flailing, etc. 18(13) 4 8 5 Unusually loud voices/ noises 19(8) 3 8 0

Subtotals 634(152)

Total cases 786

Myrtle Beach, South Carolina are also presented in Table 5. It is clear from comparison with these control data that people in the water near shark attack victims generally were doing things not appreciably different from those done by any other group of people using the sea for recreational purposes.

PRE-ATTACK SHARK BEHAVIOR

Information to at least some degree was available on the general pre-attack behavior of the shark in 530 cases. Probably the most significant finding was that in 63 % (333 attacks) of the cases where judgement was possible, the shark was not seen at all prior to contact with the victim. In interpreting the data that follow it should be kept in mind that the observed behavior may only represent that of the sharks which were actually sighted and may not necessarily reflect the actions of sharks not seen prior to the attack. And since the unseen sharks constitute the majority of attackers, we are in the uncom­ fortable position of judging the behavior of a group of animals on the basis of observations made on a minority segment of that group, with no assurance at all that that minority represents a good statistical sample of the group as a whole. Furthermore, of the 197 cases where the shark was seen immediately prior to the attack, the sighting was insufficient in 36 instances to permit any evaluation at all of what the fish was doing. Of the remaining 161 cases where attackers were observed relatively clearly, the sharks were seen making straightway approaches to the victims in 66 cases (41 % of those observed), passing close to other people in the water before striking the victims in 20 of 48 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vo\. 1, No. 2 those attacks. In only 34 cases (21 % of observed sharks) were the sharks seen to circle the victims prior to attacking. Nothing unusual was noted about the sharks' swimming behavior in 26 cases (16% of sighted sharks), while in 8 others (5%) the sharks were said to have been moving erratically, and in 12 cases (7 % ) the attackers were seen following close behind the victims immediately prior to striking them. Attacks in 15 cases (9%) followed soon after the sighting of a shark between the victim and some barrier or obstacle such as the beach, a reef, boat, etc. To generally sum up pre-attack shark behavior, it is more likely than not that the shark will not be seen at all prior to the attack. When it is, there is about an even chance that the shark will be already bearing down directly upon its victim, perhaps by now so intent upon its objective that it ignores other persons nearby, some of whom may be passed at very close range. Only very rarely will a shark be seen circling its victim. A shark, threatened by having its escape to the open sea cut off by the presence of a human in the water, is understandably a potentially aggressive, dangerous animal.

PRE-ATTACK ACTIVITIES OF VICTIMS

There was considerable information in the SAF on the general activities of victims at the time they were struck by sharks. At least some level of evaluation was possible in 961 cases, with 23 % of them involving victims who were wading or sitting in very shallow water, 44% engaged in various surface activities, 25 % associated with subsurface or diving activities, and 7% where the victims were entering or leaving the water by means other than wading. A detailed breakdown of victims' activities is presented in Table 6. Specific activities are listed under each of the general categories along with numbers of cases cited and the percentages that these cases represented of the totals under each general heading. For example, 49 victims were standing still on the bottom when they were attacked. These cases represented 22% of the 222 cases in the SAF involving victims engaged in shallow water activities. So as to compare what the victims were doing with control observations at ordinary bathing beaches, the computer was limited to the 425 cases involving attacks in beach-type situations. As expected, the emphasis then shifted away from diving towards wading and swimming: i.e. shallow water activities, 33 %; surface activities, 54%; subsurface activities, 11 %; and water entry or exit, 2 %. The primary significance that can be ascribed to the data presented in Table 6 is that shark attack victims apparently are selected from persons engaged in essentially every general type of activity associated with recreational use of the sea. Beyond that, there are only a few hints and faint suggestions in the data. 1974] BALDRIDGE: SHARK ATTACK 49

TABLE 6. Activity of victim at time of attack.

SAF Myrtle La Jolla Beach, Beach, Total Beaches, etc. S.C. California No. % No. % (No.) % (No.) % Shallow water activities 222 141 (159) (6216) Victim in shallow water, no specifics on activity 45 20 38 27 22 34 Wading 47 21 25 18 22 17 Erratic splashing I etc. 13 6 8 6 2 23 Standing still on bottom 49 22 33 23 19 11 Sitting on bottom 4 2 2 10 2 Other shallow water activities such as bobbing up and down in response to wave action 64 29 35 25 25 13

Surface activity 426 229 (169) (3902) Victim at surface, no specifics on activity 39 9 24 10 8 32 Swimming 234 55 128 56 27 15 Floating, little or no motion (inel. use of flotation gear) 20 5 8 3 24 5 Treading water (including use of flotation gear) 40 9 16 7 7 7 Snorkelling (including use of mask and fins w / wo snorkel) 17 4 6 3 4 Riding surfboard 21 5 21 9 0 24 Aboard float, raft, tube, etc. 12 3 10 4 28 5 Body surfingl planing etc. 8 2 6 3 5 7 Other surface activities 35 8 10 4

Subsurface activity 244 45 (40) (539) Subsurface or diving activity, no details 28 11 6 13 42 41 SCUBA diving 20 8 3 7 0 30 Free diving (no gear) or subsurface swimming 21 9 3 7 42 10 Free diving with mask and/ or fins, w / wo snorkel 106 43 28 62 17 19 Pearl diving 35 14 0 0 50 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2

TABLE 6. (Continued.)

SAP Myrtle La Jolla Beach, Beach, Total Beaches, etc. S.C. California No. % No. % (No.) % (No.) % Hard hat diving 7 3 0 0 Other diving/ subsurface activities 27 11 5 11

Water entry or exit other than wading 69 10 Entering water (jumping, diving, falling, etc.) 58 84 9 90 Leaving water (ladder, side of boat, etc.) 8 12 0 0 Entry or exit other than above 3 4 1 10

For shallow water activities, the victims seem to have been doing about the same things that ordinary beach-goers do, at least those at LaJolla and Myrtle Beach. Activities involving movement (wading, splashing, horseplay, etc.) seem to be no more prominent among shark attack victims than shallow water users in general. Surface activities would by their natures generally move the individuals concerned further out from the beach away from the waders and would be expected to be quite varied in form, depending greatly upon the condition of the sea surface and to some extent upon the rate of falloff of the bottom. Evidently, the waters off Myrtle Beach, South Carolina are more conducive to swimming than to surfboarding and diving, while at LaJolla, California the reverse appears true. In spite of the wide variety of possible surface activ­ ities, more shark attack victims of this category were engaged in swimming than all other forms of surface activities combined. There is in this observation the strong suggestion that movements by a swimmer may very well play a role in his being selected for attack by a shark in waters deep enough to be con­ ducive to surface activities such as are listed in Table 6. The striking thing about victims involved in subsurface activities near beaches is the heavy predominance of free divers (with mask and/ or fins, with or without a snorkel); 62 % as compared to 17 % and 19 % observed at Myrtle Beach and La Jolla, respectively. Aside from the fact that diving in general takes the participant even further away from the shore than swimming, free diving of this type is very closely associated with spearfishing. And there is little that can be denied about the high incidence of spearfishermen among shark attack victims of recent years. 1974] BALDRIDGE: SHARK ATTACK 51

TABLE 7. Headings ot people in the water.

Attack victims Control bathers Total Beach Myrtle SAF activities La Jolla Beach Facing shoreward 45% 41 % 19% 23 % Facing seaward 38 41 46 53 Facing parallel to shore 9 9 9 23 Random activities/ headings 8 9 26 1

Total number observed 53 34 3933 182

Occasionally, a victim is struck immediately upon entering the water in some very rapid manner; i.e. by jumping, falling, diving, etc. As seen in Table 6, far more people are attacked by sharks when entering the water by means other than wading than when leaving the water in some way other than walking out of it. A person's sudden entry into the water could be taken by a nearby shark as some form of threat to which it quickly responds with violently aggres­ sive behavior totally unrelated to feeding.

LENGTH OF TIME VICTIM WAS IN WATER PRIOR TO ATTACK

There appears to be no discernible relationship between the likelihood of shark attack and the length of time a person has been in the water. Where information was available, almost equal numbers of victims were attacked either immediately upon entering the water (52 cases), within minutes of entry (65), or less than one hour after getting into the water (64). Presumably because it is not usual for people to stay in the water continuously for very long periods of time, the known number of victims struck after being in the water longer than one hour was a low 31. Thinking only in terms of beach­ type activities, these numbers became: 12, struck immediately; 20, within minutes; 26, less than one hour; and 6, after more than one hour in the water. A total of 16 victims (5 at beaches) had been in the water, had exited, and then were struck within 15 minutes after re-entry into the water. Nine others (2 at beaches) were struck after having returned to the water longer than 15 minutes.

HEADINGS OF VICTIMS IN WATER

According to our observations at Myrtle Beach and La Jolla (Table 7), by far the majority of people in the water at beaches face towards the open sea. 52 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vo!. 1, No.2

In spite of this, a slightly larger percentage (45%) of attack victims were struck facing the shore as opposed to those (38 % ) facing the sea. As will be pointed out later, there was no demonstrated strong preference for striking the victim from behind as opposed to hitting him face on. The explanation more likely lies with the supposed higher chance of a cruising shark being seen by a person facing the open sea with enough time to get out of the water and thereby remove himself as a potential attack victim.

THE ATTACK

In 49 cases, the primary object of attack appeared to have been something either worn or carried by the victim rather than his person, i.e. fish tied to his waist or a shiny or brightly colored piece of equipment or clothing, etc. But in 990 cases, it was concluded that the attack was directed primarily at the body of the victim. In 32 of these cases, not much was known about the circumstances, but 868 (88%) of them involved victims who were in the water with the shark and who were not thought to have been using any kind of flotation gear. A few (51) victims were not in the water, but had their encounters aboard boats, on docks or beaches, and in laboratories. Only 12 victims were using flotation gear at the time they were attacked, but keep in mind that air and sea disasters, where use of such gear would have been likely, were not involved in this tabulation. Some (36) people were struck when only partially immersed in the water, such as while clinging to the side of a boat or some fixed structure.

NUMBER OF SHARKS DIRECTLY INVOLVED

About 94% of the attacks were considered to have been the works of solitary sharks, acting alone. In 108 cases, the attackers were not sighted by anyone, and so nothing could be concluded concerning the number of sharks involved, but there was sufficient information available on this point in 812 other cases. A solitary shark was actually sighted in association with the attack and no others were seen in the immediate areas of 691 (85 % ) of these cases. Solitary sharks were believed responsible for 60 additional attacks, even though other sharks were seen in the immediate area. In 14 other cases, physical evidence strongly indicated that solitary sharks had been responsible. Multiple numbers of sharks were sighted in 26 cases and were considered directly involved in those attacks, while physical evidence suggested multiple shark activity in 6 others. So called "schools of sharks" were cited as respon­ sible for 15 attacks. Thus, more sharks than just one were considered directly involved in only about six percent of the cases on record in the SAP. 1974] BALDRIDGE: SHARK ArrACK 53

DIRECTION OF ORIGINAL STRIKE RECEIVED BY VICTIM

In 296 cases, it was possible to determine with reasonable certainty the direction from which the original strike was delivered; from in front of the victim in 97 (33 %) cases, from behind in 93 (31 % ), from the side in 51 (17 % ), from below in 46 (16 % ), and from above in only 9 (3 %) cases. Evidently, a shark doesn't seem to be very hesitant about meeting its victim head on. Remember that almost two-thirds of attacking sharks were not seen at all before making first contact with their victims. That there would be a greater chance of seeing the shark immediately prior to its initial strike if it approached the victim from the front was evidenced by information in 82 instances of frontal approach where the sharks were seen prior to contacting the victims about 78 % of the time.

NA TURE OF INITIAL STRIKE

About three-quarters of initial contacts were in the form of sudden, violent interactions between the sharks and their victims (409 cases or 74% of the 554 attacks where information was available). In the remaining attacks, there was a minimum of turmoil with the victims very often being initially unaware that anything of significance had happened.

AGGRESSIVE CLOSE PASSES WITHOUT CONTACT

In 119 cases (Table 8), it was concluded that non-contact close passes did occur. Furthermore, in 60 cases (50% of cases where passes occurred) these passes were thought to have been very few in number, i.e. one or two. It could be said with certainty only in 30 cases that no more than one close pass occurred and in another 10 that the shark only twice passed the victim very close at hand.

NUMBER OF STRIKES BY SHARK

Information on actual numbers of contacts or strikes when they did occur (Table 8) was available with reasonable certainty in 655 cases. In by far the majority (82%) of these, there were only one or two strikes. Of the 536 cases where no more than two strikes were suspected, 449 were definitely thought to have involved only a single contact between man and shark, with two contacts concluded with certainty in 85 others. Repeated strikes, including what might be called frenzied behavior, was thought to have occurred in only 119 cases; 3-5 strikes in 82 attacks, more than 5 in 29 cases, and strikes too numerous to specify in 8 others. 54 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No.2

TABLE 8. Numbers of close passes by shark without contacting victim, numbers of strikes, and numbers of bites or similar wounds received by victim.

Close passes Strikes Bites Cases % Cases % Cases % None reported 277 70 54 8 124 16 Few (lor 2) 60 15 536 76 497 62 1 reported (30) (449 ) (413) 2 reported (10) (85 ) (53)

Several (3 to 5) 34 9 82 12 120 15 3 reported (4) (27) (23) 4 reported (0) (4) (6) 5 reported (0) (0) (0) Many (over 5) 20 5 29 4 45 6 Too numerous to specify 5 8 13 2

Total cases 396 709 799

NUMBER OF BITES OR WOUND-GROUPS

The pattern is very similar when wounds on the victim are considered; understandable since contact usually results in the victim being bitten or slashed. In Table 8, the term bite includes not only that action resulting in a double set of tooth marks on the victim but is taken also to mean any type of wound or set of wounds very likely produced by a single application of the shark's teeth to the victim. Of the 675 cases where judgement regarding numbers of bites received was possible, over 74% (497) of the victims were thought to have received only one or two discrete bites; 413 considered with certainty to have incurred only a single bite, with two bites received by 53 others. Several (3-5) wounds or associated groups of wounds were received by 120 known victims while gross multiple wounding occurred relatively rarely, i.e. in only 58 cases including 13 where the wounds were too numerous to specify.

ATTACKS AS SINGLE-EVENT ENCOUNTERS

It is clearly indicated by Table 8 that the majority of shark attacks on humans are single event encounters; that is, one strike resulting in one wound or set of wounds and possibly involving one or two close passes without the 1974] BALDRIDGE: SHARK ATTACK 55 shark making contact. This is not at all consistent with any concerted effort on the part of the shark to devour its victim. It is, however, very much in line with my contention that many or perhaps even most shark attacks on man are not motivated by hunger. Only about one-fifth of all victims or about one-fourth of those actually bitten received the numbers of wounds that would at all indicate a determined effort on the part of the shark to "eat the victim alive." It could be reasonably argued that hunger motivated the initial strike, but that either the taste or the unusual tactile characteristics of the human victim caused the shark to quickly consider this unfamiliar object unsuitable as food. Considering the assortment of things routinely found in the stomachs of coastal sharks in particular, it wouldn't appear as if they are really that particular about what they eat. On the other hand, there were of course those cases where there was little if any doubt that the sharks intended to devour the human victims, either partially or totally.

SHARK BEHAVIOR DURING ATTACK

Where the shark was observed making more than one contact with its victim, about two-thirds (61 % ) of the time (114 cases) its behavior would best be described as the making of controlled and deliberate strikes at the person. Only in 49 cases (26%) , would the term "frenzied behavior" seem to apply. In 24 attacks, the shark bit the victim, released its initial hold, and then quickly bit the person again in a manner that would be difficult to describe as distinct, separate strikes in which the shark clearly broke contact with the victim between strikes.

HIGH MORTALITY RATE OF VICTIMS OF FRENZIED SHARKS

As would be expected, the victims of attacks described as frenzied do not fare well. Wounds are often massive and extensive, with 63 % of the 49 recorded cases resulting in the death of the victim as compared to 35% for all victims. Even in these cases, it was very rare that the victim's body, or at least part of it, was reported as not recovered; only six cases out of 49 or 12% as compared to 4.2 % for victims in general.

USE OF WEAPONS AGAINST ATTACKlNG SHARKS

There were case histories where mention was made of a weapon's use without notation as to any effect of significance, one way or the other, upon the behavior of the shark. Where results of usage were reported, they were categorized as either no significant effect at all upon the shark's behavior, significant effects which were deemed desirable from the victim's point of 58 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vo!. I , No. 2

TABLE to. Diversionary actions taken by victims and other persons including rescuers and the reported resulting effects upon shark behavior.

Reported to Reported to Reported to have had have had No effect of have had significant significant Number significance insignificant desirable undesirable Action Taken by of cases reported effect effect effect Fended off shark with Victim 134(19) 33122% 65/ 42% 33/22% 22/14% hand, shark billy, etc. Others 22(2) 3/13% 8/33% 13/54% 0/ 0% Struck at shark Victim 140(20) 36/ 23 % 68/43% 50/31% 6/ 4% Others 44(2) 1/ 2% 22/ 48% 23/50% 0/ 0% Kicked at shark Victim 60(9) 21/30% 22/320/0 22/32% 4/ 6% Others 10(0) 2/ 20% 3/30% 5/ 50% 0/ 0% Poked at shark Victim 43(4) 4/ 9% 22/47% 17/36% 4/ 9% Others 20(4) 114% 13/54% 9/38% 114% Probed eyes of shark Victim 20(3) 2/ 19% 4/ 17% 17/74% 0/ 0% Others 7(3) 0/ 0% 7/70% 3/ 30% 0/ 0% Probed gills of shark Victim 4(2) 0/ 0% 0/0% 5/ 83 0/0 1117% Others 3(2) 0/ 0% 1120% 4/ 80% 0/ 0% Path of shark blocked Victim 0(2) 0/ 0% 0/ 0% 2/100% 0/ 0% with boat, etc. Others 16(2) 4/ 22% 7/39% 7/ 39% 0/ 0% Aggressive movements Victim 73(6) 7/ 9% 37/47% 31139% 4/ 5% towards shark Others 43(2) 6/13% 19/42% 19/42% 1120/0 the action taken appeared to have made matters worse. Actions taken by both the victim and others nearby were considered. In Table 10, the number of cases where the action was presumed taken with reasonable certainty is enclosed in parentheses immediately following the number of cases where that particular diversionary action was specifically mentioned in the file. The effects are expressed as fractions, the numerator being the number of cases where that particular effect was obtained and the denominator being the percentage of time that effect was produced among all those cases where the indicated diversionary action was taken. For example, Table 10 shows that kicking at the attacking shark was employed as a counter­ measure by 60 victims with another 9 very likely having done so. In 21 cases (30% of the total of 69), there was nothing in the case histories concerning any significant effect of the kicking. In another 22 cases (32% of the total), the kicking was reported to have had no real effect upon the shark or the course of the attack. For 22 victims (32 % ), kicking in some way or another produced the desired effect of lessening the intensity of the attack, perhaps even driving the shark away. But four victims (6%) found that kicking only 1974) BALDRIDGE: SHARK ArrACK 59

compounded their problems, perhaps by adding to the shark's fury or by having their leg or foot bitten by the shark during attempts to kick it. Table lOis not very encouraging at first glance, for it implies, with the notable exceptions of probing the eyes and gills of the shark, that no matter what is done in terms of fighting back more often than not there will be little or no effect upon the shark. Victims as a group employed the actions listed in the table in 539 instances, and the effects were unknown in 103 (19%) cases, insignificant in 218 (40 %) cases, classed as desirable in 177 cases (33 % ), and not to the benefit of the victims of 41 (8 %) attacks. Similar results were obtained when the actions taken by others near the victims were considered. Keep in mind, however, that statistics deal with numbers of people and not individuals, and that Table 10 does not consider the action of doing nothing. For some people, 177 to be specific, one or more of the actions listed worked for them 100% of the time, i.e. the one time when each of them was the victim of a shark attack. So, it would seem that attack victims should do something-anything-to fight back, for it just might be the right thing for the moment. Even though the numbers of cases were relatively small, it appeared particularly effective to probe the gills and eyes of the shark. For some victims, the things they tried in fighting back seemed to make things worse. However, where significant effect upon the shark's behavior was pro­ duced, the odds were better than 4-to-1 that it was a good rather than a bad one. Generally speaking, bad effects were in the nature of receiving severe bites or lacerations on the hand or arm used to fend off or strike at an attacking shark or on the leg used to kick at it. There was nevertheless an apparent general pattern of the attacker turning its attention only momentarily towards the appendage that was now kicking at it or hitting at it, giving it a retaliatory bite, and then abandoning the attack. Who's to say that the wounds received were not a reasonable price to pay for such an outcome.

OUTCOME OF ATTACK

In past considerations of the effects of shark attack upon the victims, much has been made of the percentage of cases resulting in fatalities. This is not a very meaningful statistic, since the lethal nature of wounds very often is not directly related to their severity. A relatively minor wound occasionally leads to death of the victim if dealt under circumstances where proper medical attention is not readily available. On the other hand, very severe wounds, including loss of limbs, often are survived when adequate first aid is rendered and definitive treatment is effected. Thus, older attacks and those in more remote regions of the world would be expected to involve higher levels of mortality. There are, of course, those few attacks reSUlting in wounds of such severity and nature that survival would have been impossible regardless of the availability of prompt medical attention. 62 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

TABLE 12. Photographs of significance held in case histories.

No. of cases in SAF Photos of Photos of good quality little value Site of attack 24 7 Attacker shark 21 4 Morgue photos 16 0 Scars on body of victim 14 3 Wounds, before or at time of initial treatment 25 4 Wounds, during recovery 17 Stock photo of victim 146 2 Misc. photos of significance 21 2 might well have been the intended victim except for chance circumstances including the ready availability of another, more accessible subject.

WOUND CHARACTERISTICS

Much about wound characteristics had to be concluded on the basis of written descriptions, for there were in the SAP surprisingly few photographs of significance for this purpose. Understandably, the more recent attacks were more likely to have been documented photographically, especially in terms of operating room and morgue photographs. Table 12 bears out the paucity of useful pictures of a variety of desired types.

RECOVERY OF TOOTH FRAGMENTS FROM WOUNDS. Wounds of victims were often x-rayed in the hopes of locating a fragment of tooth large enough and possessed of sufficient telltale characteristics to permit identification of the species of shark responsible for the injury. Only in 11 cases were identified fragments recovered. In 8 other attacks, tooth fragments were located in the wounds, but no mention was made in the case histories of their use in deter­ mining the species of the attackers.

BODY PARTS INJURED. Bodies of victims were reported as not recovered in 59 cases, and in about 31 other cases the wounds were so extreme that locali­ zation of damage was not possible-that is, where descriptive terms were used such as mangled, severely mauled, mutilated, etc. The number of cases where it was possible to specify damage to the victim varied with the part of the body in question, but on the average about 835 case histories held sufficient infor- 1974] BALDRIDGE: SHARK ATTACK 63

TABLE 13. Body parts injured. Percentage represents incidence of injury to body part in attacks where data were available-an average of 835 cases.

Cases prior to 1958 Cases after 1958 TotalSAF No. % No. % No. % Calf/knee 194 41 140 39 334 40 Thigh 167 37 104 29 271 33 Arms 111 24 82 23 193 23 Feet 73 16 73 20 146 18 Hands 58 12 68 19 126 15 Buttocks 56 12 32 8.8 88 10 Fingers/ toes 33 7.1 51 14 84 10 Abdomen/ stomach 40 8.6 23 6.3 63 7.6 Chest 29 6.2 17 4.7 46 5.5 Waist 20 4.3 10 2.8 30 3.6 Shoulder 17 3.6 15 4.1 32 3.8 Back 18 3.9 13 3.6 31 3.7 Genitals 9 1.9 9 2.5 18 2.2 Head 11 2.3 9 2.5 20 2.4

Total citations 836 646 1482 mation for delineation of body parts injured. The presentation in Table 13 was made in terms of attacks prior to 1958 (year of establishment of the Shark Research Panel), those that happened after 1958, and the total for the SAF as a whole. This breakdown was to see if there had been any significant changes in body regions struck by sharks that might have been attributable to simultaneous changes in such things as styles, colors and patterns of bathing costumes; the appearance upon the scene of wetsuits; the fact that a high percentage of later attacks have been against skindivers; or anything else that has changed since 1958 in the way in which man exposes himself to shark attack. It is clear from Table 13 that no such changes in injury patterns have occurred to any meaningful degree. The data in Table 13 convincingly point to appendages as the most often selected objects of attack. Occurrences of wounds to arms, hands, legs, and feet account for 78 % of those tabulated. It is not too surprising that the legs are most often damaged when it is remembered that 62 % of the attacks at beaches occurred in water waist deep or less. In addition to having been primary targets of shark attack, the hands, fingers , and arms were often injured in efforts to fight off the shark. Wounds to the head were only rarely reported. 66 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2

TABLE 15. Citations in the Shark Attack File against various species oj sharks as known attackers oj man.

Total DIVERS SAF All (Sub.)

ORDER HEXANCHIFORMES Family Hexanchidae-cow sharks Sevengill shark, no specifics 1 (1) ORDER HETERODONTIFORMES Family Heterodontidae-bullhead sharks Horn shark 1 1 ORDER SQUALIFORMES Family Orectolobidae--carpet sharks Carpet shark, no specifics 6 2(1) Nurse shark-Ginglymostoma cirratum 15 9(6) Wobbegong-Orectolobus barbatus 2 Wobbegong, no specifics 13 7(1) Family Odontaspididae-sand tigers Sand sharks, no specifics 6 3(1) Ragged-tooth shark (S. Africa)-Odontaspis taurus 3 1 (1) Ragged-tooth shark, no specifics 2 Grey nurse shark (Australia)-O. arenarius 2 2(2) Grey nurse shark, no specifics 17 8(2) O. prionodon 1 Family Alopiidae-thresher sharks Thresher shark, no specifics 1 Family Lamnidae-mackerel sharks White shark-Carcharodon carcharias 22 5(4) Great white shark 2 2(1) White pointer (Australia) 4 1 (1) Blue pointer (S. Africa) 1 White, no specifics 3 2(1) Mako shark Pacific mako--Isurus glaucus 1 (1) Blue pointer (Australian rnako) 11 2 Mako (Pacific Ocean), no specifics 3 2(1) Shortfin rnako--/. oxyrinchus 1974] BALDRIDGE: SHARK ArrACK 67

TABLE 15. (Continued.)

Total DIVERS SAF All (Sub.)

Mako (Atlantic Ocean), no specifics 2 2(1) Bonito shark (Pacific Ocean) 3 Salmon shark Mackerel shark 2 Family Carcharhinidae-requiem sharks Carcharhinid, no specifics 3 Grey sharks, no specifics 2 Whaler sharks, no specifics 11 4(1) Bull shark Bull shark-Carcharhinus leucas 8 2 Ganges River shark (India)-C. gangeticus van Rooyen's shark (S. Mrica)-C. vanrooyeni Zambezi shark (S. Africa)-C. zambezensis 4 1 (1) Bull shark (South Mrica) 2 Zambezi shark 3 Shovel-nosed grey shark Jaba (Panama) Blacktip shark, no specifics 8 7(1) Blackfin shark, no specifics Oceanic whitetip shark-C. longimanus Whitetip shark, no specifics 2 2(2) Inshore whitetip shark-C. albimarginatus Spinner shark-C. maculipinnis 2 Dusky shark-C. obscurus Cub shark 1 Tiger shark-Galeocerdo cuvieri 3 Tiger shark, no specifics 22 11(4) Leopard shark (Eastern United States) 2 Lemon shark-Negaprion brevirostris 3 Lemon shark, no specifics 3 Blue shark-Prionace glauca Blue shark, no specifics 9 1 (1) 70 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2 two estimates of 2.2 feet. The average length of shark involved in these attacks was about 8.8 feet, so the average difference of 2.2 feet between the two estimates represented about 25 % of the total estimated lengths of the sharks. The picture was not greatly improved in the 7 cases where three estimates of shark length were reported. Here the mean length of shark was 12.2 feet with an average standard deviation of +- 2.4 feet, representing about 20% of shark length. Four estimates were available in 7 cases with a mean shark length of 9.0 feet and an average standard deviation in the estimates of +- 1.3 feet, or about 14% of estimated shark length. There were two cases where five estimates of shark length were available; one where reports of length averaged 12.4 feet with a standard deviation of +- 1.7 feet, i.e. +- 14% of the mean estimated length and another averaging 9.4 feet (SD of 0.9 feet or 10% of mean).

EVALUATION OF ABILITIES TO ESTIMATE SHARK LENGTH. To get a feeling for the ability of people in general to accurately estimate the lengths of sharks and other submerged objects, a group of 17 college students visiting MML were asked to take part in an experiment. There were in our tanks a 72" bull shark, a 46" lemon shark, and, to try to remove emotionalism from the students' observations, I placed on the bottom of the tank two lengths (85.6" and 30.5") of plastic pipe. The students were asked to stand next to the 5-foot deep pool, facing away from the water. On my signal given as a shark swam by, I had them turn, look at the moving shark for 10 seconds, again face away, and record their estimates of total length without consulting with their neighbors. The same procedure was used for the immobile plastic pipes. Their mean estimates of the length of the 72" bull shark was 85.6" (range of 63 to 128 inches), with a standard deviation of 19.2" (22% of the mean estimate). The corresponding data for the 46" lemon shark were 45.5" (29-61"), SD = 8.6 (19 % ) . The plastic pipes: 85.6" actual length, 99.6" mean estimated length (range of 60-156"), SD = 26.8" (27%); 30.5", 36.7" (24-51"), 7.9" (22 % ) . The standard deviations resulting from spreads in values of the estimates made by the students at MML were of magnitudes very similar to those found with actual attackers. Note in particular the wide range of esti­ mates in each case and consider then the very poor likelihood of getting a good estimate of attacker length from a single observer. Except in the case of the small lemon shark, there was a marked tendency on the part of the students towards overestimation; 13 of 17 estimates of the 72" bull, 8 of 17 for the 46" lemon, 12 of 17 for the 85.6" pipe, and 12 of 17 for the 30.5" pipe. Four students judged too Iowan three of the four objects. Of the total of 68 obser­ vations on lengths that were made by the students, 45 or 66% were in the direction of overestimation. Therefore, even though the overall average esti­ mated length of attacker sharks in general (i.e. 6'10" or 2.1 meters) may be 1974] BALDRIDGE: SHARK ArrACK 71 a fairly accurate number statistically, one must take with a large grain of salt those estimates of shark length in individual cases which were based upon fleeting sightings by only one, or perhaps even more than one, person at the scene.

POST-ATTACK SHARK BEHAVIOR

The chance of seeing the shark after an attack is understandably much higher than seeing it before the initial strike. Almost two-thirds of attackers were not observed prior to striking their victims, while only 46 per cent (222 cases out of 482 where infonnation was available) were not seen at all after the final strike. In the 260 cases where the sharks were seen immediately after the attack, they remained nearby and thus presented a continuing threat in a majority (61 %) of the cases. Most of the time (98 out of 159 cases, i.e. 62%), the sharks that were observed remaining nearby were reported simply to have stayed in the immediate area. But in 24 (15 % ) of these cases, the attackers were seen following the victims and/ or rescuers towards shore. More often than that (37 cases, 23 %), the persistent sharks not only stayed nearby but remained affixed to their victims with their jaws holding fast even during rescue operations. In over two-thirds (70%) of these cases where the sharks held on, some form of force had to be used to cause these tenacious attackers to open their jaws and release their victims.

CAPTURE OF ATTACKER SHARK

Capture of a confirmed attacker by any means is a relatively rare thing. It would seem likely that all sorts of efforts would be made to catch the offender after a shark attack occurs in an area. Yet, in only 114 cases in the SAF was it specifically reported that serious efforts, such as intensive fishing operations, were made to capture the sharks responsible for the attacks. In only 30 such cases (i.e. 26%) could it be said with a high degree of certainty that the actual culprit was caught. These do not include the four cases where the sharks remained affixed to the victims and were captured during rescue operations. After 34 other attacks, sharks were caught which could possibly have been the attackers, but no assignment could be made with any reasonable degree of certainty. A number (93) of "attackers" were actually in captivity, and remained so, at the times of the "attacks," i.e. under such circumstances as research activities, sharks hooked on lines or caught in nets, etc.

HUMAN REMAINS IN SHARKS' STOMACHS

Other than actually bringing the shark ashore still holding onto its victim, the most damning evidence for pointing the finger of guilt at any particular 72 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No.2 captured shark is the finding of human remains in its stomach that can be positively identified as coming from the victim. Such proof has been reported in only 19 cases, thus conclusively implicating the captured sharks in either the original attacks or assaults upon the bodies of the victims at a later time. In four other cases, human remains were found in sharks captured after attacks, but these could not be identified as being from the known victims. The fact that human flesh has been found to remain in a relatively good state of preservation for extended periods of time in the stomachs of sharks indicates that there are things about the digestive systems of these animals about which we know very little.

DIVERS AS VICTIMS OF SHARK ATTACK

There is a generally accepted viewpoint, held especially by divers them­ selves, that totally submerged persons are less susceptible to shark attack than swimmers at the surface. There are at present 244 records of shark attacks on divers, over one-fifth of the entire Shark Attack File. Cases involving victims generally engaged in all forms of underwater activities were considered as diving incidents even if the victims were actually struck at the surface of the water. It is extremely interesting that only one of these case histories dealt with an attack upon a female diver. She was evidently attacked at the surface after having been diving for fish with another young girl in about 18 feet of water. Other than that, there has been no other recorded attack upon a female skindiver, SCUBA diver, free diver, or any other type of diver beneath the surface, i.e. at a depth greater than about five feet. On the other hand, there are in the SAP a total of 83 cases where attacks upon submerged male divers occurred at estimated depths up to approximately 250 feet. As stated earlier, the ratio of male-to-female victims in the SAP as a whole was found to be about 13.5-to-1 , dropping off to about 9.1-to-1 for attacks at beaches. For all forms of underwater activities, the ratio rose to an impressive 243 male diver-victims for only one female.

TRENDS WITH TIME

As would be expected in view of the relatively recent rise in popularity of all forms of sport diving, most (76.9% of 239 cases of known dates) reported attacks involving divers have occurred since 1950. They are increasing with time. During the 10-year period 1950-59, the 63 reported attacks upon divers represented 25.6% of 256 total known cases of all types for that period. There were 116 diver-cases in the similar period 1960-69, representing a rise to 29.5% of 393 total reports. It is expected that in the decade of the 1970's, attacks upon divers will average at least one-third of all reported cases. The 1974] BALDRIDGE: SHARK ArrACK 73

60.------.------,.------.------,,------,------,--.

50

40

30

20

10

\ NUIIlber of reported attacks on divers

1940 Year 1945 1950 1955 1960 1965 1970 Fig. 8. Relationships between time and number of reported shark attacks on divers. trend is evident in Fig. 8, where numbers of diver-cases were plotted against time along with the percentage of total cases represented by divers among all known attack victims. Note that in 1967, more than half of the 28 total reported attacks involved divers. The downward trend at the end of the plot is most likely due to curtailment of efforts to maintain the SAP due to lack of funding rather than to any real drop in incidence of attack. Far from connoting any degree of immunity to divers, data such as pre­ sented in Fig. 8 appear to indicate a greater danger of shark attack among divers than other potential victims. Surely, during the decades of the fifties and sixties, divers did not make up over one-fourth of all persons exposed to shark attack.

DISTANCE FROM SHORE

It is likely that the apparent increase in danger to divers is related to the greater distances from shore frequented by divers as compared to swimmers. Of the 115 cases where data were available, 76% of the diver-victims were more than 200 feet from shore, as compared to only 34% of victims engaged in beach-type activities. About 43 % of diver-victims were actually over one mile from shore. The pattern for submerged divers was essentially identical. 74 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

These data can be given only limited interpretation in the absence of infor­ mation on the distribution from shore of divers in general, including those not bothered by sharks.

DEPTHS AT WHICH ATTACKS OCCURRED ON DIVERS

Information was available on 121 cases, with 38 (31 %) of them happening at or within five feet of the surface of the water. The remainder were dis­ tributed in depth as follows: 6-10 feet, 23 cases (19%); 11-20 feet, 27 (22%);21-30 feet, 12 (10%); 31-40 feet, 8 (6.6%);41-50feet,2 (1.7%); 51-60 feet, 2 (1.7%); 61-70 feet, 3 (2.5%); 71-80 feet, 3 (2.5%); 81-100 feet, 0; 101-120 feet, 2 (1.7%); and at about 250 feet, 1 (0.8%). Interpre­ tation is limited by absence of comparable data on diver non-victims. Much like the relationship between distance from shore and incidence of attack in general, the above distribution of attacks as a function of depth is very likely due more to the depths at which greater numbers of divers operate rather than to any direct association to relative danger of shark attack. The maximum depths recorded for attacks upon free divers were from 41 to 50 feet, with assaults at greater depths involving SCUBA divers exclusively. Data on depths at which hard-hat divers have been attacked were not available.

PROVOCATION BY DIVERS

There are indications in the SAF that diver-victims may encourage shark attack by being somewhat more provocative than surface swimmers. Of 190 cases where sufficient information was available, 43 divers (23%) were con­ sidered as having overtly provoked the attack; a sizable increase over the 14 % provoked attacks for the SAP in general (19 % for submerged divers). Provocation may also have been inadvertent as well as direct, and the pro­ vocative act may well have been spearfishing. This sport is most usually engaged in either by SCUBA divers or free divers using mask, snorkel, and swim fins. These types of divers made up a combined total of 67 % of the diver-victims where sufficient data were available to ascertain the natures of their activities. Of 103 free divers where judgement was possible, 80% were engaged in spearfishing. It was possible to conclude in 72 cases that 51 % of the free-diver victims had captive fish in their possession at the times that they were attacked. SCUBA divers showed a lower incidence of spearfishing; 53 % of 19 cases, with almost all of them (50% of 18 cases) possessing cap­ tured fish. These data appeared to severely indict spearfishing as a provocative act leading possibly to shark attack. Logic supports this conclusion. How­ ever, it cannot be statistically validated in the absence of corresponding data on diver non-victims. 1974] BALDRIDGE: SHARK ArrACK 75

GENERAL ACTIVITIES OF DIVERS

The activity breakdown of 189 diver-victims where data were available was as follows: free divers using mask, snorkel, fins, etc., 106 cases (56%); pearl divers and shell divers, 35 (19 % ); SCUBA divers, 20 (11 % ); free divers using no equipment, 21 (11 %); and hard hat divers, 7 (3.7%). It is very probable that there have been numerous attacks upon pearl divers in the more remote parts of the world for which there are no records in the SAF. Only ten attacks on pearl divers have been reported in the last thirty years and none since 1959. Perhaps the reason for withdrawal of pearl divers from the ranks of shark attack victims was expressed in a New York Post article of 28 June 1950 which cast pearl fishing in the role of a dying industry. Attacks upon SCUBA divers understandably reached into greater depths than those involving free divers. This carries over into distance from shore also. Of the 14 SCUBA-victims where information was available, three were attacked 251-300 feet from shore with the remainder (79%) all out beyond one mile from land. Free divers using mask, snorkel, and usually swim fins made up the majority (57%) of 190 diver-victims whose activities at the time could be determined.

DIRECTION OF INITIAL STRIKE AGAINST DIVERS

For divers in general, the direction of the initial strike was not greatly dif­ ferent from that for shark attack victims in general: from in front of the victim, 41 % of the time as compared to 33 % for all victims in the SAF; from behind, 28 % versus 31 %; from the side, 17 versus 17; from below, 7 versus 16; and from above, 7 versus 3. The differences between below and above approaches for divers compared to swimmers is understandable. Even when considering only divers who were known to have been submerged at the time of the initial strike, the approach pattern was found to be about the same: frontal, 37%; behind, 31 %; side, 17%; below 10%; and above 6%.

CLOSE PASSES, STRIKES, AND BITES

A pattern of difference between divers, particularly those attacked when submerged, and victims in general began to develop when incidence of close passes, strikes, and bites were compared as seen in Table 17. When it was known that close passes were made by the sharks without con­ tacting the victims, multiple (i.e. more than 2) passes were reported with 58 % of the divers (an even higher 66% for submerged divers) as compared to 50% for all SAF victims. This may well be due to the increased ability of the diver, particularly when submerged, to keep the shark in view and observe his movements. 76 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [VO\. I , No. 2

TABLE 17. Comparisons between submerged divers, all divers, and all shark attack victims in terms of close passes by sharks without contacting victims, numbers of strikes made by attackers, and numbers of bites received by victims.

All cases where data were available Close passes Strikes Bites Sub. All Total Sub. All Total Sub. All Total divers divers SAP divers divers SAF divers divers SAP

None reported 52% 55% 70% 21% 14% 8% 32'}'o 29% 16% Few (1-2) 16 19 15 77 79 76 61 59 62 Several (3-5) 10 13 9 3 5 12 5 9 15 Many (over 5) 18 10 5 0 0.6 4 0 2 6 Too numerous to specify 3 3 0 0.6 1 1 2

Total number of cases considered 61 115 396 73 166 709 74 185 799

Only those cases where occurrences were reported

Few (1-2) 34% 42% 50% 97% 92% 82% 90% 84% 74% Several (3-5) 21 29 29 3 6 13 8 12 18 Many (over 5) 38 23 17 0 0.7 4 0 2 7 Too numerous to specify 7 6 4 0 0.7 2 2 2 Total number of cases considered 29 52 119 58 142 655 50 131 675

Being able to watch the shark and to take effective diversionary actions might be expected to produce highly desirable results for divers. Multiple strikes were reported against only 8 % of contacted divers (an even smaller 3 % for those submerged), much lower than the 18 % recorded for all victims. Equally impressive was the observation of 16% incidence of multiple bites on divers (10% submerged) compared to a total SAP level of 27 %.

BODY PARTS INJURED

Coppleson (1962) concluded that "most divers seem to have a charmed life as far as sharks are concerned .. ." Injuries were thought of as compara­ tively slight, leading to a lower mortality rate, than with other types of victims. The data in Table 18 do not support Coppleson's statement that injuries to the arms, hands, and upper parts of the bodies of submerged divers are twice as 1974] BALDRIDGE: SHARK ArrACK 77

TABLE 18. Body parts injured; submerged divers compared to all shark attack victims.

Incidence of injury No. of injuries Sub. divers TotalSAF Calf/knee 10 19% 40% Thigh 14 26 33 Arms 18 34 23 Feet 8 15 18 Hands 2 4 15 Buttocks 5 9 10 Fingers/ toes 9 17 10 Abdomen/ stomach 5 9 8 Chest 4 8 6 Waist 3 6 4 Shoulder 4 8 4 Back 3 6 4 Genitals 2 2 Head 3 6 2

Average cases considered 53 835 common as to the buttocks and lower limbs. Actually, in the 53 cases involving submerged divers where sufficient data were available, upper-body injury occurred less often that lower-body damage. The incidence of injury to upper regions (arms, hands, shoulder, chest, and head) was found to be only 84% of that for damage to lower parts of the body (buttocks, thigh, calf-knee, and feet). This is a sizable shift from the ratio of 50% observed for 835 cases of all types in the SAF. While the actual incidence of injury to the upper bodies of submerged divers was found to be about equal (0.58 versus 0.50 injuries per case considered) to that for SAF victims in general, damage to the lower regions occurred far less often (incidence of 0.70 compared to 1.00) among submerged divers. Thus, while shark attack on all types of victims lead primarily to lower body and leg injury, this was understandably far more true for the wader and swimmer in shallow water than for the submerged diver.

NATURE OF WOUNDS RECEIVED BY DIVERS

The data in Table 19 do appear to support Coppleson's contention that wounds received by submerged divers are generally less severe than those of victims in general. Divers were found more likely to receive wounds best described as lacerations, with no significant loss or displacement of flesh. 78 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vo\. I, No. 2

TABLE 19. Nature ot wounds; submerged divers compared to all shark attack victims.

% Victims receiving wound No. of cases Sub. Total Wound description considered divers SAF Severe lacerations/ displacement of tissue 46 63 78 Lacerations, no significant loss or displacement of tissue 38 87 71 Significant loss of tissue 46 35 55 Bone exposed 43 23 46 Bites, discontinuous tooth marks, etc. 35 54 42 Scrapes, abrasions 38 11 23 Appendage lost to shark 54 7 19 Appendage lost thru surgery 55 4 6.6 Body cavity opened 52 4 5.0 Trunk severed by shark 56 2 0.9 Body skeletonized 55 0 0.7 Swallowed whole or presumed so 56 2 0.6 Body not recovered in 4 cases

Compared to shark attack victims in general, significant tissue loss was only about 64 % as prevalent among submerged divers, and exposure of bone happened less than half as often with submerged victims. Actual bite marks were found slightly more often on divers, but scrapes and abrasions were far less frequent, probably due to the wearing of wetsuits by divers. Loss of a limb to a shark happened almost three times more often with victims in general than with submerged divers. There was no great difference between the incidence of inherently fatal wounds among divers compared to all types of victims.

MORTALITY RATE AMONG DIVER-VICTIMS

The lesser degrees to which submerged divers were injured was further evidenced by a low overall mortality rate of 12% as compared to 26% for divers in general and 35 % for victims of all types.

LENGTHS OF SHARKS ATTACKING DIVERS

Perhaps since divers generally operate much further out in the water than swimmers and bathers, there may be something different about the SIze, 1974] BALDRIDGE: SHARK ATTACK 79

TABLE 20. Shark behavior observed prior to attack; divers compared to all shark attack victims.

Divers Total Submerged TotalSAF Straightway approach to victim 300/0 35 0/0 41 0/0 Straightway, passed close to others in water 6 6 12 Circling victim 25 23 21 Swimming normally 28 19 16 Following victim closely 9 13 7 Swimming erratically 6 6 5 Shark in position between victim and barrier or obstacle such as beach, reef, boat, etc. 2 3 9 Number of cases where sharks were seen 53 cases 31 cases 161 cases Percentage of attackers not seen at all prior to contacting victim 43 0/0 41 0/0 63 0/0 species, and behavior of the sharks which they encounter. The data in 118 cases were used to calculate a median length of about 6.6 feet for attackers, ranging in size from two to 20 feet. This is not significantly different from the median of 6.8 feet found for lengths of attackers in general.

PRE-ATTACK SHARK BEHAVIOR

The better visibility and awareness of things underwater enjoyed by divers is reflected in the observation that, of 108 cases where data were available, only 43 % of the sharks were not seen at all prior to the initial encounter as compared to 63 % for all SAF cases. There was only a very slight further improvement among submerged divers; 41 % of 58 cases. There were only minor differences in observed pre-attack shark behavior between divers and other victims (Table 20). The shark was still most generally seen making a direct approach to the victim. With divers, there appeared to be a lower incidence of the shark passing close to others in its run on the victim. This may have been due simply to the higher densities of people at beaches as compared to diving environments. Divers made more numerous observations of normal and circling movements of sharks, and this is understandable considering their more directed attention to what was going on underwater. It also suggests that a lot more circling of swimmers and 80 CONTRmUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

TABLE 21. Shark behavior observed after attack; divers compared to all shark attack victims.

All divers Total SAF Shark remained in vicinity 59 cases 159 cases In immediate area 61 % 62% Followed victim/ rescuers 12% 15% Held on to victim 27% 23 0/0 (Force required to remove) (69%) (70%) Shark seen to leave area 36 cases 101 cases Shark not seen after attack 24 cases 222 cases

Total number of cases considered 119 482 bathers may go on than is indicated by actual observations. The opportunity for a shark becoming sandwiched between the victim and some barrier to its escape (beach, reef, boat, etc.) was also understandably lower with divers than with bathers and swimmers in comparatively shallow water.

NATURE OF INITIAL STRIKE

As with attack situations in general, it appeared quite often that an attack occurred after an individual diver was no longer a part of a group of other persons. Shark attack victims generally were set upon suddenly and violently (74 % of 554 cases of all types). The same held true for divers in general (73 % of 113 cases), but to an appreciably lesser extent for submerged divers, where only 63 % of 48 victims were struck violently and without warning. In the remainder of the cases, the victims were often initially unaware that they had been attacked by a shark.

SHARK BEHAVIOR DURING ATTACKS ON DIVERS

As with attackers in general, the behavior of sharks seen making mUltiple strikes on divers could be described as "frenzied" only about one-fourth of the time; 23 % of 30 cases as compared to an essentially equal 26% for 187 cases of all types. Most often (63 % of the time) the sharks made controlled and deliberate strikes at the diver-victims. A similar 61 % was found for victims of all types. In four cases, the shark very quickly bit the victim again after releasing its initial hold in a manner suggesting "getting a better grip" on the person (13 % of 30 diving cases compared to 13 % of 187 victims in general). 1974] BALDRIDGE: SHARK ATTACK 81

TABLE 22. Sharks reported to have attacked divers.

All divers Submerged divers Sharks No. % No. % Tigers 12 13 4 10 Great whites 11 12 7 17 Orectolobids (carpet, nurse, etc.) 18 19 8 19 Mako 7 7 3 7 Carcharhinids (other than tigers) 26 28 11 26 Odontaspids (sand tiger, grey nurse, etc.) 15 16 6 14 Hammerheads 3 3 2 5 Sevengill shark 1 1 2 Horn shark 0 0

Total citations 94 42

POST-ATTACK SHARK BEHAVIOR

Again, reflecting the visibility advantage afforded to divers, the data in Table 21 show that 80% of attackers in 119 cases were seen to some extent after encounters with divers (76% of 54 cases for submerged divers) as com­ pared to only 54% for the SAP as a whole. The gross post-attack behavior of sharks which remained in the general vicinity after assaulting divers was found to be essentially the same as for attackers in general.

IDENTITY OF SHARKS

The species of sharks reported as attackers of divers were listed together with those in the total SAF in Table 15. A summary listing is given in Table 22, to show differences between attackers of submerged divers and those assaulting divers in general. As was found also for the total SAP, tigers and great white sharks were the most cited attackers of divers, with tigers now leading by a very slight margin. However, the reverse was found for sub­ merged divers, where almost twice as many great whites were reported as tigers.

WATER TEMPERATURE

The apparent greater involvement of submerged divers with great white sharks brought up the question of water temperatures and whether or not they 82 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No.2 might be significantly different from those associated with attacks in general. There were 27 cases where sufficient information existed, and the number median water temperature for submerged diver-attacks was found to be 21.3°C (70.3 °F). This is not greatly different from the median of 22.rC (72.9°F) calculated for a total of 197 cases of all types, especially considering the poor quality of the data used in the computations.

RESCUERS OF DIVER-VICTIMS

It is a general safety rule with divers never to operate alone, thus placing potential rescuers close at hand to a high percentage of diver-victims. So, it was in the accounts of attacks upon divers that we saw numerous examples of rescuers not being injured even though they went immediately to the aid of the victims. They at times actually fought with the shark, and even on occasions placed themselves as barriers between the attackers and their victims.

FREQUENT AVAILABILITY OF SPEARGUN AS WEAPON

While the act of spearfishing might well increase a diver's chance for shark attack, ready availability of the speargun or spear, either as a weapon or as a billy to fend off an attacker, appeared as a factor contributing to a number of threatened divers escaping with little or no injury.

SUMMARY OF OBSERVATIONS ON ATT ACKS INVOLVING DIVERS

That over one-fourth of all records on shark attacks deal with victims gen­ erally involved in underwater activities does not at all connote any level of immunity for divers relative to swimmers and waders. The absence of females among diver-victims is of particular interest, and efforts should be devoted to finding out why. There are some obvious possibilities related to differences in behavior and activity profiles between men and women divers. The real reason may not be quite so obvious. The rising popularity of sport diving and spearfishing is reflected in the rising percentage of divers among shark attack victims. Furthermore, the average ratio of one diver-victim for every two of other types expected in the decade of the 1970's again suggests a higher hazard potential for divers as opposed to' beach users. The increased danger may well be related to the inherent need in diving for operating at greater distances offshore. There is also the very real, understandably provocative, high involvement of divers with spearfishing. Divers also at times bring disaster upon themselves by overtly provoking seemingly docile sharks. Most diver­ victims were free-divers usually using mask, snorkel, and swim fins, with SCUBA divers next among present day victims. 1974] BALDRIDGE: SHARK ATTACK 83

The attack patterns were found to be somewhat different for divers. Sharks generally approached their diver-victims from about the same directions as for swimmers. Divers, however, seemed much more aware of their sur­ roundings. Sharks were sighted by divers before attacks much more often, and close passes without contact with the victims were reported about four times as frequently. Diversionary actions and the use of weapons were there­ fore brought into play more often, resulting in a much lower incidence of actual strikes and injuries, especially multiple contacts with the sharks and repeated biting. Injury to lower parts of the bodies of divers occurred far less often than to swimmers and waders. This was understandable when it is realized that swimmers are usually splashing with their feet, water-treaders are dangling their moving legs from the surface, and waders more often than not have only the lower parts of their bodies submerged. On the other hand, injury to the upper parts of the victim's bodies were found about equally frequent for divers as for beach users. The actual injuries to divers, however, were found to be generally of a less severe nature, leading to far less loss of tissue, fewer lost arms and legs, and a mortality rate less than half that for swimmers and waders. This may have been due in part to the increased opportunity for divers to fight off the shark, particularly by using a speargun as a billy, and possibly somewhat to the usual presence close by of potential aid and assistance in the person of a diving buddy. There is also the likelihood of divers being less prone to panic thru experience underwater than would the beach user who suddenly finds himself in a crisis of life-and-death proportions. The size-spectrum of sharks striking divers is not significantly different from attackers in general. The species patterns of sharks attacking divers was found to be somewhat different from that for beach users, particularly in that there was less involve­ ment of what Stew Springer refers to as "bank loafers." Great white sharks and tigers were, as with shark attacks in general, the primary offending species. However, in the limited data available, tigers were found to be far less involved with submerged divers than with divers in general or with other types of victims. Spearguns were very effective weapons for divers, particularly when used as a billy to fend off the shark. The chance of actually driving off an attacker by spearing it didn't appear to be too great, but quite often a diver was able to use his empty speargun to hold an overly attentive shark at bay until either its ardor cooled or the opportunity for escape presented itself.

CONCLUSIONS AND RECOMMENDATIONS

By 1963, there was already enough new information available thru Navy­ sponsored research and studies of shark attack records to permit Dr. Perry 84 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2

Gilbert, Chairman of the Shark Research Panel, to formulate his often refer­ enced "Advice to Those Who Frequent or Find Themselves In, Shark-infested Waters" (Gilbert, 1963c). Although my analysis may have raised new questions about the basic nature of shark aggression towards man, it has indicated no great change in the earlier survival philosophy of Gilbert. Perhaps the fundamental reason for this is that such advice still remains predicated more on logic than on statistical validity. The following suggestions are the best that we are able to make after meticulously examining the most complete collection of infor­ mation available anywhere in the world on factors associated with predaceous shark behavior towards man. This advice is admittedly far from perfect, but to do better we must learn much more about sharks and shark attack than we now know.

ADVICE TO BATHERS AND SWIMMERS

Always swim with a companion, and no not wander away from a coherent group of other bathers and thereby isolate yourself as a prime target for attack. Do not swim in water known to be frequented by dangerous sharks. Leave the water if sharks have been recently sighted or thought to be in the area. Although not conclusively proven, human blood is highly suspect as an attractant and excitant for sharks. Keep out of the water if possessed of open wounds or sores. Women should avoid swimming in the sea during menstrual periods. It is not always convenient, but very murky or turbid water of limited under­ water visibility should be avoided if possible. In any event, a particularly watchful eye should be maintained for shadows and movements in the water. If there is any doubt, get out at once. Refrain from swimming far from shore where encountering a shark becomes more probable. Avoid swimming alongside channels or drop-offs to deeper water which provide ready access for a shark. Leave the water if fish are noticed in unusual numbers or behaving in an erratic manner. Take no comfort in the sighting of porpoises, for this does not at all mean sharks are not about. Avoid uneven tanning of the skin prior to bathing in the sea, for sharks apparently respond to such discontinuities of shading. Use discretion in terms of putting human waste into the water. Avoid swimming with an animal such as a dog or a horse, etc. Take time to look around carefully before jumping or diving into the sea from a boat. 1974] BALDRIDGE: SHARK ArrACK 85

Particularly at low tide, take notice of a nearby offshore sandbar or reef that might have entrapped a shark. Avoid swimming at dusk or at night when many species of sharks are known to be searching for food. It just might be a good idea to select other than extremely bright colors for swimwear. Never, in any form or fashion, molest a shark no matter how small it is or how harmless it might appear. Keep a wary eye out towards the open sea for anything suggestive of an approaching shark.

ADVICE TO DIVERS

NEVER DIVE ALONE. Not only might the very presence of your diving buddy deter the shark, but together you have a far better chance of becoming aware of a nearby shark in time to take effective countermeasures. Furthermore, if something did happen to you, at least there would be assistance close at hand. Do not in any way provoke even a small shark-not by spearing, riding, hanging on to its tail, or anything else that might seem like a good idea at the time. Even a very small shark can inflict serious, possibly fatal, injury to a man. Do not keep captured fish, dead or alive, about your person or tethered to you on a stringer or similar device. Remove all speared or otherwise wounded fish from the water immediately. Do not spearfish in the same waters for such extended periods of time that curious sharks may be drawn to the area by either your prolonged quick move­ ments or an accumulation of body juices from numbers of wounded fish. Leave the water as soon as possible after sighting a shark of reasonable size, even if it appears to be minding its own business. Submerged divers, as opposed to surface swimmers, have a better chance of seeing a shark making investigatory passes prior to being committed to attack. Use smooth swimming strokes, making no undue commotion, in reaching the safety of a boat or the shore. To the greatest extent possible, remain submerged where chances are greater for watching the shark and countering its charge if attack occurs. Do not count on the shark either circling or passing close at hand without contact before it makes a direct run. Use discretion in the choice of wetsuit colors in terms of conditions and sea life prevalent in the waters of intended operations. Do not take a chance on being mistaken for the area's natural prey of choice. Carry a shark billy or plan to use the butt of a speargun for this purpose if necessary. Such devices have been shown to be very effective in holding an aggressive shark at bay until its ardor cools. 86 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No. 2

Take full advantage of your submerged position and limits of visibility to be aware always of nearby movements and presences. Shark attack case histories indicate that such vigilance has played a major role in lowering injuries and mortality rates among diver-victims. Do not maneuver a shark into a trapped position between yourself and any obstacle such as the beach, reef, sandbar, or possibly even a boat. As with swimmers, do not wander away from an established group of other divers and possibly give thereby an appearance of fair game. Avoid diving at dusk and at night.

ADVICE TO VICTIMS

Try to remain calm and take full advantage of weapons available to you. Use any object at hand to fend off the shark while at the same time not intentionally provoking it further. Keep fully in mind the limitations of such devices as powerheads, gas-guns, spearguns, etc., and do not expect them to accomplish the impossible. Such weapons, if used improperly, may serve only to further agitate the shark. Use available spears and knives first to fend off the shark and attempt to wound the fish only as a last resort. Sharks often seem to react with increased vigor to efforts at sticking it with pointed objects. Discretion should be used in making aggressive movements towards a shark. One that had not yet committed itself to attack might be "turned on" by such movements if interpreted by it as a threat. On the other hand, quick move­ ments towards a shark close at hand might produce a desirable startle response. Once contact has been made or is imminent, fight the shark as best you can. Hit it with your bare hands only as a last resort. Probing the shark's eyes especially and perhaps also its gills has often turned the tide. Startle responses which at least buy valuable time have been produced occasionally by such actions as shouting underwater or blowing bubbles. Do anything that comes to mind, for the seconds or minutes of time during which the shark might withdraw as a result could be sufficient to effect your rescue. Most shark attacks produce wounds that are readily survivable. Bleeding should be controlled as quickly as possible--even before the victim has been brought ashore. Treatment by a physician is indicated even where wounds are relatively minor.

THE QUESTION OF SHARK MOTIVATION

During the lengthy, intensive review of shark attack case histories required for their analysis, a pattern became evident in terms of what sharks appeared to be trying to accomplish in their attacks upon humans. Our ideas on this 1974] BALDRIDGE: SHARK ATTACK 87 were first published several years ago in the Journal of Military Medicine, the official publication of the Association of Military Surgeons of the United States (Baldridge and Williams, 1969). The motivating factors leading to shark attack on man have been generally assumed to be closely related to hunger, for in many cases of shark predation the human victim was either partially or totally eaten by the shark. Wound characteristics have very often supported only too clearly the assumption that the intent of the predator was to feed upon the victim. The occasional findings of human remains in the stomachs of sharks have been readily accepted as prima facie evidence that hunger caused the shark to attack. Yet, other obser­ vations just as clearly indicate that this might not have been true in a number of encounters. This mounting evidence further suggests that hunger or the feeding drive may not be as important in shark attack as heretofore believed. Wounds in some attacks, and other physical damage caused by sharks, do not at all appear to have been the results of efforts on the part of the shark to bite and remove a portion of the object of the attack. It has been recognized for some time that certain species of sharks often use their razor-sharp teeth as weapons in ways and under circumstances totally unrelated to the procurement of food, i.e. to cause a cut or gash-type wound by open-mouthed use of the teeth as distinguished from a bite produced by closing of the jaws. The pos­ sibility is therefore very real that a significant fraction of shark attacks on man may well be the result of aggressive behavior directed at the victims in an attitude of fighting rather than feeding. Shark researcher G. D. Campbell (1968) suggested that a state of "petu­ lance," possibly induced by unsettling environmental conditions such as unusually high water temperatures, may have been an alternative to hunger in motivating Zambezi sharks to "half-heartedly" attack humans in South African waters. Where clean-cut unmatched gash-type wounds appeared to be the result of use of a single set of teeth, it was not usually possible to ascertain which jaw was involved. However, because of the better arrangement of the upper teeth (flat, broad, sharply pointed, with serrated edges) for the purpose of cutting in comparison to those of the lower jaw, it would be reasonable to expect that they would be more likely involved in producing gash-type wounds. It is generally assumed that blood in the water acts as a powerful attractant and excitant for hungry sharks, particularly when accompanied by movements of distress by the victims. Why then do sharks very often strike their victims only once or twice and then leave them bleeding profusely and in a state of shock or hysteria. Less than 20% of all attacks examined involved multiple strikes. Just when most sharks had their victims in a helpless condition and doing all the exciting things that are supposed to be irresistible for a hungry shark, why did they so often simply go away without further aggression? Did 88 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. 1, No.2 the sharks in those instances find the humans to be unpalatable, were they really so easily frightened away, or could it just as reasonably be that the factor motivating the sharks in such cases was not hunger at all? Sharks occasionally strike surfboards and even the relatively flat skin of a boat. An overly ambitious basking shark once even tackled a 664-ton steamer, leaving "her 200 passengers in various stages of hysteria and emotion." Surely in such cases, the sharks were not trying to eat the boats. Yet, severe damage has at times been rendered to boats of reasonably large sizes by the repeated, deliberate strikes of determined sharks. And if they were not driven by hunger, then what was the motivation and what were they trying to accomplish? In a number of cases in the SAF, relatively severe wounds have been ex­ plained as resulting from contact with the snout, skin, or edges of the fins of attacking sharks. Except for those resulting from direct "bumps" by the shark's snout, such wounds have been taken as incidental and not the results of directed action by the shark. Of course, this assumption might not be correct. In other cases, victims have reported feeling only a "tingling" sensation or scratch. Examination of the affected area often revealed a rather severe laceration. Such an effect could easily have been produced by open-mouthed raking of the area by the upper teeth of even a relatively small shark. Strange shark behavior, interpreted by some as threatening or pre-attack postures or gestures, has been observed repeatedly with aggregations and indi­ vidual specimens of C. menisorrah, the Pacific grey reef shark. Church (1961) observed a 6-foot C. menisorrah swimming very erratically just prior to its turning and suddenly attacking another diver. The pectoral fins were extended more downward (about 60 degrees from horizontal) than usual with the nose of the shark bowed up and the back area hunched. It seemed to swim stiffly with its whole body, the head moving back and forth almost as much as its tail. A high speed pass directly at the victim immediately followed this odd swim­ ming behavior. Hobson (1964) was of the opinion that such head-swinging during swim­ ming permitted the shark to maintain visual contact with any object of concern which was directly astern of it. As underwater photographers Giddings and Bergman (The Bamboo Reef, San Francisco, Calif.) gathered footage for their film entitled "The Predators," they approached aggregations of grey reef sharks, whereupon individuals would separate from the group and move in the direction of the divers with swimming patterns essentially identical to those reported by Church. As the distance between the divers and the sharks increased, the "sentinels" rejoined the group with normal swimming movements. A number of the sharks also exhibited unexplained gash-type wounds on their sides. Johnson and Nelson (1973) concluded that this behavior probably ex- 1974] BALDRIDGE: SHARK ArrACK 89 pressed defensive threat after 23 observations of such displays by grey reef sharks under approach-withdrawal conflict situations. Rapid diver-approach was shown to be a releasing stimulus. Especially intense displays occurred when escape routes for the sharks were restricted and when divers acted aggressively towards approaching sharks. Now, what were those sharks trying to communicate to those men? It would not appear that their aggressive behavior was motivated by any overpowering urge to feed upon the divers. Could it have been simply that the sharks in their own way were declaring the continuation of either the divers' presences or their threatening actions or both would leave them with no alternatives but to defend themselves? Animal behaviorists would surely express that dif­ ferently, but the sense of the thing remains that those grey reef sharks ap­ parently resented the divers being there and doing whatver it was that the sharks considered them to have been doing. With the ever expanding use of beaches and near-shore waters by swimmers and divers, the probability for encounters between man and shark will inevi­ tably increase. The observed gradual rise in incidence of shark attack is therefore both real and understandable. At the same time, it is not easy to dismiss the fact that if sharks in coastal waters fed upon people by choice in preference to their natural food, then the incidence of shark attack would be far greater than the fewer than 100 cases reported throughout the world each year. If it is as likely as it appears that the motivating factors for such attacks are not as deeply integrated with the feeding drive as has been heretofore believed, then it behooves those working with sharks to recognize whatever might be the behavioral activators and to develop means for producing their effects under controlled laboratory conditions. Some obvious alternatives other than feeding would include territorial behavior, both in the conventional sense and also as related to minimum tolerated approach distances, or even possible unintentional interference by the victims in courtship patterns of the sharks. Unfortunately, there is not sufficient information on actual attackers to determine whether or not a single sex of sharks is responsible for the majority of attacks on man. It would be consistent with the findings of this analysis if as many as 50 to 75% of all recorded attacks by sharks upon man were motivated by a drive or drives other than feeding. Remember that only about one-fourth of the victims received sufficient numbers of wounds to indicate determined efforts on the part of the sharks to devour them. Only slightly over half the victims suffered significant loss of tissue, and only about one in five lost a limb. Furthermore, actual bites as evidenced by the presence of discontinuous tooth marks were received by less than half of the victims. These, of course, did not include those where the bites were sufficiently severe to go beyond the pro- F'

90 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

duction of puncture wounds, resulting usually in severe lacerations accom­ panied at times by loss of tissue. There were only one or two strikes made against 76% of all the victims in the SAF. Even where repeated strikes were made, only about one-fourth of the attackers behaved in a "frenzied" manner. The majority of shark attacks on humans appear as single event encounters; one violently delivered strike resulting in one wound or set of wounds and possibly involving one or two close passes without the shark making contact. Most of all, it can not be ignored that these extremely efficient predators kill only about 35 % of their human victims, a figure which truly indicates that they were not really trying most of the time. In line with the general past acceptance of shark attacks as results of efforts by sharks to feed upon man, shark repellents and other anti-shark measures have been routinely tested under feeding conditions. The principal procedure currently employed by the Navy (Gilbert and Springer, 1963) for evaluating chemical repellents involves determination of feeding inhibition elicited by the chemical in the presence of attractive food. Furthermore, the sharks are usually intentionally starved to insure the presence of a powerful hunger drive. If further laboratory and field studies indicate that factors other than feeding may truly be responsible for a significant fraction of shark attacks, then it would appear in order to consider appropriate modification of those testing procedures for repellents and repellers which rely heavily upon discouraging the taking of attractive food by starved sharks. Answers to such questions can best be provided by intensification of research on sharks both in their natural environment and under laboratory conditions so delicately controlled that a reasonable approach to "normal" shark behavior can be realized.

REMARKS CONCERNING STATISTICAL TREATMENTS The ideal result of this analysis would have been the development of strong correlations between the occurrence of shark attack and a number of readily observed environmental and/ or behavioral factors and to show with a quanti­ tative measure of statistical significance that victims differed in these corre­ lations from non-victims. Such findings would bear directly upon the principal matter before us, i.e. what recognizable values or forms of these factors, either singly or in combination, determine the probability of an individual being at­ tacked by a shark. The questions of prime importance are what settings of the environmental stage and/ or modes of shark behavior make an attack likely and what in the makeup or actions of a particular person makes him the likely victim. All of these considerations require not only detailed information on attack situations but also equally good data for the same parameters under normal, non-attack, control conditions. Without the latter, it is meaningless to speak in terms of cause-and-effect relationships. The description of any developed correlations in data dealing with only attack conditions must always 1974] BALDRIDGE: SHARK ArrACK 91 be appropriately qualified; "When attacks occurred, this or that parameter was related to some other factor associated with attack conditions." This is only half the picture of shark attack causative factors, and no equation has ever been solved by considering only the parameters on one side of the equal sign. Consequently, no great amount of time was devoted to testing the intensity of correlations among attack data in the absence of control information on non-attack situations. Even with water temperature where an element of control was introduced, it was not felt that the data on non-victims were com­ plete enough to develop a measure of "exposed population" from which victims were drawn on a worldwide basis. Nevertheless, the data have been presented in sufficient detail for statisticians who may wish to pursue such a course. In response to publicity given to my analysis in the popular press, William Borucki of the NASA Ames Research Center, Moffett Field, California, re­ quested access to the data. He wished to test the applicability of his newly developed technique for identifying statistical dependencies among two or more variables in a many variable data space. With the generous cooper­ ation of the Naval Aerospace Medical Center at Pensacola, Mr. Borucki was supplied with all available data in forms directly useable for his purposes. Here again, the approach was to search for correlations in the data on attack situations and not to examine directly the matter of differences in statistical significance between attack and non-attack conditions. As dimensions of examined correlations increased, Borucki found himself dealing with numbers of applicable cases that were far too low to justify continuation of his efforts. His general conclusion was that SAF information as presently constituted is essentially "data noise" from which signals in the form of multidimensional correlations are far too weak to permit their isolation with any meaningful level of statistical significance. My program has remained, therefore, primarily one of data assimilation and reduction followed by analyses predicated more upon logic and intuitive reasoning than upon quantitative statistical treatments. The justification for my non-mathematical analytical approach lies principally in the lack of suitable control data against which statistical significance could be tested. The logical approach is not without its place and value, for a detailed mathe­ matical examination would not appear indicated when one is faced with problems not unlike examining the suitability of round pegs for filling square holes.

THE CASE FOR CONTINUING AND EXPANDING THE SAF

One of the stated objectives of this program was to evaluate the present methods for gathering meaningful data on shark attack and to examine re­ quirements for maintaining such an effort in the future. 92 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

It is very clear that past data collecting efforts leaned far too heavily upon popular sources, especially newspapers. If truly meaningful information is to be generated, it must be gleaned from as close to the source as possible and as near in time as practical. Interviews by trained observers at the attack site and with the persons directly involved would be the only way to get infor­ mation of such a nature and degree of accuracy to be of any real value in subsequent analyses. Adequate means must also be devised for gathering and treating control data. The present program has clearly indicated a number of areas to which particular attention should be directed in gathering infor­ mation on future attacks. The Shark Attack File, appropriately modified in terms of data collecting procedures, should definitely be continued and expanded. Support would logi­ cally come from some agency of the federal government. Shark attacks will continue to occur, possibly with rising frequency as man turns increasingly towards the sea for recreational, economic, and military pursuits. We are only now beginning to recognize some hints as to the causes of shark attack; motivations other than hunger, for example. The time to cease collecting data is certainly not just when we are beginning to learn which data to collect. In terms of factors affecting the morale and performance of Navy personnel and civilian swimmers and divers alike, it is very important that the matter of shark attack be kept in its proper perspective. This can only be done by making available to concerned individuals and governmental agencies up-to­ date valid information on known instances of attack on man. The Navy and other services during World War II found themselves faced with shark-induced morale problems which seriously degraded the performance of men operating in warm water regions of the world. The shark repellent which was hastily developed to combat more the problem of morale than shark attack has now been clearly demonstrated to be of little value as a deterrent to shark ag­ gression. So, the problems of shark attack and what might be done to counter it are still there-in clearer focus to be sure than in the days before the Office of Naval Research set up the Shark Research Panel and the Shark Attack File­ but still there, nevertheless.

ACKNOWLEDGMENTS

The efforts of a number of people over almost a 1S-year period are grate­ fully acknowledged. There was first the foresightedness of Dr. Sidney Galler and Helen Hayes of the Office of Naval Research in setting up and energetically furthering the aims of the Shark Research Panel; the members of the panel under the chairmanship of Dr. Perry Gilbert; Dr. Leonard Schultz and Marilyn Malin of the Smithsonian Institution who along with Dr. Gilbert labored so )

1974] BALDRIDGE: SHARK ATTACK 93

long compiling the case histories making up the Shark Attack File-and, of course, the contributors of information to the SAP from all parts of the world; Carolyn Gugino of the Navy's Bureau of Medicine and Surgery and other BUMED statisticians who with others from the Smithsonian took the first giant steps towards developing procedures for assimilating data in the SAP. At the Mote Marine Laboratory there were Joy Williams who for over a year worked on reducing all sorts of information to code numbers and letters in addition to accomplishing innumerable filing chores; Beth Arthur who as a volunteer worker not only edited and filed SAP data but also for over a year made beach counts in good weather and bad which served as control data for some of our analyses; my son David Baldridge III and Randy Wells who helped with manual tabulations of data; Patricia Bird who edited and copied most of the code sheets; Mary Kumpe who coded the responses to a number of geographical questions; Jennifer Thorner who accomplished hundreds of manual tabulations of code sheets; and Donna Johnson who with Mrs. Thorner prepared the manuscript. Voluntary assistance in gathering control data at bathing beaches was gen­ erously provided by Edward Broedel of Myrtle Beach, South Carolina; F. G. Wood of the Navy Undersea Center, San Diego, California; Captain Robert Nelson and members of the San Diego Lifeguard Service; and the late Dr. Earl Herald of the Steinhart Aquarium, San Francisco, California. Very helpful printouts of data from IBM cards were gratuitously prepared by Mildred Blake, Head of the Data Processing Department, Sarasota County Vocational Technical School, Sarasota, Florida. All machine manipulations of data were accomplished at no cost to my project by the Data Processing Department, Naval Aerospace Medical Center, Pensacola, Florida. For that, I am very indebted to RADM E. P. Irons, Medical Corps, U.S. Navy and later to his successor RADM Oscar Gray MC USN, and to CDR Harry Boone MSC USN, CDR Tom MacConnell, and especially to Betsy White, Kathie Bevins, and other computer scientists who actually did the work. Attempts by Mr. William Borucki of the NASA Ames Research Center to apply advanced programming techniques to SAP data are gratefully acknowl­ edged. Financial support for this program of data analysis was provided by Oceanic Biology Programs (Code 484) of the Office of Naval Research, Department of the Navy, under Contract N00014-73-C-0252, Work Unit No. NR 104-148.

LITERATURE CITED

BALDRIDGE, H. D. 1966. Reaction of sharks to a mammal in distress. Milit. Med. 131: 440-446. ---. 1968. A proposed reference shark. ibid. 133: 654-662. 94 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

1969a. International shark attack file data assimilation program. Final Report. Submitted to the Smithsonian Institution by the Mote Marine Laboratory, Sarasota, Florida, 15 March 1969, 58 pp. 1969b. Analytic indication of the impracticability of incapacitating an at­ tacking shark by exposure to waterborne drugs. Milit. Med. 134: 1450-1453. ---, AND J. WILLIA MS. 1969. Shark attack: feeding or fighting? ibid. 134: 130- 133. CAMPBELL, G. D. 1968. Does a shark problem now exist in South Africa? Docu­ menta Geigy Nautilis, Geigy Pharmaceuticals, Ardsley, N. Y. June 1968, p. 2. CHURCH, R. 1961. Shark Attack. Skin Diver. June 1961 , pp. 30-31. CLARK, E. 1963. The maintenance of sharks in captivity. Part II. Experimental work on shark behavior. Bull. Inst. Oceanogr. Monaco, N° Special 1D: 1-10. ---, AND K. VON SCHMIDT. 1965. Sharks of the central gulf coast of Florida. Bull. Mar. Sci. 15: 13-83. COPPLESON, V. M. 1962. Shark attack, 269 pp. Angus and Robertson, Sydney. 1963. Patterns of shark attack for the world. In: Sharks and survival, pp. 389-421. P. W. Gilbert, ed., D. C. Heath and Co., Boston. DAVIES, D. H . 1965. About sharks and shark attack, 237 pp. Brown, Davis, and Platt Ltd., Durban. GILBERT, P. W. 1963a. The AIBS shark research panel. In: Sharks and survival, pp. 505-507. P. W. Gilbert, ed., D. C. Heath and Co., Boston. 1963b. The visual apparatus of sharks. ibid. pp. 283-326. 1963c. Advice to those who frequent, or find themselves in, shark-infested waters. ibid. pp. 501-503. 1966. Feeding and attack patterns of sharks. Proceedings of the Eleventh Pacific Science Congress, Tokyo, Japan. 7 : 32-33. 1968. Report on the use of the NUWC shark screen as a deterrent to sharks. NUWC-TP-52. Naval Undersea Center, San Diego. 30 pp. 1972. Response patterns of porpoises and elasmobranch fishes. Final Report. Office of Naval Research, Contract No. NOOOI4-69-C-0340. 13 pp. ---, B. IRVINE, AND F. H. MARTINI. 1971. Shark-porpoise behavioral interactions. Am. Zool. 11 : 636. ---, AND S. SPRINGER. 1963. Testing shark repellents. In: Sharks and survival, pp. 477-494. P. W. Gilbert, ed., D. C. Heath and Co., Boston. GRUBER, S. H. 1969. The physiology of vision in the lemon shark, N egaprion brevi­ rostris (Poey): A behavioral analysis. Doctoral Dissertation. Univ. of Miami, 113 pp. ---, D. 1. HAMASAKI, AND C. D. B. BRIDGES. 1963. Cones in the retina of the lemon shark. Vision Res. 3: 397-399. HAMASAKI, D. I., AND C. D. B. BRIDGES. 1965. Properties of the electroretinogram in three elasmobranch species. Vision Res. 5 : 483-496. HOBSON, E. S. 1964. Sharks increasing visual field. Underwater Naturalist. 2: 29. IRVINE, B., R. S. WELLS, AND P. W. GILBERT. 1973. Conditioning an Atlantic bottle­ nosed dolphin, Tursiops truncatus, to repel various species of sharks. J. Mammal. 54(No. 2) : 503-505. JOHNSON, C. S., AND E. McFADDEN. 1971. Color and reflectivity of sea survival equipment as related to shark attack. Presented at the Aerospace Medical Meeting, Houston, Texas, 26 April. JOHNSON, R. H., AND D. R. NELSON. 1973. Agonistic display in the grey reef shark, Carcharhinus menisorrah, and its relationship to attacks on man. Copeia 1973 : 76-84. MATHEWSON, R., AND P. W. GILBERT. 1967. Report on shark-porpoise experiments at the Lerner Marine Laboratory. In: Amer. Inst. BioI. Sci., Conference on the shark-porpoise relationship. Amer. Inst. BioI. Sci., Washington, D. C., pp. 17-24. 1974] BALDRIDGE: SHARK ArrACK 95

MCCORMACK, H . W., T. ALLEN, AND W. E. YOUNG. 1963. Shadows in the sea. Chilton Books, New York. 415 pp. MOLNAR, G. W. 1946. Survival of hypothermia by men immersed in the ocean. J.A.M.A. 131, No. 13: 1046-1050. MYRBERG, A. A. 1970. The behavior and sensory physiology of sharks. Final Report. Office of Naval Research, Contract No. NONR 4008(10) . Univ. of Miami. 8 pp. SCHULTZ, L. P. 1967. Predation of sharks on man. Chesapeake Science 8: 52- 62. ---, AND M. H . MALIN. 1963. A list of shark attacks for the world. In: Sharks and survival, pp. 509-567. P. W. Gilbert, ed., D. C. Heath and Co., Boston. SPEALMAN, C. R. 1945. Unpublished data. Naval Medical Research Institute, Be­ thesda, Maryland. SPRINGER, S. 1963. Field observations on large sharks of the Florida-Caribbean re­ gion. In: Sharks and survival, pp. 95-113. P. W. Gilbert, ed., D. C. Heath and Co., Boston. TESTER, A. L. 1963. The role of olfaction in shark predation. Pacific Sci., 17(2): 145-170. ---, G. J. NELSON, AND C. I. DANIELS. 1968. Test of NUWC shark attack deter­ rent device, NUWC-TP-53. Naval Undersea Center, San Diego. 46 pp. WOOD, F. G., D. K . CALDWELL, AND M. C. CALDWELL. 1970. Behavioral interactional between porpoises and sharks. In: Investigations on Cetacea, II: 264-277. G. Pilleri, ed., Benteli AG, Berne-Biimpliz. 96 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vol. I, No. 2

APPENDIX SUMMARY LISTING OF LOCALITIES OF 1165 SHARK ATTACKS FOR WmCH DATA WERE CODED AND ANALYZED Note the heavy predominance of attacks reported from English-speaking parts of the world, perhaps connoting a language barrier in the procedures for gathering information on such hap­ penings in other localities. AFRICA, ATLANTIC OCEAN Liberia 2 Senegal 3 Sierra Leone 1 AFRICA, INDIAN OCEAN Kenya 2 Mozambique 10 AFRICA, REPUBLIC OF SOUTH AFRICA Cape of Good Hope 25 Natal, Durban 42 Natal, Other than Durban 32 ASIA, INDIAN OCEAN Aden 2 India 15 Iran 17 ASIA, PACIFIC OCEAN China, Taiwan 5 China, Tsingtao 2 Hong Kong 5 Japan 8 Malaysia 3 Philippine Islands 18 Singapore 6 Southeast Asia 2 AUSTRALIA No specifics 2 Great Barrier Reef 9 New South Wales 137 Northern Territory 5 Queensland 57 South Australia 12 Tasmania 10 Torres Strait (Australia/ New Guinea) and Islands 43 Victoria 15 Western Australia 29 CENTRAL AMERICA Costa Rica, Pacific coast 2 Panama, Caribbean coast 9 Panama, Pacific coast 8 EUROPE, NORTH ATLANTIC England 3 1974] BALDRIDGE: SHARK ArrACK 97

ISLANDS, ATLANTIC OCEAN Bermuda 6 Cape Verde Islands 1 ISLANDS, INDIAN OCEAN Adaman/Nicobar Islands 1 Madagascar 2 Mascarene Islands 1 ISLANDS, PACIFIC OCEAN Admiralty Islands 4 Bismark Archipelago 28 Caroline Islands 4 Fiji Islands 22 Galapagos Islands, Ecuador 2 Guam 1 Hawaii, United States 31 Indonesia 1 Johnston Island 2 Line Islands 1 Marshall Islands 5 Midway Island 2 New Caledonia 3 Kwajalein Atoll 1 Samoa 3 Society Islands 3 Solomon Islands 19 Tonga Islands 2 Wake Island 3 ISLANDS, WEST INDIES Bahama Islands 11 Cuba 15 Curacao 2 Dominican Republic 2 Granada 5 Grand Turk 1 Haiti 3 Jamaica 4 Martinique 1 Puerto Rico 9 Virgin Islands 1 Windward Islands, no specifics 1 MEDITERRANEAN SEA African coast, no specifics 1 Egypt 4 Greece 4 Italy 7 Israel 1 Malta 1 Yugoslavia 12 MEXICO Mexico, no specifics 1 Mexico, Caribbean coast 3 Mexico, Gulf of Mexico 13 Mexico, Pacific coast 11 98 CONTRIBUTIONS FROM THE MOTE MARINE LABORATORY [Vo\. I, No. 2

NEW GUINEA 53 NEW ZEALAND 29 OPEN SEA 32 PERSIAN GULF 6 RED SEA Gulf of Aqaba 1 Sudan 1 SOUTH AMERICA Argentina 1 Columbia 1 Venezuela 4 UNITED STATES California, Point Conception and north 20 California, South of Point Conception 26 Connecticut 1 Delaware 2 Florida, no specifics 2 Florida, Atlantic coast 61 Florida Keys 17 Florida, Gulf of Mexico 27 Georgia 6 Massachusetts 4 Mississippi 1 New Jersey 17 New York 8 North Carolina 2 South Carolina 23 Texas 6 Virginia 2 LOCATION UNKNOWN 10 TOTAL 1165