Can Animals Sense Earthquakes Before They Happen? The Evidence
Stories of restless dogs and fleeing snakes before earthquakes go back thousands of years. Here is what tagged farm animals, seismology and the USGS actually say.
Animals can sense an earthquake a few seconds before people feel strong shaking, because many detect the fast, weak P wave that arrives before the more damaging waves. Whether they sense quakes hours or days ahead is unproven: the US Geological Survey says no consistent, reliable pre-quake behavior or mechanism has been established, so animals are not a usable prediction method.
The stories are vivid. A dog paces and whines, cattle bunch together in a corner of the field, birds leave a tree all at once, and then the ground starts to move. Accounts like these go back at least to ancient Greece. The question scientists ask is whether such behavior happens reliably before earthquakes, and not just in the memories of people who lived through one.
What do animals actually detect seconds before an earthquake?
The clearest explanation is the P wave. An earthquake sends out several kinds of seismic waves. The primary, or P, wave is the fastest and arrives first, but it is usually weak. The slower S waves and surface waves that follow carry most of the strong shaking.
The USGS explains that animals with keener senses can apparently feel the P wave seconds before the larger S wave arrives, while few people notice it. That fits the many reports of animals reacting just before the shaking: they are responding to the start of the earthquake, not predicting it.
The same physics powers modern earthquake early warning. Seismometers detect the P wave near the source, and computers send alerts that can outrun the strong shaking by a few seconds. Our article on whether earthquakes can be predicted explains how systems like ShakeAlert do this.
What did the 2020 study of tagged farm animals find?
The best-known recent study tracked cows, sheep and dogs on a farm in central Italy during the 2016 to 2017 earthquake sequence that included the magnitude 6.6 Norcia earthquake. Researchers fitted the animals with collars carrying motion sensors and recorded their activity continuously through the period of repeated earthquakes.
The team, led by animal-tracking researcher Martin Wikelski, reported in the journal Ethology in 2020 (doi.org/10.1111/eth.13078) that the animals’ collective activity rose before some earthquakes. The apparent warning time ranged from about 1 to 20 hours and was longer the closer the farm was to the quake. The pattern showed up while the animals were kept in a stable, not while they were out on pasture.
The study covered one earthquake sequence and one farm. Other scientists published a critique of its statistics in Ethology, and the authors published a reply. It is an interesting signal that needs testing in many more places.
Why does the USGS say animals cannot be used to predict earthquakes?
The USGS says the evidence so far is anecdotal and inconsistent. For a prediction method to work, the same behavior would have to show up before most earthquakes, rarely happen without one and be explained by a real physical signal. None of that has been shown.
- Reports are collected after the fact. People remember the strange behavior that came before a quake and forget the many times animals acted oddly with no quake at all.
- Timing varies wildly. Reported warnings range from seconds to weeks, which makes them hard to use.
- No confirmed mechanism. If animals sense something days ahead, scientists would first need to show that a measurable precursor exists. As the USGS puts it, without real precursor signals, any connection is irrelevant.
The USGS also cites a test of a popular theory from California: that lost-pet ads in newspapers rose before Bay Area earthquakes. A statistical analysis published in 1988 found no such correlation.
What signals could animals be picking up?
Several ideas have been proposed, but none has been confirmed as a reliable earthquake precursor:
- Small foreshocks. Some large earthquakes follow smaller ones that animals may notice more than people do. Foreshocks can only be identified as foreshocks after the main quake happens.
- Changes in groundwater or gases. Researchers have looked at whether rock stress releases gases or shifts water levels before a quake. Results have been inconsistent.
- Electrical or magnetic changes. Some studies have looked for electromagnetic signals before earthquakes, without a consistent pattern that holds up across events.
The USGS page also mentions a 2000 paper by Kirschvink suggesting that an escape response to earthquakes could have evolved from existing instincts to flee predators. That idea remains a hypothesis, and the agency notes that much research is still needed.
How do scientists study the question now?
Modern studies use tags with accelerometers and GPS, which record animal movement every few seconds for weeks or months. That makes it possible to compare behavior before earthquakes with behavior on ordinary days, which anecdotes cannot do.
| Approach | What it can show | Main limitation |
|---|---|---|
| Eyewitness reports | That people noticed odd behavior | Collected after the quake; no comparison with normal days |
| Zoo and farm records | Behavior logged before and after events | Few earthquakes near any one site |
| Motion-sensor tags (e.g., Italy 2016 to 2017) | Continuous activity data to test against quake times | Small number of animals and places so far |
| Seismometers (USGS) | The P wave arriving seconds before strong shaking | Measures the quake itself, not a precursor |
To judge any claim, it helps to know how earthquake size and shaking are measured; our explainer on how earthquakes are measured covers magnitude and intensity.
Bottom line
Animals can react a few seconds before strong shaking, most likely because they feel the P wave. Claims that they sense earthquakes hours or days in advance rest on anecdotes and a handful of small studies, such as the 2020 Italian farm study, which its own authors say needs much wider testing. The USGS does not consider animal behavior a usable way to predict earthquakes.
FAQ
Can dogs predict earthquakes?
There is no reliable evidence that dogs can predict earthquakes days ahead. Dogs may react seconds before strong shaking because they notice the first, weaker P wave.
Should I watch my pets for earthquake warnings?
No. Pets act strangely for many reasons, and the USGS does not consider animal behavior a reliable warning. Official alerts, where available, are based on seismometers.
Are animals better earthquake detectors than seismometers?
No. Seismometers record even tiny P waves, measure them precisely and feed warning systems in seconds. Near coasts, strong shaking is itself a natural warning; our article on how tsunamis form explains the signs people should act on.
