You just store the entire tree? Then how do you get a subtree? I don't think you're solving anything by just storing the whole tree as one thing. I mean you can do that in SQL too if you want.
MongoDb lets you query inside the document/tree. It's sort of like how some databases (eg Postgres: http://www.postgresql.org/docs/current/static/xml2.html) let you store XML in a blob, then query inside that using XPath
And yes, XML/XPath support in SQL databases allows them to act as schema-free document stores. However, they aren't optimized for that, so indexing inside the document is limited. OTOH, SQL DB vendors might be able to add that quicker than NoSQL vendors can improve tool support and querying. OTOH you have to deal with XML instead of JSON. OTOH...
