It might not be, anyway.
Scientists are musing that the Martian shield volcanoes might be dormant, not dead. On Earth, continental drift will shift the eruption spot of a volcano. We see this in the Hawaiian Islands: the crust of that part of the Pacific Ocean is moving northwest, so as one travels southeast on the surface, the landforms are more recent. Indeed, only the Big Island has active volcanoes today.
The Mars Express orbiter has uncovered evidence that the most recent deposits on Ascraeus Mons might be just a few million years old. The thinking is that the subcrustal hot spot might itself be drifting. The conventional wisdom today is that continental drift requires the lubrication of oceans of water. Mars doesn’t seem to have drift; its crust appears to be in one piece.
Of course, the crust itself could be slowly moving over a liquid interior layer. I don’t think that has been definitively ruled out. It’s just not in the model we have. We think that’s what’s going on at Jupiter’s moon Europa: a thin ice crust floating above a liquid ocean. That water in turn covers a “bedrock” another several dozen miles below the bottom of the ice.
So, maybe, that’s where the water went? Hmm. That would/could jibe with the plume that was recently found indicating that water periodically seeps up.