METHODS

I attempted to document all attacks from 1 January 1890 through 31 December 1990. I searched in both scientific and popular literature, including hunter's magazines and newspapers, for reports of unprovoked attacks by wild cougars on humans. During 1897-1925, the magazine Outdoor Life encouraged letters on this topic from readers and followed most such letters with the results of the editor's inquiries to local authorities. C. Hart Merriam of the U.S. Biological Survey prepared a file (now in Dep. For. Resour. Manage., Univ. Calif., Berkeley) of newspaper and magazine clippings on attacks during 1909-1932. I found additional leads in monographs written by Young (1946), Barnes (1960), and Anderson (1983). I also obtained information from wildlife agencies in each of 12 western states (those at least partly west of 105° west longitude) and 2 provinces (British Columbia and Alberta). In each of these states and provinces, I also contacted 1 or more biologists studying cougars.

I define an attack as an incident in which the cougar bit, clawed, or knocked down a human; only attacks were tabulated. I excluded maulings by captive cougars and cases in which a person (e.g., a cougar hunter) deliberately approached or harassed a wild cougar. I define a near-attack as a cougar advancing toward a person at close range without making contact, or crouching beside a trail as if to pounce. Reports of near-attacks were relatively common, but usually impossible to verify. It was also difficult to determine if the cougar would have attacked if the person had not taken action. Therefore they were not tabulated. However, I qualitatively evaluated credible near-attacks for human behaviors that prevented physical contact.

I include a report only if it was verified by a newspaper or other published account that included statements from medical personnel or law enforcement officers, or if it was a report of the state or provincial wildlife management agency or the National Park in which the incident occurred.

Adult cougars ( 2.0 years old) were classified as underweight if they were more than 2 standard deviations below the mean body mass for their sex and subspecies (or the nearest subspecies with data) using body masses summarized by Anderson (1983:21). In general, adults were judged underweight if males and females weighed <45 and <30 kg, respectively. Males and females recorded as 12-17 months old or simply as "yearlings" were judged underweight if they weighed <30 and <20 kg, respectively (Robinette et al. 1961). When age was estimated to be 18-23 months, cutoffs of 37 and 25 kg were used. Thus only markedly underweight animals were so classified. Percentages (e.g., the proportion of victims that were children) were computed as the proportion of individuals for which the appropriate attribute (e.g., age) was recorded.

RESULTS

Temporal and Geographic Trends
I documented 9 fatal attacks and 44 nonfatal attacks resulting in 10 human deaths and 48 nonfatal injuries
(Table 1). The greater number of victims occurred because there were 2 victims in each of 5 attacks.

I believe that I discovered all fatal attacks reported since 1890 and all nonfatal attacks since 1970. During 1970-1990, there were 31 nonfatal and 5 fatal attacks (6.2:l). A lower ratio of nonfatal to fatal attacks in the period from 1890 through 1969 (3.25:l) suggests that some nonfatal attacks before 1970 escaped my attention. If the ratio of nonfatal : fatal attacks during 1890-1969 was comparable to that for 1970-1990, then I did not document about 12 nonfatal attacks during the earlier period.

Cougar attacks have clearly increased during the last 2 decades, despite some possibly undocumented nonfatal attacks during the early years. There were more fatal attacks during the last 20 years (5) than during the previous 80 years (4). Also, C. Hart Merriam documented only 3 attacks (1 nonfatal, 2 fatal) in the 23 years from 1909 to 1932 versus 36 attacks (31 nonfatal, 5 fatal) reported during the last 21 years.

There were 8 attacks during December-February, 15 attacks during March-May, 21 attacks during June-August, 6 attacks during September-November, and 3 with no month recorded. This seasonal pattern of attacks may reflect increased human activity in wildlands in warmer months. The diel pattern of attacks also resembles the diurnal activity pattern of humans rather than the nocturnal activity pattern of cougars (Beier, unpubl. data). For the 32 cases in which hour of day could be determined, there were 6 attacks during 0630-1130, 15 attacks during 1130-1630, 11 attacks during 1630-2130, and no attacks during the remaining 9 hours. Of the other 21 attacks, 2 occurred at night, 6 occurred during daylight, and in 13 cases the records do not indicate time.

Twenty of the 53 attacks (38%) occurred on Vancouver Island (British Columbia), a 30,000-km² island with 300,000 human residents. There were 10 attacks in mainland British Columbia, 5 in Texas, 4 in California, 3 each in Alberta and Colorado, 2 each in Arizona, Montana, and Washington, and 1 each in New Mexico and Nevada.

Victims
Thirty-seven of 58 victims (64%) were children (16 years old); the other 21 (36%) were 24 years old. Using 5-year age classes, the modal age class of known-age victims was 5-9 years (19 victims). Group associations were known for 54 victims. Of 37 children, 35% were alone, 43% were in groups of children, and 22% were accompanied by adults. Eleven of 17 adult victims were alone at the time of attack. Of those victims classified as alone (i.e., out of sight of other humans), 6 of 13 children and 4 of 11 adults were within earshot of other humans. Except for 1 adult and 1 child who died of probable rabies resulting from a single attack Table 1: (Morgan Hill, Calif., 1909), all fatalities were children unaccompanied by adults.

Fifteen of 54 victims (28%) were attacked close to a house or cabin, or just outside a motor home within a developed recreational area. These attacks would have been visible if someone had been watching from the home or vehicle. In several cases the house was occupied but the occupants were unaware of the attack occurring outside. The 3 most intrusive cases involved a cougar crashing through the window of an isolated cabin to attack a telephone lineman (Table 1: Kelsey Bay, B.C., 1951), a cougar who attacked a 2-year-old boy in the garage of his home in a village of 250 persons (Table 1: Lewis, Colo., 1970), and a cougar attack on a 6-year-old boy in a residential area (Table 1: Hinton, Alta., 1962). Not included in these 15 cases were 2 attacks on persons in tents or sleeping bags (Table 1: Big Bend, Tex., 1978; Strathcona Park, B.C., 1972: Hurford case); the records do not indicate whether the campsites were within developed campgrounds. In yet another case a boy was attacked while riding a bicycle (Table 1: Holberg, B.C., 1983).

Behaviors That Might Prevent an Imminent Attack
Most victims (24 of 32 for which data exist, excluding 2 who were injured after they came to the assistance of a companion) did not see the cougar before being clawed or bitten; thus, no preventive action was taken. The other 8 victims, by definition, failed to prevent the attack. In all near-attacks, actual attack was averted, presumably because the person had time to react. In some near-attacks, the cougar was shot as it approached. In most other near-attacks, aggressive responses by the human (shouting, swinging a stick, waving arms above the head, throwing rocks) clearly deterred the cougar from carrying out an attack. I discovered only 1 credible near-attack in which the intended victim escaped by a panicked flight. In that case, a 16-year-old boy fled after encountering a cougar at 25 feet. The cougar was gaining ground rapidly when the boy's boot fell off and the cougar attacked and ate the boot. This story was supported by the presence of boot fragments in the stomach of the cougar when it was shot an hour later (Vancouver Province: 18 Jun 1966).

Behaviors That Did Not Avert Attacks
Eight victims did have at least a few seconds between seeing the animal and physical contact. Three of these victims watched the cougar quietly, apparently uncertain of its intentions or too surprised to act. Four victims attempted to run away, but only 1 appeared to benefit from this response. In that case a man who had stopped to repair his bulldozer was rushed by a cougar; he had to run only a few steps to mount his rig and was lightly clawed as he did so (Table 1: Squamish, B.C., 1951). In at least 2 cases, running appeared to stimulate the cougar to select the victim out of a larger group (Table 1: Sooke, B.C., 1985; Big Bend, Tex., 1987). Tracks in the snow indicate that a 13-year-old boy tried to outrun a cougar for about 100 m before being killed (Table 1: Olema, Wash., 1924). In the eighth case, the victim initially stood her ground, shouted, and waved her arms, but when the 2 cougars continued to approach her, she scrambled up an embankment and climbed a tree. The cougars climbed after her, and 1 of them raked her leg once. She then struck 1 animal with her foot and the other with a stick. The cougars immediately left the tree and abandoned the site shortly afterward
(Table 1: Boulder, Colo., 1990).

Human Behavior After Attacks Began
For only 29 victims is it meaningful to examine the victim's response after physical contact occurred. In the other cases, such details were not recorded (23 victims), the child victim may have been in shock because his head was in the cougar's jaws (4), the cougar fled at once (l), or the victim was able to drive away at once (1). The 4 cases in which a child's head was in the cougar's jaws (possibly inducing shock) seem equivalent to playing dead (see below). In all such cases, the cougar continued to bite the victim's head or drag him by the head until another person came to assist. Twenty-six of the 29 remaining victims fought back with bare hands, a stick, a knife, a jacket, or a rock. These efforts usually succeeded in repelling the attack. In several cases, even children unassisted by adults were able to repel the cougar by fighting back. Most victims also shouted loudly, and loud shouts apparently did intimidate the cougar; the noise also often brought other persons to assist the victim.

I documented only 3 cases in which alternate responses were tried. Two cases involved "playing dead." A 9-year-old boy said that he had been advised to play dead if he ever encountered a wild animal. When a cougar pounced on him as he was walking out of a lake, he followed this advice. However, the cougar continued to bite him and drag him away until his father kicked gravel at the cougar. The cougar then rushed at his 7-year-old sister; she did not play dead but screamed, causing the cougar to turn and run (Table 1: Glacier Park, Mont., 1990). In the second case (Table 1: Campbell Lake, B.C., 1972), the older brother of the 8-year-old being attacked by a cougar "shouted for him to be quiet--to play 7 dead--but the cougar started carrying him off into the bush. Then the mother came up screaming at the cougar. It dropped the boy and made off" (Vancouver Sun: 27 Jul l972).

In the third case (Table 1: Kootenay Park, B.C., 1970) a cougar attacked a 50-year-old solo hiker, clawing her arm and knocking her down. As she fell she set up her backpack as a shield, faced the cougar and (in her words) "began talking to her the way you would if you were trying to soothe a dog or cat." She continued soothing talk for 30 minutes until she heard other hikers nearby. Then she yelled for help, an approaching hiker blew a whistle, and the cougar retreated. This strategy apparently did prevent the cougar from continuing its attack. However, it was a loud shout and a loud whistle that eventually caused the cougar to retreat.

Offending Animals
Although most offending cougars were promptly shot and killed, few data on these animals were recorded. Twelve of 31 offending cougars were estimated to be 12-23 months of age; 1 was under 12 months of age. Seven of 9 attacking yearlings, 7 of 17 attacking adults, and 3 of 4 attackers of unknown age were markedly underweight.

Only 2 offending cougars were documented to have had a disease or physical disability. One was probably rabid and caused the single double fatality (Table 1: Morgan Hill, Calif., 1909). This is the only documentation of apparent transmission of rabies from a cougar to humans; both victims died of the disease, not from the physical injuries (Storer 1923). The other diseased cougar (Table 1: Cowichan Lake, B.C., 1916) had cataracts.

One yearling cougar attacked and bit a person in Big Bend National Park 4 months after being chased, treed, drugged, and radio-collared (Table 1: Big Bend, 1987). Aversive conditioning was tried deliberately on another cougar; that animal, shot with rock salt at close range after a near-attack, returned to aggressive behavior (without physical attack) only 2 weeks later and had to be removed (C. M. Fleming and R. Skiles, Big Bend National Park, unpubl. data).

DISCUSSION

Temporal and Geographic Trends
Schmidt (J. E. Schmidt, Mountain lion attacks on humans, unpubl. rep., Wildl. Ext., Univ. Calif., Davis, 1986) double-counted several cases in tabulating 17 fatal attacks in North America during 1890-1986. Schmidt's 1986 list of nonfatal attacks was not comparable because it included provoked attacks and encounters lacking physical contact.

Each year in the U.S. there are about 12 human deaths resulting from over 5,000 bites by rattlesnakes (Crotalus spp.), 40 deaths due to bee (Hymenoptera) stings, and 3 deaths due to bites of black widow spiders (Latrodectus spp.) (Weiss 1990). Dogs annually kill 18-20 people and inflict suture-requiring injuries on 200,000 US. residents (Sacks et al. 1989). In a single recent year (1979) there were 86 U.S. deaths due to lightning strikes (Natl. Center Health Stat. 1984:33-35). Thus cougar attacks are much rarer than other hazards from animals or nature. Nonetheless, these attacks have increased in the last 2 decades, probably because of increased numbers of cougars and humans during that time. Cougar populations have increased markedly in recent years in British Columbia (Hebert 1989), California (Mansfield 1986), Colorado (Anderson and Tully 1989), Nevada (Stiver 1989), Texas (Russ 1989), and Wyoming (Shorma 1989). Although other states have not estimated population trend (Smith 1989), cougar populations throughout the West probably increased during 1965-1980 as each state and province changed the legal status of the cougar from bountied predator to game species subject to controlled hunting or (in California) full protection. Simultaneously, human use of wildlands has grown, increasing the potential for encounters.

It is also possible that the decreased persecution of cougars, along with the establishment of large wilderness areas free of hunting, may allow cougars to habituate to humans as a non-threatening part of their environment. However, there is no evidence that cougars are more likely to attack humans in unhunted areas. Indeed, 57% of the attacks occurred in British Columbia, where about 200 cougars are killed annually by hunters and predator control agents (Hebert 1989).

There is no compelling explanation for the striking concentration of attacks on Vancouver Island. One speculative line of reasoning (which I raise but do not advocate) stems from the observation that several prey species taken by cougars in other parts of North America are absent from Vancouver Island. The absent species include porcupine (Erethizon dorsatum), Virginia opossum (Didelphis virginiana), cottontail rabbits (Sylvilagus spp.), coyote (Canis latrans), bobcat (Lynx rufus), badger (Taxidea taxus), and spotted and striped skunks (Spilogale putorius, Mephitis mephitis) (Burt and Grossenheider 1976). Although large ungulates are the cougar's main prey throughout its range, porcupine are important in many areas (Russell 1978). In the Santa Ana Mountains of California, porcupine are absent but opossum, cottontail, and coyotes make up over 30% of prey items and about 8% of prey biomass (Beier, unpubl. data). Because a cougar typically hunts for several days after consuming 1 deer until killing the next deer (Beier, unpubl. data), small prey may be important in sustaining a cougar between deer kills. A lack of small prey may be especially critical for a yearling animal less proficient at taking deer, and may contribute to the increased attacks on humans on Vancouver Island.

Victims
Children are more vulnerable than adults, even though the proportion of humans in cougar habitat that are children is unknown. Children unaccompanied by adults are probably also at increased risk. The proportion of children in cougar habitat that are not supervised by adults is unknown, but is probably smaller than the 78% of child victims not in sight of adults when attacked. This increased vulnerability is especially clear when only fatal attacks are considered. Except for the adult and child who died of probable rabies, all fatalities were children who were either alone or accompanied by other children. The increased vulnerability of children to attacks can be minimized by keeping them within sight of an adult, who may not prevent but can repulse an attack.

Appropriate Human Behaviors
Aggressive responses appear to be effective in averting an imminent attack. The records do not support the notion that one should avoid loud shouting or avoid eye contact with the cougar when attack appears imminent. Running away from an aggressive cougar seems particularly futile unless one is only a few steps from the safety of a home or vehicle.

An aggressive response may also be effective in causing a cougar to retreat after it initiates physical contact. There is no empirical support for the efficacy of "playing dead" or curling up in a fetal position once a cougar attack has begun.

Offending Animals
The data suggest that underweight yearlings may have a propensity to attack humans. Juveniles (0-24 months old) compose under 50% of most wild cougar populations (Seidensticker det al. 1973; Ashman et al. 1983; Hemker et al. 1984; Logan et al. 1986; Beier, unpubl. data), implying that on average yearlings (12-23 months old) are less than 25% of the population. However, nearly 40% of the attacking cougars were classified as yearlings. At this age a young cougar increasingly hunts without maternal assistance, and by 14-24 months of age it moves into a new and often unfamiliar home range (Seidensticker et al. 1973, Ashman et al. 1983). Under these stresses, some yearlings may have difficulty capturing wild prey. The low body mass of most yearling attackers suggests that this may be an important factor. Two of the underweight yearling attackers also had porcupine quills in their throats. One attacker, a 14-month-old radio-collared male, weighed 29.5 kg, 2.3 kg less than at his capture 3 months earlier; he smelled strongly of skunk odor and was not yet independent of his dam (R. Skiles and C. M. Fleming, Big Bend Natl. Park, Tex., unpubl. data).

Management Implications
Attacks by cougars are rare but increasing. It is unlikely that sport hunting will remove enough cougars to reduce the risk. The high aesthetic value of cougars may preclude reduction of cougar populations by other means (e.g., bounties, control programs). Managers of wildlands, in consultation with legal staff, might consider using information reported here to offer advice that may reduce risk to the human visitors.

In 6 of the cases tabulated here, the offending cougars were promptly killed by wildlife conservation officers but no data were recorded; in other cases data were estimated instead of measured, or necropsy records could not be located. In the future, all cougars shot after attacking humans should receive a careful postmortem examination, and the results should be filed so as to make them accessible.

SUMMARY

I examined historical records of unprovoked attacks by cougars on humans in the U.S. and Canada during the last century (1890-1990) to determine historical trends and characteristics of victims and offending cougars. There were 9 attacks resulting in 10 human deaths and at least 44 nonfatal attacks. Attacks on humans increased markedly during the last 2 decades, during which cougar numbers and human use of cougar habitats also increased. Most victims (64%) were children; the modal age class was 5-9 years. Of 37 child victims, 35% were alone, 43% in groups of children, and 22% were accompanied by adults; 11 of 17 adult victims were alone at the time of attack. The data suggest that yearlings and under-weight cougars were most likely to attack humans. Aggressive responses on the part of intended victims may avert an impending attack and repel an attack in progress.

Acknowledgments.--I am most indebted to J. Bone, who provided leads to most of the Vancouver Island cases. The Orange County Chapter of Safari Club International generously defrayed the publication costs. This research was funded by the County of Orange, California, the California Department of Fish and Game, and California Agricultural Experiment Station Project 4326-MS. I thank all those persons who provided me with information I may not otherwise have gained; these include L. Fitzhugh, C. Fleming, J. Gonzales, H. Hash, D. Janz, M. Jalkotzy, P. Koepp, T. Malmsbury, D. R. McCullough, D. Pemble, J. Rieck, D. J. Spalding, M. Sanders, R. Skiles, S. J. Stiver, and D. Turner. F. Lindzey, R. C. Belden, N. Holler, and an anonymous reviewer made constructive comments on the manuscript.

LITERATURE CITED

• ANDERSON, A. E. 1983. A critical review of literature on Duma (Felis concolor). Colo. Div. Wildl. Spec. Rep. 54. 91pp.
• _____. AND R. J, TULLY. 1989. Status of the mountain lion in Colorado. Proc. Mountain Lion Workshop 3:19-23.
  
• ANONYMOUS. 1909. Further details regarding the California lion attack. Outdoor Life 30:598-600.
• _____. 1917. Children attacked by cougar. Outdoor Life 38:193-194.
• _____. 1925. An old question. Outdoor Life 46:113.
• ASHMAN, D. L., G. C. CHRISTENSEN, M. L. HESS, G. K. TSUKAMOTO, AND M. S. WICKERSHAM. 1983. The mountain lion in Nevada. Nev. Dep. Wildl., Reno. 75pp.
• BARNES, C. T. 1960. The cougar or mountain lion. The Ralton Co., Salt Lake City, Ut. 176pp.
• BURT, W. H., AND R. P. GROSSENHEIDER. 1976. Field guide to the mammals. Third ed. Houghton Mifflin Company, Boston, Mass. 289pp.
• FINLEY, W. L. 1925. Cougar kills a boy. J. Mammal. 6:197-199.
• HALL, F. S. 1925. Killing of a boy by mountain lion. Murrelet 6:33-37.
• HEBERT, D. 1989. The status and management of cougar in British Columbia. Proc. Mountain Lion Workshop 3:ll-14.
• HEMKER, T. P., F. G. LINDZEY, AND B. B. ACKERMAN. 1984. Population characteristics and movement patterns of cougars in southern Utah. J. Wildl. Manage. 48:1275-1284.
• HIBBEN, F. C. 1975. Terror at Arroyo Seco. Petersen's Hunting. March 1975:22-25.
• HOWARD, C. B. 1915. An instance of a mountain lion's attack upon a boy. Outdoor Life 36:162-163.
• LOGAN, K. A,, L. I. IRWIN, AND R. SKINNER. 1986. Characteristics of a hunted mountain lion population in Wyoming. J. Wildl. Manage. 50:648-654.
• MANSFIELD, T. M. 1986. Mountain lion management in California. Trans. North Am. Wildl. and Nat. Resour. Conf. 51:178-182.
• NATIONAL CENTER FOR HEALTH STATISTICS. 1984. Vital statistics of the United States 1979, Vol. 2. Mortality, Part A-5. U.S. Gov. Print. Off., Washington, D.C.
• ROBINETTE, W. L., J. S. GASHWILER, AND 0. W. MORRIS. 1961. Notes on cougar productivity and life history. J. Mammal. 42:204-217.
• RUSS, W. B. 1989. Status of the mountain lion in Texas. Proc. Mountain Lion Workshop 3:30-31.
• RUSSELL, K. R. 1978. Mountain lion. Pages 207-225 in J. L. Schmidt and D. L. Gilbert, eds. Big game of North America. Stackpole Books, Harrisburg, Penn. 512pp.
• SACKS, J. J., R. W. SATTIN, AND S. E. BONZO. 1989. Dog-bite fatalities from 1979 through 1988. J. Am. Medical Assoc. 262:1489-1492.
• SEIDENSTICKER, J. C., JR., M. G. HORNOCKER, W. V. WILES, AND J. P. MESSICK. 1973. Mountain lion social organization in the Idaho primitive area. Wildl. Monogr. 35. 60pp.
• SHORMA, G. 1989. Status of the mountain lion in Wyoming. Proc. Mountain Lion Workshop 3:38-39.
• SMITH, R. H., editor. 1989. Proc. Mountain Lion Workshop 3:l-88.
• STIVER, S. J. 1989. Status of mountain lions in Nevada. Proc. Mountain Lion Workshop 3:26-29.
• STORER, T. I. 1923. Rabies in a mountain lion. Calif. Fish and Game 9:45-48.
• WEISS, R. 1990. Researchers foresee antivenin improvements. Sci. News 138:360-362.
• YOUNG, S. P. 1946. History, life habits, economic status, and control. Pages 1-173 in S. P. Young and E. A. Goldman. The puma, mysterious American cat. The Am. Wildl. Inst., Washington, D.C. 358pp.

Received 26 September 1990.
Accepted 8 May 1991.