I thought the main driver for collecting telemetry and having Mars Odyssey (an orbiter) in sight to relay data is due to the earlier loss of Mars Polar Lander (http://en.wikipedia.org/wiki/Mars_Polar_Lander#Communication...).
The review board analyzing the loss of MPL could not narrow down the cause for certain, although they did identify a most likely cause. It was resolved in the future to have as much data collected as possible so that an unexplained loss could not occur again.
One key ingredient in this is line of sight to a relay orbiter.
But this is not a science driver, it is an engineering driver.
"But this is not a science driver, it is an engineering driver."
Fair point.
Basically you have to ask the question "What science can I do at this stage?" so that during every possible outcome you extract the most data you can.
I find that a fascinating way to structure your thinking about a project.