ABNORMAL ANIMAL BEHAVIOR AND THE PREDICTION OF EARTHQUAKES


Introduction

The prediction of earthquakes is a relatively new branch of seismology. Just two decades ago, earthquake prediction was not even considered to be serious science; rather, it was left to astrologers, mystics, and religious zealots (Press & Siever, 1978). The emergence of earthquake prediction as a truly scientific discipline has taken place as a result of a program of fundamental observations begun in the late 1960's and supplemented in 1973 and 1976. The research effort has from its inception in the United States consisted both of attempts to develop a systematic catalog of precursory phenomena through field measurements and a program of fundamental studies of the physical basis for the occurrence and nature of earthquakes and of the pre-earthquake failure process (Raleigh, 1980).
Earthquake premonition by animals, while having a long history and persistence in literature, does not seem to fit into the rational world of science, and as a phenomenon, it has the added disadvantage in that it cannot be examined at will. Skepticism of anomalous animal behavior before earthquakes by many scientists in the Western world is largely based upon this unfavorable image, as any scientist interested in studying this problem risks not only his professional reputation, but also any chance of getting research funds (Tributsch, 1982). Beginning


in the mid-1970's, however, the perspective of the possible connection between animals and earthquakes began to change, when it was revealed that in 1975, China successfully predicted a major earthquake, using observations of abnormal animal behavior, as well as geophysical and geochemical precursors. The entire population of Haicheng, a large city, was evacuated before a magnitude 7.3 earthquake struck, saving thousands of lives. This imminent prediction was the first prediction of a major earthquake in recorded history. U.S. seismological teams visiting China were impressed by China's methods and accomplishments in earthquake prediction, although the Chinese admitted some failures among their successes, and much more study still needs to be done. In 1976, the U.S. Geological Survey sponsored a conference on "Abnormal Animal Behavior Prior to Earthquakes", held in Menlo Park, California. This conference discussed and reviewed accounts of unusual animal behavior and the various geophysical precursors which might stimulate the animals. In 1979, a second U.S.G.S. conference was held at the University of Texas in Galveston. This conference introduced experimental data on the effects of earthquake-related geophysical and geochemical phenomena on animals, under controlled conditions. A survey, summary, and analysis of the literature concerning abnormal animal behavior before earthquakes, and its possible usefulness in predicting future earthquakes, is the focus of this paper.

CHAPTER 1
ANECDOTAL EVIDENCE

Sources of Data

The oldest recorded observations of abnormal animal behavior connected with earthquakes date back some 3000 years, a continuous record containing an estimated 50 millions words in the official dynasty records of China. However, their usefulness to modern scientists is somewhat limited by the sheer volume of material, and by the style of ancient classical Chinese writing - it is not certain whether the observed unusual behavior took place before or after the earthquakes (Lee et al., 1976).
Although many examples of unusual animal behavior have been reported from nearly all parts of the world, many scientists have regarded them as anecdotal folklore concocted by untrained observers, and not to be taken seriously (Lee et al., 1976). Nearly all reported accounts of abnormal animal behavior are second-hand, that is, told by laymen observers to scientists after the earthquake occurred (termed "retrospective recollections"), and thus are subjective interpretations which may be inaccurate or exaggerated. Attempts have been made by Lott et al. (1977, 1979, 1980) to introduce more rigorous controls on data collection methodology; behavioral specialists collected eyewitness accounts of several earthquakes with carefully worded questions (the respondents were not pressured to answer).
A different direction of investigation was started by biologists, biophysicists, and animal behaviorists. Their laboratory studies of the effects of geophysical and geochemical phenomena known to be present prior to earthquakes provided new experimental data of animal sensitivities to these stimuli under controlled conditions. (Many of these studies were initially concerned with isolated phenomena not necessarily related to earthquakes, such as the effect of sound waves on fish.)
A major difficulty of relating abnormal animal behavior to earthquakes is that unlike mechanical measuring devices, the behavior of animals is not well understood. Also, the behavior of the many types of animals in the data is not uniform, either between individuals of one species or between groups of different species. Finally, abnormal animal behavior may also be caused by non-seismic "noise" factors such as weather, magnetic storms, disease, insects, or other conditions causing stress in animals (McClellan, 1980).
Therefore, any meaningful analysis of the literature must consider these questions:
1. How reliable is the data?
2. Can non-seismic noise factors be eliminated from the data set of a particular earthquake?
3. What is the "normal" behavior of animals under non-earthquake conditions?
4. Can a particular geophysical stimulus be shown to cause the observed behavior?
5. What are the sensory thresholds (limits of detection by the senses) of the animals under study to different stimuli?
6. How many individual animals within a population tend to be more sensitive to the earthquake phenomena?


Anecdotal Accounts

By far, the largest volume of data concerning abnormal animal behavior before earthquakes consists of innumerable anecdotal observations made by laymen, classified as folklore or legends (Lee et al., 1976). The earliest accounts date back several thousand years to the time of Pliny the Elder in Rome, and to the Chou Dynasty (circa 1000 B.C.) in China. However, folklore accounts have long been ignored as irrelevant by scientists, as expressed by Richter (1958) in Elementary Seismology :
"...The legendary material of seismology includes many stories of horses and other domestic animals being uneasy during the hours preceding a large earthquake. It is impossible to judge such evidence scientifically. If there is any explanation beyond coincidence and incomplete reporting, it probably rests on the occurrence of small foreshocks, not noticed by humans, but disturbing to sensitive animals. During earthquakes, animals are seen to react as they do to almost any sudden and unexpected event." (Foreshocks are defined as small tremors that commonly precede a larger earthquake by an interval ranging from seconds to weeks, and that originates at or near the focus of the larger earthquake.)
Rikitake (1976) believes that while not all of these legends may be true, it is nevertheless important for scientists to look into things which might contain some truth, without being biased. Lee et al. (1976) compiled a summary of literature on this subject, (some of which was translated from foreign journals not readily available in the Western world), which they organized into three geographical sections: China, Japan, and Western countries, as the history and approach towards studying the problem is somewhat different in each of these three regions.


Chinese Data on Animal Behavior Anomalies

Modern records suggest that China endures the most severe seismic activity on earth. Because the majority of major tremors strike densely populated areas, their impact has been more lethal than in most other countries. China averages six quakes of at least 6.0 on the Richter scale each year, whereas the United States averages two or three per year, most of them in the Aleutian Islands of Alaska. Thus, the philosophy of China's government is to include as much information as can be gathered to aid in prediction, including the abnormal behavior of animals (Mead, 1976).
Earthquakes have played an important role in China's political history. The emperors of the ancient dynasties were said to have a "mandate from Heaven" to rule, on the condition that the emperor was benevolent to his subjects. The regimes of many a corrupt emperor were toppled by the civil unrest caused by the destruction of a great earthquake, as the suffering peasants symbolically interpreted the earthquake as a sign from Heaven that "the emperor had lost his mandate," and revolted. The belief that great earthquakes herald a change in the nation's leadership is still widely held in China even during modern times, as the 1976 Tangshan earthquake happened to occur just months before the death of Chairman Mao Tse-tung. The Tangshan earthquake killed more than half a million people; unfortunately, a short-term prediction could not be made, although a long-range earthquake forecast had been made for that area of China. For political as well as practical reasons, the Chinese have been using animal anomalies as one of the key indicators to predict earthquakes. With over 80% of the population living in farming areas in close association with animals, this approach has great public appeal (Lee et al., 1976). In 1966, spurred by two severe earthquakes which struck Hsingtai, the government of China mobilized its citizenry and declared a "People's War" on earthquakes, by way of a program of scientific research, education of the general population, and the construction of a nationwide seismic monitoring network (employing 10,000 professional seismologists and 20,000 technicians and amateur observers) overseen by a National Bureau of Seismology, with the goal of successfully predicting future earthquakes. The effort has been described as the "Chinese equivalent of the American Apollo space program (Mead, 1976)."
The following accounts of abnormal animal behavior before earthquakes in China, Japan, and the Western world will concentrate mostly upon descriptions, and not on their possible causes, which will be discussed in a later chapter:
Historical records said that farmers in China could tell that something was vastly wrong in the earth beneath them when normally placid horses reared up and ran away. Dogs howled continuously for no apparent reason. Fish leaped from the water into the air. Animals which normally hid from humans, such as snakes and rats, suddenly appeared in droves (Mead, 1976). Cattle, sheep, mules, and pigs refused to enter their corrals. Frightened pigeons flew continuously and refused to nest. Rabbits raised and twitched their ears spasmodically, jumped aimlessly and bumped into things. Most dramatically, hibernating snakes left their burrows early, sometimes in the dead of winter, only to freeze to death on the icy ground (Lee et al., 1976).
From Lee et al., (1976) the following observations of abnormal animal behavior before an earthquake were recorded at the Tientsin Zoo by the Chinese government:
A giant panda refused to play and was observed holding its head with its forearms, crying strangely. A tiger became depressed and sluggish. Loaches (a type of bottom-dwelling fish), leeches, and turtles suddenly swam up and down and rolled over continuously. A Tibetan yak refused to eat, and rolled over continuously on the ground. Water-loving birds such as ducks and swans stayed on land far away from the shore, avoided the water, and pointed their feet towards the sky. Among large zoo animals, those with young were observed to be more sensitive to earthquake precursors than those without young.
More observations were recorded in historical Chinese documents (Lee et al., 1976): sudden calls of pheasants, abundant catch of fish, disappearance of rats, fish jumping from the water onto the shore, bees building their nests in low places, cocks crowing at night, disappearance of cats, dogs barking loudly and continuously, birds singing at night, normally placid catfish swimming actively, and snakes and rabbits moving their homes.
In addition, the Geological Bureau of China in 1975 reported that these animals also behaved anomalously before earthquakes (Lee et al., 1976): tigers, wolves, deer, eels, frogs, donkeys, geese, monkeys, sparrows, eagles, and insects. People living in areas near the earthquake epicenter observed that burrowing animals such as rats and snakes, etc. were more sensitive than animals living above the ground, and small animals appeared to be more sensitive than large animals. According to the survey of four major regions of China by the combined team from the Biophysics and Zoology Institutes of the Academia Sinica, preliminary results (dated 1975) show that 58 species of wild and domestic animals have relatively reliable anomalous reactions before earthquakes.
About one and a half months before the February 4, 1975 Haicheng earthquake (magnitude 7.3 on the Richter scale), hibernating snakes came out of their underground burrows and froze to death. One or two days before the earthquake, pigs refused to eat, climbed up the walls of their stalls, and butted the doors. Small pigs bit each other, and a dozen or so small pigs chewed off their tails and ate them. Cows became restless, pawed the ground, fought with each other and butted horns. Deer at a deer farm suddenly ran madly, trying to escape from their pens. One deer reportedly broke two of its legs while running away in a panic (Lee et al., 1976). Birds at a municipal aviary attempted to relocate their nests; while carrying their eggs, many of them fell to the ground and broke. Many caged birds, such as sparrows, attempted to fly inside their cages, as if trying to escape; many died from injuries suffered from striking the sides of their cages (Deng and Jiang, 1981).
To summarize the above accounts, the Group of Earthquake Research of the Institute of Biophysics, Academia Sinica (1979) made the following analysis of the data:
Table 1A, Table 1B
Figure 1
Figure 2

1. Most animals in the seismic area become increasingly restless, and a number of them fall into a state of anxiety.
2. These features may appear from a few minutes to as long as ten days before the earthquake, but usually the precursor time is within 24 hours of the quake.
3. These phenomena appear to have a non-random and non-uniform regional distribution, occurring mostly in particular belts of the seismic area and in certain regions. These places correlate to some degree with the strike of active faults, as well as with the bends, branches or terminal points of faults. It appears that they tend to be in the epicentral area or high- intensity zones of the impending earthquake (Fig. 1).
4. These remarkable changes in animal behavior generally precede a destructive earthquake of magnitude 5 or greater.
5. Although there is great variety in the cases of unusual behavior, generally speaking, this behavior falls in to the category of increased restlessness - being startled, extremely nervous, and panicky, as if the animals were on the brink of meeting their natural enemies. A small number of animals may become depressed or apathetic.
6. Macroscopic (detectable without instrumentation) changes in animal behavior constitute only one of many precursors of earthquakes. Some characteristics, such as precursor time, regional distribution, animal species, total number of events per day, etc. are linked with origin-time, the epicentral location and the magnitude of the coming quake. So far as the time of occurrence is concerned, it can be used as a means of extremely short-term forecasting. Combined with data from other disciplines, animal observations may be of some value in making an overall judgment.
7. Anomalous animal behavior has also been noticed during the earthquake, as well as before.
8. The number of animals of any given species that behave abnormally prior to earthquakes is not necessarily a high proportion of the total population (usually, only a fraction of a given species shows unusual behavior). Some show no anomalous behavior; on the other hand, some anomalous animal behavior may not be followed by an earthquake. Furthermore, it must be noted that much of the data were collected after the earthquakes had occurred.
9. The unusual behavior is not necessarily related only to earthquakes. Non-seismological factors can sometimes cause similar behavior. Although such interfering factors have been investigated, it is almost impossible to eliminate them completely.


Japanese Data on Animal Behavior Anomalies

Ancient Japanese legends attributed earthquakes to giant catfish moving underground (Fig. 3), a tradition said to have started sometime in the eighteenth century. After that time, some examples of catfish associated with earthquakes were reported in historic documents. In particular, many colored wood-block prints with catfish legends appeared after the 1855 Edo earthquake. During the late 1800's, J. Milne described some examples of unusual behavior of fish before earthquakes, but no remarkable scientific work was done until the 1920's (Lee et al., 1976).
The Kanto earthquake of 1923, which destroyed much of Tokyo, led several seismologists to pay attention to the problem of animal anomalies. For example, in 1932, Hatai and Abe investigated the responses of the catfish, Parasilurus asotus to earthquakes. During that same year, Hatai, Kokubo and Abe studied earth currents in relation to the response of catfish. Terada in 1932 demonstrated an apparently good correlation between the number of fish caught and the number of earthquakes felt near the time of the 1930 Ito earthquake swarm. More than 20 examples of unusual behavior of fish and sea life were reported by people on the coast of Japan, as well as several
Figure 3
examples of unusual behavior of rats and birds. Musha in 1935 summarized a great deal of data on unusual animal behavior and earthquakes (Lee et al., 1976).
After the 1930's, however, research on animal behavior in Japan gradually diminished. Possible reasons are: (1) World War II intervened, (2) no large destructive earthquake has occurred in Japan since 1949, and (3) no scientific progress in the mechanism of animal anomalies has been advanced. More recently, Japanese scientists renewed their interest in unusual animal behavior and earthquakes, especially after the Chinese success in predicting the Haicheng earthquake of February 4, 1975 (Lee et al., 1976).
Most of the animal anomalies reported in Japan are concerned with fish. This is not surprising, because 80% of Japanese earthquakes occur in the oceanic areas surrounding Japan, and fish are a staple food for the Japanese. Some of the more dramatic accounts regarding fish and marine life describe sudden changes of habitat: crabs and catfish crawling out of the water onto land, deep sea fish swimming upwards to surface waters and being caught by fisherman, dolphins appearing offshore, and the discovery of benthic (bottom-dwelling) diatoms and mud in the stomachs of sardines, which normally ingest planktonic diatoms (Lee et al., 1976).
During the period 1913 to 1916, Omori conducted experiments to determine the ability of pheasants to detect earth tremors. Working in a quiet house where he could hear pheasants crowing in a neighboring garden, he took note of the time of every perceptible earthquake and comparing it with the crowing of the pheasants and also later checking the tromometer, an instrument for recording minute earth tremors. In 11 out of 23 cases, the pheasants were actually more sensitive than Omori; they either crowed before the scientist felt the quake, or crowed when he felt no quake but found that one had been registered by the instruments. Passing vehicles which shook the ground did not cause the pheasants to crow (Anderson, 1973).
Since most animal anomalies observed in Japan are based on second-hand reports, their accuracies vary greatly. Japanese reviewers of this subject tend to collect all available information regardless of accuracy (Lee et al., 1976).


Western Data on Animal Behavior Anomalies

The semi-scientific literature of the Western world contains many anecdotal reports of supposed animal responses to forthcoming earthquakes. Most of the reports on record were made during the nineteenth century; they are so far separated culturally and in time from the more objective present-day reports that little credence can be given to them (Lee et al., 1976). Perhaps the best known of all animal-earthquake stories dates back to the early twentieth century; the dogs of San Francisco were reported to have barked the night before the great earthquake struck the city in the early morning of April 18, 1906 (Anderson, 1973). Cats also became very excited during the San Francisco earthquake. Some ran around wildly with their fur standing on end; others hid in dark corners or behaved abnormally in other ways. In some cases, they disappeared for several days after the quake. During the aftershocks, cats seemed to register the tremors before people did. They cowered in fear and ran. Cows were reported fleeing in a panic before the earthquake; some were seen lowing and mooing during the earthquake. Horses snorted, trembled, and galloped in a frightened panic (Tributsch, 1982).
Von Hentig (1923) summarized the following observations which were reported without reference to particular earthquakes in the older literature:
1. Dogs bark frantically, howl, growl, or whine. Dogs and foxes are restless, whine and howl, and rush into the open; or they are lethargic, somnolent, and either hide or stay by their masters. Cats draw back ears, bristle fur, and mew pitifully.
2. Pigs and other animals show signs of suffering for 10 days prior to an earthquake; pigs bite each other like dogs.
3. Swallows abandon nests and seek refuge under roofs on the eve of an earthquake.
4. Fish flee from the banks toward the center of rivers; fish migrate strangely.
5. Mice, rats, moles, lizards and snakes leave burrows and wander restlessly.
6. Alligators vocalize loudly, leave the water and seek refuge in the woods.
Werner (1974) mentions these observations in one of the few Russian reports available on abnormal animal behavior before earthquakes: ants pick up their eggs and move out of anthills in a mass migration, pheasants crow, goats and antelopes refuse to go into indoor pens for months before earthquakes, while tigers and other big cats do the same several weeks before earthquakes. Additionally, it was reported that a dog saved its master's life by dragging her out of the house before a large 1966 earthquake struck, and the natives of Kamchatka were said to observe the behavior of bears to predict volcanic eruptions. There were also reports of a dramatic surge in complaints of heart patients prior to a 1948 earthquake. To Russian scientists, this suggested that human hearts can somehow detect small variations in geophysical factors such as static and magnetic fields caused by the buildup toward an earth tremor, and implied that the same natural device may also warn animals.
In April and May of 1976, there were earthquakes of magnitude 7 in the Uzbek Soviet Republic. One day before the catastrophe, great swarms of bats were observed flying around the day. (Bats are normally nocturnal.) Such observations have also been made in Turkey (Tributsch, 1982).
Tributsch (1978) reported the following unusual animal behavior which occurred before the May 6, 1976 earthquake epicentered in Friuli, Italy:
(1) Deer formed flocks, (2) cats left houses in the village, (3) mice and rats left their hiding places, (4) fowl refused to roost, (5) cattle panicked in their barns, (6) dogs barked without apparent reason, (7) birds uttered calls at unusual times, (8) caged birds attempted to fly, and (9) a caged bird kept wetting its plumage.
According to Tributsch (1978), this type of behavior was essentially identical to that reported from China.
Tributsch (1982) summarized the essential information from the following folk legends, which, if valid, would indicate an imminent earthquake:
(1) when four-legged animals show great nervousness without apparent reason (Peru, Venezuela),
(2) when birds become excited and give forth unusual calls (Peru),
(3) when dogs howl in unison (Chile),
(4) when roosters crow persistently at night (Italy, Chile),
(5) when roosting chickens leave theirs roost and cackle excitedly (Chile),
(6) when wild animals appear to be tame and intimidated (Chile),
(7) when normally abundant flies suddenly disappear (Venezuela),
(8) when snakes leave their lairs and flee outdoors (Cuba),
(9) when birds flock together, fly high, and circle conspicuously (Greece),
(10) when bears leave their winter lairs prematurely (Kamchatka, U.S.S.R.),
(11) when cats cry nervously, run around houses excitedly, and then flee outdoors (Italy, Chile).
In many respects, the above list resembles accounts reported in China (Tributsch, 1982). Tributsch (1982) believes that folk wisdom about earthquake prediction by animals deserves respect. Many of the popular sayings are, after all, not just the fruits of individual observations, but have been repeatedly submitted to tests through generations. It would be senseless and pointless for people who depend so much on nature to observe the animals for no apparent purpose or benefit. It seems unlikely that China's thousands of years of tradition, and the earnest efforts and observations of hundreds of millions of people would be devoted to the apparition of a phantom. Yet to Western scientists, the lack of explanation of the mechanisms behind these phenomena seems sufficient enough reason to reject the folklore altogether, despite evidence that some of these observations can be understood and used to predict earthquakes (Tributsch, 1982).

CHAPTER 2
PATTERNS IN THE ANIMAL BEHAVIOR PHENOMENA


Context of Abnormal Animal Behavior

From the voluminous historical catalog of anecdotal observations of animal behavior, certain essential patterns may be found. Most unusual animal behavior is not totally abnormal for the animal's repertoire, but has been observed in the species under other circumstances (Table 2); in other words, the behavior is only "abnormal" when it occurs out of its normal context. Since earthquakes are relatively rare events, it is unlikely that animals have evolved a specific instinctual response to pre-earthquake signals; thus, it is speculated that animals may misinterpret various geophysical changes and respond in a confused, but frightened manner. The behavior sometimes resembles startle movements in response to a sudden stimulus; in other cases, it resembles the orientation movements that animals use to investigate or avoid a stimulus (Buskirk et al., 1981). The behavior reported in post-earthquake interviews resembles fear or escape reactions and ranges from mild response to bizarre behavior (Lott et al., 1980). Investigators familiar with the full range of behavior for a species will often recognize that reported "abnormal" behavior is actually species-typical behavior which may be triggered by a variety of stimuli not necessarily related to earthquakes. For example, behavior such as cats
Table 2
hiding and pigs biting their tails also appears during times of stress unrelated to geophysical changes (Buskirk et al., 1981).
Much of the behavior resembles that reported for animals before geophysical events other than earthquakes, such as thunderstorms or sudden volcanic eruptions. For instance, dogs sniffed and pawed at the earth two to four days prior to the eruption in 1955 of Mt. Kilauea, Hawaii (Buskirk et al., 1981). Cows abandoned their pastures two weeks before Volcan Arenal, Costa Rica erupted in 1968, and dogs barked incessantly for minutes to hours prior to the 1965 eruption at Tall, Philippines (Anderson, 1973).

Time Patterns of Abnormal Animal Behavior

For a few major earthquakes, reports of unusual animal behavior have been widespread enough that it was possible to study the timing of the behavior (Fig. 4). For the Tangshan, China, earthquake (M = 7.8, July 28, 1976), reports of fish, rodents, and wolves were cited as early as a month or two before the event. In the epicentral area of Tangshan, over 70% of the reported incidents took place within 1 day before the earthquake. Most incidents (70%) also occurred in areas which were to experience the great Mercalli intensities (Buskirk et al., 1981). Although in other earthquakes, unusual behavior of cows and horse has been noted in the seconds or minutes prior to the shock (Lee et al., 1976; Tributsch, 1978), at Tangshan only 10% of the reported incidents for horse, donkeys, and cows occurred
Figure 4
immediately before the earthquake. At Tangshan there were reports of earthquake lightning and changes in telluric currents the days before the earthquake, but there were no reported foreshocks (Buskirk et al., 1981).
As early as a month prior to the 1975 earthquake in Haicheng, China, unusual behavior in fish, rodents, and snakes was observed. However, most of the unusual behavior took place within two days of the main shock. Numerous foreshocks and obvious groundwater changes also occurred 1 or 2 days before this event (Buskirk et al., 1981).
Rikitake (1978) considered the temporal relationship between geophysical and behavioral precursors for the Izu, Japan, earthquake (M = 7.0, January 14, 1978), shown in Fig. 5. Nearly all physical precursors measured for this and other earthquakes occurred at least 2 days prior to the main shock, while most of the 129 behavior incidents happened within 24 hours of the earthquake (Fig. 6). Most behavioral precursors coincided with a swarm of foreshocks a few hours prior to the main shock (Rikitake, 1978).
Rikitake's statistical analysis (1978) attempted to compare the nature of animal precursors to those obtained by geophysical and geochemical methods. He classified these precursors within three time frames: long-range (measured in years), moderately short-range (measured in days), and extremely short-range (measured in hours). The conclusions of the statistical analysis are outlined as follows:
Figure 5, Figure 6
1. No long range animal precursors having times on the order of years (long-range) have been reported.
2. Most of unusual animal behavior seems to belong to moderately short-range precursors which have a mean precursor time amount to 0.4 day, if they are indeed precursors at all. If this conclusion is true, it can be said that animal precursors are complementary to geophysical ones, that is, the distribution of animal precursors exhibits a maximum around 0.4 days. Meanwhile, the frequency distribution of geophysical precursors indicates a minimum around that value. Although it is not certain whether or not the above conclusion is a mere coincidence, it seems worthwhile to pay attention to abnormal animal behavior which might have a spectrum of precursor times different from that of geophysical precursors (Fig. 7, 8, and 9).
3. It seems likely, as far as the existing data are concerned, that animals sometimes present extremely short-range premonitory signals on the order of a few hours, just like some geophysical disciplines such as land deformation, resistivity, underground water, and so on (Fig. 10 and 11). No significant difference in precursor time distribution between geophysical and animal precursors is found as far as the extremely short-range precursors are concerned.
Figure 7
Figure 8
Figure 9
Figure 10, Figure 11
4. The existing data on unusual behavior prior to an earthquake indicate that animal precursors cannot be ruled out although most data are based on non-scientific observations. Although no firm proof that animal behavior reflects signals forerunning an earthquake has been established, it is not fair to rule it out in the search for earthquake precursors (Rikitake, 1978).
The great variability of animal behavior before earthquakes is apparent from post-earthquake interviews such as the standardized studies of Lott et al., (1979). Individual animals of the same species, even when located in the epicentral area, did not respond in the same way before the 1977 Willits, California, earthquake (M = 4.7) (Lott et al., 1979).
This variability has two sources: behavioral differences among individual animals, even within the same species, and geophysical differences between earthquakes (Buskirk et al., 1981). Earthquakes have individual characteristics: modifications occur in acoustic waves, air pressure levels, tilt of the land, electrical conductivity, electromagnetic fields, electrostatic discharges, gas emissions, groundwater level, and temperature, for example, but these events do not always appear in identical patterns. The time of onset, frequency, duration, and magnitude of each may vary, as well as which particular physical events are present. Often, the physical changes are so small that they fall within the range of normal fluctuations regularly experienced by the animal (Shaw, 1977). In addition, it is clear from comparative studies (Lott et al., 1980) that unusual behavior is observed before some earthquakes but not others. The same type of interview study was conducted after four California area earthquakes which occurred during the late 1970's: Willits, November 22, 1977 (M = 4.7); Landers, March 15, 1979 (M = 5.5); Coyote Lake, August 6, 1979 (M = 5.4); and Mexicali, October 19, 1979 (M = 6.9). All four had strike-slip faults with rather shallow epicenters located in rural areas. Only the Willits earthquake had a significant number of behavior precursors (Lott et al., 1980).
Figure 12 (Buskirk et al., 1981) displays available data for 36 earthquakes on four continents. The following generalizations may be made about the reported unusual animal behavior:
1. Most, but not all, of the animal behavior precursors occur close to the epicenter within 1 or 2 days of the earthquake. The species primarily reported are domestic mammals, such as dogs, probably because of their close association with humans, and animals of commercial importance, such as horses and chickens (Buskirk et al., 1981).
2. Some, but not all, of the behavior precursors occur within a few minutes before the earthquake (Tributsch, 1978, 1982). For these precursors, it is difficult to dismiss the hypothesis that the animals are sensing the vibrations of the P waves arriving from the earthquake,
Figure 12
while humans sense only the latter and stronger S waves or surface waves (Buskirk et al., 1981). (Animals such as horses and pheasants responded about 5 to 10 seconds before humans felt the earthquake in an aftershock sequence in Chile.) However, phenomena that may coincide with the P wave arrival are placed in a separate category from earlier behavior precursors by some researchers (Rikitake, 1976; Lott et al., 1979).
3. A few of the behavioral precursors actually are reported days to weeks before the earthquakes, and some of these occur at a considerable distance from the epicenter (Buskirk et al., 1981). The animal species most often mentioned in these reports are fish (Rikitake, 1976) and rodents (Lee et al., 1976).

CHAPTER 3
OVERVIEW OF POSSIBLE SENSORY MECHANISMS IN ANIMALS


Several problems complicate an analysis of reports concerning anomalous animal behavior prior to earthquakes: the variability of animal behavior, the unreliability of human observations, and the existence of uncontrolled physical factors such as weather (McClellan, 1980). These problems are so significant that scientists have taken a serious look at the phenomena only recently. During the late 1970's and early 1980's, several types of investigations were undertaken to document the nature of animal behavior responses (Buskirk et al., 1981). These investigations include systematic post-earthquake interviews (Lott et al., 1979), a network of observers reporting by telephone (Otis and Krautz, 1980), and biological activity monitors on individual animals under controlled conditions in a field laboratory (Kenagy and Enright, 1980; Lindberg et al., 1981).
Because anecdotal observations are unable to explain the mechanisms of what the animals might be sensing prior to an earthquake, Frey (1980) advocates the use of psychophysical and psychophysiological tests on the animals in the laboratory, where geophysical events and stimuli might be recreated. In the final analysis, the animal is simply a measuring device. The measuring devices used by humans, such as tilt meters, are nothing more than extensions of our own sensory systems which are acting as measuring devices. From the anecdotal reports, it would appear that the animals are measuring something that we are either not measuring or, are interpreting the stimuli differently than we are (Frey, 1980).
Buskirk et al. (1981) compared available data on geophysical precursors to the sensory thresholds of humans and other animals (obtained from laboratory studies), particularly those mentioned in the anecdotes. Much of the published biological research relevant to this area of study is unfamiliar to most geophysicists, and the gap between the biological and geological disciplines was bridged somewhat by the U.S.G.S. Conferences on Abnormal Animal Behavior Prior to Earthquakes, held in 1976 and in 1979. For example, the second U.S.G.S. conference reviewed biological and biophysical research on: detection of sound and vibration by birds (Kreithen, 1980); vibrotactile responses in animals and man (Verrillo, 1980); low frequency vibration detection in fishes (Fay and Patricoski, 1980); seismic wave detection by fishes (Frohlich and Buskirk, 1980); electric and magnetic field detection by animals (Kalmijin, 1980; Medici, 1980; Frey, 1980); changes in air ions on animals before earthquakes (Yost, 1980); and earthquake odor detection by animals (Moulton, 1980).
Buskirk et al. (1981) cited a number of geophysical phenomena that have been observed prior to earthquakes might also stimulate unusual animal behavior, including:
(1) sound with an intensity and frequency outside the range of human perception,
(2) variations in local magnetic or electric fields,
(3) ground vibrations or foreshocks,
(4) changes in groundwater level,
(5) electromagnetic waves, and
(6) the release of gases usually trapped beneath the surface.

Animals probably do not sense some types of reported geophysical precursors, including ground tilt, variations in the velocities of P and S waves, changes in electrical resistivity, and gravity anomalies (Buskirk et al., 1981).



CHAPTER 4
ANIMAL SENSITIVITIES TO GEOPHYSICAL STIMULI


Buskirk et al. (1981) attempted to correlate the biological literature on animal sensitivities to geophysical stimuli commonly reported prior to earthquakes. To strengthen their correlations, they considered (1) the background level of the stimulus, (2) data from the geophysical literature suggesting that the stimulus may occur occasionally as an earthquake precursor, and (3) biological research (in summary form) on the sensitivity of animals to that type of stimulus.


Sounds and Vibrations (Foreshocks)

Before discussing these studies, it is pertinent to mention some basic background information regarding acoustics and human hearing:
Sound is created by the movement of energy through a medium, such as a gas (air), liquid (water), or solid (rock), which causes it to vibrate as a pressure wave. The speed of vibration (usually measured in cycles per second, or Hertz, abbreviated as "Hz") is called the frequency. The size of a wave cycle, or wavelength, has an inverse relationship with the frequency. The measurement unit of sound pressure level (loudness) is usually the decibel, abbreviated "dB." Because the perception of sound is not linear, but logarithmic, a 10 decibel increase in sound roughly corresponds to double the loudness. It has been demonstrated that human hearing is not linear in its perception of the frequency spectrum. While the range of human hearing extends from 20 - 20,000 Hz, the ear is more sensitive to the mid-range frequencies (from above 100 Hz to about 10,000 Hz), and weak at the lower and upper frequencies. Very low frequency sounds (below 20 Hz) are usually not heard, but felt, while sensitivity to high frequencies drops off sharply past 15,000 Hz; with age, this loss of sensitivity is more pronounced. Also, there exists a relationship between the frequency range of sound, the energy required to produce it, and its ability to travel through a medium. High frequency sounds have shorter wavelengths, and thus require less energy to produce them. Because they tend to travel in a straight line, their relatively low energy is easily absorbed by obstacles (high frequency seismic sound waves would be attenuated by surrounding rock). Low frequency sounds are more energetic, travel in a radiating pattern (rather than a straight line), are less prone to attenuation, and can travel great distances. Seismic waves commonly have a frequencies between 0.01 and 10 Hz (Buskirk et al., 1981).
Biologists have shown that many species of animals possess a greater frequency range of hearing (down to as low as 1 Hz, or as high as 100,000 Hz), and a greater sensitivity in the ranges where human hearing is poor. If sounds are indeed generated before earthquakes, animals which are sensitive to either very low or very high frequencies may be hearing something that humans cannot detect (Buskirk et al., 1981).


Background levels
Animals that can sense low-amplitude sounds or vibrations are exposed to many types of earth noise below 100 Hz in frequency. If unusual animal behavior is caused by foreshocks, then the seismic signals must somehow differ from background noise because of its frequency, amplitude, or pattern. Several kinds of earth noise have been measured. Microquakes have a predominant frequency lower than 1 Hz. Infrasound (ultra-low frequency sound below 50 Hz) generated by thunderstorms has the highest sound pressure levels at frequencies below 100 Hz. Microearthquakes in active seismic areas create a low-level background noise at frequencies below 50 Hz. Man-made noises by machines also have been shown to contribute to background noise in many locations; many of these sounds reach as low as 5 Hz (which can be detected by pigeons). However, most of these man-made sounds are extremely regular in amplitude and frequency content, quite unlike earthquake-related sounds (Buskirk et al., 1981).



Acoustic Precursors and Coseismic Signals
Sounds that precede or accompany the shaking of the ground have been reported for a number of earthquakes in the literature (Buskirk et al., 1981).
Because more than 4000 seismograph stations are currently operating throughout the world, foreshocks are probably the most commonly observed precursors of large earthquakes. Although not all earthquakes have detectable foreshocks, their existence provides a reasonable explanation for much of the observed unusual animal behavior. According to Buskirk et al., (1981) at least 60% of major earthquakes (M > 7.0) had foreshocks large enough (M > 4.0) to be located by the International Seismological Centre (Fig. 13).
Buskirk et al. (1981) addressed an important misconception concerning foreshocks and precursory animal behavior. It is often incorrectly assumed that sound waves below the human hearing threshold (i.e. below about 50 Hz) cannot explain the animal behavior because these signals would be detected by seismographs. In fact, for two reasons it is quite possible that foreshocks which stimulate unusual animal behavior often are not detected by seismographs. First, conventional seismographs are most sensitive to frequencies around 0.01 - 10 Hz, and most of the energy radiated by microearthquakes and especially by microfractures is at much higher frequencies. Second, in well-instrumented areas, conventional seismographs routinely detect earthquakes with magnitudes as small as 1.5, but as is shown in Figure 13 and Table 3, over most of the earth's surface, earthquakes as large as M = 4 routinely go undetected. Thus, many foreshocks detectable by animals may go undetected by existing seismic networks (Buskirk et al., 1981).
Sound is produced in all major earthquakes (Fig. 14); the seismic waves themselves (0.1 - 10 Hz), the microquakes (up to several hundred Hz), and the sounds produced by breaking rock (up to 3000 Hz). The latter source of sound was confirmed by laboratory experiments on rock fracturing; before rocks break, fine hairline cracks appear. In this process, ultrasound signals are emitted at frequencies that can reach 100,000 - 1,000,000 Hz. The frequency varies with the size of the tears that are being produced and thus with the kind of rock in which the cracks appear. However, most of this high frequency sound is selectively muffled by the surrounding rock, leaving mostly the lower frequency earthquake sounds to escape to the earth's surface (Tributsch, 1982).
Audible earthquake sounds with frequencies of 40 to 70 Hz were fortuitously tape recorded in 1975 in the Imperial Valley of California (Hill et al., 1976). These authors calculated that P waves could account for audible sounds a few seconds before the perceptible S waves for small earthquakes. While earthquakes can generate low-frequency pressure waves (infrasound) which can travel great distances, higher frequency sound waves (audible to humans and animals) are attenuated by the rock overlying the
Table 3
Figure 13
Figure 14
earthquake focus. Instruments sensitive to very low frequency infrasound (0.1 - 5 Hz) recorded such sounds from the 1964 Alaska earthquake as far away as Washington, D.C. In addition, loud noises resembling thunder have been reported days to months before earthquakes, even in the absence of foreshocks (Buskirk et al., 1981).


Sound Detection By Animals
In reviewing animal sensitivity data, Buskirk et al. (1981) divided the sound spectrum into three ranges: below 100 Hz, 100 - 10,000 Hz (the range of best human sensitivity), and above 10,000 Hz. Most of the data reviewed are from tests performed by biologists involving behavior responses of animals to a sound stimulus.
According to Buskirk et al. (1981), it is unlikely that sounds in the frequency range 100 - 5000 Hz are responsible for the anecdotes about unusual animal behavior, because in this frequency range, most animals are no more sensitive than humans (Fig. 15). Within this range, common animals such as cats and dogs are generally more sensitive than humans above 5000 Hz. The most sensitive mammals are no more than an order of magnitude more sensitive than humans in this frequency range (Buskirk et al, 1981). Some birds, such as owls, have more sensitive hearing than man in the mid-frequency range of 1000 - 10,000 Hz, a threshold of 10 decibels below the normal 0 dB reference for human hearing (Kreithen, 1980).
Figure 15
Above 10,000 Hz, human hearing is clearly inferior to that of many small mammalian species. Behavioral as well as neurophysiological data show that laboratory mice and rats are extremely sensitive to sounds in the range 30,000 - 80,000 Hz. These rodents "speak" to each other within this frequency range, probably at close range, since ultrasound over 20,000 Hz tends to attenuate very quickly with distance. In anecdotal observations, some of the rats in a nearby cage displayed frightened behavior before rock failure during laboratory tests (Buskirk et al., 1981).
Although many mammals are more sensitive than humans in this high-frequency range, sounds of this frequency are probably not the cause of the unusual animal behavior observed before earthquakes. Only rather small cracks (with lengths of a few centimeters or less) would produce sounds with most of their energy at frequencies above 10,000 Hz. If the cracks are small, the amount of energy they radiate is also small. In addition, it has been calculated that these high-frequency sounds would attenuate too quickly to explain abnormal animal behavior before earthquakes (Buskirk et al., 1981). Similarly, Hill (1976) showed that attenuation would prevent animals from sensing most foreshocks that produce ultrasounds. Animals could perceive such high-frequency signals only from earthquakes with depths shallower than 0.01 km (Hill, 1976).
One of the best candidates for explaining observations of unusual animal behavior before earthquakes is infrasound (sounds below 50 Hz in frequency). A number of studies of birds and mammals show that many animals respond to extremely low sound pressure levels at frequencies far below 50 Hz (Buskirk et al., 1981). Kreithen (1980) showed that birds are highly sensitive to low frequency sounds and vibrations, and their ability to analyze complex signals far exceeds the ability of humans (Fig. 16). The general capacity of birds to filter signals buried in noise may exceed the abilities of our best instruments. Pigeons have low frequency auditory responses down as low as 3 cycles per minute (0.05 Hz). Their infrasonic sensitivity is 40 - 50 dB better than most other animals, including man (Kreithen, 1980). Furthermore, pigeons are able to discriminate between very small frequency differences (as low as 3%) in the range 1 - 20 Hz; for example, pigeons can tell the difference between a 2.00 Hz and a 2.06 Hz signal. This means that a flying bird can determine the direction of a very long wavelength sound merely by performing a slight detour in its flight path. The ability to use doppler shifts may not be crucial for earthquake detection, but it does point out that pigeons have a precise ability to process low frequency signals (Kreithen, 1980).
Birds possess three types of vibration receptors: (1) skin receptors in the legs and elsewhere, which decode surface vibrations and are most sensitive between 400 - 800 Hz, (2) inner ear receptors, which include organs which provide a horizontal reference plane for the bird's head, and (3) semicircular canals, with a fluid-filled chamber which is very
Figure 16
Figure 17
sensitive to angular accelerations and are responsible for the bird's balance (Kreithen, 1980). In laboratory experiments, pigeons who had parts of their inner ears surgically removed were unresponsive to infrasound, while unaltered pigeons were visibly frightened (Buskirk et al., 1981). Birds have a very different pattern of sound and vibration sensitivity than do humans. Birds can "hear" ground vibrations as low as 2 Hz (whereas humans can only hear down to 20 Hz); below 2 Hz, birds feel rather than hear the vibrations. For frequencies between 0.5 and 20 Hz, birds are from one to two orders of magnitude more sensitive than man, as shown in Figures 16 and 17 (Kreithen, 1980).
Buskirk et al. (1981) also cited kangaroo rats as being more sensitive than humans to infrasound. Whereas pigeons may use infrasound as homing signals, kangaroo rats may detect this sound to escape from predators.
Most earlier investigations of animal perception of earthquake-generated sounds concentrated on mammals, and did not consider aquatic species, or discuss low frequency sounds. However, Frohlich and Buskirk (1980) extended the analysis of Hill (1976) and considered data collected from fish. They concluded that fish could sense earthquake-generated pressure waves at least 2 Richter magnitudes smaller than those detectable by humans. In addition, an extrapolation of these hearing curves indicates that fish sensitivity to seismic sounds becomes increasingly better than human sensitivity as the frequency gets lower (Buskirk et al., 1981).
Fish are apparently more sensitive than most mammals to infrasound (50 Hz and below). They possess different sensory organs for perceiving sound: (1) the lateral line system, which is an array of hair cells extending along the side of the fish, apparently most sensitive to frequencies 200 Hz and below; (2) the inner ear otolith, located in the fish's skull; and (3) the air-filled swim bladder. Besides the greater hearing and vibration sensitivity of fish, their aquatic environment efficiently propagates sound and seismic waves, with relatively little attenuation. Seismic waves travelling through water are several orders of magnitude more intense than similar waves travelling through air, because water is a denser medium. Experiments have shown that fish can also discriminate low-level seismic signals from background noise (Frohlich and Buskirk, 1980).


Vibration Reception By Animals
A number of animals possess various sensitive external organs which detect ground vibrations directly, independent of their sense of hearing. The vibration detection capability of humans has been more carefully studied than those of any other animal, so they provide a good comparison with animal studies (Buskirk et al., 1981).
Humans are able to detect mechanical vibrations, using skin receptors occurring on the body, covering a range from below 1 Hz to over 100,000 Hz, which far exceeds the range of human hearing. The most sensitive receptors are located in the fingertips or the eardrum; their sensitivity is comparable to most other mammals. However, because humans seldom have their fingertips or their eardrums in contact with the ground, they cannot detect earthquake vibrations felt by terrestrial and burrowing animals of comparable sensitivity (Buskirk et al., 1981).
Recalling the anecdotal reports from China of snakes leaving their burrows more than a month before a major earthquake (Lee et al., 1976), these phenomena may be explained by evidence that snakes are physiologically more sensitive to low frequency vibrations below 100 Hz, even more so in the range below 10 Hz (Buskirk et al., 1981).
Chinese investigators have also reported tests in which the skin mechanoreceptors of 50 pigeons were severed, while the same number in a control group were left intact. Prior to a small earthquake, the normal pigeons panicked and flew around, while the altered birds behaved as usual (Lee et al., 1976).
Buskirk et al. (1981) cited a number of studies on insects, in which they reported much greater sensitivities to vibrations relative to humans.
In summary, the biological literature indicates that many animals are more sensitive to sounds and vibrations than man, in terms of frequency range, threshold response, and noise-filtering capacity. In view of the tests in different frequency ranges, low-frequency infrasound seems to be the best candidate among acoustical precursors detectable by animals before earthquakes, because it is less likely to be attenuated by the surrounding rock. However, as pointed out by Tributsch (1982), since dogs, cats, rats and mice are no more sensitive to infrasound than are humans, their panic reactions preceding earthquakes cannot be accounted for by infrasound.

Electromagnetic Phenomena

Electromagnetic phenomena which occur prior to an earthquake
include: electric fields (changes in the earth's background level), changes in the geomagnetic field, electromagnetic radiation caused by electrical phenomena in rock, and the creation of air ions (Buskirk et al., 1981).
Any electric and magnetic precursor within the animal's sensitivity and frequency range is a potential agent in evoking anomalous behavior in advance of earthquakes. Since there is no evidence of adaptive behavior in response to precursor phenomena, the animals that do respond seem either to misinterpret the stimuli or exhibit an inappropriate normal behavior. As animals are likely to detect precursor phenomena, the study of anomalous behavior may very well contribute, if only indirectly, to the geophysicist's ability to forecast earthquakes (Kalmijin, 1980).


Electric Field Changes
The earth's level of background electric fields varies considerably in time and in space. Local electrical storms, rainfall, and magnetic storms can cause large fluctuations. Measurements of electric field changes prior to earthquakes stand above background variations. The mechanisms behind these electrical changes have been attributed to: increased ionization due to the release of radioactive radon gas, or by the movement of groundwater (Buskirk et al., 1981).
Aquatic animals show extraordinary sensitivity to weak direct current electric fields, with thresholds well below electric fields reported before some major earthquakes (Fig. 18). Electric fish such as the elephant fish (freshwater Gymnarchidae ) and electric catfish have specialized organs which can detect minute electrical changes in the water (Buskirk et al., 1981). Sharks can home in on prey in the absence of any odor cues, using their well-developed electrical sensitivity. Catfish, rays, and eels also possess this ability (Tributsch, 1982).
A correlation between catfish activity and electrical earthquake precursors has been reported by Japanese researchers. During the 1930's, Hatai and Abe observed the catfish Parasilurus asotus in an aquarium filled with water flowing in from a freshwater creek. Several hours before a small earthquake, the fish became unusually sensitive to tapping on the table on which their aquarium rested. Fluctuations in the electric field were measured in the creek corresponding to the time of the catfishes' reactions. When the water flow (and the electric field fluctuations) from the creek was cut off, the catfish were no longer able to "predict" earthquakes (Buskirk et al., 1981).
Among terrestrial animals, mice, monkeys, and hamsters have been exposed to artificially manipulated electric fields (Fig. 18). Hamsters have been shown to be very sensitive to electric field changes. In laboratory experiments, hamsters were allowed to comfortably establish their nests in glass terrariums; then, an external electric field was alternated at irregular intervals, simulating the changes in a thunderstorm. The behavioral changes were remarkable; the hamsters wandered aimlessly from nest to nest (Tributsch, 1982). Changing electric fields are believed to affect nerve membranes, hormones, enzymes, and electrical activity in the brains of animals (Buskirk et al., 1981).
In general, terrestrial animals are much less sensitive (by about 4 or more orders of magnitude) than aquatic animals to electric field changes. It thus seems unlikely that terrestrial animals would respond to the levels of electric fields reported prior to earthquakes (Buskirk et al., 1981). Rapidly alternating electrical fields seem to have the most effect on land animals, whereas slowly changing electric fields would be significant only if they departed greatly from the background level (Tributsch, 1982).


Magnetic Field Changes
A major problem in interpreting the significance of animal response to magnetic changes is the great spatial and temporal variation in the earth's magnetic field. The normal geomagnetic field strength is 0.5 gauss, or 50,000 gammas. Typical temporal
Figure 18
Figure 19
variations in the field are about 30 gammas. Most geomagnetic changes associated with most recent earthquakes have been no greater than about 20 gammas (Rikitake, 1976).
Some animals respond to extremely small changes in the magnetic field, including some changes of the order of those observed before some earthquakes (about 10 gammas). However, Buskirk et al. (1981) have concluded that it is unlikely that magnetic field variations cause unusual animal behavior before earthquakes. The documentation of animal response to changes in magnetic field is of three types: the impairment of normal orientation movements of moving animals during short-term magnetic fluctuations, the response of stationary animals to the reversal or cancellation of the geomagnetic field, and changes in animal behavior occurring during exposure to larger artificial fields. The greatest sensitivities to magnetic changes has been demonstrated in experiments concerning animal orientation or homing. As indicated in Fig. 19, several species, including migrating birds, homing birds, and honeybees, have been shown to respond to magnetic changes as small as 10 gammas. However, as the chart shows, documented geomagnetic changes fall below the range of sensitivity of most animals (Buskirk et al., 1981).
In summary, magnetic field changes are probably not detected by animals before earthquakes because of the following:
1. Magnetic precursors are extremely small, about 20 gammas or less. Biological evidence indicates that such small changes are either undetectable or just barely detectable by animals.
2. Normal cyclical variations in the geomagnetic field, along with changes due to storms, are in the range of 20 gammas or so. This makes it difficult to tell whether the magnetic changes are due to an impending earthquake.
3. Magnetic precursory changes tend to occur over a period of days or weeks, while animals are more likely to respond to changes over minutes or hours. Therefore, there is little synchronicity of magnetic changes and abnormal animal behavior (Buskirk et al., 1981).


Microwave And Other Electromagnetic Radiation
Although microwave radiation has been mentioned as a possible cause of unusual animal behavior before earthquakes (Frey, 1980), no mechanism that produces microwaves during an earthquake has been proposed, and there are no published field data on microwave precursors. In addition, even if microwaves were produced in the hypocentral region of an earthquake, it is difficult to see how they could be transmitted to the surface, since the earth is opaque to microwaves (Buskirk et al., 1981).
Soviet scientists during the early 1980's reported that broadband electromagnetic radiation signals occurred prior to 12 shallow focus earthquakes located at distances of up to 1000 km away from the detecting devices. These precursors occurred from a few hours to a week before the earthquake (usually 1 to 2 days before), and were not detected prior to any earthquakes with a focus deeper than 30 km. Some Soviet scientists believe that this is a type of ionization phenomena which originates in the atmosphere above the epicentral region of shallow earthquakes, caused by some electromechanical or electrostatic mechanisms in the earth's crust, and may also explain "earthquake lights" (Buskirk et al., 1981).
The body of information about the biological effects of long-wave electromagnetic radiation is still full of gaps. Nevertheless, it can be assumed with some certainty that the long waves can have effects on living creatures that are at least as great as those caused by the much weaker rapid alternating electric fields (Tributsch, 1982). Symptoms of depression, insomnia, and general malaise have been reported by workers after being exposed for several hours daily in the radiation field. Migratory birds were reported to have flown off course because of the effect of a powerful transmitting antenna set up by the U.S. Navy; this antenna used extremely low frequency radio waves to communicate with submerged submarines. Therefore, long-wave electromagnetic radiation from the ground should not be completely eliminated as a possible earthquake precursor detectable by animals (Tributsch, 1982).


Phenomena Related To Air Ions
For terrestrial animals mentioned in the anecdotal reports, changes in the density of air ions (electrically-charged particles) may cause strange behavior. Tributsch (1982) believes that these charged aerosol particles are more abundant prior to earthquakes, and may explain other phenomena as well, such as earthquake lightning and earthquake fog.
Electrostatic phenomena are observed during earthquakes and may be implicated in precursory animal behavior (Yost, 1980). The release of charged particles into the atmosphere prior to an earthquake could cause significant changes in the ambient levels of small air ions (Table 4), which many organisms may be capable of detecting. (Large air ions are believed to be biologically inactive.) Biological literature contains many reports of air ions influencing animal behavior under laboratory conditions, but caution must be exercised when extrapolating these results to the field. Ambient ion levels are strongly influenced by weather, which could modify or mask any ion changes due to earthquakes (Yost, 1980).
Several mechanisms which may produce air ions before earthquakes have been postulated, although few have been tested in the field (Table 5). Radon outgassing of the soil near earthquake faults is believed to be responsible for a 50 - 100% increase in the background level of ions present. Such outgassing may begin several weeks before an earthquake, and the rate of outgassing may reach two or three times the normal background level. This could be caused by movement of the water-mineral interface in the soil, which creates an electrostatic release of positive charge into the air. Another mechanism proposed is a piezoelectric effect due to rocks placed
Table 4, Table 5
under stress in the fault zones, which can create a large release of charge (a result confirmed in laboratory stress tests of granite and ceramics). However, the high conductivity of the soil would probably absorb this charge (Yost, 1980).
Some of the anecdotal reports of abnormal animal behavior (Lee et al., 1976) resembles similar reactions reported by biologists who studied the effects of excessive levels of positive ions. Positive ions seem to inhibit test performance of laboratory rats (Yost, 1980), increase susceptibility to disease in mice, and retard growth in microbes (Buskirk et al., 1981). Long-term depletion of ions and an excess of positive ions can cause physiological changes and a heightened emotional state in animals. In mammals, an excess of positive air ions is known to increase the level of the neurohormone serotonin, which causes irritable behavior, changed habits, and physiological deterioration. Weather-sensitive humans possess high serotonin levels during periods of hot, dry winds, when high levels of positive ions are measured. Their symptoms include headaches, tension, and swelling. These symptoms disappear when humans are exposed to antistatic treatments, or high levels of negative air ions (Buskirk et al., 1981). Commercial negative-ion generators have been sold during the past few years, claiming to produce physiological and psychological well-being by removing the deleterious positive air ions from the living and working environment. The "good smell" of air after a rainstorm is due to the presence of large amounts of negative ions, which has a soothing effect (Tributsch, 1982).
Under certain geological and atmospheric conditions, positive airborne ions (charged aerosols) seem to appear in abundance before an earthquake. Some typical individual behavior patterns can simply be interpreted as the animals' flight responses before charged clouds of particles streaming out of the ground. If such clouds appeared before earthquakes and penetrated the burrows of snakes, mice, and other small animals, it would explain their flight from underground. The agitation and flight of birds could also be explained by airborne clouds of positive ions. It also explains why so many different animals are reported to have fled any enclosed buildings - enclosed spaces possess much higher charge densities than open areas, and the animals seek to escape the electrostatic effects which must cause them great discomfort (Tributsch, 1982).
In summary, air ions are still a possible candidate to explain unusual animal behavior prior to earthquakes. However, with the geophysical and biological data available at present, it is not yet possible to make a quantitative interpretation of these phenomena (Buskirk et al., 1981).


Phenomena Related to the Opening or Closing of Small Cracks
Changing stress conditions near the hypocentral region can open or close tiny cracks prior to the occurrence of an earthquake, changing the permeability of nearby aquifers, affecting groundwater levels, and affecting measures of electrical resistivity. In addition, the opening of cracks can influence the rate of release of gases contained within the earth (Buskirk et al., 1981).


Precursory Changes in Groundwater Level
Irregular water levels, artesian flow, and muddy wells have been correlated with many historic earthquakes, and reportedly were used extensively in predicting the 1975 Haicheng, China, earthquake (Group of Earthquake Research, 1979). In the epicentral area of Haicheng, wells became artesian 12 hours before the earthquake, and had been muddy and oily the previous day. Waterspouts, muddy wells, floods, and interruptions in the flow of hot springs had been reported within 50 km of the epicenter within the previous month (Buskirk et al., 1981). In a study of water table changes in the area of San Andreas, California, small but significant water level minima were followed by earthquakes on the fault, while no significant water level changes were associated with two more distant earthquakes. Striking changes in well levels are most commonly observed one or two days before earthquakes (Rikitake, 1978). Unfortunately, a great many water level variations result from seasonal changes and short-term changes caused by rains. Rains also affect measurement of tilt, electrical resistivity, and radon concentration (Rikitake, 1976).
The presence of groundwater changes seems to correlate with anomalous animal behavior during the 3 months before the 1975 Haicheng, China, earthquake in an analysis by Deng and Jiang (1981). From the geographical spacing and timing of the data within 150 km of the epicenter, there appears to be a clustering of the two types of anomalies, with perhaps a slight lag in the animal behavior observations. Both the number of groundwater changes and the reports of abnormal animal behavior began to increase dramatically on the day of the first recorded foreshock, 3 days before the magnitude 7.3 earthquake (Deng and Jiang, 1981).
It is possible that groundwater changes of seismic origin could explain a few observations of unusual behavior in animals with burrows or nests underground; for example, snakes emerging from hibernation in midwinter and rats moving their dens (Lee et al., 1976). However, these groundwater changes are often small, and it is therefore difficult to predict how they could affect animal behavior. In addition, many of the behavioral anomalies could also be caused by other stimuli, such as the presence of odorous gases. Furthermore, changes in water level of non-seismic origin occur commonly without affecting animal behavior. For these reasons, it seems unlikely that groundwater changes could explain unusual animal behavior before earthquakes (Buskirk et al., 1981).


Earth Gases
Changing stress conditions which open tiny cracks in rocks before an earthquake may change the release rate for gases trapped beneath the ground. Measurements of radon concentration in soil gases and in groundwater provide the strongest evidence that gases are released before earthquakes (Buskirk et al., 1981).
There are several difficulties with using radon anomalies to predict earthquakes. Radon changes which accompany earthquakes cannot always be attributed solely to changes in tectonic stress, because rainfall and seasonal climate changes also affect radon levels, as shown in Fig. 20 (Buskirk et al., 1981).
Although radon is easily measured because of its radioactivity, other gases are also released, such as hydrogen sulfide, ozone, and phosphorus. In Japan, the levels of elements such as helium, argon, nitrogen, fluorine, iodine, mercury, and uranium in groundwater were observed to change prior to earthquakes. Prior to the earthquake in Haicheng, China, the precursory changes in the levels of dissolved nitrates, sulfates, chloride, and fluoride in groundwater were reported (Buskirk et al., 1981).
Figure 20


Gaseous Odorants
Historical accounts from different parts of the world reported unusual smells before earthquakes. Many of these events coincided with unusual animal behavior (Moulton, 1980), often resembling animal responses to strange smells, such as dogs barking and sniffing the air (Buskirk et al., 1981). Before the 1975 earthquake in Haicheng, China, citizens in the epicentral area reported smelling unusual odors such as sulfides, phosphorus, and ozone about a month or two before the earthquake. Just before the earthquake, there was a dense, stratified fog with a strange odor, and dogs were seen sniffing and digging in the snow (Tributsch, 1982). These sulfurous odors are most likely to include one or more of the following: hydrogen sulfide, carbon disulfide, cobalt sulfide, sulfur dioxide, and dimethyl sulfide (Moulton, 1980). Tributsch (1982) speculates that electrical phenomena before earthquakes may electrolyze groundwater and create new chemical compounds which animals smell, but do not recognize, and cause reactions of fear. This hypothesis may explain why non-seismic emissions of gases from the earth (such as swamp methane) do not cause the same reactions from animals who encounter the gases (Tributsch, 1982).
If we wish to examine the question of whether animals smell gases that leak out of the ground before earthquakes and are alarmed by them, we must take note that all living things except apes and most birds have a keener sense of smell than man, as
Figure 21
shown in Fig. 21. Compared to humans, dogs have a sense of smell roughly one million times more sensitive. Male silk moths are able to detect a single air-borne molecule of pheromone (a sexual attractant) secreted by female moths over a range of 7 miles (Tributsch, 1982). However, the measurement of olfactory thresholds is complex, and it is difficult to compare different species quantitatively. Laboratory tests of olfactory sensitivity on dogs were inconclusive, because naturally-occurring odorants could not be used, as they made decontamination of the test chambers difficult (Buskirk et al., 1981).
The behavior of fish prior to earthquakes is also consistent with known behavior responses to olfactory stimuli observed in other circumstances. Eels, minnows, catfish, and trout seem to have a sense of smell equal to or better than humans. Homing and orientation movements of salmon, eels, and bullhead catfish apparently require chemical cues. Alarm behavior is triggered by chemicals released from the injured skin of other fish, especially those of the same species. This fright behavior is prevalent only in the Ostariophysini species of fish which are mentioned most often in the anecdotes (Buskirk et al., 1981). Deep sea fishes and bottom-dwelling fishes have been reported to behave strangely before earthquakes (Moulton, 1980). Moulton (1980) believes that sulfides or other odorants may be released into the seawater prior to earthquakes, especially if there is a
fault zone or hydrothermal vent nearby. In quantitative terms, however, it is difficult to compare the sense of smell in fish to that in land animals (Buskirk et al., 1981).
Given that most animals possess better olfactory sensitivities than humans, the release of odorous gases before earthquakes seems to be a good candidate for some behavioral anomalies in animals (Buskirk et al., 1981).
Table 6 (Buskirk et al., 1981) summarizes the data for geophysical precursors and biological sensitivities of animals.
Table 6
CHAPTER 5
EARTHQUAKE PRECURSORS BEST DETECTED BY ANIMALS


Humans usually report animal behavior before earthquakes as being unusual only if they themselves sense no precursory stimuli. In order to isolate the causes of the animal behavior, it is therefore necessary to identify those precursory stimuli which are detectable by animals, but not by humans. Among the geophysical precursors discussed in the previous chapter, the most promising candidates include: (1) infrasound (audible low-frequency sound of about 10 - 40 Hz) and ground vibrations, (2) electric field changes, (3) positively-charged aerosol particles, and (4) earthquake gases smelled by animals. It must be noted that none of these stimuli alone can explain all of the observed behavioral anomalies in animals (Buskirk et al., 1981; Tributsch, 1982). Also, Tributsch (1982) notes that not all geophysical stimuli cause the reaction of fear in animals. Each of the four geophysical factors listed above will be discussed and compared in an attempt to find which one(s) best explain abnormal animal behavior before earthquakes.


Audible Low-Frequency Sound/Ground Vibrations
Several quite diverse species, including birds, rodents, and fish, are known to respond to sounds with frequencies of about 40 Hz and below that are 2 to more orders of magnitude weaker than those detectable by humans. Animals can also detect foreshocks about 2 Richter magnitudes smaller or correspondingly more distant than those detectable by man (Buskirk et al., 1981).
Stierman (1980) and a team of seismologists from the California Institute of Technology recorded two moderate (M = 4.9 and 5.2) earthquakes, followed by an earthquake swarm, in the Mojave Desert in March of 1979. Their field observations noted that many small earthquakes generated audible booming noises, which was accompanied by the barking of dogs. Stierman noted that the dogs responded to virtually every booming noise by a brief but vigorous sequence of barking about 4 to 10 seconds after the shock (the dogs did not bark before the noise). In several cases, the dogs responded to a seismic aftershock not felt or heard by the human observers, but was recorded by their seismograph. Stierman speculated that the vibrations generated by small foreshocks would be misidentified as thunder by animals dwelling underground. If thunderstorms typically signal spring rains in a region, for hibernating animals (snakes, frogs, worms, etc.), seismic vibrations which mimic thunder could cause their untimely awakening (Stierman, 1980). Foreshocks are the most probable explanation for reports of unusual animal behavior that occurs very far from the epicenter (Buskirk et al., 1981). Many of the distant behavior reports concern fish, particularly in Japan (Lee et al., 1976). The relatively low attenuation of sound waves propagating in water may allow seismic waves from foreshocks to travel greater distances through water than through air (Frohlich and Buskirk, 1980).
It has been shown in laboratory studies that sounds in the 100- to 5000- Hz range are detected nearly as well by humans as by animals; therefore, it is unlikely that seismic sounds caused by foreshocks in the mid-range frequencies can explain their unusual behavior. Although many animals are much more sensitive than humans to high-frequency sound above 5000 Hz, it is also known that higher frequency seismic sounds are easily attenuated within a short distance from the earthquake hypocenter (Hill, 1976).
Depending upon the species, below a certain threshold frequency, low frequency vibrations are no longer heard, but felt. For this reason, ground vibrations may be overlapped with audible low frequency sounds and treated as more or less a single geophysical precursor. Frequencies from about 50 Hz and below may collectively be called infrasound. A number of animal species, including fish, snakes, birds, rodents, and insects, are much more sensitive than humans to vibrations in the 10 - 100 Hz frequency range. It is important to note that seismographs are most sensitive in the 0.1 - 10 Hz range, meaning that animals may respond to earthquake shocks having frequencies from 10 - 100 Hz which are not recorded by seismographs. Although humans possess the capability to feel seismic vibrations through the palms of their hands and the soles of their feet, this requires close contact to the ground, which is an unnatural situation. Also, humans rely mostly on eyesight, and tend to ignore vibrations sensed by animals, who use vibrations to find food or avoid predators (Buskirk et al., 1981).
The advantages of citing infrasound for explaining anomalous pre-quake animal behavior are:
1. Infrasound is capable to travelling great distances without appreciable attenuation, which explains abnormal animal behavior which occurs far from the earthquake epicenter.
2. It accounts for the anomalous behavior of underground and burrowing animals, such as snakes and rats, who flee their hiding places often months before the earthquake occurs.
3. It accounts for the anomalous behavior of fish and other sea creatures, who generally are much more sensitive to low frequency vibrations, which propagate with greater speed and intensity under water than in air (Frohlich and Buskirk, 1980).

Infrasound does not explain, however, the panic reactions of dogs, cats, and other mammals that are not appreciably more sensitive than humans to low frequency vibrations . Also, birds in the branches of trees are unlikely to feel the intensity of seismic vibrations felt in the ground (Tributsch, 1982).


Electric Field Changes
Although many animals have been found to be quite sensitive to small changes in electric fields, there is laboratory data on only a few species. What is currently known about this area is that aquatic animals, such as sharks and some other fish, are much more sensitive than land animals to electric field changes. However, electric field changes before earthquakes have seldom been documented, and those were measured only on land (Buskirk et al., 1981). According to Buskirk et al. (1981) it seems unlikely from available data that animals can sense changes in electric fields before earthquakes.


Positively-Charged Ions
The effects of positively-charged ions on animal physiology and behavior have been well documented. An excess of airborne cations (positively-charged aerosols) can cause general irritation and excitability in animals, evidenced by a measurable increase in the body levels of serotonin (a powerful neurohormone affecting sleep, moods, and the transmission of nerve impulses), which is known to cause such reactions (Tributsch, 1982). Tributsch (1982) favors this hypothesis for explaining abnormal animal behavior before earthquakes. The following observations and circumstantial evidence are in favor of citing charged aerosols as the culprit for pre-quake animal behavior:
1. The unusual behavior of flying birds (who would not feel seismic vibrations as intensely as ground-dwellers), and the abandonment of enclosed shelters (which hold distressful concentrations of static charges) by animals such as cats may be explained by electrostatic charging of the air.
2. Shortly before the 1976 earthquake epicentered in Friuli, Italy, Tributsch observed an anomalous electrostatic phenomenon in a watchmaker's shop - the watchmaker was unable to repair a watch because the small stainless steel parts kept jumping apart from each other, implying that they had somehow acquired the same electrical charge (like charges repel each other). The metal objects could only have acquired this charge by charged aerosol particles expelled by the earth, as there was no sign of inclement weather that day.
3. The frequent anecdotal reports of earthquake fog and earthquake lights (dating back to the time of Aristotle) occurring under cloudless skies may be explained by the electrostatic charging of aerosol particles. In China, earthquake fog and/or earthquake lights are frequently reported along with abnormal animal behavior (Lee et al., 1976).
4. Many qualitative similarities are found between animal behavior before earthquakes and before storms; this suggests that there are common geophysical stimuli sensed by animals. It is possible that animals could misinterpret charged aerosols of seismic origin for electrical changes prior to an impending storm. Although animals probably have not evolved an actual "earthquake sense" because of the rarity of earthquakes, their "storm sense" may sometimes serve the same purpose prior to earthquakes (Buskirk et al., 1981).

The major disadvantage with the charged aerosol hypothesis is the lack of hard evidence. There are apparently no reliable observations proving that the level of air ions change prior to an earthquake (Buskirk et al., 1981).


Earthquake Gases Smelled By Animals
It is quite possible that some unusual animal behavior occurs as a response to the odor of gases released prior to the occurrence of an earthquake. The evidence that gases are released before some earthquakes is convincing, although few quantitative data are available for gases other than radon. If current studies suggesting the presence of both long-term and short-term radon anomalies are correct, the release of deep-earth gases could explain behavior anomalies of several weeks before the event was well as several days before. Because of technical problems, it is quite difficult to determine quantitatively the comparative threshold of sensitivity of animals to various odorants. Nevertheless, the available information seems to confirm that many species of animals are remarkably sensitive to particular odorants and that a few species, such as dogs, seem to be more sensitive than humans for almost any olfactory stimulus. Humans may also be culturally insensitive to strong odors (they tend to ignore smells which they are physically capable of detecting). In contrast, animals use their keen sense of smell to find food, to avoid predators, and to communicate (Buskirk et al., 1981).

CHAPTER 6
CONCLUSIONS REACHED BY RESEARCHERS


By far, the most important unanswered scientific question about unusual animal behavior before earthquakes is, "Does it exist at all?" A typical report is made after the event by an inexperienced, biased, and excited observer, and no attempt is made to quantify what constitutes "normal" or baseline level animal behavior (Buskirk et al., 1981). Other significant factors, such as the techniques of measurement or interview (Lott et al., 1979) and weather changes (McClellan, 1980) are usually ignored. Otis and Kautz (1980) designed a research program that continuously collects reports of unusual animal behavior behavior, via a telephone hotline. Lindberg et al. (1981) designed an outdoor animal monitoring facility that continuously collected data indicating the level of activity by the test animals (kangaroo rats and pocket mice). To date, these studies have shown animal response to only small earthquakes, and there seemed to be only slight differences in animal behavior before and after the earthquakes.
Another important question concerning the evaluation of unusual animal behavior in laboratory studies is, "Why would animals react to a geophysical signal by exhibiting unusual animal behavior, and how can they distinguish the signal from the enormous amount of background noise?" One possible explanation is provided by a number of physiological studies of animals suggesting that an "alerting" stimulus makes the central nervous system more responsive to a second stimulus. In other words, the presence of one precursory change, such as the level of air ions, may make animals uneasy or irritable, so that they are more likely to sense an alarm response to the sound or vibration of a subsequent foreshock (Buskirk et al., 1981).
The variability in animal behavior reported from earthquake to earthquake and from animal to animal has been considered an obstacle to explaining the biological precursors. Instead, it should remind us that different geophysical mechanisms are responsible for the phenomena. It is not possible to generalize about abnormal animal behavior prior to earthquakes based upon the studies of a single animal species, nor from geophysical data from a single geographical area (Frohlich and Buskirk, 1980).
Animals may behave "abnormally" as a result of misleading or insufficient sensory information, misinterpretation of the available stimuli, or a combination thereof (Kalmijin, 1980). Animals have evolved sensory systems, from very simple to very complex, for detecting environmental energies, and it may be assumed that this evolution is linked to their capacity to survive. It seems unlikely that animals have evolved sensory systems for the purpose of detecting impending earthquakes, because it has no evolutionary survival value. Earthquakes are simply too infrequent relative to the life span of most animals and seldom catastrophic in terms of population sizes (Lindberg et al., 1981). Animals do not "predict" earthquakes; they respond to sensory input. Humans may be able to predict earthquakes on the basis of animal responses (Verrillo, 1980). From the conclusions reached at the U.S.G.S. Conference on Abnormal Animal Behavior Prior to Earthquakes in 1979, it is now clear that some animals are significantly better equipped than humans to sense almost any geophysical stimulus which may precede an earthquake. There are many avenues for further research, including:
1. Acoustic waves and vibrations in the frequency range 10 - 50 Hz : More measurements of animal behavior to sound both in air and under water are needed.
2. Electric field changes and air ion monitoring : More measurements of electric fields are needed to compare terrestrial and aquatic environments. Monitoring equipment for changes in air ion levels should also be set up.
3. Earthquake gases other than radon : Radon is the most often measured gas before earthquakes; however, radon is chemically inert, and animals probably are insensitive to it. Methane, ozone, and sulfur compounds are likely candidates for further monitoring of gases released prior to earthquakes.
4. Olfactory thresholds for odors : Most of the data on animal sensitivities to natural odorants are not quantitative. Quantitative data are necessary to test the effects of gaseous geochemical precursors on animals.
5. Responses of common domestic animals : Dogs, cows, horses, and chickens are the animals most commonly mentioned in the anecdotes. Surprisingly, there have been few or no controlled studies of the sensitivities of these animals to low-frequency sound, vibrations, electric fields (with the exception of dogs), and odorous gases (Buskirk et al., 1981).

The consensus of most Western researchers is that there is, as yet, no conclusive scientific evidence for believing that abnormal animal behavior is in any way related to the occurrence of earthquakes. There is, however, general agreement among scientists that a great many of the anecdotal accounts are entirely plausible in light of what is known about animal sensory capabilities and the level of many seismic precursors. The correlations between animal sensitivities and the measured levels of geophysical precursors have provided geophysicists clues about what signals they should measure to predict earthquakes (Buskirk et al., 1981).
Although the mechanisms behind the animals' responses to seismic precursors may not be worked out for years to come, it is possible to use the existing information towards aiding the efforts to predict earthquakes. For cultural and practical reasons, the Chinese have heeded observations of abnormal animal behavior for hundreds of years, even though they now have a nationwide seismic monitoring network using the latest measuring instruments. China has chosen to take anecdotal observations seriously, even if they cannot be explained, because earthquakes have killed millions throughout her history; false alarms are a small price and inconvenience to suffer if many lives can be saved by warning people that an earthquake is imminent.
It is important for scientists to cast aside prejudices held by many skeptics against using controversial but potentially useful information; abnormal animal behavior constitutes the largest collection of earthquake information in existence (Tributsch, 1982). Since the Western world relies almost exclusively on instrumental measurements, Stewart (1977) cited the advantages of using observations of abnormal animal behavior as a complementary part of an interdisciplinary effort to successfully predict earthquakes. Existing Western techniques have the following limitations: (1) earthquakes cannot be predicted where there are no instruments; (2) instrumental surveillance of all possible seismic areas is too expensive; and (3) instruments cannot tell or estimate when an earthquake will occur. Rikitake (1978) has concluded from his statistical analysis of abnormal animal behavior that such behavior may sometimes be used as an extremely short-term seismic precursor.
Ongoing research may reveal additional details about sensory capabilities and geophysical parameters, but short of controlled observations of animals before a major earthquake, there may never be a conclusive breakthrough that will be satisfactory to the Western scientific establishment. The major advantage of the interdisciplinary (biology, biophysics, geophysics, and even psychology) approach is that the weaknesses of one discipline may be covered by the strengths of another. There is no reason to believe that a combined approach toward predicting earthquakes, including the use of data on abnormal animal behavior, would be impractical. The surge in interest in abnormal animal behavior prior to earthquakes in the Western world was spurred by the world's first successful short-term earthquake prediction (Haicheng, China) in 1975. Perhaps successful results can best dictate which methods are practical or impractical.
BIBLIOGRAPHY


An, Q. L.; Chen, X. L.; Song, S. Q.; and Gu, P.; 1987. Activities of the Chinese Seisomological Delegation; attendance of the Second International Earthquake Conference and visit to the United States. Recent Developments in World Seismology, vol. 7(103), pp. 1-6.

Anderson, C. J., 1973. Animals, earthquakes, and eruptions. Field Museum of Natural History Bulletin, vol. 44, No. 5, pp. 9-11.

Buskirk, R. E.; Frohlich, C.; and Latham, G. V.; 1980. Animal behavior prior to earthquakes: a review of possible sensory mechanisms. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 1-12.

Buskirk, R. E.; Frohlich, C.; and Latham, G. V., 1981. Unusual animal behavior before earthquakes: a review of possible sensory mechanisms. Reviews of Geophysics & Space Physics, vol. 19, No. 2, pp. 247-270.

China, Working Group for the Research of Earthquake Prediction, 1986. 20 years' research on comprehensive earthquake prediction. Earthquake Research in China, Vol. 2, No. 4, pp. 13-23.

Davies, D., 1975. Earthquake prediction in China. Nature, vol. 258, pp. 286-287.

Deng, Q.; and Jiang, P., 1981. A preliminary analysis of reported changes in ground water and anomalous animal behavior before the 4 February 1975 Haicheng earthquake. In Earthquake Prediction: An International Review , American Geophysical Union, Washington, pp. 543-565.

Deshpande, B. G., 1986. Earthquakes, animals, and man; Chapter IV; What can we do? Proceedings of the Indian National Science Academy, Part B: Biological Sciences, vol 52, No 5, pp. 585-618.

Ding, G. Y.; Ma, T. C.; and Mei, S. J.; 1984. Methods of quake prediction. In Earthquake Prediction: Proceedings of the International Symposium on Earthquake Prediction , Terra Sci. Publ. Co., Tokyo, Japan, p. 453-465.

Dubrov, A. P., 1978. The Geomagnetic Field and Life:Geomagneto- biology , Plenum Press, New York, 318 p.

Fay, R. R.; and Patricoski, M. L., 1980. Sensory mechanisms for low frequency vibration detection in fishes. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 63-91.

Feng, D. Y., 1986. Time dependency and spatial distribution of earthquake precursors. Earthquake Prediction Research, Vol. 4, No. 1-2, pp. 13-37.

Frey, A. H., 1980. Approaches to earthquake prediction: implications from biophysical research with electromagnetic fields. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 141-147.

Frohlich, C.; and Buskirk, R. E., 1980. Can fish detect seismic waves? Geophysical Research Letters, vol. 7, No. 8, pp. 569-572.

Golden, F., 1983. The Trembling Earth: Probing & Predicting Earthquakes , Charles Scribner's & Sons, New York, 175 p.

Group of Earthquake Research (Beijing, China), 1979. Some characteristics of behavior prior to earthquakes. In Earthquake Prediction: Proceedings of the International Symposium on Earthquake Prediction , Terra Sci. Publ. Co., Tokyo, Japan, pp. 243-251.

Hill, D. P.; Fisher, F. G.; Lahr, K. M.; and Coakley, J. M., 1976. Earthquake sounds generated by body-wave ground motion. Bull. Seismol. Soc. Am., vol. 66, pp. 1159-1172.

Jacobson, M. L.; and Rodriguez, T. R. (compilers), 1987. National Earthquake Hazards Reduction Program; summaries of technical reports, Volume XXIV. U.S. Geol. Survey Open File Report 87-0374, 714 p.

Jiang, J. C.; Sun, W.; Xu, M. L.; Nie, S. F.; Liu, X. Q.; Liu, L. J., and Wang, Q.; 1985. Attenuation characteristics of earthquake premonitory geo-sound and its biological effects. Acta Seismologica Sinica, vol. 7, No. 2, pp. 202-211.

Kalmijin, A. J., 1980. Research on electric and magnetic field detection. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 110-113.

Kerr, R. A., 1984. How to catch an earthquake. Science, vol. 223, p. 38.

Kenagy, G. J.; and Enright, J. T., 1980. Animal behavior as a predictor of earthquakes? An analysis of rodent activity. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 178-197.

Kreithen, M. L., 1980. Detection of sound and vibration by birds. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 13-29.

Lee, W. H. K.; Ando, H.; and Kautz, W. H., 1976. A summary of the literature on unusual animal behavior before earthquakes, U.S. Geol. Survey Open-File Report 76-816.

Lindberg, R. G.; Skiles, D. D.; and Hayden, P., 1981. Can animals predict earthquakes? A search for correlations between changes in activity patterns of two fossorial rodents and subsequent seismic events. U.S. Geol. Survey Open-File Report 81-385, pp. 1-37.

Lott, D. F.; Hart, B. L.; and Howell, M., 1978. Unusual animal behavior prior to the Willets, CA earthquake of November 22, 1977 (abstract). Trans. Am. Geophys. Union, vol. 59, No. 4, p. 329.

Lott, D. F.; Hart, B. J.; and Howell, M.; 1979. Is unusual animal behavior observed before earthquakes? Yes and no. Geophysical Research Letters, vol. 6, No. 9, pp. 685-687.

Lott, D. F.; Hart, B. J.; and Howell, M.; 1979. A search for reports of unusual animal behavior prior to the Coyote Lake earthquake of 6 August 1979 (abstract). EOS, vol. 60, No. 46, pp. 891-892.

Lott, D. F.; Hart, B. L.; and Howell, M.; 1981. Retrospective studies of unusual animal behavior as an earthquake predictor. Geophysical Research Letters, vol. 8, No. 12, pp. 1203-1206.

Lott, D. F.; Hart, B. J.; and Howell, M.; 1981. Animal behavior and earthquake prediction. In Baseline Studies of the Feasibility of Using Animal Behavior as a Component in the Prediction of Earthquakes , U.S. Geol. Survey Open-File Report 81-378, pp. 121-144.

Lu, Z. H., 1986. The status quo and prospect in research on earthquake prediction at home and abroad. Recent Development in World Seismology , vol. 2, pp. 21-26.

McClellan, P. H., 1980. Pre-earthquake animal behavior: a closer look for alternative causes. Geophysical Research Letters, vol. 7, No. 5, pp. 333-336.

Mead, D. F., 1976. How the Chinese predict earthquakes. Science Digest, March 1976, pp. 57-61.

Medici, R. G., 1980. Methods of assaying behavioral changes during exposure to weak electric fields. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 114-140.

Mulilis, J. P.; and White, M. H., 1986. Behaviors of the catfish Corydoras aeneus for use in earthquake prediction. Earthquake Prediction Research, Vol. 4, No. 1-2, pp. 47-67.

Moulton, D. G., 1980. Odorant emissions preceding earthquakes and odor detection and recognition in animals. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 156-177.

Otis, L. S.; and Kautz, W. H.; 1979. Biological precursors of earthquakes: a validation study. In Earthquake Prediction: Proceedings of the International Symposium on Earthquake Prediction , Terrapub, Tokyo, pp. 253-262.

Panel on Earthquake Prediction of the Committee on Seismology, 1976. Predicting Earthquakes: A Scientific And Technical Evaluation With Implications For Society , National Academy of Sciences, Washington, 62 p.

Papadopoulos, G. A., 1988. A note on the prediction of the September 13, 1986 strong earthquake in Kalamata, Southwest Peloponnesus, Greece. Tectonophysics, vol. 145, No. 3-4, pp. 337-341.

Press, F.; and Siever, R.; 1978. Earth , 2nd ed., W.H. Freeman & Co., San Francisco, 649 p.

Raleigh, C.B., 1980. Status of and recent progress in earthquake prediction, in Proceedings of Conference XII: Earthquake Prediction , U.S. Geol. Survey Open-File Report 80-843, pp. 138-141.

Richter, C. B., 1958. Elementary Seismology , W. H. Freeman & Co., San Francisco, 768 p.

Rikitake, T., 1976. Earthquake Prediction , Elsevier, Amsterdam, 357 p.

Rikitake, T., 1978. Biosystem behavior as an earthquake precursor. Tectonophysics, 51, pp. 1-20.

Rikitake, T., 1982. Earthquake Forecasting And Warning , vol. 3 of Developments in Earth and Planetary Sciences , Center for Academic Publications, Tokyo, 402 p.

Rikitake, T., 1982. Earthquake prediction work in Sichuan Province, China, with special reference to the Songpan- Pingwu earthquakes. In Earthquake Research in China, 3 , Wu, F. T. (ed.), Chinese Geophysics, vol. 2, No. 1(3), pp. 139-156.

Shaw, E., 1977. Can animals anticipate earthquakes? Natural History, vol. 86, No. 9, pp. 14-25.

Stewart, D. M., 1977. A master plan for earthquake prediction; physics, biology, and psychology combined; a proposal to use instruments, animal behavior, and psychic perceptions together (abstract). Geol. Soc. Am., Abstr. Programs, vol. 9, No. 4, 509 p.

Stierman, D. J., 1980. Earthquake sounds and animal cues; some field observations. Bull. Seismol. Soc. Am., vol. 70, No. 2, pp. 639-643.

Tributsch, H., 1978. Do aerosol anomalies precede earthquakes? Nature, vol. 276, pp. 606-607.

Tributsch, H., 1982. When The Snakes Awake: Animals & Earthquake Prediction , MIT Press, Cambridge, 248 p.

Verillo, R. T., 1980. Some aspects of vibrotactile responses in animals and man. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 30-62.

Verosub, K. L.; Lott, D. F.; and Hart, B. L.; 1981. The use of indigenous observers to detect abnormal animal behavior prior to earthquakes. In Baseline Studies of the Feasibility of Using Animal Behavior as a Component in the Prediction of Earthquakes , U.S. Geol. Survey Open-File Report 81-378, 101-115.

Von Hentig, H., 1923. Reactions of animals to changes in physical environment, I. Animals and earthquakes. J. Compar. Psychology, vol. 3, pp. 61-71.

Werner, A., 1974. Russian scientists use animals to predict disasters, in California Superquake , P. James, ed., Exposition Press, Hicksville, N.Y., pp. 40-42.

Wu, F. T. (editor), 1982. Earthquake research in China, III. Chinese Geophysics, vol. 2, No. 1(3), 197 p.

Wu, F. T.; and Han, D.; 1982. Possible precursors of the 1976 Tangshan earthquake, in Earthquake research in China, 3, (Wu, F. T., ed.), Chinese Geophysics, vol. 2, No. 1(3), pp 157-172.

Yost, M. G., 1980. Air ions as a mediator in animal behavior prior to earthquakes. In Abnormal Animal Behavior Prior to Earthquakes, II , U.S. Geol. Survey Open-File Report 80-453, pp. 148-155.

Zhu, C. Z., 1984. Earthquake prediction research and earthquake precursors. J. of Seismol. Res., vol 7, no. 5, pp. 607-610.

Zhu, F. M.; and Zhong, Y. Z.; 1984. Anomalous macroscopic phenomena and their significance in the prediction of strong earthquakes. In Earthquake Prediction: Proceedings of the International Symposium on Earthquake Prediction , Terra Sci. Publ. Co., Tokyo, Japan, pp 193-201.
ÿ