Unusual Animal Behavior Preceding the 2011 Earthquake off the Pacific Coast of Tohoku, Japan: A Way to Predict the Approach of Large Earthquakes
Summary
WIVVI summary (paraphrased): This study examined possible animal-related changes before the 11 March 2011 magnitude 9.0 Tohoku earthquake in Japan. Researchers retrospectively surveyed dog and cat owners about unusual behavior observed before the earthquake and separately analyzed routinely recorded milk yields from dairy facilities at different distances from the epicenter. Of 1,259 dog owners and 703 cat owners, 236 and 115, respectively, reported unusual behavior, most commonly restlessness and predominantly within one day before the earthquake. Milk yield decreased at the facility closest to the epicenter, approximately 340 km away, during several days preceding the event, while comparable decreases were not identified at the more distant facilities. The observations generated testable hypotheses but do not establish that the animals anticipated the earthquake or that either measure can provide reliable prospective forecasts.
Source
Animals
Keywords
unusual animal behavior, companion animals, dogs, cats, dairy cows, milk yield, Tohoku earthquake, retrospective questionnaire, owner reports, earthquake precursors, earthquake occurrence, distance analysis, observational study
Source Type
research article
Key Findings
A retrospective survey found that 236 of 1,259 dog owners and 115 of 703 cat owners reported unusual behavior before the 2011 Tohoku earthquake. Restlessness was the most frequently reported behavioral change, and most reports concerned behavior occurring within one day before the earthquake. The analysis reported an association between the occurrence of unusual behavior and distance from the epicenter. Separately, dairy-cow milk yield decreased during several days before the earthquake at the studied facility closest to the epicenter, approximately 340 km away, whereas the two more distant facilities did not show comparable decreases. These results document reported and recorded anomalies associated with one major earthquake, but they do not demonstrate prospective forecasting accuracy.
Limitations
The companion-animal evidence was collected through an internet questionnaire approximately nine to ten months after the earthquake and was therefore vulnerable to recall bias, selective participation, altered memories, media influence, and awareness of the event. Owners were not prospectively recording behavior against predefined criteria, and the analysis lacked comparable reporting data from periods without a subsequent major earthquake. “Unusual behavior” included diverse behaviors and subjective judgments rather than one standardized biological measure.
The milk-yield analysis concerned one earthquake and only three facilities, with the reported decrease confined to the closest facility. Milk production can be affected by lactation stage, health, feed, weather, husbandry, farm operations, and other stressors. The study did not use a prospectively declared alarm rule or quantify sensitivity, specificity, missed events, or false-alarm frequency across independent earthquakes. Consequently, the findings cannot establish causation, a sensing mechanism, or reliable earthquake prediction.
Methods Summary
Researchers conducted an internet questionnaire in December 2011 and January 2012 asking dog and cat owners about unusual behavior observed before the 11 March 2011 Tohoku earthquake. Responses from 1,259 dog owners and 703 cat owners were categorized by behavior type and reported timing. Statistical models were used to examine whether reports varied with factors including distance from the earthquake epicenter.
The researchers also obtained routinely recorded daily milk-yield data from dairy facilities in Ibaraki, Kanagawa, and Shizuoka prefectures. Records preceding the earthquake were examined for deviations from expected milk production, with adjustments intended to account for ordinary production changes. Results were compared among facilities at different epicentral distances. Both components were retrospective observational analyses rather than prospective forecasting trials.