Agonistic Attacks on Divers and Submersibles by Gray Reef Sharks, <I>Carcharhinus Amblyrhynchos</I>: Antipredatory O

BULLETIN OF MARINE SCIENCE, 38(1): 68-88, 1986

AGONISTIC ATTACKS ON DIVERS AND SUBMERSIBLES BY GRAY REEF SHARKS, CARCHARHINUS AMBLYRHYNCHOS: ANTIPREDATORY OR COMPETITIVE?

Donald R. Nelson, Robert R. Johnson, James N. McKibben and Gregory G. Pittenger

ABSTRACT Pursuit by a small wet submersible was used to release exaggerated-swimming display and subsequent attack in order to determine proximate causative factors. A total of 57 trials were conducted which tested the effect of: I) species of shark, 2) bait vs. no bait, 3) grouping type- lone vs. aggregated, and 4) location on the reef. Ten high-speed strikes on the vehicle were elicited, all by gray reef sharks, Carcharhinus amblyrhynchos. Neither attack nor full display could be elicited from silvertip sharks, C. albimarginatus; blackfin reef sharks, C. melanopterus; or reefwhitetip sharks, Triaenodon obesus. Gray reef sharks attacked in both baited and unbaited situations, but apparently more readily in the latter. Lone individuals (unbaited) seemed more prone to display and attack than those in aggregations. Attack probability was positively related to degree of cornering by reef or bottom. An oriented pursuit (following shark's every move), even at quite low speeds, seemed of primary importance. Motivation for display and attack remains uncertain and may differ with circumstance, but the sharks most likely regard the pursuing submersible as a predator (anti predatory motivation), rather than as a competitor (defense of rank or territory), and certainly not as food (predatory motivation).

Among the reef sharks inhabiting the coral atolls of the tropical Indo-Pacific, the gray reef shark, Carcharhinus amblyrhynchos, is clearly the boldest and most aggressive towards humans. It has attacked both free and scuba divers as well as small diver-piloted submersibles on a number of occasions, some of which have been reviewed by Nelson (1981; 1983). Gray reef shark attacks can be separated into two distinct situational categories, feeding and nonfeeding. The first type occurs when the shark is highly aroused by opportunistic feeding stimuli, such as the sounds and odors of a freshly speared fish, and especially when there is competition as in a feeding frenzy. As we discuss later, these attacks are not prefaced by the type of threat display described below. The second type of attack occurs mainly in nonfeeding situations, appears agonistic in motivation, and consists of a sudden, high-speed strike always prefaced by a distinct exaggerated- swimming display. The display, in tum, is usually triggered by certain provocative (sometimes inadvertent) actions on the part of the diver or the submersible. It is the second (agonistic) type of attack that is the principal subject of this paper. The exaggerated-swimming behavior was initially studied at Enewetak by John- son and Nelson (1973), who concluded that it represents an agonistic (threat) display, probably defensive in nature. They described it as consisting of a tense, laterally exaggerated swimming, often with rolling and/or spiral looping, accom- panied by the postural elements of arching of the back, lifting of the snout (often with some mouth opening), and lowering of the pectoral fins. Using a diver- approach-shark technique in nonbaited situations, they determined that a rapid, aggressive approach by a diver would frequently release the display, and that its intensity increased with the degree to which the shark was cornered by reef structures. They noted that the displays occurred in situations which potentially put the shark into approach-withdrawal conflict. Subsequent observations at Rangi-

68 NELSON ET AL.: ATTACKS BY ORA Y REEF SHARKS 69 roa, French Polynesia (Johnson, 1978; Nelson and Johnson, 1980) indicated that gray sharks in feeding situations, e.g., approaching or circling bait, were less likely to display at approaching divers than sharks in nonfeeding situations, e.g., investigating diver in absence of bait. This evidence was considered as support for a nonfeeding motivation for the display and subsequent attacks. Additional hy- potheses on the causation, motivation, and derivation of the gray shark display behavior have been offered by Barlow (1974), Klimley (1974), Myrberg and Gruber (1974), McNair (1975), Johnson (1978), Starck and Anderson (1979), and Nelson (1981). Some of these ideas will be discussed later. The first published account of an agonistic attack by a gray reef shark on a diver is Church's (1961) description of the attack on Jim Stewart at Wake Island. Church and Stewart were free diving on the ocean reef when they made moves at the passing shark-which then displayed "erratic and jerky" swimming motions and "turning and twisting and rolling back and forth in the water." The shark then turned and made a "lightning speed" attack, inflicting two severe bites on Stewart's arm. Fellows and Murchison (1967) reported an incident at Johnston Island in which they were free diving and chased a small gray. The shark developed an "exaggerated swimming motion" and soon attacked, making a "very rapid dash at Fellows' arm." Fellows avoided injury by dodging the first pass and kicking away the shark on the second. At Enewetak, scuba diver Shot Miller was attacked by a gray shark that appeared and started displaying while he and John Randall were diving near East Channel, a deep entrance pass to the lagoon. Miller struck the attacking shark with his powerhead, but the shell misfired and the shark continued on in and slashed him on one side of his head (account by Randall in Ellis, 1975). Another incident occurred at Enewetak in 1978 and resulted in two scuba divers being bitten (and hospitalized) by one gray reef shark. While diving on a lagoon pinnacle, Michael deGruy took a flash photo of the displaying shark and this apparently triggered the attack. The shark charged in, bit him severely on the arm, then circled back and bit a chunk of rubber out of his swim fin, and then bit Phil Light on the hand as he moved in to ward off the shark with a shark billy (deGruy, pers. comm. in Nelson, 1981). Prior to the present study, several gray shark attacks occurred on a two-person wet submersible operated by Walter A. Starck II. One such attack was described by McNair (1975), who was with Starck in the submersible one day at Enewetak when they pursued a shark, causing it to develop exaggerated-swimming display: "The shark moved slowly ahead of us in this attitude for perhaps 30 seconds before exploding into an incredibly fast back loop which brought it crashing straight down onto the half-inch thick plexiglass hood a few inches over our heads. The deep scratches on the hood clearly showed that both upper and lower teeth bit the plexiglass ... :' Further cases of threats and attacks on Starck's submarine are described in popularized books by Doak (1975) and Starck and Anderson (1979). It was evident that further study of agonistic display and attack in the gray reef shark would contribute to an understanding of the overall problem ofshark attack on humans. It was also clear that close study of these behaviors was too risky for unprotected divers, so we designed and constructed a protective diver vehicle- the Shark Observation Submersible-specifically for studies of shark aggression. This paper reports our initial experiments at Enewetak in 1977 and 1978, using the submersible as a model to determine the proximate causative factors which lead to threat and attack in the gray reef shark. 70 BULLETIN OF MARINE SCIENCE. VOL. 38. NO. I, 1986

Figure 1. The Shark Observation Submersible (SOS II) as used at Enewetak in 1978 for experimentally inducing shark attacks. The craft contains an internally mounted scuba air supply, and is entered underwater by removing the forward acrylic-dome view port (photo by J. McKibben).

METHODS AND MATERIALS

Study Area. - The research was conducted at Enewetak Atoll, 11°30'N, 162°15'E, in the northwestern Marshall Islands, Micronesia. Enewetak was chosen for the study because of its optimum underwater conditions, large and accessible populations of reef sharks (including gray reef sharks known to display and attack), and the logistical support provided by the Mid-Pacific Research Lab (MPRL). Located on Enewetak Island, MPRL provided room and board, small boats, and diving support and served as the base of operations for the present study and for a concurrent telemetry study of patterns of movement and grouping of gray reef sharks (McKibben and Nelson, 1986). The experimental trials took place at several ocean-reef sites from the mouth of East Channel clockwise along the atoll perimeter to West Spit. Depths ranged from about 10 to 35 m, and topography varied from the relatively flat coral and sand bottom of South Channel to the nearly vertical ocean reefface at West Spit. Gray reefsharks, Carcharhinus amblyrhynchos, were the most common at these sites, but other species were frequently present, including the silvertip, C. albimarginatus; the blackfin (blacktip) reef shark, C. melanopterus; and the reefwhitetip, Triaenodon obesus. and some trials were conducted on these sharks as well. Shark Observation Submersible. -Prior to the 1977 visit, we designed and built a portable diver protection vehicle to facilitate the safe study of aggressive behaviors of gray reef sharks. The craft was designed for both (I) observation alone, and (2) experimentation, in which the vehicle itself would serve as a model to elicit responses from sharks. Dubbed the Shark Observation Submersible (SOS), the one-person submersible was 8 feet long overall, 2 feet in maximum diameter, and weighed about 200 Ib in air without its scuba tank. The streamlined fiberglass hull had an acrylic-dome entry port at the forward end, which also provided good visibility for the prone diver within. The scuba air supply and battery pack were mounted inside, and neutral buoyancy was achieved by internal syntactic foam with a small air bladder for trim. For the 1978 visit, the submersible hull was modified to increase the pilot's elbow room, a side window was installed to improve lateral visibility, and the fixed rear thruster was replaced by a rotatable one mounted on the tail fin. A hole was cut in the dome to allow handling oflines, etc., and a secondary small scuba bottle was mounted externally for emergency use. With its three electric motors, two forward "pectoral" fins, and aft tail fin all independently controllable, the new submersible (SOS II, Fig. I) had excellent maneuverability with the ability to turn 360° in its own length. On most days the submersible was carried athwartship aboard MPRL's 21-foot Boston Whaler outboard on a special aluminum-rail platform which was used for launch and recovery. On other days, the submersible was launched by davit from a 26-foot inboard boat. Experimental Techniques. - The submarine was used as a model in several experimental modes- stationary, pass-by, and pursuit-to determine the immediate releasing stimuli and other proximate NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 71

Table I. Responses of Enewetak reef sharks to pursuit by submersible

Sudden Swiml Moderatel Strong display accelerate maneuver Mild threat strong + Species away away display display attaek Total trials

Gray reef 5 6 5 12 10 38 Silvertip 5 1 6 B1ackfin 2 2 Whitetip 7 4 11

factors affecting agonistic behaviors. Trials were organized to test the effects of (1) species of shark, (2) grouping type-lone vs. aggregated, (3) location on the reef, and (4) presence or absence of bait. In baited situations, the submersible rested on the bottom near the bait (speared fish). When a shark arrived, usually attracted up the odor corridor, the submersible would take off and commence pursuit. In non baited trials, the submersible was suspended on a short line about 3 m below the boat as it drifted with the current over the reef. When a shark was sighted, the pilot released the vehicle from the line, descended towards the shark, and began pursuit. In both situations, there were also control periods in which the submersible remained stationary in the presence of a shark, or moved past the shark without directly following it. Observational data were taken by the submarine pilot (D. Nelson) and, in most cases, by other divers in the water. Some of the displays and attacks were recorded on l6-mm movie film by ourselves, or by underwater cameramen which were with us for documentary film productions. These films were used for later analysis.

RESULTS A total of 57 test trials were conducted in which the submersible approached and pursued a shark (Table 1). Of these, 38 were on gray reef sharks, and only this species responded by full display and, in 10 cases, subsequent attack. All of the gray shark attacks were sudden, high-speed strikes, prefaced by obvious exaggerated-swimming display, usually of maximum intensity. Of the 10 attacks elicited, 5 were filmed, including those in Figures 2, 3, and 4. Pre-attack Behavior. - In these experiments, the primary releaser for exaggerated- swimming display was an "oriented pursuit" - the submersible following the shark's every move, turning with it, etc., even at relatively slow speeds. This was especially effective if the shark became somewhat cornered. Of the 38 gray shark pursuits, display resulted in 27 cases (71%) as shown in Table 1. If the submersible was moving in the vicinity of a shark, or moved past it but not oriented at it, this was much less effective at releasing display, and never released attack. We observed only one case of mild display in apparent response to the submersible when stationary (at rest on the bottom), although sharks sometimes made investigative approaches in this situation. We never saw display from more than one shark at a time; even if several sharks were in the immediate area or crossing the submersible's path, only the shark we were pursuing showed display. Once released, the display was rather persistent, lasting up to 30 sec or more- usually until the submersible either stopped pursuit or was attacked. Display continued even if the chase led to an open-water area well away from cornering reef structures, e.g., 10 m out from the vertical reefface at West Spit. While being actively followed by the submersible, a displaying shark would either (1) gradually outdistance the submersible, (2) develop a circling or "carouselling" with the submersible, or (3) slow down, intensifying display as the submersible neared. The latter case most readily led to attack, with the strike being triggered when the submersible had closed to some critical distance. Just prior to attack, the shark was often in a very tense, contorted posture with its forward motion nearly stopped. 72 BULLETIN OF MARINE SCIENCE, VOL. 38, NO. I, L986

Figure 2. Attack no. 2 (1977) by a gray reef shark on the approaching submarine. Note the extremely tense posture, with more than 45° rolling, just prior to attack. Shark hit the right forward motor, biting it twice, breaking the polycarbonate plastic propellor (from l6-mm footage by R. Nelson, courtesy of Don Meier Productions). NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 73

Figure 3. Attack no. 4 (1978). The shark avoided direct frontal attack, instead circling around and striking from the rear. After its bite, the shark immediately departed at high speed, as shown in the last frame (from 16-mm footage by A. Giddings, courtesy of Hessischer Rundfunk, Frankfurt, West Germany). 74 BULLETIN OF MARINE SCIENCE, VOL. 38, NO. I. 1986

Figure 4. Left. A gray reef shark performing exaggerated-swimming display while carouselling (circling) with the pursuing submarine. Several complete circles were made, with the shark making no apparent effort to escape. Right. Attack no. 9 (1978), at the moment of impact on the upper right side of the submarine hull (from l6-mm footage by J. McKibben, taken along the ocean-reef dropoff at West Spit).

In this attitude it usually appeared off balance, either rolled distinctly on one side (perhaps to allow it to see the submersible) and/or with a head-up inclination of the body axis (perhaps to retard its sinking). In its locomotor-slowed posture, of course, the denser-than-water shark tends to sink, and therefore many of the sharks were somewhat below the approaching submersible just prior to attack. Attack. - The actual attacks (Table 2) came after about 5-20 sec of pursuit, starting with a sudden acceleration towards the submersible. Usually this was in a straight line directly at the submersible, but in two or three cases, the shark rapidly circled around to the side or behind and then made its high-speed strike from there- which was out ofthe pilot's view. The observed attacks were launched at estimated shark-to-submersible distances of 1.5-5.0 m (mean, 2.7 m). With one exception, the final attack approaches were very fast-probably at the shark's maximum burst speed-and resulted in heavy impacts on the submersible. In one filmed strike (Fig. 2), the shark took only 8 frames (0.33 sec) to reach the submersible and bite the forward motor. All of the initial hits were on the forward half of the vehicle, four on the acrylic dome, four on the upper fiberglass hull, one on the right forward motor, and one on the left forward fin. In two attacks, the shark delivered a second hit immediately after the first, within about one second. Many deep tooth scratches were inflicted, a propellor was broken, and one forward fin was badly bent and cracked, partially disabling the submersible. However, the sub withstood the attacks well and in no case was the diver-pilot inside in any danger. One filmed attack (Fig. 3) was unusual in that the shark aborted its direct approach at the dome, circled around to the side, and then attacked from above and behind-biting the upper, forward part of the hull. This attack appeared somewhat tentative, at less than maximum speed, as if the shark were close to the balance point between attacking and fleeing. After the bite, it did flee at high speed. We could not always determine whether the attacks involved bites, slashes, or NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 75

..• C'-' !l" » » » <':l <':l ~ <':l !~i5. i5. ~ ~ i5. ~ ~o t; t; ~ •... 8::; ~'" '" 0 '" ~] ~ g g 3 g'" ,a1i d ~ '(;j :s 1:: 1:: '(;j 1:: 1:: 1:: 1:: <':l <':l ~ o:l <':l <':l o:l o:l "e e e o. 0. e 0. 0. 0. 0. .§ .., .., e- .., .., .., .., .., .., ~ ~ 0 0 0 ~ 0 0 0 0

~ ,,-Ie -;- o:l ~GJ -;- -;- > c j .0 .0'" .0 .0 .g :J~ d) '5~ = = = <.l <.l C,) --- ~ a <.l z0"," ~ ~ ~ ~N 8 ~ ~ .,. .,. N 0 0 0 N N

~ ~ "0- ~ ~ <'"\ <'"\ "c <'"\ <'"\ -ll ~ ~ I I :b c c ~ ~ '-' ~ I~'0 ] ] ] 'g .•~~.• .., ~ •... •... ,,~O.c 'i c ' .., .., .., .., .3~ j I c:: c:: c:: c:: ~ I -

_c "il II *.., *.., .., .., 0 0 0 0 0 0 ""~ '"» '"» '"» '"» c:: c:: c:: c:: c:: c:: "'"

~ "i:'.., .c E' E' E' E' is. 0 0 ~ -= ~ 0 0 0. 0. "0" ~ §; ~ 1t •... 0. 0. "0 '-' '" c .., .., .., <':l e e .• ] c:: c:: ] So S S ~ ~ c c:: c c:: c:: •... •... ao:l~ o:l~ ~- cu- cu,-.. c::~ c::~ ,.c::1I"I ,.c::1I"I ,.c::N,.c::II"I,.c::<'"\ f::lN f::lo 5r;;-'~~'~i;;' <.l~ <.l~ o~o~u:::, =- =N ~:::.. .:::, .~ J vi vi vi vi vi ~'-' ~'-' ~ ~ ~

"' Co •.. 0"- ""o:l "0" B •.. ,.c:: 8:§ '3 '0 .., '3 "3 ~Ie ,.c:: ~ ,.c:: ....l •.. .§ ,.c:: •.... e .., " 0 ..,'" .••",""e ..; 0_ -=<.l= .., .., •... .., .., .., 0 ..; :g .., !lo"-" :9 8'3 0.= :g e e e •.. ..s o,.c:: o,.c:: o 8 0 ..s 0 » «-~ ....l'" ~ 0 f-o ~'" 0 0 ....l ~'" 0 ~ » ci!! .0.., z:a N N :E '~ 81l .,., 0 0 0 "l 0 0 "l .,., .., ...; ...; ...; ~ 8 ~~~t§: N N .,; <'"\ N .0 :.0; = ..•~ c::'" ~.:~-tj- 0 r-- 0 0 0 0 .,., 0 0 0 § 0 .r!e~'Bf!~ N N N N N .• ";:::.0 ClS- ~ - - S <.l'" ~ 0 S "';...J- "l "l "l "": "": "": t-: "l "l 'r: co .r!•..s ~ •.... c"'" 0 e ~ ~ ~ 'IS rJl L1.. L1.. "', "', L1.. ~ ~ ~ ~ ~ ~ .., "0 0 e" r-- r-- 1)l N r-- r-- 00 00 00 00 00 00 00 00 c " ..... r-- r-- r-- r-- r-- r-- r-- r-- 0 .., ;;( " :E 8 >:: >:: >:: >:: >:: >:: >:: >:: >:: 'il '-0 '-0 N .,., .,. r-- r-- ~ N- N N N N "" 76 BULLETIN OF MARINE SCIENCE, VOL. 38, NO. I, 1986

Table 3. Responses of gray reef sharks to submarine pursuit as related to three situational variables; grouping type (lone vs. aggregated), presence of feeding stimuli (baited YS. nonbaited), and type of reef habitat (horizontal flats YS. dropoffs)

Sudden Swiml Moderate! Strong display accelerate maneuver Mild threat strong + Situation away away display display attack Totallrials

Lone I 3 3 4 II Aggregated 4 6 2 9 6 27 Baited I I 2 6 4 14 Nonbaited 4 5 3 6 6 24 Horizontal reefs (flats) I I 3 8 6 19 Dropoffs 4 5 2 4 4 19 Nonbaited lone 1 I 4 6 Nonbaited aggregated 4 5 2 5 2 18 hits, although several attacks were seen to be open-mouthed just prior to contact. Two hits on the dome resulted in multiple tooth scratches apparently just from the lower jaw, which may have "caught" as the shark glanced upwards off the smooth dome. Another struck the dome without leaving scratches and thus may have been a closed-mouth hit. Films showed that at least one shark (Fig. 2) made two distinct "bites" in the process of breaking the propellor on the right forward motor. Post-attack Behavior. - After striking the submersible, most of the attacking sharks quickly departed the area. At least two fled at high speed, including the one shown in Figure 3. Three or four, however, remained in the vicinity-still in exaggerated- swimming display-for at least a short while after the attack. None showed any obvious physical injury from their impact with the hard surface of the submarine. Characteristics of Attacking Sharks. - The gray reef sharks which attacked (Table 2) were of both sexes (5 males, 3 females, 2 uncertain), of the usual adult sizes (estimated total lengths, 1.4-1.7 m). Near maximum size did not appear to be a requirement for the display and attack. However, the smallest shark chased (estimated TL of 0.8 m) did not display and eventually accelerated away. Other Causative Factors. -One of our objectives was to relate display/attack behaviors to other possible causative factors besides the immediate releasing stimuli. We tested grouping type, presence of feeding stimuli, and location on the reef. Although we ended up with fewer, less evenly distributed trials than desired, the data do suggest that one or more of these factors had an effect (Table 3). GROUPINGTYPE. As described by McKibben and Nelson (1986), the gray reef sharks at Enewetak exhibited three intergrading grouping types: (1) lone individuals, (2) aggregations, and (3) polarized schools. In the present study, we were able to conduct submarine trials on both lone and aggregated sharks, but not on schooled sharks. Our results suggest that the lone individuals were more prone to display and attack than were individuals in loose aggregations. Overall, of 11 lone sharks pursued, 10 (91%) displayed and 4 (36%) attacked, while of27 aggregated sharks, 17 (63%) displayed and 6 (22%) attacked. This effect is clearer when we consider only unbaited sharks. All 6 (100%) unbaited lone individuals displayed, of which 4 (67%) attacked. Of 18 unbaited aggregated sharks, only 9 (50%) displayed and 2 (11%) attacked. This finding relates to the question of motivation, specifically territoriality, and is discussed later. NELSON ET AL.: ATfACKS BY GRAY REEF SHARKS 77

FEEDINGSTIMULI.Although we gained the general impression during the study that baited sharks were less inclined to display when approached, the overall numerical data do not show this. Of 14 baited sharks pursued, 12 (86%) displayed and 4 (29%) attacked, while of the 24 unbaited sharks, 15 (63%) displayed and 6 (25%) attacked. However, of the 24 unbaited individuals, the 9 that did not display were all aggregated ones from the West Spit. If we look only at lone individuals, then all six unbaited ones (100%) displayed and four (67%) attacked, while ofthe five baited ones, four (80%) displayed and none (0%) attacked. We also had the impression that pursuit more quickly triggered display in unbaited lone individuals than in baited ones, although time measurements were not made. A further complication is that two of the four baited attacks were in a situation different from the rest of the baited trials. In these two, during our first day of submarine trials in 1977, the bait had all been consumed some 10 min or more prior to the two attack trials and the sharks at the time were circling about the general area, not newly arrived at the bait as in other baited trials. These sharks, therefore, might have been behaving more like unbaited sharks-and if treated as so, would shift the numerical data to support greater aggressiveness in unbaited sharks. LoCATIONON THE REEF. The sites where we conducted the submarine trials were of two general types: (1) flats-relatively horizontal reef (or sand and reef) bottoms, e.g., S. Channel, and (2) dropojfs-vertical reef faces or relatively steep reef slopes, e.g., ocean reefs off West Spit, Ikuren, and the outer E. Channel. Data indicate that sharks from the flats were more prone to display and attack. Of 19 flats individuals, 17 (89%) displayed and 6 (32%) attacked. Of 19 dropoff sharks, 10 (53%) displayed and 4 (21%) attacked. It thus appeared that, on the average, sharks from the flats were more aggressive. However, nine of the flats sharks were lone individuals, whereas only two of the dropoff sharks were lone. Since lone individuals also appeared more aggressive, this confounds the comparison between flats and dropofflocations. Another possible bias is that it was probably easier for the sub to corner or crowd a shark against a relatively shallow horizontal bottom than along a vertical dropoffwhere the shark could more easily escape outward or downward into water too deep for the sub to follow. Species Comparisons. -Neither attack nor full display could be elicited from species other than the gray reef shark (Table 1). Silvertip sharks (6 trials) kept ahead of the pursuing sub, sometimes swimming slightly stiffly, and once with a brief pectoral-fin-down display just prior to an accelerated move away. Reef whitetips (11 trials) paid relatively little attention to the sub until it was very close, then maneuvered away among the coral or withdrew with an accelerated burst of swimming or explosive glide. Blackfin reef sharks (2 trials) circled briefly with the sub, then fled with a high-speed burst of swimming. These species comparisons generally agree with those of Starck (Doak, 1975; Starck and Anderson, 1979) for responses to his larger, two-person submersible.

DISCUSSION Exaggerated-swimming Display. -Johnson and Nelson (1973) described agonistic display of the gray reef shark as a graded phenomenon, varying from mild to intense depending on the releasing circumstances. The most intense displays were elicited by rapid diver approach, especially when the shark was cornered, or when the shark itself was on an initial investigatory approach to the diver. Displays of 78 BULLETIN OF MARINE SCIENCE, VOL. 38, NO.1, 1986 lesser intensity were seen under a wider variety of circumstances, including diver presence, diver entering water with a splash, boat anchoring, and during test playbacks of low-frequency, pulsed sounds (Nelson and Johnson, 1972), All of these situations may present some degree of novelty (to the shark), and a potential for an approach-avoidance conflict. Starck, during his own submarine pursuits, recognized degrees of intensity of the display and described it as being oftwo stages. It was initially an "exaggerated lateral swimming motion" beginning as far away as 40 or 50 feet from the approaching sub. Upon closer approach, the shark developed a "distinct arching of the back and downward angling of the pectoral fins." In some cases the display became so exaggerated that "forward swimming motion ceased and the animal assumed a diagonal, head-up position in the water." Johnson and Nelson (1973) noted a similar head-up component as part of the "spiral looping" form of the display. In a shark whose forward motion has nearly stopped, such an upward angle may serve (with some body or tail action) to keep the shark from sinking to the bottom. Starck also described the pattern we call "carouselling" - with the displaying shark turning in towards the sub in such a way that the shark and the pursuing sub end up "making a circle chasing each other." In such carouselling, and in other cases of locomotor slowdown in full display, the shark is in effect showing a distinct reluctance to withdraw from the submarine. In discussing the evolutionary derivation of the gray reef shark threat display, Barlow (1974) suggests that it results from an attack-flee conflict, with its form derived from the act of biting large prey. The head-up and mouth open component is said to be incomplete biting, and the lowered pectoral fins provide resistance to body displacement, as when the head rips laterally through prey. Starck and Anderson (1979) also commented on the similarity of the display to this type of feeding. Barlow, however, points out that while the display may be derived from the feeding act, this "need not mean the threat and subsequent attack are motivated by hunger."

General Aggressiveness. -Gray reef sharks in general are regarded as more aggressive and excitable than the other reef sharks, boldly approaching divers, making quick movements, vigorously competing for bait, etc. Hobson et at. (1961) noted that the Enewetak grays were "far more aggressive" than the other two common lagoon species, the blackfin and the whitetip. During bait-stimulated situations, Hobson (1963) commented that grays in particular seemed "tense and highly responsive to subsequent stimulation." Movements just before attack on bait were "markedly abrupt; (the shark's) body often appeared stiff, with back slightly arched and head extending straight out and slightly upward. The pectoral fins were characteristically pointed noticeably downward." These sharks thus showed at least some components of agonistic display. They may have been directing this behavior towards other sharks (feeding competitors), or perhaps it was triggered by a conflict between approaching the bait and avoiding the relatively novel experimental set-up (divers, apparatus, etc.). During the present study, we also observed this type of aggressiveness in non- pursuit situations. For instance, gray sharks aggregated along the ocean-reef dropoff at West Spit were often attracted to us when scuba diving, circling excitedly with abrupt movements, and showing brief fragments of the agonistic behavior, i.e., stiffened body, lowered pectoral fins. In this situation the sharks, while perhaps in some conflict, were relatively more confident than when being pursued by the submarine. They behaved as if investigating the divers as potential opportunistic NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 79 food, or perhaps they were mobbing the divers as potential predators. They did not exhibit the same form of display as an individual shark does when pursued by the submarine, e.g., the prolonged exaggerated-swimming leading to intense, locomotor-slowed display. Other Submarine Attacks. - To our knowledge, the only other submarine attacks comparable to ours were those of Walter Starck. Prior to our study, he elicited attacks with his larger, two-person submersible-a model called the Shark Hunter, built by Perry Submarine Co., West Palm Beach, Florida. Although apparently not published in scientific format, several accounts have appeared (McNair, 1975; Doak, 1975; Starck and Anderson, 1979), in which six attacks are described, five at Enewetak, and one at Truk. The Starck-Anderson book (written by Anderson from tape recordings provided by Starck) has the most detailed accounts and discussions of these attacks. With one exception, Starck's observations generally agree with our own. He pursued the same four species as we did, getting full display and attacks only from gray reef sharks. About one in five of his chases resulted in attack, of which about half were repeated attacks. One shark struck a total of five or six times, while being chased between each hit. Attacks were very fast, came from the front, side, and rear, and hit all parts of the submersible, e.g., the plexiglas canopy, fiberglass hull, and rear propellor. Starck felt, however, that the sharks showing the most intense, exaggerated displays were less likely to launch an attack than those responding by somewhat milder displays. He described an incident where a smaller gray (1.2 m TL) developed such a severe posture that it was "virtually disabled by its display," but that it did not attack even though the submersible repeatedly circled and crowded it, "moving within 2 or 3 feet," and "practically bumping into it." Starck hy- pothesized that display in this extreme form represents a "powerful inhibition against attacking," a conflict between drives to attack and to flee, with neither being expressed. We did not observe an attack inhibition of this type. All of our sharks which developed the most intense, locomotor-slowed display readily attacked the submersible when it came sufficiently close. We had other attacks from less intensely displaying, more rapidly moving individuals, but we concluded that the most severe, locomotor-slowed (or stopped) form was most likely to lead to attack. It also appears that our pursuits resulted in a higher percentage of attacks (26% overall, 67% for lone, unbaited sharks) than Starck's pursuits, quoted in Doak (1975) as about (20%), and apparently from unbaited, mainly lone individuals. The reason for this difference, if it is significant in view of the small sample sizes, is not clear. Although both studies occurred at Enewetak, there were differences in grouping type, presence of bait, and location on the reef-as well as in the characteristics of the submarines. Starck's chases appear to have been unbaited, mainly lone individuals, along the ocean reef dropoff or in the deeper parts of the passes. His submersible was larger (16' long, 5' high, 8' wingspan), less maneuverable, and more brightly painted (yellow) than our more streamlined, countershaded (blue on top to white below) vehicle. Perhaps his submersible being larger and brighter, appeared more novel and formidable and thus inhibited attack more than our submersible. Perhaps our submersible, being more shark-like in appearance and actions, may have looked to the shark more like a real predator or competitor intent on attack-and thus a more appropriate target for retaliatory attack. 80 BULLETIN OF MARINE SCIENCE, VOL. 38, NO.1, 1986

To some degree this idea fits Archer's (1976) generalized model of aggression and fear in vertebrates. He suggests that either fear or attack can occur in a number of basic situations, one of which is "individual distance intrusion-a moving object coming too close," and this object need not be a conspecific or even another animal. Another is the "presence of a novel object in a familiar area," which includes, but is not limited to, the typical territorial situation. A commonality to all of these situations is a discrepancy between the observed and the expected. Furthermore, the degree of discrepancy affects whether the response will be attack or fear; "at low levels, attack is more likely, but this tends to be replaced at higher levels by fear behavior." To the gray reef sharks, perhaps both submersibles represented a sufficient discrepancy from what they normally expected, in both appearance and actions, from the usual inhabitants of the reef-and thus they sometimes attacked. However, perhaps Starck's sub, being the greater discrepancy, elicited relatively less attack and more fear than did our smaller submersible. Motivation. - Why do gray reef sharks sometimes respond aggressively to divers and submersibles? We now have a fair documentation ofthe releasing stimuli for display and attack, and are beginning to understand some of the other proximate factors involved, but the motivation for these behaviors remains an open question. We discuss below the pros and cons of three general motivational categories for attack: predatory, antipredatory, and social. Since we believe the sharks were not trying to eat the submersible, the real question is whether they regarded the sub as a predatory danger to themselves, as a competitor for food or other resources, or as some combination of both. We should also distinguish between the motivation for the shark's initial investigative approach (or general aggressiveness) and the motivation for its attack on the submarine. The shark's initial approach could very well be a general investigation of the commotion-for the purpose of either obtaining food or defending an area. However, when the shark gets close enough to see the submersible as an unexpected large and novel object, it might then change its motivation to antipredatory defense and begin its threat-attack sequence. This situation temporarily creates an approach-avoidance conflict-a factor recognized by both Johnson and Nelson (1973) and Barlow (1974) as significant in the release of the threat display. According to Barlow, "The prominent display can be expected in sharks brought into conflict between approaching and avoiding. Ap- proach might be stimulated by hunger or by social aggression whereas avoidance will be elicited by potential danger." PREDATORY(attacking to obtain food). The sharks were clearly not trying to eat the submersible or its occupants. There are at least three reasons for this: (1) The submersible does not at all resemble the gray reef sharks normal food (small reef fishes, etc.), and a shark would need very poor discrimination indeed to mistake the submersible for an edible object. (2) Predators do not forewarn prey of inten- tion to attack; thus the exaggerated-swimming display would be inappropriate in a predatory situation. (3) When aroused to a feeding mode-by presence of bait- sharks appear far less inclined to display and attack than during non-feeding trials. In one case a shark at the bait pile was actually rammed by the submersible, but it did not respond aggressively and just swam off, bait in mouth. By comparison, one unbaited displaying shark was contacted from behind (dome touched its tail); it immediately circled around and attacked from the side (Fig. 4). Some years ago, Baldridge and Williams (1969) questioned the generally held assumption that shark attacks were feeding motivated. In their paper "Shark NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 81 attack: Feeding or fighting?," they pointed out that many attacks were ofthe slash type, seemingly to inflict damage, but not to remove flesh. They proposed that as many as 50 to 75% of the cases in the International Shark Attack File could have been non-hunger motivated, in the form of "aggressive behavior directed at victims in an attitude of fighting rather than feeding." While the display-prefaced attacks of gray reef sharks certainly fall into this category, it should be noted that not all were of the slash type. In the present study, at least one attack was clearly a full bite (Fig. 2) and several others probably were-although there may also have been open-mouthed slashes or closed-mouth hits. It is impossible to tell whether a bite or a slash was intended when an open- mouthed shark collides with and glances off the smooth, rounded submersible hull, leaving tooth scratches from one jaw. The attacks on both Jim Stewart (Church, 1961) and Michael deGruy (pers. comm.) were quick bites, although minimal flesh was removed. The attack on Shot Miller was described as a "slash with the upper jaw" (Randall in Ellis, 1975). It thus appears that both bites and slashes do occur in agonistic attacks. Furthermore, Barlow (1974) questioned a motivational dichotomy between biting and slashing, and suggested that a slash may result from a conflict between biting and avoiding a potentially dangerous animal-either prey or rival. He pointed out that a slash attack, as opposed to a longer-lasting bite, would quickly open the distance between the shark and its target, thereby reducing the danger of a counterbite by the target animal. One or more slashes might therefore be a safer way, initially, for a shark to disable large prey. However, gray reef shark stomach contents indicate that they feed primarily on small reef fishes (Wass, 1971; Randall, 1980). While they certainly have the teeth for gouging pieces out of larger prey-and do feed this way opportunistically on large baits-it is unknown to what extent they feed on living prey large enough to pose a danger to them. Gray reefsharks have attacked without obvious exaggerated-swimming display, but these have been in situations where the sharks were highly aroused by opportunistic feeding stimuli, e.g., the sounds and odors of wounded fish (Nelson, 1983). Hobson et al. (1961) described two incidents in which gray sharks appeared in response to spearfishing, then charged a diver who, in each case, succeeded in fending off or kicking away the shark. In the Coral Sea, scuba diver Valerie Taylor was bitten twice within seconds, probably by two individuals, while filming gray reef sharks frenzying over a speared grouper. One bite was on the leg, one on the face. She was wearing a chainmail armored suit and escaped with a few tooth punctures in her chin where the suit did not cover her (Taylor, 1981). Attacks such as these may be feeding motivated, with the excited shark mistaking the diver for food, or the shark may be regarding the diver as a feeding competitor, but not likely as a predator. ANTIPREDATORY(attacking to escape being eaten). There are several reasons why display-prefaced gray shark attacks might represent anti predatory retaliatory aggression: (1) The submarine, although rather novel, looks and behaves more like a hunting predator, e.g., a larger shark, than either a food object or a typical competitor such as a conspecific. (2) Cornering increases the probability of threat and attack, which is a characteristic of antipredatory attacks in general. (3) At least some of the sharks flee rapidly after the attack. A number of prey animals will tum and attack a pursuing predator, especially if cornered (Edmunds, 1974; Ratner, 1975). The concept of "flight and fight distances" applies here, as originally described by Hediger (1950; 1955) and re- 82 BULLETIN OF MARINE SCIENCE, VOL. 38, NO.1, 1986 viewed by McBride (1971). A typical sequence goes as follows: When relatively far the prey animal notices the predator but does not move away. At some closer distance, the prey may "freeze" to avoid detection by the predator. If the predator moves closer still- to a certain "flight distance" - the prey then withdraws. If, by taking the prey by surprise, or by cornering it during a chase, the predator reaches an even closer "fight distance" (critical distance), the prey animal then faces the predator and launches an all-out defensive attack on the predator-hoping to startle it sufficiently to allow escape. Blanchard and Blanchard (1981) described the defensive behaviors of wild rats towards human "predators" who approached and cornered them in a test chamber. As the distance decreased, initial freezing gave way to flight, then to defensive threats (rear up, face person, display teeth), and then at about 15 cm the rat would leap towards and bite the person. In comparing the above anti predatory sequence to the gray shark sequence, one difference stands out. The individuals that attacked did not seem to have a flight distance, only a threat and a fight distance. As the submersible moved towards a relatively distant shark (perhaps 20-30 m away, near the limit of visibility) no responses occurred until the gradual development of exaggerated-swimming display, which became more intense as the submersible neared, eventually leading to attack. Even when not cornered, the sharks made no real attempt to flee, thereby allowing the submersible to get closer. One possibility, perhaps, is that even at first detection the shark regards the submersible as a danger already too close to flee-and instead chooses threat display. Of course the shark could have easily outdistanced the submersible, by either gradual acceleration or an explosive burst of swimming. But the shark does not know this and perhaps perceives the submersible as a large predatory animal that could outswim it, as perhaps some of the larger sharks can. After striking the submersible, some of the attacking sharks did quickly depart the area, a few at high speed, as would be expected for an antipredatory attack. Some, however, did not. In our case, 3 or 4 of the 10 sharks remained at least a short while, still in display. In Starck's attacks, only 2 of the 6 fled after the initial strike, the other 4 remaining long enough to permit further submarine chases and more attacks. The data is thus mixed in this regard, perhaps indicating mixed motivations. There is another possible explanation, also antipredatory, for why gray reef sharks behave aggressively towards divers or submarines prior to the initiation of pursuit. Perhaps they are mobbing the object to harrass it out of the area-as aggregations of songbirds mob hawks and owls. Mobbing is effective mainly when a predatory animal is discovered at a time when it is not an immediate danger, i.e., it is not actively in pursuit of prey. Possibly the sharks regard the diver or submarine as something which may later become a hazard to either themselves or their young. Since pupping in gray reef sharks appears to occur mainly on lagoon pinnacles (McKibben and Nelson, 1986), this may explain why the grays there are particularly aggressive to divers. What natural predators at Enewetak might endanger gray reef sharks, either adults or young? Tiger sharks, Galeocerdo cuvieri, and Galapagos sharks, Car- charhinus galapagensis, are known to feed significantly on other sharks (Wass, 1971; Randall, 1977). Both occur at Enewetak and are considerably larger than gray reef sharks. The silvertip shark is also larger than the gray, though not as large as the above two, and Randall (1980) found a juvenile gray in the stomach of a silvertip from Enewetak. Some predation therefore does occur on gray reef sharks, but its extent is unknown. One must also ask whether the exaggerated-swimming display, presumably NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 83 communicated visually, would be effective in deterring these predators-especially if they hunt at night as they supposedly do. There are no observations of attempted predation on gray reef sharks, although there is one account of a gray shark apparently displaying to an approaching large hammerhead shark, Sphyrna mo- karran, at Rangiroa, French Polynesia (Johnson, 1978). SOCIAL/COMPETITIVE (attacking to defend resources). There are both pro and con arguments for competition as a motivation for display and attack, either in the form of territoriality (defense of an area) or dominance (defense of social rank). In either case, we presume that the submarine (or diver) appears novel enough to the shark to be regarded as a much different "species" -and thus the shark- submersible interactions would be more comparable to interspecific competition than to the intraspecific type. Territoriality is frequently mentioned in the popular press as a reason for shark attack but, in fact, it has not yet been demonstrated in any species of shark (Barlow, 1974; Johnson, 1978). To what degree the evidence is positive depends on exactly what is meant by territorial behavior. In the usual narrow sense, a territorial resident actively defends an area, proclaiming it to potential intruders by threats or other advertisement and, if necessary, aggressively expelling intruders. In the broader sense (Archer, 1976), any aggression towards a "novel object in a familiar area" might be called territorial. This borders on the concept of site-related dominance, e.g., what Brown (1975) calls a "dominion" -an area where an individual exhibits elevated dominance and aggression towards others, but does not attempt to expel them from the area. The one evidence for territoriality-in the broad sense-is that gray reef sharks at times show some apparent site-related aggressiveness. For example, a lone shark (usually on a flats reef or lagoon pinnacle) will sometimes make a moderately fast investigative approach to a newly arrived diver, exhibiting mild to moderate display without any apparent provocation by the person. It behaves as if it was a resident treating the diver as an intruder, although other interpretations are possible. A shark in this situation can be relatively easily stimulated to attack. We also know from telemetry studies of diel movements that lagoon pinnacle sharks are more day-to-day site-attached than their more nomadic ocean-reef counterparts (McKibben and Nelson, 1986). McNair (1975) made some observations at Enewetak that relate to this point. He was able to identify certain gray reef sharks by scars or other markings, and observed that there was a "pre- dictable difference in temperament when the same individual is encountered in different areas. On one piece of reef, the shark may be aggressive, for example, while half a mile away it may be docile or shy, indicating that the first place was 'home.' After seeing this pattern repeat itself with the same individual several times, territoriality seems the only logical conclusion." In this study, we were unable to make the repeated observations necessary to either support or differ with McNair's conclusion. What is needed is a telemetry study to allow test encounters with individual sharks in different parts of their known home areas. There are several arguments against territoriality: (1) the lack of observed display or attack directed towards conspecifics, (2) that pursuit by the submersible is necessary to release attack, not just presence, and (3) that flight is often the post- attack behavior. The apparent lack of aggression towards other gray reef sharks is negative evidence because animals usually defend territories more readily against their own species than against different species (especially the much-different species which the submarine must represent). This is because conspecifics usually represent the most serious competitors for the resources in question, in this case 84 BULLETIN OF MARINE SCIENCE, VOL. 38, NO. I, 1986 probably food. When interspecific territoriality does occur, as in the coral reef damselfishes studied by Myrberg and Thresher (1974), the resident still more readily defends against its own species. When presented with an intruder consisting of a fish in a glass bottle, Eupomacentrus planifrons attacked conspecifics at greater distances from its residence that congenerics, and the latter at greater distances than less related species. We therefore think it unlikely that an object the size and appearance of the submarine would be treated by the shark as a food competitor. A shark the size of the submersible would probably not compete for the same type of prey as the gray reef shark. A possibility that must be considered is that the lack of aggression between conspecific sharks is because of mutual familiarity-that they belong to the same social group. Would a strange conspecific shark from the outside be recognized and attacked? This seems unlikely, considering the variable composition of gray reefshark aggregations, the distances these sharks normally travel, and the number of other individuals they must encounter. Telemetered individuals from the outer reefs, for example, made location changes of up to 15 km/day, and the more site- attached lagoon individuals still covered relatively large areas at night (McKibben and Nelson, 1986). The behavior of the shark towards the pursuing submarine is also evidence against a territorial motivation for display and attack. A territorial resident normally regards just the presence of an intruder as a releaser for threat and attack, i.e., it does not require pursuit by the intruder. Furthermore, if one attack fails to expel the intruder, the resident attacks again and again. In our trials, however, the shark usually departed after attacking even though the submarine had not moved away. If any form of territoriality does exist in gray reef sharks, one would not nec- essarily expect it in all individuals at all times. If there are territorial and non- territorial phases, we would expect the lone individuals from the fiats or lagoon pinnacles to be the territorial ones, with the loosely aggregated ones from the dropoff to be less so. But more evidence is needed to clarify this point. Several authors (Barlow, 1974; Klimley, 1974; McNair, 1975) have suggested that the sharks might be defending an individual distance-a moving "personal sphere" about themselves, rather than a fixed territory. While this could be a proximate explanation of what releases aggression-another way of saying that attack results if something comes too close-it does not address the question of antipredatory vs. competitive motivations. As McBride (1971) pointed out, the antipredatory "fight distance" is the interspecific equivalent of the intraspecific "personal distance." Furthermore, in gray reef sharks, whether or not the submarine is attacked within such a distance depends on several factors, e.g., the type of "intruder," what the intruder is doing, and what the shark is doing, such as feeding. Johnson (1978), in reviewing possible functions of the agonistic behavior, suggested that the display occurs "under such a variety of circumstances, it seems no single motivation can adequately account for all situations." He further argues that many, perhaps most, of the incidents of exaggerated-swimming display can be accounted for on the basis of a general interspecific dominance-such as for the purpose of intimidating food competitors of other species. He related one incident from the Cook Islands in which a speared fish was about to be eaten by a moray eel. "A gray reef shark appeared and was observed to direct a display posture toward the eel, which persisted in its efforts to feed on the fish. The shark abruptly terminated the display and attacked the eel, leaving a noticeable white NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 85 slash." In bait situations where grays and reef whitetip sharks are competing, the grays do appear dominant-in the sense of usually getting the food first, and occasionally causing give-ways or withdrawals by the whitetips (Nelson and John- son, 1980). However, in nearly all cases, the gray shark competitive behaviors seem to be of the scramble type (each individual directing its efforts at the bait itself), not of the contest type involving threats or attacks on each other. Domi- nance is thus expressed subtly, if at all, in the usual competitions for bait that we and others have observed. The stiff, abrupt movements that Hobson (1963) reported for gray sharks feeding on bait may have had this function, or they may have been in response to the novel experimental situation. In any event, the moderate to intense displays such as given to the submarine are rarely, if ever, seen in competitive feeding situations where sharks are interacting only with other sharks. Other Species. -Of the four species of sharks tested, why does only the gray reef shark "stand and fight" rather than withdraw at the approach of the submarine? Assuming an anti predatory motivation, a speculation can be made based on the habitats of the four species. Whitetip sharks stay close to the reef bottom, readily enter caves, and perhaps can seek refuge there from predators. Blackfin sharks typically inhabit very shallow areas, such as the upper reef flats, and perhaps could elude large predators by entering water too shallow for them to follow. Silvertip sharks inhabit the deeper, unprotected waters along the ocean-reef dropoffs, but perhaps their larger size makes them less vulnerable to predation. The smaller gray reef sharks, also inhabiting open unprotected areas, are thus perhaps the most vulnerable to predation and in need of an aggressive antipredatory mech- anism. Stark speculated on the function of these differences by postulating that the gray reef sharks are "in a far more competitive intraspecific feeding situation than the whitetip or the blacktip (blackfin)." The whitetip, foraging in reef caves and crevices, and the blacktip, chasing small fishes in very shallow water, probably feed more or less alone, without as much direct competition from conspecifics as the gray sharks which group in more open water. He also felt that the grays, being very responsive to distress stimuli, may be more dependent on sick or wounded prey which might attract several sharks and thus form a highly competitive situation (Starck and Anderson, 1979). Although the gray reef shark is the only species known to exhibit the very conspicuous, intense form of exaggerated-swimming display, and to follow it with attack, a number of apparently aggressive behaviors have been seen in other sharks (Nelson, 1981). Hobson (1964) noted the early stages of display-the laterally exaggerated swimming with head swinging-in both the gray reef shark and in the Galapagos shark, Careharhinus galapagensis, at Cocos Island in the eastern Pacific. He suggested that this behavior helps the shark maintain visual contact with objects directly behind them, such as divers following them. He also noted the connection between this behavior (in gray sharks) and the attack on Jim Stewart, although he did not suggest a communicative threat function. Myrberg and Gruber (1974) noted the similarity between the "hunch" posture of bonnethead sharks, Sphyrna tiburo, and the postural part (arched back, snout up) of the gray shark display. The exaggerated swimming part was not seen. The hunch, although seldom seen, occurred in potentially agonistic situations, such as when a newcomer shark or diver-observer entered the pool where the sharks were kept. They also reported the hunch in captive blacknose sharks, C. aero notus, and free- ranging silky sharks, C. faleiformis, in related circumstances. Probable threat 86 BULLETIN OF MARINE SCIENCE, VOL. 38, NO. I, 1986 behaviors were also noted in the lemon shark, Negaprion brevirostris, in response to a diver dressed in apparel resembling a killer whale. The shark responded by agitated swimming in tight circles and figure-eights, and repeated openings and closings of the mouth (Klimley in Nelson, 1981). Whether such behavior could lead to attack is unknown, but the lemon shark has made retaliatory attacks when sufficiently provoked, such as when speared, harpooned, or chased by small boats.

CONCLUSIONS The gray reefshark is the most aggressive reefshark at Enewetak, and apparently elsewhere in the Indo-Pacific. It will readily threaten and subsequently attack divers or submersibles which approach or pursue too closely. The primary releasing stimulus is an "oriented" pursuit, especially if the shark is cornered to some degree. The other common sharks of the area, the blackfin (blacktip) reefshark, silvertip shark, and reef whitetip, do not attack in this manner. When pursued by a submarine, they withdraw without obvious display. The threat display of the gray reef shark-called exaggerated-swimming display-is a graded behavior, varying in intensity from mild (laterally exaggerated swimming) to intense (back arched, snout up, pectoral fins down; sometimes with rolling, head-up inclination, and locomotor slowdown). It can be initiated at distances as far as 15 m or more from the approaching submarine, and persists for a relatively long time. It has not been observed between con specifics. An apparently related agonistic behavior-tense, abrupt swimming with some pectoral-fin lowering-occurs under different, non-pursuit circumstances such as competitive feeding on bait, or close circling of divers by investigating aggregations of sharks. It does not have the prolonged laterally exaggerated swimming component. All of these agonistic behaviors occur in situations which potentially put the shark in a conflict between approaching (for investigating potential food or rivals) and avoiding the novel, potentially dangerous submarine, diver, or other apparatus. Attacks usually follow high-intensity display, and are launched when the pursuing submarine closes to a critical distance (ca. 2-3 m in our trials). The attacks are high-speed strikes, usually directly at the target object. However, the attacking shark will sometimes rapidly circle around to the side or behind and strike from there. At contact, the sharks are usually (possible always) open-mouthed, and can inflict either bites or slashes, leaving deep tooth scratches in the submarine. After hitting the submarine, some of the sharks rapidly flee the area, while others remain for a while still in exaggerated-swimming display. The latter individuals may attack again if chased. Experiments suggest that display/attack readiness is related to several proximate causative factors: Grouping type (lone individuals more so than aggregated ones), presence of bait (unbaited individuals more so than baited ones), and location on the reef (sharks from the flats and lagoon pinnacles more so than from the ocean- reef dropoft). The most likely sharks to display and attack appear to be unbaited lone individuals from the flats and lagoon pinnacles. The motivation for the display/attack-in the context of the submarine pursuits-is most likely antipredatory retaliatory defense, with the shark regarding the submarine as an immediate "predatory" danger to itself. The motivation for a shark's initial approach in an unbaited situation is prob- NELSON ET AL.: ATTACKS BY GRAY REEF SHARKS 87 ably general investigation, with the shark regarding the commotion caused by the submarine or diver as either a potential feeding opportunity, an intruding competitor, or possibly as a predator to be harrassed in the mobbing sense. The usefulness of the gray reef shark's agonistic behaviors under natural conditions remains uncertain, but is probably multifunctional-of value in both antipredatory and competitive social situations. Gray sharks appear generally dominant over other reef species during competitive feeding. There is no evidence that gray sharks are territorial in the strict sense (aggressively expelling intruders), although this possibility cannot be ruled out. There is some evidence for territorial- like behaviors such as site-related dominance, but this needs confirmation.

ACKNOWLEDGMENTS

We are indebted to many persons associated with the Mid-Pacific Research Laboratory (Enewetak) of the University of Hawaii for help in the financial and logistical aspects of this study. Special thanks are due P. Allen, M. deGruy, G. Long, C. Visser, and E. Reese. Additional financial support was provided by two documentary film companies, Don Meier Productions, Chicago, IL; and the Hessischer Rundfunk, Frankfort, West Germany, and we thank underwater cameramen R. Nelson and A. Giddings for taking film footage of some of the shark attacks. We also acknowledge the Office of Naval Research through contract NOOO14-77-CoO113, under project NR-I 04-062, for supporting the program of shark- behavior research of which this study is a part.

LITERATURE CITED

Archer, J. 1976. The organization of aggression and fear in vertebrates. Pages 231-298 in P. P. Bateson and P. H. Klopfer, eds. Perspectives in ethology, Vol. 2. Plenum Press, New York. Baldridge, H. D., Jr. and J. Williams. 1969. Shark attack: feeding or fighting? Military Medicine 134: 130-133. Barlow, G. W. 1974. Derivation of threat display in the gray reef shark. Mar. Behav. Physiol. 3: 71-81. Blanchard, R. J. and D. C. Blanchard. 1981. The organization and modeling of animal aggression. Pages 529-563 in P. F. Brain and D. Benton, eds. The biology of aggression. Sijthoffand Noordholf, Rockville, Maryland. Brown, J. L. 1975. The evolution of behavior. W. W. Norton and Co., New York. 761 pp. Church, R. 1961. Shark attack. Skin Diver Mag. June: 30-31. Doak, W. 1975. Sharks and other ancestors. Hodder and Stoughton, Sydney. 333 pp. Edmunds, M. 1974. Defence in animals: a survey of anti predator defences. Longman Group, Ltd., New York. 357 pp. Ellis, R. 1975. The book of sharks. Grosset and Dunlap, New York. 320 pp. Fellows, D. P. and E. A. Murchison. 1967. A noninjurious attack by a small shark. Pacific Sci. 21: 150-151. Hediger, H. 1950. Wild animals in captivity: an outline of the biology of zoological gardens. But- terworth Scientific Publications, London. 207 pp. --. 1955. Studies of the psychology and behaviour of captive animals in zoos and circuses. Criterion Books, New York. 166 pp. Hobson, E. S. 1963. Feeding behavior in three species of sharks. Pacific Sci. 17(2): 171-194. -_. 1964. Sharks increasing visual field. Underwater Naturalist 2(3): 29. --, Mautin, F. and E. S. Reese. 1961. Two shark incidents at Eniwetok Atoll, Marshall Islands. Pacific Sci. 15(4): 605-609. Johnson, R. H. 1978. Sharks of Polynesia. Editions du Pacifique, Papeete, Tahiti. 170 pp. -- and D. R. Nelson. 1973. Agonistic display in the gray reef shark, Carcharhinus menisorrah, and its relationship to attacks on man. Copeia 1973: 76-84. Klimley, A. P. 1974. An inquiry into the causes of shark attacks. Sea Frontiers 20(2): 66-75. McBride, G. 1971. Theories of animal spacing: the role of flight, fight, and social distance. Pages 53-68 in A. H. Esser, ed. Behavior and environment. Plenum Press, New York. McKibben, J. N. and D. R. Nelson. 1986. Patterns of movement and grouping of gray reef sharks, Carcharhinus amblyrhynchos, at Enewetak, Marshall Islands. Bull. Mar. Sci. 38: 89-110. McNair, R. 1975. Sharks I have known. Skin Diver Mag. 24(1): 52-57. Myrberg, A. A., Jr. and S. H. Gruber. 1974. The behavior of the bonnethead shark, Sphyrna tiburo. Copeia 1.974: 358-374. 88 BULLETINOFMARINESCIENCE,VOL.38, NO.1. 1986

and R. E. Thresher. 1974. Interspecific aggression and its relevance to the concept of territoriality in reef fishes. Amer. Zool. 14: 81-96. Nelson, D. R. 1981. Aggression in sharks: is the gray reef shark different? Oceanus 24(4): 45-55. --. 1983. Shark attack and repellency research: An overview. Pages 11-74 in B. J. Zahuranec, ed., Shark repellents from the sea: new perspectives. AAAS Selected Symposium Series. Westview Press, Boulder, Colorado. --- and R. H. Johnson. 1972. Acoustic attraction of Pacific reef sharks: effect of pulse inter- mittencyand variability. Compo Biochem. Physiol. 42(lA): 85-96. --- and ---. 1980. Behavior of the reef sharks of Rangiroa, French Polynesia. National Geographic Soc. Research Reports 12: 479-499. Randall, J. E. 1977. Contribution to the biology of the whitetip reef shark (Triaenodon obesus). Pacific Sci. 31: 143-164. ---. 1980. A survey of ciguatera at Enewetak and Bikini, Marshall Islands, with notes on the systematics and food habits ofciguatoxic fishes. Fish Bull. 78(2): 201-249. Ratner, S. C. 1975. Animal's defenses: fighting in predator-prey relations. Pages 175-190 in P. Pliner, L. Krames and T. Alloway, eds., Nonverbal communication of aggression. Plenum Press, New York. 196 pp. Starck, W. A. and A. Anderson, Jr. 1979. The blue reef. Alfred A. Knopf, New York. 259 pp. Taylor, V. 1981. Anti-shark armor that works. Skin Diver Mag. 30(10): 23-27. Wass, R. C. 1971. A comparative study of the life history, distribution and ecology of the sandbar shark and the gray reef shark in Hawaii. Ph.D. thesis, Univ. Hawaii. 219 pp.

DATEACCEPTED: August 12, 1985.

ADDRESS: Department of Biology, California State University, Long Beach. California 90840.