While there exists certain patterns of travel between the midwest and the south or the southwest, there is nothing that even comes close to approaching the very unique demographics that is prevalent on the Autotrain. I don't believe that any other route could provide a sufficient demand to be successful.
When I went to Florida for a week in March of ... 2001? ... my traveling friend and I kept a count of license plates from each state and province that we saw -- not just a checklist to make sure we saw one of each, but a tally of how many of each. I've long since lost the tally, but I can tell you the top three:
Florida
Michigan
Ontario
Cars from the Mid-Atlantic and New England were certainly in the top-ten -- I'm pretty sure New York and Virginia were represented there -- but so were cars from other Midwestern states like Ohio and Illinois. I think the only states we didn't at all were Arizona and maybe New Mexico, for obvious reasons. (Yes, we saw Alaska and Hawai`i, but no, I don't think we saw some of the western Canadian provinces either.)
The difference may well be that a lot of those Michiganders own second homes and come down, stay for two months and return, while a lot of East Coasters may own time shares or just get hotel rooms, and come down for a week and return. The East Coast model would produce a much higher Auto Train ridership than the Midwest model, since people are constantly rotating in and out from the East Coast. But the demographics may change with a train, too. If it becomes easier for more Midwesterners to get to Florida, more might come for weeklong trips. (Historically, this trend held true as the trains in the 1930s-60s were always longer, more frequent, and sometimes greater in number in the winter than in the summer, as Bill Haithcoat has shown us in the past. Whereas in that same era the big East Coast train like the Silver Meteor saw some variation, but I think not nearly as much variation, seasonally?)
But it's been a long time since we've had a Midwest Auto Train, or even a Floridian, long enough that I think people's ideas of where, when, and how long to holiday for have changed, and data from 20-30 years ago may no longer accurately reflect the demographics of today. (There's been almost 100% turnover in the population of people 60+ since there was last a Midwest Auto Train! And I would imagine retirees make up a large portion of potential riders. There's been 100% turnover in the population of parents of "Disney-aged" children, too, for that matter.) People have different amounts of leisure time and different amounts of money and place different values on their time and money than they used to. Marketing is different. Airlines are
very different. The price of gas is different. The attractions in Florida are different now than they were then, too. Using the old marketing data to say "that train didn't have the ridership to succeed" shouldn't make it an open-and-shut case not to re-start the train. That's the sort of argument that still has a lot of strength after 5-10 years, but not 20-30 years.