If their statistics showed it to be a relevant discovery then it would be a discovery. The mechanism behind it would be considered "unknown" until such time as theories were presented, tested, peer-reviewed etc.Yes maybe, but they would still claim it to be mere chance, until they did discovered the mechanism.
and in the mean time we end up with a lot of dead rats!
The discovery would be the observation. A single observation would likely not be considered statistically relevant except where current theories suggest that any such observation should be impossible.
Since in your example "mere chance" suggests that it would occur given enough time and enough dead rats, then a single observation is not going to be deemed significant.
When they think there are enough observations to say that yes, there is something going on that "mere chance" can not explain (ever heard of the 5-sigma standard) then it is deemed a discovery. But to understand the mechanism would likely require more analysis for sure.