Can Animals Predict Earthquakes?
Summary
WIVVI summary (paraphrased): This review evaluated published claims that abnormal animal behavior can precede earthquakes. The authors examined 180 publications and assembled 729 reported observations from 44 of them, representing more than 130 animal species and 160 earthquakes. Reported anticipation times ranged from seconds to months, and reported distances ranged from a few kilometers to hundreds of kilometers. However, most evidence consisted of isolated, retrospective observations rather than continuous controlled records; only 14 published time series were identified, and these were generally short or incompletely presented. The timing distribution of reported animal anomalies resembled the timing of foreshocks, suggesting that some animals may have reacted to weak seismic events that people did not notice rather than to a distinct pre-earthquake process. The review found insufficient evidence that animal behavior can reliably forecast earthquakes.
Source
Bulletin of the Seismological Society of America
Keywords
animal behaviour, earthquake prediction, earthquake precursors, abnormal animal behavior, systematic review, foreshocks, seismic sensing, retrospective reports, observational bias, time-series monitoring
Source Type
review article
Key Findings
The review identified 729 reports of abnormal behavior involving more than 130 species and 160 earthquakes, but the reports varied enormously in claimed anticipation time, distance, species, behavior, and earthquake magnitude. Only 14 continuous time series were found; most evidence consisted of single retrospective observations without documented baseline behavior or comparable periods when no earthquake occurred. Many published records lacked sufficiently defined prediction criteria, controls, or complete datasets. The temporal distribution of reported animal anomalies closely resembled the distribution of foreshocks, supporting the possibility that some animals responded to weak ground motion after an earthquake sequence had already begun. The available literature did not demonstrate a consistent, reproducible animal behavior that could reliably predict an earthquake’s time, location, and magnitude.
Limitations
The review was constrained by the quality and heterogeneity of the underlying literature. Reports used different species, behavioral definitions, observation methods, earthquake thresholds, distances, and anticipation periods, preventing a conventional quantitative meta-analysis of forecasting performance. Most observations were retrospective and vulnerable to selective reporting, recall bias, missing negative observation,s and uncertainty about the animals’ normal behavior. Published time series were scarce, short, or incompletely displayed. The review could not determine what proportion of unreported earthquakes had no preceding animal anomaly or how often unusual animal behavior occurred without a subsequent earthquake. Its conclusions show that reliable predictive ability had not been demonstrated; they do not prove that no animal can ever detect foreshocks or other earthquake-associated physical changes.
Methods Summary
The authors reviewed 180 publications concerning abnormal animal behavior before earthquakes. From 44 publications containing sufficiently extractable observations, they compiled 729 reports associated with 160 earthquakes and more than 130 animal species. The records were evaluated for claimed anticipation time, distance from the earthquake, earthquake magnitude, species, type of behavior, and observation quality. The analysis examined proposed magnitude–distance relationships, compared the timing of animal anomalies with the statistical timing of foreshocks, and assessed whether the publications supplied continuous time series, baseline observations, controls, and sufficiently defined criteria for testing predictive performance.