> Turns out there is an answer: the garter snakes sequester the tetrodotoxin, storing it in their livers. This makes them toxic to their own predators.
But this doesn't seem as immediate as the newt's defense where it's on the skin and thus causes potential predators to spit them out or even to seize up - meaning that at least some attacked newts survive the encounter. Eating the liver means the snake is dead. And since it's going to be impossible to tell if a particular snake is immune (and is thus potentially toxic) how would this deter predators? (Especially given the limited range of snakes with this immunity and the probability that there are predators of the snakes that don't necessarily have this same limited range - ravens, raptors, etc.)
If the predator species have heritable differences in prey selection, no matter how slight, natural selection can work with that.
> how would this deter predators?
Maybe the predator's carcass next to a half eaten garter snake is meant to serve as a lesson to other potential predators.
Or perhaps the aim is not to deter but to simply take one natural predator down with them for the good of their species.
Eating a snake might not kill a predator outright.
And higher predators (like mammals) also have food preferences. They don't always eat stuff indiscriminately, so predators that don't _like_ snakes will preferentially survive. Eventually, this can get established as a genetic trait.
Or as a behavioral one, if parents don't teach cubs to hunt snakes.