Here you go. Another post will follow.

Can Dogs Predict Earthquakes? A Possible Auditory Answer
Stanley Coren1

The suggestion that animals can sense earthquakes before they occur was first recorded in Greece in 373 BC, when dogs howled and many rats, weasels, snakes, and centipedes moved to safety several days before a destructive earthquake. In China and Japan animals are considered to be an integral part of their national earthquake warning systems. One early indication of their usefulness occurred in 1975, when officials in the Chinese city of Haicheng were alarmed by odd and anxious behaviors of dogs and other animals. These observations led them to order 90,000 residents to evacuate the city. Only a few hours later a 7.3 magnitude earthquake destroyed nearly 90 percent of the city’s buildings. Since then a number of studies have suggested that animals can anticipate seismic events(1,2).
A serendipitous event provided an opportunity to explore the ability of dogs to predict earthquakes. Behavioral data on variations in canine moods was being collected from 200 dogs. Twice a week owners were sent an e-mail and asked to rate activity level and signs of anxiety in their dog over the previous day using a nine point scale(3). In general there was little day to day variability in group averages, however one particular day (02/27/2001) showed a sharp increase in activity (+1.85, t[192]=19.02, p<0.001) and anxiety (+1.38, t[192]=9.47, p<0.001). Of the 193 dogs reported that day, 47% (91) were more than two standard deviations above their usual baseline for activity and 49% (95) for anxiety. By chance, I had captured data on dog behaviors on the day before a 6.8 earthquake shook the Pacific Northwest. With a focus not far from Vancouver, Canada, it shook the city with enough intensity to cause damage to some structures.
It has been suggested that one possible explanation for animals “predicting” earthquakes is that their superior auditory sensitivity allows them to hear seismic activities that precede earthquakes(1,2). In the sample 14 dogs were reported as having hearing impairments. Consistent with possible auditory cuing, only one showed any significant increase in anxiety or activity, and it was living with a normal hearing dog (not in the sample).
To further explore the possibility of an auditory cue the sample was divided into dogs with lopped ears whose ear flaps attenuate sounds somewhat (N=122) compared to dogs with pricked ears (N=71). Dogs with pricked ears showed greater mean increases in activity (+2.97 versus +1.19, t[191]=11.35, p<0.001) and anxiety (+2.47 versus +0.75, t[191]=6.20, p<0.001).
Sound attenuation by an ear flap is greater for high frequencies since low frequencies bend around objects. Thus a beagle’s ear flap interposed between a 90 dB, 1000 Hz tone source attenuated the sound reaching a sensor 2 cm behind it by 6.2 dB, compared to 31.1 dB for a 14000 Hz tone. Ear flap differences could mean that higher frequency sounds are critical. To test this, dogs were rank ordered according to head size (using average inter-aural distance per breed) and divided into thirds. Smaller head sizes are associated with better sensitivities for high frequency sounds(4,5). Consistent with the importance of high frequencies, the 65 dogs with smallest head sizes showed greater increases than those with the largest head sizes for pre-earthquake activity level (+2.88 versus +1.16, t[128]=8.26, p<0.001) and anxiety (+2.22 versus +0.96, t[128]=3.72, p<0.001).
Taken together these results are suggestive. It appears that there is an increase in observable activity and anxiety in dogs in the 24 hours preceding an earthquake. Observations that dogs with poor hearing are not affected, and dogs with lopped ears are less affected, hint that animals are responding to an auditory cue. The fact dogs with smaller head size are more responsive is consistent with a presumption that higher frequency sounds serve as the signal predicting an impending earthquake, perhaps from rocks scraping or breaking underground,.
Obviously, analysis based upon a single event, even though the sample size is reasonably large, can not be considered a definitive proof. In addition statistical sensitivity was augmented by the fact that baselines were extremely stable because they were computed from 54 data points per dog. However the results are internally consistent and suggest that canine sensitivity to higher frequencies allows them to serve as bioindicators of future seismic events.

References and Notes
1. M. Ikeya, Earthquakes and animals: from folk legends to science. (World Scientific Publishing Co., Singapore, 2004).
2. T. Miller, Earthquake Prediction Handbook. (Info-Pub, San Bernadino, CA, 1996).
3. Materials and methods are available as supporting material on Science Online.
4. B. Masterton, H. Heffner, R. Ravizza, J.Acous. Soc. Am., 45, 966 (1969).
5. R.R. Fay, Hearing in Vertebrates: a Psychophysics Databook. (Hill-Fay Associates, Winnetka IL, 1988).