This sort of thing (HT) is actually quite a challenge to evolution. Imagine what a species has to go through to select for earthquake detection: either most individuals that can't detect earthquakes have to die, or else the detectors have to breed disproportionately. With the reliability these toads display, the faculty would have to be selected for extremely strongly - earthquakes massacring toads over and over or somehow causing a huge offspring disparity over several generations.
Both are problematic. Toads are small, thus light and can take to the water. An earthquake doesn't have much purchase on particular individuals. Their response to the quake actually takes them away from spawning, so I have a hard time imagining the earthquake-dodging toads breed a great deal more.
Secondly, every generation without an earthquake is going to damage to earthquake-detecting genes. Genetic drift will reliably erode them. Except for places with very frequent shocks like the San Andreas fault, the ability should die out long before it is put to use.
What all this adds up to is that toads are dodging earthquakes without ever having been selected to dodge earthquakes.
A word on anecdotal evidence. It's still evidence. Don't ignore evidence because it's unfashionable.