Most pitvipers vibrate their tails when aggressed against - this is a primitive condition for the clade, and the main exceptions to the rule are some arboreal species. On many substrates, this vibration will cause a rustling noise w/o the aid of a rattle.
If this tail vibration has an appreciable aposematic effects (i.e., at least some predators or potential "crushers" take it as a cue to avoid the snake), then any increase in the volume of the noise might lead to increased effectiveness, and thus increased fitness for the bearer (assuming no other costs). Thus, the first snake that retained the terminal tail cone after a shed and made more noise with it would have been at an advantage. This would apply irrespective of habitat, so long as it confers some survival advantage in a confrontational situation.
In other words, under this scenario, the rattle evolved per chance in the common ancestor of Sistrurus and Crotalus because, due to some quirk, a snake that used tail vibration as a defensive strategy evolved a derived character that enhanced its effectiveness. It could have happened anywhere, but did in the common ancestor of Crotalus and Sistrurus. No need to invoke any special circumstances to explain it.
