1. Introduction
Many countries including, most recently, Japan, have suffered extensive damage due to earthquakes (EQs). On 11 March 2011, the EQ (Mw 9.0) occurred in Japan, which took about twenty thousand lives together with missing and killed persons. There are several studies on the short-term prediction of EQ occurrence [
1,
2,
3]. Those studies have mainly focused on pre-seismic unusual physical and/or chemical variations near the epicenters, such as electromagnetic signals, ionospheric propagations, and radon gasses, emerging prior to EQs [
1,
2,
3]. These anomalies have often been observed within a few weeks before EQs.
Additionally, posteriori surveys concerning anomalous phenomena have been conducted. Macroscopic anomalies include unusual animal behaviors (UABs), abnormal sounds [
4], EQ lights [
5], EQ clouds [
6], ground deformation [
4], and abnormalities in the ground water [
7]. Wadatsumi [
8] investigated macroscopic anomalies preceding the Kobe EQ in Japan, on 17 January 1995, and found that the anomalous phenomena associated with animals constituted more than half of the total reported (872/1519). A similar ratio of UABs to the total macroscopic anomalies was reported after another EQ [
9]. Many UABs had been observed within a week before the EQs [
10]. The locomotive activities of mice drastically increased one day prior to the Kobe EQ [
11], while the circadian rhythm of mice locomotion disappeared in the days before the Wenchuan EQ in China (12 May 2008, M = 8.0) [
12]. Rikitake [
13] noted that small animals and insects showed UABs first and then larger animals (birds, rodents and mammals) up to the hour before the EQ.
The UABs of dogs and cats were observed most frequently within the 24 hours prior to the Kobe EQ [
8]. These behaviors included “barking loudly”, “being panicked”, or “biting owners” in dogs, and “hiding”, “being restless”, “meowing pathetically”, “taking the kitten outside”, “climbing a high tree”, or “disappearing” in cats [
14].
The hearing range is 67 to 44,000 Hz in dogs, 55 to 79,000 Hz in cats, and 31 to 17,000 Hz in humans [
15]. Thus, dogs and cats can hear ultrasounds that humans cannot. The number of smell receptors, olfactory cells, of dogs and cats are over an order of magnitude more numerous when compared with the 12 million of humans [
16]. The olfactory bulb contains approximately 280 million cells in dogs, 67 million in cats, and 5 to 20 million in humans [
17,
18]. Humans can detect odor concentrations from 10
−4.5 molar (M) to 10
−5.0 M. Remarkably, dogs can detect a concentration of 10
−17 M [
19,
20,
21]. With these superior senses, dogs and cats show a greater sensitivity to small changes in smell and/or sound in their environment than humans. The occurrence of behaviors is incomplete without stimuli. For UABs prior to EQs, possible candidate stimuli include changes in atmospheric pressure, changes in gravity, ground deformation (ground uplift and tilt changes), acoustic signals and vibrations due to the generation of micro cracks, ground water level changes, and emanations of gases and chemical substances [
22]. With their extraordinary sensory abilities, it is possible that animals sense and respond to such stimuli. Animals might show UABs as results of which they felt anxiety for EQ precursors as described by Lott
et al. [
23]. In addition, there are some considerable theoretical and experimental evidences that animals may be responding to some physical or chemical anomalies prior to EQs; (1) massive amounts of positive airborne ions which cause changes of the stress hormone concentration in animals and humans as described by Grant
et al. [
24], Freund [
25], and Freund and Stolc [
26]; (2) large amounts of toxic gasses such as the carbon monoxide, which is odorless but deadly [
26,
27]; (3) electromagnetic anomalies such as ultralow frequency and extremely low frequency field which cause physiological effects [
22,
26]. However, the mechanism by which these stimuli are sensed remains undetermined. Another controversial point is the anecdotal and retrospective nature of the reports regarding UABs prior to EQs. At present; however, the best course of action is to collect as much information on unusual physical and biological phenomena before EQs as possible.
The milk yield of dairy cows, although it is not a behavior, could be useful as an EQ predictor because it is often measured daily by instruments in the animal industry. In addition, milk yields are decreased by various stressors [
28,
29]. There are also reports that cows showed UABs prior to EQs [
30,
31]. If cows feel unusual physical or chemical variations prior to EQs, then milk yields could decrease. It may help to reveal the relationship between the seismic activities and UABs prior to EQs. The information on UABs prior to EQs could assist in predicting the next EQ occurrence. The aim of this study is to categorize the information on UABs prior to the massive 2011 Tohoku EQ in Japan and analyze the relationship between the information and seismic characteristics.
4. Discussion
The clear findings of this survey were that UABs of pets and decreased milk yields from cows occurred about one week before the EQ. Most of the UABs of dogs and cats were observed within one day before the EQ, and the decreased milk yield occurred in Ibaraki Prefecture, 340 km from the EQ’s epicenter.
The numbers of reports in which pet owners observed UABs of their pets were 18.6% for dogs and 16.6% for cats. The most frequently observed UAB in dogs and cats was restless behavior and “stuck close to the owner” was also indicated as a UAB in both species. Cats were described as “hiding” or “escaped” more often than dogs; however, this might be due to the natural difference between the species’ behaviors in which cats are more able to act three-dimensionally and go through narrower interspaces than dogs [
41]. Although the options of diarrhea, vomited, and no appetite were used for visually apparent behavioral changes, they were rarely observed in either animal.
The ratio of total UABs increased in both animals as the time of the EQ occurrence neared. Many UABs prior to the Christchurch EQ in New Zealand (4 September 2010, Mw = 7.1) occurred within one hour of the EQ [
42]. Although the exact time of the UABs were not asked for in this study, the variation in the UABs occurrence with time seemed similar to that of the Christchurch EQ. UABs were observed from a few seconds to minutes before the EQ, which could include responses to P-waves as described by Kirschvink [
43]. In the Tohoku EQ, P-waves arrived at the closest area from the epicenter about 20 seconds before the arrival S-waves [
44]. Thus, there was sufficient time to notice the behavioral changes of pets by their owners. In fact, three out of 84 respondents replied that “behavioral changes were shown before EQ sounds”. This suggests the presence of acoustic stimuli that could be discerned even by humans. No seismic swarms, as foreshocks of the Tohoku EQ, were observed before 9 March 2011; however, a number of large and small EQs occurred from 9 March 2011 (11:45 JST) to the Tohoku EQ on 11 March 2011 (14:46 JST), including 56 EQs of more than 3.5 M and 18 EQs of more than 5.0 M. Furthermore, 18 EQs satisfied the Dobrovolsky radius condition in the area nearest the epicenter. The UABs of dogs and cats after 9 March 2011, might include stress responses to shaking caused by these EQs. Therefore, the real ratio of UABs as precursor of the Tohoku EQ were observed after 9 March 2011, might be lower than indicated by the results of our survey. However, we consider that the UABs of pets occurred mostly in the few hours preceding the largest EQ (Mw = 9.0).
The UABs of dogs increased as their distance from the epicenter decreased, except from two to three days and from four to five days before the EQ, whereas cats showed the same correlation from only two to three days before the EQ. Previous research found the same relationship between the number of UABs and the distance from the epicenter for many EQs [
4,
36]. UABs prior to EQs are considered as stress or emotional responses to physical or chemical variations. Some physical or chemical variations prior to the Tohoku EQ emerged during various time periods, and the emerged anomalies did not necessarily remain until the EQ occurred [
45,
46,
47]. The existence or non-existence of a relationship between UABs and the epicentral distance depended on precursory terms. These irregularities of physical or chemical states might cause the UABs identified in this study, although we do not know what triggers the UABs prior to EQs. However, significant relationships between epicentral distance and UABs in dogs “1 day”, “from 1 to a few hours” and “from a few seconds to minutes” before the Tohoku EQ, and those in cats “from 2 to 3 days” before the Tohoku EQ might include effects from stress responses to shaking caused by foreshocks after 9 March 2011.
There are some suggestions for possible stimuli, which cause UABs prior to EQs [
22,
24,
26]. For example, Freund and Stolc have suggested that animals might respond to massive amounts of positive airborne ions, massive amounts of toxic gases and electromagnetic waves of ultralow and extremely low frequency [
26]; however, the clear mechanism for UABs has remained unknown. At present time, to examine whether UABs are useful in EQ predictions, longitudinal observations of UABs are important and essential. It may be an effective prediction method since the objective behavior is observed everyday automatically. In fact, changes in the locomotive activities of mice before large EQs were reported by Yokoi
et al. [
11] and Li
et al. [
12]. Lott
et al. [
48] reported that ratios of UABs prior to EQs differed between some events, even if these occurred with similar distance, depth and magnitude. Longitudinal observations could enable to know that animals show UABs more frequently prior to what kind of EQs. In this study, restless behavior was the most frequently observed UAB in dogs and cats. It may be useful to quantify the frequency of this behavior. However, this result is based on only one case from an extremely large EQ, so further studies are required.
The milk yield from milking facilities at institutes in Shizuoka and Kanagawa Prefecture were not significantly affected by the coming EQ, but the facility in Ibaraki showed lowered milk production 6 days before the EQ. The decrease in the milk yield continued for four days. This might be because Ibaraki was the closest of the three institutes to the epicenter. If so, milk yield might be useful as an EQ precursor. Furthermore, these decreases of milk yields were probably not caused by fear responses to the EQ’s shaking, because no seismic swarms that satisfied the Dobrovolsky radius condition [
40] occurred near the location of the institute in Ibaraki Prefecture from the 5 to 8 March 2011. The 7.3 M EQ on 9 March 2011, did not influence milk yields, because there were not significant decreases after 9 March 2011, at the three institutes. Milk yields on 10 February 2011, also decreased compared with the basal period. This decrease might have been a precursor for the EQ (M = 5.4) that occurred the same day, although the possibility that this decrease was a precursor of the Tohoku EQ cannot be excluded. As the milking work was already finished at the time of this EQ (22:03 JST), the decrease was not a response to the shaking. However, milk yields before other EQs that satisfied the Dobrovolsky radius condition did not show significant decreases. Tests for relationships between milk yields and more EQs using a longer data period could reveal differences between EQs that affect milk yields and those that do not.
Our study revealed the time periods in which the UABs of pets and changes in the milk yields of dairy cows occurred. About 80% of pet UABs were observed within one day of the EQ, whereas cow milk yields decreased six days before the EQ, which preceded the UABs of dogs and cats. Milk yields in each individual have been measured by many institutes in the animal husbandry industry to manage and improve productivity, and the UABs of pets have been observed directly by owners. These two phenomena could contribute to EQ predictions if people carefully observe whether pets show UABs, especially restless behavior, and when dairy cow milk yields decrease, as EQ precursors.