One of the benefits to the first option however is it leaves you more room for variable length link text. Depending on your use case the benefit of having indentation might be mitigated by the negative effect of wrapping text:

download bmml source – Wireframes created with Balsamiq Mockups
I think option 1 could be effective if you use something other than size to communicate the hierarchy to the user:

download bmml source
This option maximizes available space for link text, which can matter depending on the viewport through which your user accesses the site.
Additionally, the user's eye is able to travel along a single vertical path to evaluate their navigation options.
My suspicion is contrast is a "primary" perceptive function, not a reflective function on the user's part. They just see the contrast without having to think about it.
Option 2, by using white space to create delineation, might require "secondary" perception. The user's brain would engage in cortical reasoning to group the items together.
I don't know enough about human psychology to categorically state whether there is any meaningful difference between the two visual representations and cognitive load, so if anyone can provide a link or set me straight that would be great.
I would just prototype the two using the longest navigation labels you think could possibly appear and do some usability testing. That data would show if there is any meaningful difference. It's possible they're both effective.